Background
Prostate biopsy and postbiopsy complications represent important risks of prostate-specific antigen (PSA) screening. Although landmark randomized trials and updated guidelines have challenged routine PSA screening, it is unclear whether these publications have affected rates of biopsy or postbiopsy complications.
Objective
To evaluate whether publication of the 2008 and 2012 US Preventive Services Task Force (USPSTF) recommendations, the 2009 European Randomized Study of Screening for Prostate Cancer and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial, or the 2013 American Urological Association (AUA) guidelines was associated with changes in rates of biopsy or postbiopsy complications, and to identify predictors of postbiopsy complications.
Design, setting, and participants
This quasiexperimental study used administrative claims of 5 279 315 commercially insured US men aged ≥40 yr from 2005 to 2014, of whom 104 584 underwent biopsy.
Interventions
Publications on PSA screening.
Outcome measurements and statistical analysis
Interrupted time-series analysis was used to evaluate the association of publications with rates of biopsy and 30-d complications. Logistic regression was performed to identify predictors of complications.
Results and limitations
From 2005 to 2014, biopsy rates fell 33% from 64.1 to 42.8 per 100 000 person-months, with immediate reductions following the 2008 USPSTF recommendations (−10.1; 95% confidence interval [CI], −17.1 to −3.0; p < 0.001), 2012 USPSTF recommendations (−13.8; 95% CI, −21.0 to −6.7; p < 0 .001), and 2013 AUA guidelines (−8.8; 95% CI, −16.7 to −0.92; p = 0.03). Concurrently, complication rates decreased 10% from 8.7 to 7.8 per 100 000 person-months, with a reduction following the 2012 USPSTF recommendations (−2.5; 95% CI, −4.5 to −0.45; p = 0.02). However, the proportion of men undergoing biopsy who experienced complications increased from 14% to 18%, driven by nonsepsis infectious complications (p < 0.001). Predictors of complications included prior fluoroquinolone use (odds ratio [OR]: 1.27; 95% CI, 1.22–1.32; p < 0.001), anticoagulant use (OR: 1.14; 95% CI, 1.04–1.25; p = 0.004), and age ≥70 yr (OR: 1.25; 95% CI, 1.15–1.36; p < 0.001). Limitations included the retrospective design.
Conclusions
Although there has been an absolute reduction in rates of biopsy and 30-d complications, the relative morbidity of biopsy continues to increase. These observations suggest a need to reduce the morbidity of biopsy.
Patient summary
Absolute rates of biopsy and postbiopsy complications have decreased following landmark publications about prostate-specific antigen screening; however, the relative morbidity of biopsy continues to increase.
In 2009, two landmark randomized trials—the European Randomized Study of Screening for Prostate Cancer [ERSPC] [1] and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial [PLCO] [2]—reported smaller than anticipated reductions in prostate cancer (PCa) mortality and prompted reexamination of prostate-specific antigen (PSA) screening recommendations. The US Preventive Services Task Force (USPSTF), which in 2008 had recommended against PSA screening in men aged >75 yr [3], updated its position in 2012 to recommend against systematic PSA screening in the general population [4]. The American Urological Association (AUA) likewise updated its PSA screening guidelines in May 2013, recommending shared decision making for men aged 55–69 yr [5]. Both the 2012 USPSTF and 2013 AUA guidelines cited prostate biopsy and its potential complications as a significant harm associated with PSA screening.
Although several studies have noted a reduction in PSA screening rates following these publications [6], [7], [8], [9], [10], [11], and [12], there are little data on whether this has translated into the intended reductions in screening-related morbidity in the form of prostate biopsy and postbiopsy complications [13]. We evaluated whether publication of the 2008 USPSTF recommendations, 2009 ERSPC/PLCO trials, 2012 USPSTF recommendations, or 2013 AUA guidelines was associated with reductions in prostate biopsy or postbiopsy complications. Biopsy-related complications also remain understudied both with regard to incidence as well as identification of risk factors [14], [15], and [16]. Consequently, we also sought to characterize postbiopsy complications and evaluate predictors using a contemporary population-based cohort.
We conducted a retrospective analysis of administrative claims from the Optum Labs Data Warehouse that contains data on >100 million individuals enrolled in private health plans and Medicare Advantage Plans, from geographically diverse regions across the United States, as described in Supplementary Methods [17]. Because this study involved analysis of preexisting deidentified data, it was exempt from institutional review board approval.
We identified all prostate biopsies performed in men aged ≥40 yr between January 1, 2005, and September 30, 2014, using Current Procedural Terminology-4 code 55700 alone (11%) or in combination with 76872 (84%) or 76942 (0%), or International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 60.11 (5%). The first biopsy was selected as the index biopsy when more than one biopsy was performed. Patients with a prior diagnosis of PCa (ICD-9 code 185) were excluded. We restricted the cohort to patients with continuous medical and pharmacy enrollment for at least 12 mo prior to biopsy and 30 d after biopsy to ascertain comorbidities and 30-d complications (Supplementary Fig. 1).
We recorded sociodemographic characteristics, comorbid conditions, anticoagulant use within 30 d of biopsy, and fluoroquinolone use in the 12 mo prior to biopsy as described in Supplement 1. The Charlson Comorbidity Index was calculated as previously described [18]. Prebiopsy PSA values within 60 d of biopsy were available for a subset of patients.
The primary outcomes were crude and adjusted rates of prostate biopsy and 30-d complications. Complications were identified using inpatient and outpatient administrative claims as summarized in Supplementary Table 1.
Baseline demographic features were compared across study periods using the Kruskal-Wallis test for continuous variables and the chi-square test for discrete variables. To evaluate the impact of each publication on rates of prostate biopsy and 30-d complications, we used an interrupted time series (ITS) approach as described in Supplementary Methods [19] and [20]. Rates were estimated using all men aged ≥40 yr with at least 12 mo of continuous medical and pharmacy enrollment as the denominator. As secondary analyses, we estimated rates of 30-d complications among men who underwent biopsy. Two separate ITS analyses were performed: (1) unadjusted analyses, using crude rates of biopsy and complications aggregated for each month, and (2) analyses standardized for case mix, to adjust for population shifts over time as described in Supplementary Methods [20] and [21].
Logistic regression was used to evaluate predictors of 30-d complications. Multivariable models were adjusted for age at biopsy, year of biopsy, race, anticoagulant use, history of fluoroquinolone use, and the presence of the five most prevalent Charlson comorbidities.
Statistical analyses were performed using SAS v.9.3 (SAS Institute Inc., Cary, NC, USA). All tests were two sided with p < 0.05 considered statistically significant.
Between January 2005 and September 2014, a total of 104 584 men underwent prostate biopsy (Table 1). Median age at biopsy was 63.0 yr (interquartile range [IQR]: 57.0–70.0). Overall, 18% of men had used fluoroquinolones in the 12 mo preceding biopsy, and 2.8% were on anticoagulants. Median prebiopsy PSA was 5.1 ng/dl (IQR: 4.0–7.1) for the 18 749 men (18%) for whom this was available. Over the time periods, men undergoing biopsy were older, had increased prevalence of anticoagulant use and comorbid conditions, and had a higher PSA (Table 1 and Supplementary Fig. 2).
Table 1
Patient demographics for men who underwent prostate biopsy (January 2005 to September 2014) stratified by pre/postintervention time periods (n = 104 584)
| Baseline Jan 2005–Jul 2008 (n = 37 217) |
2008 USPSTF Aug 2008–Feb 2009 (n = 6672) |
ERSPC/PLCO Mar 2009–Apr 2012 (n = 36 603) |
2012 USPSTF May 2012–Apr 2013 (n = 10 772) |
2013 AUA May 2013–Sep 2014 (n = 13 320) |
p value | Total (n = 104 584) | |
|---|---|---|---|---|---|---|---|
| Age at biopsy, yr | |||||||
| Median (IQR) | 62.0 (56.0–68.0) | 62.0 (56.0–68.0) | 63.0 (57.0–70.0) | 65.0 (59.0–71.0) | 66.0 (59.0–71.0) | <0.001 | 63.0 (57.0–70.0) |
| Age group, yr (%) | |||||||
| 40–49 | 2269 (6.1) | 416 (6.2) | 1737 (4.7) | 420 (3.9) | 467 (3.5) | <0.001 | 5309 (5.1) |
| 50–59 | 12 366 (33) | 2201 (33) | 10 342 (28) | 2644 (25) | 3098 (23) | 30 651 (29) | |
| 60–69 | 14 671 (39) | 2697 (40) | 15 105 (41) | 4426 (41) | 5546 (42) | 42 445 (41) | |
| ≥70 | 7911 (21) | 1358 (20) | 9419 (26) | 3282 (31) | 4209 (32) | 26 179 (25) | |
| Race (%) | |||||||
| White | 25 508 (69) | 4643 (70) | 26 187 (72) | 7795 (72) | 9407 (71) | <0.001 | 73 540 (70) |
| African American | 3695 (9.9) | 780 (12) | 4423 (12) | 1283 (12) | 1589 (12) | 11 770 (11) | |
| Hispanic | 2549 (6.8) | 538 (8.1) | 2911 (8.0) | 837 (7.8) | 1186 (8.9) | 8021 (7.7) | |
| Asian | 1133 (3.0) | 236 (3.5) | 1142 (3.1) | 306 (2.8) | 387 (2.9) | 3204 (3.1) | |
| Unknown/Missing | 4332 (12) | 475 (7.1) | 1940 (5.3) | 551 (5.1) | 751 (5.6) | 8049 (7.7) | |
| Census region (%) | |||||||
| Northeast | 7094 (19) | 1196 (18) | 6120 (17) | 1886 (18) | 2424 (18) | <0.001 | 18 720 (18) |
| Midwest | 10 094 (27) | 1562 (23) | 8837 (24) | 2742 (26) | 3627 (27) | 26 862 (26) | |
| South | 15 330 (41) | 2889 (43) | 16 371 (45) | 4698 (44) | 5538 (42) | 44 826 (43) | |
| West | 4660 (13) | 993 (15) | 5147 (14) | 1442 (13) | 1728 (13) | 13 970 (13) | |
| Other/Missing | 39 (0.1) | 32 (0.5) | 128 (0.3) | 4 (0) | 3 (0) | 206 (0.2) | |
| PSA, ng/d | n = 5174 | n = 548 | n = 7345 | n = 2211 | n = 3471 | n = 18 749 | |
| Median (IQR) | 5.1 (3.9–7.1) | 4.9 (3.8–6.8) | 5.0 (4.0–6.8) | 5.3 (4.2–7.4) | 5.4 (4.2–7.8) | <0.001 | 5.1 (4.0–7.1) |
| Anticoagulant use (%) | 928 (2.5) | 184 (2.8) | 1103 (3.0) | 299 (2.8) | 415 (3.1) | <0.001 | 2929 (2.8) |
| History of fluoroquinolone use in prior 12 mo (%) | 6797 (18) | 1267 (19) | 6557 (18) | 2017 (19) | 2504 (19) | 0.054 | 19 142 (18) |
| Charlson Comorbidity Index (%) | |||||||
| 0 | 22 354 (60) | 3887 (58) | 20 269 (55) | 5758 (54) | 7039 (53) | <0.001 | 59 307 (57) |
| 1 | 8199 (22) | 1453 (22) | 8252 (23) | 2503 (23) | 2999 (23) | 23 406 (22) | |
| 2 | 3232 (8.7) | 664 (10) | 3520 (9.6) | 1108 (10) | 1359 (10) | 9883 (9.4) | |
| ≥3 | 3432 (9.2) | 668 (10) | 4562 (13) | 1403 (13) | 1923 (14) | 11 988 (12) | |
| Comorbidities (%) | |||||||
| Myocardial infarction | 763 (2.1) | 144 (2.2) | 812 (2.2) | 227 (2.1) | 273 (2.0) | 0.6 | 2219 (2.1) |
| Congestive heart failure | 1443 (3.9) | 223 (3.3) | 1548 (4.2) | 467 (4.3) | 586 (4.4) | <0.001 | 4267 (4.1) |
| Peripheral vascular disease | 2054 (5.5) | 399 (6.0) | 2660 (7.3) | 783 (7.3) | 1071 (8.0) | <0.001 | 6967 (6.7) |
| Cerebrovascular disease | 2007 (5.4) | 379 (5.7) | 2358 (6.4) | 681 (6.3) | 947 (7.1) | <0.001 | 6372 (6.1) |
| Dementia | 370 (1.0) | 63 (0.9) | 442 (1.2) | 141 (1.3) | 198 (1.5) | <0.001 | 1214 (1.2) |
| Chronic obstructive pulmonary disease | 4251 (11) | 758 (11) | 4293 (12) | 1254 (12) | 1686 (13) | 0.004 | 12 242 (12) |
| Ulcer | 362 (1.0) | 60 (0.9) | 297 (0.8) | 84 (0.8) | 102 (0.8) | 0.07 | 905 (0.9) |
| Mild liver disease | 1182 (3.2) | 223 (3.3) | 1332 (3.6) | 413 (3.8) | 515 (3.9) | <0.001 | 3665 (3.5) |
| Diabetes mellitus | 6475 (17) | 1290 (19) | 7823 (21) | 2442 (23) | 2973 (22) | <0.001 | 21 003 (20) |
| Diabetes with organ damage | 1105 (3.0) | 225 (3.4) | 1600 (4.4) | 492 (4.6) | 734 (5.5) | <0.001 | 4156 (4.0) |
| Hemiplegia | 128 (0.3) | 25 (0.4) | 150 (0.4) | 51 (0.5) | 56 (0.4) | 0.3 | 410 (0.4) |
| Moderate/severe renal disease | 1133 (3.0) | 254 (3.8) | 1950 (5.3) | 674 (6.3) | 908 (6.8) | <0.001 | 4919 (4.7) |
| Moderate/severe liver disease | 27 (0.1) | 7 (0.1) | 31 (0.1) | 9 (0.1) | 12 (0.1) | 0.9 | 86 (0.1) |
| Metastatic solid tumor | 175 (0.5) | 37 (0.6) | 199 (0.5) | 58 (0.5) | 80 (0.6) | 0.4 | 549 (0.5) |
| AIDS | 82 (0.2) | 16 (0.2) | 101 (0.3) | 27 (0.3) | 46 (0.3) | 0.2 | 272 (0.3) |
| Rheumatologic disease | 449 (1.2) | 71 (1.1) | 522 (1.4) | 158 (1.5) | 196 (1.5) | 0.007 | 1396 (1.3) |
| Cancer | 1736 (4.7) | 337 (5.1) | 1796 (4.9) | 571 (5.3) | 701 (5.3) | 0.02 | 5141 (4.9) |
| Net worth range (%) | |||||||
| <$25 000 | 1831 (4.9) | 404 (6.1) | 2101 (5.7) | 561 (5.2) | 695 (5.2) | <0.001 | 5592 (5.3) |
| $25 000–149 000 | 5188 (14) | 1042 (16) | 5608 (15) | 1591 (15) | 2046 (15) | 15 475 (15) | |
| $150 000–249 000 | 4754 (13) | 837 (13) | 5182 (14) | 1499 (14) | 1862 (14) | 14 134 (14) | |
| $250 000–499 000 | 9694 (26) | 1779 (27) | 10 048 (28) | 3027 (28) | 3636 (27) | 28 184 (27) | |
| ≥$500 000 | 10 910 (29) | 2030 (30) | 10 860 (30) | 3208 (30) | 3852 (29) | 30 860 (30) | |
| Unknown/missing | 4840 (13) | 580 (8.7) | 2804 (7.7) | 886 (8.2) | 1229 (9.2) | 10 339 (9.9) | |
| 30-d complications (%) | |||||||
| Any | 5973 (16) | 1151 (17) | 6120 (17) | 1863 (17) | 2451 (18) | <0.001 | 17 558 (17) |
| Bleeding | 2547 (6.8) | 499 (7.5) | 2660 (7.3) | 860 (8.0) | 1083 (8.1) | <0.001 | 7649 (7.3) |
| Nonsepsis infection | 2978 (8.0) | 590 (8.8) | 2990 (8.2) | 875 (8.1) | 1236 (9.3) | <0.001 | 8669 (8.3) |
| Sepsis | 191 (0.5) | 50 (0.7) | 349 (1.0) | 105 (1.0) | 134 (1.0) | <0.001 | 829 (0.8) |
| Urinary retention | 980 (2.6) | 161 (2.4) | 1138 (3.1) | 341 (3.2) | 422 (3.2) | <0.001 | 3042 (2.9) |
| Hospitalization | 1019 (2.7) | 181 (2.7) | 1056 (2.9) | 320 (3.0) | 367 (2.8) | 0.6 | 2943 (2.8) |
AIDS = acquired immunodeficiency syndrome; AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; IQR = interquartile range; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer [screening trial]; PSA = prostate-specific antigen; USPTF = US Preventive Services Task Force.
Overall, 30-d complications occurred in 17% of biopsies. The most prevalent was nonsepsis genitourinary infections (8.3%), followed by bleeding complications (7.3%) and urinary retention (2.9%) (Table 1). The 30-d hospitalization rate was 2.8%.
We first examined unadjusted rates of prostate biopsy among men aged ≥40 yr (Fig. 1 and Table 2). From January 2005 to September 2014, biopsy rates fell 33% from 64.1 to 42.8 per 100 000 person-months. There was substantial seasonal variation in biopsy rates with reductions near the end of each calendar year. With the ITS approach, each intervention is potentially associated with an immediate change in event rates (ie, level shift or change in intercept) as well as a change in the rate of growth (ie, change in slope). At baseline, there was a statistically significant growth in prostate biopsy rates of 2.9 biopsies per 100 000 person-months per year (95% confidence interval [CI], 0.96–4.8; p < 0.001). Following the 2008 USPSTF recommendations, there was an immediate decrease (ie, level shift) in biopsy rates of −10.1 biopsies per 100 000 person-months (95% CI, −17.1 to −3.0; p < 0.001). Likewise, there were immediate decreases in biopsy rates following the 2012 USPSTF recommendations (−13.8 biopsies per 100 000 person-months; 95% CI, −21.0 to −6.7; p < 0.001) and 2013 AUA guidelines (−8.8 biopsies per 100 000 person-months; 95% CI, −16.7 to −0.92; p = 0.03). Analyses adjusted for case mix were similar, although the baseline growth rate was no longer significant, indicating this was related to population shifts over time (Table 2, model 2).
Fig. 1
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 2
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315)
| Prostate biopsy | Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Model 1* | ||||||||||||||
| Baseline rate | 57.8 (53.4–62.3) |
<0.001 | 8.6 (7.5–9.7) |
<0.001 | 4.0 (3.3–4.6) |
<0.001 | 0.17 (0.03–0.32) |
0.02 | 2.8 (2.3–3.3) |
<0.001 | 1.5 (1.3–1.6) |
<0.001 | 1.4 (1.1–1.7) |
<0.001 |
| Baseline trend in rate | 2.9 (0.96–4.8) |
<0.001 | 0.84 (0.48–1.3) |
<0.001 | 0.72 (0.48–1.1) |
<0.001 | 0.12 (0.12–0.12) |
<0.001 | 0.36 (0.12–0.48) |
<0.001 | 0.12 (0.00–0.12) |
<0.001 | 0.00 (−0.12 to 0.12) |
0.5 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.1 (−17.1 to −3.0) |
<0.001 | −0.61 (−3.1 to 1.9) |
0.6 | −1.4 (−2.8 to −0.03) |
0.053 | 0.04 (−0.35 to 0.42) |
0.8 | 0.56 (−0.70 to 1.8) |
0.4 | −0.06 (−0.69 to 0.57) |
0.9 | 0.16 (−0.58 to 0.90) |
0.7 |
| Trend change | 7.9 (−12.7 to 28.4) |
0.5 | −1.0 (−9.1 to 7.2) |
0.8 | −0.12 (−4.1 to 4.0) |
0.98 | −0.60 (−1.9 to 0.72) |
0.4 | −2.0 (−6.5 to 2.4) |
0.4 | −1.1 (−3.4 to 1.3) |
0.4 | −1.0 (−3.6 to 1.6) |
0.5 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.6 (−12 to 4.5) |
0.4 | −0.74 (−3.8 to 2.4) |
0.6 | −0.12 (−1.8 to 1.5) |
0.9 | 0.29 (−0.20 to 0.78) |
0.2 | −0.04 (−1.6 to 1.6) |
0.97 | 0.59 (−0.23 to 1.4) |
0.2 | 0.51 (−0.43 to 1.5) |
0.3 |
| Trend change | −11.2 (−32.0 to 9.6) |
0.3 | 0.0 (−8.2 to 8.0) |
0.99 | −0.96 (−5.0 to 3.1) |
0.7 | 0.36 (−0.96 to 1.8) |
0.5 | 1.7 (−2.8 to 6.1) |
0.5 | 1.0 (−1.4 to 3.4) |
0.4 | 0.84 (−1.7 to 3.5) |
0.5 |
| 2012 USPSTF | ||||||||||||||
| Level change | −13.8 (−21.0 to −6.7) |
<0.001 | −2.5 (−4.5 to −0.45) |
0.02 | −1.7 (−3.1 to −0.31) |
0.02 | −0.12 (−0.42 to 0.19) |
0.5 | −0.61 (−1.5 to 0.28) |
0.2 | −0.75 (−1.1 to −0.44) |
<0.001 | −0.25 (−0.77 to 0.27) |
0.4 |
| Trend change | 8.2 (−5.8 to 22.0) |
0.3 | 3.0 (−0.36 to 6.4) |
0.09 | 3.1 (0.72–5.5) |
0.01 | 0.12 (−0.36 to 0.60) |
0.7 | 0.84 (−0.72 to 2.3) |
0.3 | 0.48 (0.00 to 1.1) |
0.08 | 0.24 (−0.60 to 1.2) |
0.6 |
| 2013 AUA | ||||||||||||||
| Level change | −8.8 (−16.7 to −0.92) |
0.03 | −1.4 (−3.8 to 1.1) |
0.3 | −1.9 (−3.5 to −0.22) |
0.03 | −0.02 (−0.41 to 0.37) |
0.9 | −0.19 (−1.3 to 0.91) |
0.7 | −0.38 (−0.77 to 0.01) |
0.06 | −0.01 (−0.66 to 0.63) |
0.96 |
| Trend change | −10.4 (−30.0 to 9.2) |
0.3 | −4.1 (−7.9 to −0.24) |
0.04 | −2.9 (−6.0 to 0.12) |
0.07 | −0.24 (−0.72 to 0.36) |
0.5 | −1.8 (−3.5 to 0.0) |
0.046 | −0.60 (−1.2 to 0.12) |
0.09 | −0.60 (−1.7 to 0.36) |
0.2 |
| Model 2† | ||||||||||||||
| Baseline rate | 86.8 (82.1–91.4) |
<0.001 | 16.2 (14.9–17.6) |
<0.001 | 6.4 (4.9–7.9) |
<0.001 | 0.39 (−0.13 to 0.91) |
0.2 | 6.1 (4.6–7.7) |
<0.001 | 3.6 (2.8–4.4) |
<0.001 | 3.0 (2.2–3.8) |
<0.001 |
| Baseline trend in rate | 0.24 (−1.8 to 2.3) |
0.8 | −0.24 (−0.72 to 0.24) |
0.4 | 0.48 (−0.12 to 1.1) |
0.09 | 0.12 (−0.12 to 0.24) |
0.4 | −0.24 (−0.72 to 0.24) |
0.4 | −0.24 (−0.48 to 0.00) |
0.054 | −0.12 (−0.36 to 0.12) |
0.3 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.8 (−17.8 to −3.8) |
<0.001 | −5.4 (−9.6 to −1.2) |
0.01 | −2.5 (−5.6 to 0.71) |
0.1 | −0.61 (−1.9 to 0.71) |
0.4 | 0.14 (−3.8 to 4.1) |
0.9 | 1.1 (−0.98 to 3.1) |
0.3 | 0.14 (−1.9 to 2.1) |
0.9 |
| Trend change | 10.6 (−10.1 to 31.2) |
0.3 | 6.1 (−8.4 to 20.8) |
0.4 | −1.7 (−12.2 to 8.8) |
0.8 | 1.7 (−2.9 to 6.2) |
0.5 | 1.7 (−12.0 to 15.5) |
0.8 | −4.7 (−11.8 to 2.5) |
0.2 | −0.84 (−7.8 to 6.1) |
0.8 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.9 (−12 to 4.1) |
0.3 | −0.24 (−5.6 to 5.1) |
0.9 | 1.9 (−2.1 to 5.8) |
0.4 | −0.32 (−2.0 to 1.4) |
0.7 | −0.99 (−6.0 to 4.0) |
0.7 | 1.8 (−0.84 to 4.4) |
0.2 | 0.87 (−1.7 to 3.4) |
0.5 |
| Trend change | −16.2 (−37.1 to 4.7) |
0.1 | −6.8 (−21.4 to 7.8) |
0.4 | 0.36 (−10.2 to 10.9) |
0.95 | −1.9 (−6.5 to 2.8) |
0.4 | −1.6 (−15.2 to 12.2) |
0.8 | 4.7 (−2.4 to 11.9) |
0.2 | 0.48 (−6.4 to 7.4) |
0.9 |
| 2012 USPSTF | ||||||||||||||
| Level change | −11.0 (−18.1 to −4.0) |
0.001 | −2.8 (−6.0 to 0.38) |
0.09 | 0.53 (−1.9 to 3.0) |
0.7 | −0.02 (−0.95 to 0.91) |
0.97 | −1.8 (−4.6 to 0.96) |
0.2 | −1.5 (−3.1 to 0.17) |
0.08 | 0.16 (−1.3 to 1.6) |
0.8 |
| Trend change | 7.0 (−7.2 to 21.1) |
0.3 | 1.7 (−3.6 to 7.0) |
0.5 | 0.12 (−4.1 to 4.2) |
0.96 | 0.12 (−1.4 to 1.8) |
0.9 | −0.48 (−5.3 to 4.3) |
0.9 | 1.6 (−1.3 to 4.3) |
0.3 | 0.12 (−2.4 to 2.6) |
0.9 |
| 2013 AUA | ||||||||||||||
| Level change | −8.2 (−15.9 to −0.38) |
0.04 | −1.0 (−4.8 to 2.8) |
0.6 | −0.44 (−3.2 to 2.3) |
0.8 | −0.15 (−1.3 to 0.99) |
0.8 | 0.28 (−3.2 to 3.7) |
0.9 | −0.50 (−2.7 to 1.7) |
0.7 | 0.47 (−1.3 to 2.3) |
0.6 |
| Trend change | −4.2 (−24.5 to 16.1) |
0.7 | −0.7 (−7.6 to 6.2) |
0.9 | 1.1 (−4.2 to 6.2) |
0.7 | −0.12 (−1.8 to 1.7) |
0.9 | −0.60 (−5.9 to 4.8) |
0.8 | −1.8 (−4.9 to 1.2) |
0.3 | −0.48 (−3.1 to 2.2) |
0.7 |
* Model 1: Base model with constant, time trend, intervention indicator variables. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
† Model 2: Adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, and anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPTF = US Preventive Services Task Force.
Next, we examined 30-d postbiopsy complications among men aged ≥40 yr as a measure of the absolute morbidity of biopsy (Fig. 1, Supplementary Fig. 3, and Table 2). Across the study periods, overall complication rates decreased 10% from 8.7 to 7.8 per 100 000 person-months. In unadjusted ITS analyses, there was a baseline growth in the overall complication rate (0.84 per 100 000 person-months per year; 95% CI, 0.48–1.3; p < 0.001), driven by growth in the rate of nonsepsis infections (0.72 per 100 000 person-months per year; 95% CI, 0.48–1.1; p < 0.001). Only the 2012 USPSTF recommendations were associated with a significant change in complication rates, following which there was an immediate decrease in the overall complication rate of −2.5 per 100 000 person-months (95% CI, −4.5 to −0.45; p = 0.02), driven by reductions in nonsepsis infections (−1.7 per 100 000 person-months; 95% CI, −3.1 to −0.31; p = 0.02) and urinary retention (−0.75 per 100 000 person-months;95% CI, −1.1 to −0.44; p < 0.001). Following the 2013 AUA guidelines there was also a reduction in the overall complication baseline growth rate (−4.1 per 100 000 person-months per year; 95% CI, −7.9 to −0.24; p = 0.04). In analyses adjusted for case mix (Table 2, model 2), effects were overall attenuated, and only the 2008 USPSTF recommendations were associated with a reduction in overall complication rates (−5.4 per 100 000 person-months; 95% CI, −9.6 to −1.2; p = 0.01).
To examine whether there was a change in the relative morbidity of biopsy, we performed a secondary analysis examining complication rates among men undergoing biopsy (Fig. 2). The proportion of men undergoing biopsy who experienced a complication increased from 14% in January 2005 to 18% in September 2014, an increase of 36%. There was again a baseline growth in overall complication rates (0.72% per year; 95% CI, 0.36–1.1; p < 0.001), driven by an increase in nonsepsis infections and sepsis (Supplementary Table 2). None of the interventions affected this growth. Similar results were obtained after adjusting for case mix (Table 3).
Fig. 2
Interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 3
Adjusted* interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Baseline rate | 15.6 (14.3–16.8) |
<0.001 | 7.1 (6.5–7.7) |
<0.001 | 0.34 (0.11–0.57) |
<0.001 | 6.7 (5.9–7.5) |
<0.001 | 2.8 (2.5–3.2) |
<0.001 | 3.2 (2.8–3.6) |
<0.001 |
| Baseline trend in rate | 0.60 (0.12–1.1) |
0.02 | 0.72 (0.60–0.96) |
<0.001 | 0.12 (0.12–0.24) |
<0.001 | 0.12 (−0.12 to 0.36) |
0.3 | −0.12 (−0.24 to 0.00) |
0.3 | −0.12 (−0.24 to 0.00) |
0.0498 |
| 2008 USPSTF | ||||||||||||
| Level change | 1.1 (−2.0 to 4.1) |
0.5 | −1.3 (−3.2 to 0.58) |
0.2 | 0.13 (−0.45 to 0.72) |
0.7 | 1.9 (−0.11 to 3.9) |
0.07 | 0.46 (−0.50 to 1.4) |
0.4 | 0.27 (−0.69 to 1.2) |
0.6 |
| Trend change | −3.5 (−13.4 to 6.6) |
0.5 | −1.8 (−8.4 to 4.7) |
0.6 | −1.0 (−3.0 to 0.96) |
0.3 | −4.3 (−11.4 to 2.6) |
0.2 | −1.6 (−4.9 to 1.9) |
0.4 | −0.72 (−4.0 to 2.6) |
0.7 |
| ERSPC/PLCO | ||||||||||||
| Level change | −0.14 (−3.9 to 3.6) |
0.9 | 1.1 (−1.3 to 3.4) |
0.4 | 0.55 (−0.19 to 1.3) |
0.2 | 0.39 (−2.2 to 3.0) |
0.8 | 0.75 (−0.47 to 2.0) |
0.2 | 0.65 (−0.56 to 1.9) |
0.3 |
| Trend change | 2.9 (−7.1 to 13.0) |
0.6 | 0.72 (−5.8 to 7.2) |
0.8 | 0.84 (−1.2 to 2.9) |
0.4 | 4.2 (−2.8 to 11.3) |
0.2 | 1.7 (−1.8 to 5.0) |
0.4 | 0.72 (−2.6 to 4.0) |
0.7 |
| 2012 USPSTF | ||||||||||||
| Level change | −0.27 (−2.7 to 2.1) |
0.8 | −0.56 (−2.0 to 0.87) |
0.5 | 0.16 (−0.31 to 0.63) |
0.5 | 1.1 (−0.38 to 2.5) |
0.2 | −0.38 (−1.0 to 0.28) |
0.3 | 0.44 (−0.34 to 1.2) |
0.3 |
| Trend change | 1.3 (−2.6 to 5.3) |
0.5 | 2.9 (0.60–5.3) |
0.02 | −0.24 (−1.1 to 0.60) |
0.6 | −0.72 (−3.2 to 1.7) |
0.5 | 0.24 (−0.96 to 1.3) |
0.7 | −0.48 (−1.8 to 0.84) |
0.5 |
| 2013 AUA | ||||||||||||
| Level change | 0.25 (−2.6 to 3.1) |
0.9 | −0.64 (−2.3 to 1.1) |
0.5 | 0.22 (−0.40 to 0.83) |
0.5 | 1.1 (−0.68 to 2.8) |
0.2 | −0.29 (−1.1 to 0.48) |
0.5 | 0.47 (−0.55 to 1.5) |
0.4 |
| Trend change | −1.6 (−6.1 to 2.9) |
0.5 | −2.5 (−5.5 to 0.60) |
0.1 | 0.12 (−0.72 to 1.1) |
0.8 | −0.36 (−3.1 to 2.4) |
0.8 | 0.24 (−1.2 to 1.7) |
0.8 | 0.12 (−1.3 to 1.6) |
0.9 |
* Model adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 person-months. Baseline trend in rate and trend change per 100 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial]; USPTF = US Preventive Services Task Force.
Given the increase in the relative morbidity of biopsy, we performed logistic regression to identify predictors of each complication category (Supplementary Table 3 and Table 4). Due to the large sample size, we focused on those features associated with the greatest magnitude of effect. A prior diagnosis of cancer was associated with an increased risk of all complication categories including any complication (odds ratio [OR]: 1.30; 95% CI, 1.22–1.39; p < 0.001) and hospitalization (OR: 1.59; 95% CI, 1.39–1.82; p < 0.001). Similarly, prior fluoroquinolone use was associated with an increased risk of all complication categories including any complication (OR: 1.27; 95% CI, 1.22–1.32; p < 0.001) and sepsis (OR: 1.55; 95% CI, 1.33–1.81; p < 0.001). Age at biopsy ≥70 yr was associated with a nearly 4-fold increased risk of urinary retention. Although anticoagulant use was associated with increased risks of bleeding (OR: 1.28; 95% CI, 1.14–1.45; p < 0.001), sepsis (OR: 1.48; 95% CI, 1.06–2.07; p = 0.02), and hospitalization (OR: 1.39; 95% CI, 1.16–1.67; p < 0.001), the increase in the overall complication risk was modest (OR 1.14; 95% CI, 1.04–1.25; p = 0.004).
Table 4
Multivariable analysis* of 30-d postbiopsy complications among men aged ≥40 yr undergoing prostate biopsy (January 2005 to September 2014) (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Hospitalization | Urinary retention | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Age at biopsy, yr | ||||||||||||
| 40–49 | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| 50–59 | 1.01 (0.94–1.09) | 0.8 | 0.94 (0.85–1.04) | 0.2 | 1.16 (0.83–1.63) | 0.4 | 1.01 (0.90–1.14) | 0.8 | 1.12 (0.93–1.36) | 0.2 | 1.54 (1.20–1.97) | <0.001 |
| 60–69 | 1.05 (0.97–1.13) | 0.2 | 0.94 (0.85–1.03) | 0.2 | 0.98 (0.70–1.37) | 0.9 | 1.05 (0.94–1.18) | 0.4 | 1.16 (0.95–1.40) | 0.1 | 2.16 (1.68–2.77) | <0.001 |
| ≥70 | 1.25 (1.15–1.36) | <0.001 | 1.03 (0.93–1.15) | 0.6 | 1.05 (0.74–1.49) | 0.8 | 1.29 (1.15–1.46) | <0.001 | 1.31 (1.07–1.60) | 0.008 | 3.98 (3.08–5.15) | <0.001 |
| Race | ||||||||||||
| White | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| Asian | 1.27 (1.15–1.40) | <0.001 | 1.27 (1.11–1.44) | 0.0003 | 1.65 (1.19–2.29) | 0.003 | 1.36 (1.18–1.56) | <0.001 | 0.94 (0.76–1.17) | 0.6 | 1.15 (0.93–1.42) | 0.2 |
| African American | 1.09 (1.03–1.15) | 0.002 | 1.05 (0.97–1.13) | 0.2 | 1.34 (1.09–1.64) | 0.005 | 1.17 (1.09–1.27) | <0.001 | 1.13 (1.01–1.27) | 0.03 | 1.15 (1.03–1.28) | 0.02 |
| Hispanic | 1.32 (1.24–1.40) | <0.001 | 1.36 (1.26–1.47) | <0.001 | 1.52 (1.21–1.90) | <0.001 | 1.33 (1.22–1.45) | <0.001 | 1.12 (0.97–1.28) | 0.1 | 1.37 (1.21–1.56) | <0.001 |
| Unknown/missing | 0.99 (0.93–1.06) | 0.8 | 1.07 (0.99–1.17) | 0.1 | 1.23 (0.94–1.61) | 0.1 | 0.96 (0.88–1.05) | 0.4 | 0.94 (0.81–1.08) | 0.4 | 1.02 (0.88–1.18) | 0.8 |
| Diabetes mellitus | 1.03 (0.99–1.07) | 0.09 | 1.06 (1.00–1.11) | 0.045 | 1.08 (0.92–1.28) | 0.4 | 1.01 (0.96–1.07) | 0.7 | 1.20 (1.10–1.31) | <0.001 | 0.99 (0.91–1.08) | 0.8 |
| COPD | 1.14 (1.09–1.20) | <0.001 | 1.07 (1.00–1.14) | 0.04 | 1.23 (1.01–1.49) | 0.04 | 1.14 (1.06–1.22) | <0.001 | 1.36 (1.23–1.51) | <0.001 | 1.14 (1.03–1.25) | 0.01 |
| PVD | 1.24 (1.16–1.31) | <0.001 | 1.1 (1.02–1.2) | 0.02 | 1.19 (0.93–1.51) | 0.2 | 1.28 (1.18–1.40) | <0.001 | 1.31 (1.15–1.48) | <0.001 | 1.26 (1.12–1.42) | <0.001 |
| Cerebrovascular disease | 1.11 (1.04–1.19) | 0.001 | 1.12 (1.03–1.22) | 0.01 | 1.21 (0.93–1.57) | 0.2 | 1.03 (0.94–1.14) | 0.5 | 1.26 (1.10–1.44) | <0.001 | 1.30 (1.15–1.47) | <0.001 |
| Cancer | 1.30 (1.22–1.39) | <0.001 | 1.16 (1.06–1.27) | 0.001 | 1.37 (1.05–1.79) | 0.02 | 1.40 (1.27–1.53) | <0.001 | 1.59 (1.39–1.82) | <0.001 | 1.32 (1.16–1.52) | <0.001 |
| Anticoagulant use | 1.14 (1.04–1.25) | 0.004 | 1.01 (0.89–1.14) | 0.9 | 1.48 (1.06–2.07) | 0.02 | 1.28 (1.14–1.45) | <0.001 | 1.39 (1.16–1.67) | <0.001 | 1.11 (0.92–1.33) | 0.3 |
| Prior fluoroquinolones | 1.27 (1.22–1.32) | <0.001 | 1.37 (1.31–1.45) | <0.001 | 1.55 (1.33–1.81) | <0.001 | 1.21 (1.14–1.28) | <0.001 | 1.25 (1.14–1.36) | <0.001 | 1.27 (1.16–1.38) | <0.001 |
* Models adjusted for year of biopsy and the features listed in the table.
CI = confidence interval; COPD = chronic obstructive pulmonary disease; PVD = peripheral vascular disease.
We conducted sensitivity analyses where the postintervention run-in period was varied from 2 to 4 mo (Supplementary Table 4). Results overall were similar, although effects were attenuated for the 2013 AUA guidelines on biopsy rates and the 2012 USPSTF recommendations on overall complication rates, likely due to fewer time points for analysis. In a separate sensitivity analysis, we examined bleeding complications excluding microhematuria from 2009 to 2014, following introduction of a separate code for microhematuria. Bleeding rates paralleled those of the primary bleeding end point (Supplementary Fig. 4), and results of both ITS and logistic regression analyses were similar (Supplementary Table 5 and 6). As an exploratory analysis, we examined PCa diagnosis rates within 60 d following biopsy (Supplementary Fig. 5). These appeared to increase steadily across the study period.
To our knowledge, this is the largest study to examine the impact of landmark PSA screening publications on rates of prostate biopsy and the first to examine their impact on postbiopsy complications. Overall, biopsy rates have declined by 33%, which appears attributable to the publication of the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines. Interestingly, the magnitude of reduction in biopsy rates exceeded previously reported decreases in PSA screening rates [6], [7], [8], [9], [10], [11], [12], [22], and [23], which suggests that either the threshold for referral to a urologist or the threshold for performing a biopsy has changed. More importantly, there was a concurrent 22% decrease in 30-d complication rates that appears largely attributable to the 2012 USPSTF recommendations. Still, the reduction in complication rates was less than the reduction in biopsy rates due to an increase in the relative morbidity of biopsy, driven by growth in infectious complications. This may in part be related to a shift to older patients with greater comorbidity undergoing biopsy, although similar results were obtained in analyses adjusted for case mix, which suggests additional factors may exist.
To this end, we evaluated predictors of postbiopsy complications to identify actionable targets for further risk reduction. Some of these features represent modifiable risk factors (eg, anticoagulant use), and others represent opportunities for altering periprocedural management. For instance, prior fluoroquinolone use may identify patients who would benefit from prebiopsy rectal swabs to guide periprocedural antibiotics [24] and [25]. Nonmodifiable risk factors, including specific comorbidities such as cancer and peripheral vascular disease, may encourage reconsideration of the risk–benefit ratio before proceeding with biopsy.
It is also important to consider the concurrent impact of landmark PSA screening publications on rates of PCa diagnosis to contextualize the observed changes in biopsy and its morbidity. An exploratory analysis suggested that cancer diagnosis rates have steadily increased, although further studies are necessary to examine this issue comprehensively.
A number of studies have examined the impact of the 2008 USPSTF recommendations [7], [8], [9], and [22], ERSPC/PLCO trials [6] and [9], and 2012 USPSTF recommendations [6], [10], [11], [12], and [23] on PSA screening rates, generally noting reductions ranging from 0% to 8%. However, much less data are available on the impact of these publications on biopsy rates. One single-center study noted a 31% reduction in the number of biopsies performed following the 2012 USPSTF recommendations [13]; another noted no change in the proportion of referred men who underwent biopsy [26]. A report from a Canadian academic health network noted a reduction in biopsy rates from 58.0 to 35.5 biopsies per month following the 2012 USPSTF recommendations [27]. Our findings therefore confirm a reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines.
Limited data exist regarding secular trends in postbiopsy complications, and prior studies have focused exclusively on the relative morbidity among men undergoing biopsy rather than absolute complication rates. A Canadian population-based study observed that the 30-d hospitalization rate increased from 1.0% in 1996 to 4.1% in 2005, driven by an increase in infectious complications [28]. A Surveillance, Epidemiology and End Results–Medicare study also noted an increase in hospitalizations related to infectious complications over time [14]. Although we observed a similar increase in 30-d overall complications among men undergoing biopsy, likewise driven by infectious complications, there was an absolute 22% reduction in 30-d complications over the study period. Taken together, these findings support a reduction in the morbidity of biopsy following publication of revised PSA screening guidelines but highlight the continued need to reduce the relative morbidity of biopsy.
A number of studies have described complication rates following prostate biopsy [15], [16], [29], and [30]. Hematuria has been reported in 10–84% of biopsies, rectal bleeding in 1–45%, infectious complications in up to 6.3%, urinary retention in up to 1.7%, and hospitalization in up to 6.9% [14] and [29]. The wide range in complication rates is due to differences in definitions, ascertainment of end points, and variation in clinical practice. Given the heterogeneity in postbiopsy complication rates, our findings have two advantages. First, they represent a contemporary, geographically diverse population across the entire United States. Second, use of claims-based ascertainment limited complications to those cases in which patients sought medical care, and although this may underestimate the comprehensive morbidity of biopsy, it more accurately reflects the economic burden of serious complications.
This study has several limitations. Most importantly, a control population of men who were not exposed to the PSA screening publications was not available. This study is retrospective, and additional, unrecognized changes in policy or clinical practice may possibly have confounded the impact of the interventions examined. However, the ITS approach has been described as the strongest quasi-experimental design to evaluate policy changes when a randomized trial is not possible [19]. Because we relied on administrative claims for ascertainment of complications, minor complications that did not generate a claim were not captured, thus underestimating these end points. Although we adjusted for case mix, we were unable to adjust for PSA. However, population shifts in PSA may be confounded by changes in screening practices, and PSA adjustment may therefore under- or overestimate the impact of interventions. We were also unable to adjust for other factors that may be associated with biopsy or complication rates, such as prostate magnetic resonance imaging (MRI), ancillary biomarker testing, biopsy approach (eg, MRI-fusion biopsy, transperineal biopsy), indication for anticoagulation, use of periprocedural bridging, or operator biopsy volume/experience.
We observed an immediate reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines, with a smaller reduction in absolute 30-d complication rates. Still, the relative morbidity of biopsy has increased over time, driven by infectious complications. Accordingly, if health policy aims to reduce the morbidity of PSA screening related to biopsy, further efforts should focus on improving patient selection and periprocedural processes of care.
Author contributions: Boris Gershman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Gershman, van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Acquisition of data: Van Houten, Shah.
Analysis and interpretation of data: Gershman, Van Houten, Shah, Karnes.
Drafting of the manuscript: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Critical revision of the manuscript for important intellectual content: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Statistical analysis: Van Houten.
Obtaining funding: Karnes, Shah.
Administrative, technical, or material support: Karnes, Shah.
Supervision: Karnes, Shah.
Other (specify): None.
Financial disclosures: Boris Gershman certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: This publication was made possible by funding from the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery.
Funding/Support and role of the sponsor: None.
In 2009, two landmark randomized trials—the European Randomized Study of Screening for Prostate Cancer [ERSPC] [1] and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial [PLCO] [2]—reported smaller than anticipated reductions in prostate cancer (PCa) mortality and prompted reexamination of prostate-specific antigen (PSA) screening recommendations. The US Preventive Services Task Force (USPSTF), which in 2008 had recommended against PSA screening in men aged >75 yr [3], updated its position in 2012 to recommend against systematic PSA screening in the general population [4]. The American Urological Association (AUA) likewise updated its PSA screening guidelines in May 2013, recommending shared decision making for men aged 55–69 yr [5]. Both the 2012 USPSTF and 2013 AUA guidelines cited prostate biopsy and its potential complications as a significant harm associated with PSA screening.
Although several studies have noted a reduction in PSA screening rates following these publications [6], [7], [8], [9], [10], [11], and [12], there are little data on whether this has translated into the intended reductions in screening-related morbidity in the form of prostate biopsy and postbiopsy complications [13]. We evaluated whether publication of the 2008 USPSTF recommendations, 2009 ERSPC/PLCO trials, 2012 USPSTF recommendations, or 2013 AUA guidelines was associated with reductions in prostate biopsy or postbiopsy complications. Biopsy-related complications also remain understudied both with regard to incidence as well as identification of risk factors [14], [15], and [16]. Consequently, we also sought to characterize postbiopsy complications and evaluate predictors using a contemporary population-based cohort.
We conducted a retrospective analysis of administrative claims from the Optum Labs Data Warehouse that contains data on >100 million individuals enrolled in private health plans and Medicare Advantage Plans, from geographically diverse regions across the United States, as described in Supplementary Methods [17]. Because this study involved analysis of preexisting deidentified data, it was exempt from institutional review board approval.
We identified all prostate biopsies performed in men aged ≥40 yr between January 1, 2005, and September 30, 2014, using Current Procedural Terminology-4 code 55700 alone (11%) or in combination with 76872 (84%) or 76942 (0%), or International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 60.11 (5%). The first biopsy was selected as the index biopsy when more than one biopsy was performed. Patients with a prior diagnosis of PCa (ICD-9 code 185) were excluded. We restricted the cohort to patients with continuous medical and pharmacy enrollment for at least 12 mo prior to biopsy and 30 d after biopsy to ascertain comorbidities and 30-d complications (Supplementary Fig. 1).
We recorded sociodemographic characteristics, comorbid conditions, anticoagulant use within 30 d of biopsy, and fluoroquinolone use in the 12 mo prior to biopsy as described in Supplement 1. The Charlson Comorbidity Index was calculated as previously described [18]. Prebiopsy PSA values within 60 d of biopsy were available for a subset of patients.
The primary outcomes were crude and adjusted rates of prostate biopsy and 30-d complications. Complications were identified using inpatient and outpatient administrative claims as summarized in Supplementary Table 1.
Baseline demographic features were compared across study periods using the Kruskal-Wallis test for continuous variables and the chi-square test for discrete variables. To evaluate the impact of each publication on rates of prostate biopsy and 30-d complications, we used an interrupted time series (ITS) approach as described in Supplementary Methods [19] and [20]. Rates were estimated using all men aged ≥40 yr with at least 12 mo of continuous medical and pharmacy enrollment as the denominator. As secondary analyses, we estimated rates of 30-d complications among men who underwent biopsy. Two separate ITS analyses were performed: (1) unadjusted analyses, using crude rates of biopsy and complications aggregated for each month, and (2) analyses standardized for case mix, to adjust for population shifts over time as described in Supplementary Methods [20] and [21].
Logistic regression was used to evaluate predictors of 30-d complications. Multivariable models were adjusted for age at biopsy, year of biopsy, race, anticoagulant use, history of fluoroquinolone use, and the presence of the five most prevalent Charlson comorbidities.
Statistical analyses were performed using SAS v.9.3 (SAS Institute Inc., Cary, NC, USA). All tests were two sided with p < 0.05 considered statistically significant.
Between January 2005 and September 2014, a total of 104 584 men underwent prostate biopsy (Table 1). Median age at biopsy was 63.0 yr (interquartile range [IQR]: 57.0–70.0). Overall, 18% of men had used fluoroquinolones in the 12 mo preceding biopsy, and 2.8% were on anticoagulants. Median prebiopsy PSA was 5.1 ng/dl (IQR: 4.0–7.1) for the 18 749 men (18%) for whom this was available. Over the time periods, men undergoing biopsy were older, had increased prevalence of anticoagulant use and comorbid conditions, and had a higher PSA (Table 1 and Supplementary Fig. 2).
Table 1
Patient demographics for men who underwent prostate biopsy (January 2005 to September 2014) stratified by pre/postintervention time periods (n = 104 584)
| Baseline Jan 2005–Jul 2008 (n = 37 217) |
2008 USPSTF Aug 2008–Feb 2009 (n = 6672) |
ERSPC/PLCO Mar 2009–Apr 2012 (n = 36 603) |
2012 USPSTF May 2012–Apr 2013 (n = 10 772) |
2013 AUA May 2013–Sep 2014 (n = 13 320) |
p value | Total (n = 104 584) | |
|---|---|---|---|---|---|---|---|
| Age at biopsy, yr | |||||||
| Median (IQR) | 62.0 (56.0–68.0) | 62.0 (56.0–68.0) | 63.0 (57.0–70.0) | 65.0 (59.0–71.0) | 66.0 (59.0–71.0) | <0.001 | 63.0 (57.0–70.0) |
| Age group, yr (%) | |||||||
| 40–49 | 2269 (6.1) | 416 (6.2) | 1737 (4.7) | 420 (3.9) | 467 (3.5) | <0.001 | 5309 (5.1) |
| 50–59 | 12 366 (33) | 2201 (33) | 10 342 (28) | 2644 (25) | 3098 (23) | 30 651 (29) | |
| 60–69 | 14 671 (39) | 2697 (40) | 15 105 (41) | 4426 (41) | 5546 (42) | 42 445 (41) | |
| ≥70 | 7911 (21) | 1358 (20) | 9419 (26) | 3282 (31) | 4209 (32) | 26 179 (25) | |
| Race (%) | |||||||
| White | 25 508 (69) | 4643 (70) | 26 187 (72) | 7795 (72) | 9407 (71) | <0.001 | 73 540 (70) |
| African American | 3695 (9.9) | 780 (12) | 4423 (12) | 1283 (12) | 1589 (12) | 11 770 (11) | |
| Hispanic | 2549 (6.8) | 538 (8.1) | 2911 (8.0) | 837 (7.8) | 1186 (8.9) | 8021 (7.7) | |
| Asian | 1133 (3.0) | 236 (3.5) | 1142 (3.1) | 306 (2.8) | 387 (2.9) | 3204 (3.1) | |
| Unknown/Missing | 4332 (12) | 475 (7.1) | 1940 (5.3) | 551 (5.1) | 751 (5.6) | 8049 (7.7) | |
| Census region (%) | |||||||
| Northeast | 7094 (19) | 1196 (18) | 6120 (17) | 1886 (18) | 2424 (18) | <0.001 | 18 720 (18) |
| Midwest | 10 094 (27) | 1562 (23) | 8837 (24) | 2742 (26) | 3627 (27) | 26 862 (26) | |
| South | 15 330 (41) | 2889 (43) | 16 371 (45) | 4698 (44) | 5538 (42) | 44 826 (43) | |
| West | 4660 (13) | 993 (15) | 5147 (14) | 1442 (13) | 1728 (13) | 13 970 (13) | |
| Other/Missing | 39 (0.1) | 32 (0.5) | 128 (0.3) | 4 (0) | 3 (0) | 206 (0.2) | |
| PSA, ng/d | n = 5174 | n = 548 | n = 7345 | n = 2211 | n = 3471 | n = 18 749 | |
| Median (IQR) | 5.1 (3.9–7.1) | 4.9 (3.8–6.8) | 5.0 (4.0–6.8) | 5.3 (4.2–7.4) | 5.4 (4.2–7.8) | <0.001 | 5.1 (4.0–7.1) |
| Anticoagulant use (%) | 928 (2.5) | 184 (2.8) | 1103 (3.0) | 299 (2.8) | 415 (3.1) | <0.001 | 2929 (2.8) |
| History of fluoroquinolone use in prior 12 mo (%) | 6797 (18) | 1267 (19) | 6557 (18) | 2017 (19) | 2504 (19) | 0.054 | 19 142 (18) |
| Charlson Comorbidity Index (%) | |||||||
| 0 | 22 354 (60) | 3887 (58) | 20 269 (55) | 5758 (54) | 7039 (53) | <0.001 | 59 307 (57) |
| 1 | 8199 (22) | 1453 (22) | 8252 (23) | 2503 (23) | 2999 (23) | 23 406 (22) | |
| 2 | 3232 (8.7) | 664 (10) | 3520 (9.6) | 1108 (10) | 1359 (10) | 9883 (9.4) | |
| ≥3 | 3432 (9.2) | 668 (10) | 4562 (13) | 1403 (13) | 1923 (14) | 11 988 (12) | |
| Comorbidities (%) | |||||||
| Myocardial infarction | 763 (2.1) | 144 (2.2) | 812 (2.2) | 227 (2.1) | 273 (2.0) | 0.6 | 2219 (2.1) |
| Congestive heart failure | 1443 (3.9) | 223 (3.3) | 1548 (4.2) | 467 (4.3) | 586 (4.4) | <0.001 | 4267 (4.1) |
| Peripheral vascular disease | 2054 (5.5) | 399 (6.0) | 2660 (7.3) | 783 (7.3) | 1071 (8.0) | <0.001 | 6967 (6.7) |
| Cerebrovascular disease | 2007 (5.4) | 379 (5.7) | 2358 (6.4) | 681 (6.3) | 947 (7.1) | <0.001 | 6372 (6.1) |
| Dementia | 370 (1.0) | 63 (0.9) | 442 (1.2) | 141 (1.3) | 198 (1.5) | <0.001 | 1214 (1.2) |
| Chronic obstructive pulmonary disease | 4251 (11) | 758 (11) | 4293 (12) | 1254 (12) | 1686 (13) | 0.004 | 12 242 (12) |
| Ulcer | 362 (1.0) | 60 (0.9) | 297 (0.8) | 84 (0.8) | 102 (0.8) | 0.07 | 905 (0.9) |
| Mild liver disease | 1182 (3.2) | 223 (3.3) | 1332 (3.6) | 413 (3.8) | 515 (3.9) | <0.001 | 3665 (3.5) |
| Diabetes mellitus | 6475 (17) | 1290 (19) | 7823 (21) | 2442 (23) | 2973 (22) | <0.001 | 21 003 (20) |
| Diabetes with organ damage | 1105 (3.0) | 225 (3.4) | 1600 (4.4) | 492 (4.6) | 734 (5.5) | <0.001 | 4156 (4.0) |
| Hemiplegia | 128 (0.3) | 25 (0.4) | 150 (0.4) | 51 (0.5) | 56 (0.4) | 0.3 | 410 (0.4) |
| Moderate/severe renal disease | 1133 (3.0) | 254 (3.8) | 1950 (5.3) | 674 (6.3) | 908 (6.8) | <0.001 | 4919 (4.7) |
| Moderate/severe liver disease | 27 (0.1) | 7 (0.1) | 31 (0.1) | 9 (0.1) | 12 (0.1) | 0.9 | 86 (0.1) |
| Metastatic solid tumor | 175 (0.5) | 37 (0.6) | 199 (0.5) | 58 (0.5) | 80 (0.6) | 0.4 | 549 (0.5) |
| AIDS | 82 (0.2) | 16 (0.2) | 101 (0.3) | 27 (0.3) | 46 (0.3) | 0.2 | 272 (0.3) |
| Rheumatologic disease | 449 (1.2) | 71 (1.1) | 522 (1.4) | 158 (1.5) | 196 (1.5) | 0.007 | 1396 (1.3) |
| Cancer | 1736 (4.7) | 337 (5.1) | 1796 (4.9) | 571 (5.3) | 701 (5.3) | 0.02 | 5141 (4.9) |
| Net worth range (%) | |||||||
| <$25 000 | 1831 (4.9) | 404 (6.1) | 2101 (5.7) | 561 (5.2) | 695 (5.2) | <0.001 | 5592 (5.3) |
| $25 000–149 000 | 5188 (14) | 1042 (16) | 5608 (15) | 1591 (15) | 2046 (15) | 15 475 (15) | |
| $150 000–249 000 | 4754 (13) | 837 (13) | 5182 (14) | 1499 (14) | 1862 (14) | 14 134 (14) | |
| $250 000–499 000 | 9694 (26) | 1779 (27) | 10 048 (28) | 3027 (28) | 3636 (27) | 28 184 (27) | |
| ≥$500 000 | 10 910 (29) | 2030 (30) | 10 860 (30) | 3208 (30) | 3852 (29) | 30 860 (30) | |
| Unknown/missing | 4840 (13) | 580 (8.7) | 2804 (7.7) | 886 (8.2) | 1229 (9.2) | 10 339 (9.9) | |
| 30-d complications (%) | |||||||
| Any | 5973 (16) | 1151 (17) | 6120 (17) | 1863 (17) | 2451 (18) | <0.001 | 17 558 (17) |
| Bleeding | 2547 (6.8) | 499 (7.5) | 2660 (7.3) | 860 (8.0) | 1083 (8.1) | <0.001 | 7649 (7.3) |
| Nonsepsis infection | 2978 (8.0) | 590 (8.8) | 2990 (8.2) | 875 (8.1) | 1236 (9.3) | <0.001 | 8669 (8.3) |
| Sepsis | 191 (0.5) | 50 (0.7) | 349 (1.0) | 105 (1.0) | 134 (1.0) | <0.001 | 829 (0.8) |
| Urinary retention | 980 (2.6) | 161 (2.4) | 1138 (3.1) | 341 (3.2) | 422 (3.2) | <0.001 | 3042 (2.9) |
| Hospitalization | 1019 (2.7) | 181 (2.7) | 1056 (2.9) | 320 (3.0) | 367 (2.8) | 0.6 | 2943 (2.8) |
AIDS = acquired immunodeficiency syndrome; AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; IQR = interquartile range; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer [screening trial]; PSA = prostate-specific antigen; USPTF = US Preventive Services Task Force.
Overall, 30-d complications occurred in 17% of biopsies. The most prevalent was nonsepsis genitourinary infections (8.3%), followed by bleeding complications (7.3%) and urinary retention (2.9%) (Table 1). The 30-d hospitalization rate was 2.8%.
We first examined unadjusted rates of prostate biopsy among men aged ≥40 yr (Fig. 1 and Table 2). From January 2005 to September 2014, biopsy rates fell 33% from 64.1 to 42.8 per 100 000 person-months. There was substantial seasonal variation in biopsy rates with reductions near the end of each calendar year. With the ITS approach, each intervention is potentially associated with an immediate change in event rates (ie, level shift or change in intercept) as well as a change in the rate of growth (ie, change in slope). At baseline, there was a statistically significant growth in prostate biopsy rates of 2.9 biopsies per 100 000 person-months per year (95% confidence interval [CI], 0.96–4.8; p < 0.001). Following the 2008 USPSTF recommendations, there was an immediate decrease (ie, level shift) in biopsy rates of −10.1 biopsies per 100 000 person-months (95% CI, −17.1 to −3.0; p < 0.001). Likewise, there were immediate decreases in biopsy rates following the 2012 USPSTF recommendations (−13.8 biopsies per 100 000 person-months; 95% CI, −21.0 to −6.7; p < 0.001) and 2013 AUA guidelines (−8.8 biopsies per 100 000 person-months; 95% CI, −16.7 to −0.92; p = 0.03). Analyses adjusted for case mix were similar, although the baseline growth rate was no longer significant, indicating this was related to population shifts over time (Table 2, model 2).
Fig. 1
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 2
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315)
| Prostate biopsy | Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Model 1* | ||||||||||||||
| Baseline rate | 57.8 (53.4–62.3) |
<0.001 | 8.6 (7.5–9.7) |
<0.001 | 4.0 (3.3–4.6) |
<0.001 | 0.17 (0.03–0.32) |
0.02 | 2.8 (2.3–3.3) |
<0.001 | 1.5 (1.3–1.6) |
<0.001 | 1.4 (1.1–1.7) |
<0.001 |
| Baseline trend in rate | 2.9 (0.96–4.8) |
<0.001 | 0.84 (0.48–1.3) |
<0.001 | 0.72 (0.48–1.1) |
<0.001 | 0.12 (0.12–0.12) |
<0.001 | 0.36 (0.12–0.48) |
<0.001 | 0.12 (0.00–0.12) |
<0.001 | 0.00 (−0.12 to 0.12) |
0.5 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.1 (−17.1 to −3.0) |
<0.001 | −0.61 (−3.1 to 1.9) |
0.6 | −1.4 (−2.8 to −0.03) |
0.053 | 0.04 (−0.35 to 0.42) |
0.8 | 0.56 (−0.70 to 1.8) |
0.4 | −0.06 (−0.69 to 0.57) |
0.9 | 0.16 (−0.58 to 0.90) |
0.7 |
| Trend change | 7.9 (−12.7 to 28.4) |
0.5 | −1.0 (−9.1 to 7.2) |
0.8 | −0.12 (−4.1 to 4.0) |
0.98 | −0.60 (−1.9 to 0.72) |
0.4 | −2.0 (−6.5 to 2.4) |
0.4 | −1.1 (−3.4 to 1.3) |
0.4 | −1.0 (−3.6 to 1.6) |
0.5 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.6 (−12 to 4.5) |
0.4 | −0.74 (−3.8 to 2.4) |
0.6 | −0.12 (−1.8 to 1.5) |
0.9 | 0.29 (−0.20 to 0.78) |
0.2 | −0.04 (−1.6 to 1.6) |
0.97 | 0.59 (−0.23 to 1.4) |
0.2 | 0.51 (−0.43 to 1.5) |
0.3 |
| Trend change | −11.2 (−32.0 to 9.6) |
0.3 | 0.0 (−8.2 to 8.0) |
0.99 | −0.96 (−5.0 to 3.1) |
0.7 | 0.36 (−0.96 to 1.8) |
0.5 | 1.7 (−2.8 to 6.1) |
0.5 | 1.0 (−1.4 to 3.4) |
0.4 | 0.84 (−1.7 to 3.5) |
0.5 |
| 2012 USPSTF | ||||||||||||||
| Level change | −13.8 (−21.0 to −6.7) |
<0.001 | −2.5 (−4.5 to −0.45) |
0.02 | −1.7 (−3.1 to −0.31) |
0.02 | −0.12 (−0.42 to 0.19) |
0.5 | −0.61 (−1.5 to 0.28) |
0.2 | −0.75 (−1.1 to −0.44) |
<0.001 | −0.25 (−0.77 to 0.27) |
0.4 |
| Trend change | 8.2 (−5.8 to 22.0) |
0.3 | 3.0 (−0.36 to 6.4) |
0.09 | 3.1 (0.72–5.5) |
0.01 | 0.12 (−0.36 to 0.60) |
0.7 | 0.84 (−0.72 to 2.3) |
0.3 | 0.48 (0.00 to 1.1) |
0.08 | 0.24 (−0.60 to 1.2) |
0.6 |
| 2013 AUA | ||||||||||||||
| Level change | −8.8 (−16.7 to −0.92) |
0.03 | −1.4 (−3.8 to 1.1) |
0.3 | −1.9 (−3.5 to −0.22) |
0.03 | −0.02 (−0.41 to 0.37) |
0.9 | −0.19 (−1.3 to 0.91) |
0.7 | −0.38 (−0.77 to 0.01) |
0.06 | −0.01 (−0.66 to 0.63) |
0.96 |
| Trend change | −10.4 (−30.0 to 9.2) |
0.3 | −4.1 (−7.9 to −0.24) |
0.04 | −2.9 (−6.0 to 0.12) |
0.07 | −0.24 (−0.72 to 0.36) |
0.5 | −1.8 (−3.5 to 0.0) |
0.046 | −0.60 (−1.2 to 0.12) |
0.09 | −0.60 (−1.7 to 0.36) |
0.2 |
| Model 2† | ||||||||||||||
| Baseline rate | 86.8 (82.1–91.4) |
<0.001 | 16.2 (14.9–17.6) |
<0.001 | 6.4 (4.9–7.9) |
<0.001 | 0.39 (−0.13 to 0.91) |
0.2 | 6.1 (4.6–7.7) |
<0.001 | 3.6 (2.8–4.4) |
<0.001 | 3.0 (2.2–3.8) |
<0.001 |
| Baseline trend in rate | 0.24 (−1.8 to 2.3) |
0.8 | −0.24 (−0.72 to 0.24) |
0.4 | 0.48 (−0.12 to 1.1) |
0.09 | 0.12 (−0.12 to 0.24) |
0.4 | −0.24 (−0.72 to 0.24) |
0.4 | −0.24 (−0.48 to 0.00) |
0.054 | −0.12 (−0.36 to 0.12) |
0.3 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.8 (−17.8 to −3.8) |
<0.001 | −5.4 (−9.6 to −1.2) |
0.01 | −2.5 (−5.6 to 0.71) |
0.1 | −0.61 (−1.9 to 0.71) |
0.4 | 0.14 (−3.8 to 4.1) |
0.9 | 1.1 (−0.98 to 3.1) |
0.3 | 0.14 (−1.9 to 2.1) |
0.9 |
| Trend change | 10.6 (−10.1 to 31.2) |
0.3 | 6.1 (−8.4 to 20.8) |
0.4 | −1.7 (−12.2 to 8.8) |
0.8 | 1.7 (−2.9 to 6.2) |
0.5 | 1.7 (−12.0 to 15.5) |
0.8 | −4.7 (−11.8 to 2.5) |
0.2 | −0.84 (−7.8 to 6.1) |
0.8 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.9 (−12 to 4.1) |
0.3 | −0.24 (−5.6 to 5.1) |
0.9 | 1.9 (−2.1 to 5.8) |
0.4 | −0.32 (−2.0 to 1.4) |
0.7 | −0.99 (−6.0 to 4.0) |
0.7 | 1.8 (−0.84 to 4.4) |
0.2 | 0.87 (−1.7 to 3.4) |
0.5 |
| Trend change | −16.2 (−37.1 to 4.7) |
0.1 | −6.8 (−21.4 to 7.8) |
0.4 | 0.36 (−10.2 to 10.9) |
0.95 | −1.9 (−6.5 to 2.8) |
0.4 | −1.6 (−15.2 to 12.2) |
0.8 | 4.7 (−2.4 to 11.9) |
0.2 | 0.48 (−6.4 to 7.4) |
0.9 |
| 2012 USPSTF | ||||||||||||||
| Level change | −11.0 (−18.1 to −4.0) |
0.001 | −2.8 (−6.0 to 0.38) |
0.09 | 0.53 (−1.9 to 3.0) |
0.7 | −0.02 (−0.95 to 0.91) |
0.97 | −1.8 (−4.6 to 0.96) |
0.2 | −1.5 (−3.1 to 0.17) |
0.08 | 0.16 (−1.3 to 1.6) |
0.8 |
| Trend change | 7.0 (−7.2 to 21.1) |
0.3 | 1.7 (−3.6 to 7.0) |
0.5 | 0.12 (−4.1 to 4.2) |
0.96 | 0.12 (−1.4 to 1.8) |
0.9 | −0.48 (−5.3 to 4.3) |
0.9 | 1.6 (−1.3 to 4.3) |
0.3 | 0.12 (−2.4 to 2.6) |
0.9 |
| 2013 AUA | ||||||||||||||
| Level change | −8.2 (−15.9 to −0.38) |
0.04 | −1.0 (−4.8 to 2.8) |
0.6 | −0.44 (−3.2 to 2.3) |
0.8 | −0.15 (−1.3 to 0.99) |
0.8 | 0.28 (−3.2 to 3.7) |
0.9 | −0.50 (−2.7 to 1.7) |
0.7 | 0.47 (−1.3 to 2.3) |
0.6 |
| Trend change | −4.2 (−24.5 to 16.1) |
0.7 | −0.7 (−7.6 to 6.2) |
0.9 | 1.1 (−4.2 to 6.2) |
0.7 | −0.12 (−1.8 to 1.7) |
0.9 | −0.60 (−5.9 to 4.8) |
0.8 | −1.8 (−4.9 to 1.2) |
0.3 | −0.48 (−3.1 to 2.2) |
0.7 |
* Model 1: Base model with constant, time trend, intervention indicator variables. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
† Model 2: Adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, and anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPTF = US Preventive Services Task Force.
Next, we examined 30-d postbiopsy complications among men aged ≥40 yr as a measure of the absolute morbidity of biopsy (Fig. 1, Supplementary Fig. 3, and Table 2). Across the study periods, overall complication rates decreased 10% from 8.7 to 7.8 per 100 000 person-months. In unadjusted ITS analyses, there was a baseline growth in the overall complication rate (0.84 per 100 000 person-months per year; 95% CI, 0.48–1.3; p < 0.001), driven by growth in the rate of nonsepsis infections (0.72 per 100 000 person-months per year; 95% CI, 0.48–1.1; p < 0.001). Only the 2012 USPSTF recommendations were associated with a significant change in complication rates, following which there was an immediate decrease in the overall complication rate of −2.5 per 100 000 person-months (95% CI, −4.5 to −0.45; p = 0.02), driven by reductions in nonsepsis infections (−1.7 per 100 000 person-months; 95% CI, −3.1 to −0.31; p = 0.02) and urinary retention (−0.75 per 100 000 person-months;95% CI, −1.1 to −0.44; p < 0.001). Following the 2013 AUA guidelines there was also a reduction in the overall complication baseline growth rate (−4.1 per 100 000 person-months per year; 95% CI, −7.9 to −0.24; p = 0.04). In analyses adjusted for case mix (Table 2, model 2), effects were overall attenuated, and only the 2008 USPSTF recommendations were associated with a reduction in overall complication rates (−5.4 per 100 000 person-months; 95% CI, −9.6 to −1.2; p = 0.01).
To examine whether there was a change in the relative morbidity of biopsy, we performed a secondary analysis examining complication rates among men undergoing biopsy (Fig. 2). The proportion of men undergoing biopsy who experienced a complication increased from 14% in January 2005 to 18% in September 2014, an increase of 36%. There was again a baseline growth in overall complication rates (0.72% per year; 95% CI, 0.36–1.1; p < 0.001), driven by an increase in nonsepsis infections and sepsis (Supplementary Table 2). None of the interventions affected this growth. Similar results were obtained after adjusting for case mix (Table 3).
Fig. 2
Interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 3
Adjusted* interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Baseline rate | 15.6 (14.3–16.8) |
<0.001 | 7.1 (6.5–7.7) |
<0.001 | 0.34 (0.11–0.57) |
<0.001 | 6.7 (5.9–7.5) |
<0.001 | 2.8 (2.5–3.2) |
<0.001 | 3.2 (2.8–3.6) |
<0.001 |
| Baseline trend in rate | 0.60 (0.12–1.1) |
0.02 | 0.72 (0.60–0.96) |
<0.001 | 0.12 (0.12–0.24) |
<0.001 | 0.12 (−0.12 to 0.36) |
0.3 | −0.12 (−0.24 to 0.00) |
0.3 | −0.12 (−0.24 to 0.00) |
0.0498 |
| 2008 USPSTF | ||||||||||||
| Level change | 1.1 (−2.0 to 4.1) |
0.5 | −1.3 (−3.2 to 0.58) |
0.2 | 0.13 (−0.45 to 0.72) |
0.7 | 1.9 (−0.11 to 3.9) |
0.07 | 0.46 (−0.50 to 1.4) |
0.4 | 0.27 (−0.69 to 1.2) |
0.6 |
| Trend change | −3.5 (−13.4 to 6.6) |
0.5 | −1.8 (−8.4 to 4.7) |
0.6 | −1.0 (−3.0 to 0.96) |
0.3 | −4.3 (−11.4 to 2.6) |
0.2 | −1.6 (−4.9 to 1.9) |
0.4 | −0.72 (−4.0 to 2.6) |
0.7 |
| ERSPC/PLCO | ||||||||||||
| Level change | −0.14 (−3.9 to 3.6) |
0.9 | 1.1 (−1.3 to 3.4) |
0.4 | 0.55 (−0.19 to 1.3) |
0.2 | 0.39 (−2.2 to 3.0) |
0.8 | 0.75 (−0.47 to 2.0) |
0.2 | 0.65 (−0.56 to 1.9) |
0.3 |
| Trend change | 2.9 (−7.1 to 13.0) |
0.6 | 0.72 (−5.8 to 7.2) |
0.8 | 0.84 (−1.2 to 2.9) |
0.4 | 4.2 (−2.8 to 11.3) |
0.2 | 1.7 (−1.8 to 5.0) |
0.4 | 0.72 (−2.6 to 4.0) |
0.7 |
| 2012 USPSTF | ||||||||||||
| Level change | −0.27 (−2.7 to 2.1) |
0.8 | −0.56 (−2.0 to 0.87) |
0.5 | 0.16 (−0.31 to 0.63) |
0.5 | 1.1 (−0.38 to 2.5) |
0.2 | −0.38 (−1.0 to 0.28) |
0.3 | 0.44 (−0.34 to 1.2) |
0.3 |
| Trend change | 1.3 (−2.6 to 5.3) |
0.5 | 2.9 (0.60–5.3) |
0.02 | −0.24 (−1.1 to 0.60) |
0.6 | −0.72 (−3.2 to 1.7) |
0.5 | 0.24 (−0.96 to 1.3) |
0.7 | −0.48 (−1.8 to 0.84) |
0.5 |
| 2013 AUA | ||||||||||||
| Level change | 0.25 (−2.6 to 3.1) |
0.9 | −0.64 (−2.3 to 1.1) |
0.5 | 0.22 (−0.40 to 0.83) |
0.5 | 1.1 (−0.68 to 2.8) |
0.2 | −0.29 (−1.1 to 0.48) |
0.5 | 0.47 (−0.55 to 1.5) |
0.4 |
| Trend change | −1.6 (−6.1 to 2.9) |
0.5 | −2.5 (−5.5 to 0.60) |
0.1 | 0.12 (−0.72 to 1.1) |
0.8 | −0.36 (−3.1 to 2.4) |
0.8 | 0.24 (−1.2 to 1.7) |
0.8 | 0.12 (−1.3 to 1.6) |
0.9 |
* Model adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 person-months. Baseline trend in rate and trend change per 100 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial]; USPTF = US Preventive Services Task Force.
Given the increase in the relative morbidity of biopsy, we performed logistic regression to identify predictors of each complication category (Supplementary Table 3 and Table 4). Due to the large sample size, we focused on those features associated with the greatest magnitude of effect. A prior diagnosis of cancer was associated with an increased risk of all complication categories including any complication (odds ratio [OR]: 1.30; 95% CI, 1.22–1.39; p < 0.001) and hospitalization (OR: 1.59; 95% CI, 1.39–1.82; p < 0.001). Similarly, prior fluoroquinolone use was associated with an increased risk of all complication categories including any complication (OR: 1.27; 95% CI, 1.22–1.32; p < 0.001) and sepsis (OR: 1.55; 95% CI, 1.33–1.81; p < 0.001). Age at biopsy ≥70 yr was associated with a nearly 4-fold increased risk of urinary retention. Although anticoagulant use was associated with increased risks of bleeding (OR: 1.28; 95% CI, 1.14–1.45; p < 0.001), sepsis (OR: 1.48; 95% CI, 1.06–2.07; p = 0.02), and hospitalization (OR: 1.39; 95% CI, 1.16–1.67; p < 0.001), the increase in the overall complication risk was modest (OR 1.14; 95% CI, 1.04–1.25; p = 0.004).
Table 4
Multivariable analysis* of 30-d postbiopsy complications among men aged ≥40 yr undergoing prostate biopsy (January 2005 to September 2014) (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Hospitalization | Urinary retention | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Age at biopsy, yr | ||||||||||||
| 40–49 | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| 50–59 | 1.01 (0.94–1.09) | 0.8 | 0.94 (0.85–1.04) | 0.2 | 1.16 (0.83–1.63) | 0.4 | 1.01 (0.90–1.14) | 0.8 | 1.12 (0.93–1.36) | 0.2 | 1.54 (1.20–1.97) | <0.001 |
| 60–69 | 1.05 (0.97–1.13) | 0.2 | 0.94 (0.85–1.03) | 0.2 | 0.98 (0.70–1.37) | 0.9 | 1.05 (0.94–1.18) | 0.4 | 1.16 (0.95–1.40) | 0.1 | 2.16 (1.68–2.77) | <0.001 |
| ≥70 | 1.25 (1.15–1.36) | <0.001 | 1.03 (0.93–1.15) | 0.6 | 1.05 (0.74–1.49) | 0.8 | 1.29 (1.15–1.46) | <0.001 | 1.31 (1.07–1.60) | 0.008 | 3.98 (3.08–5.15) | <0.001 |
| Race | ||||||||||||
| White | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| Asian | 1.27 (1.15–1.40) | <0.001 | 1.27 (1.11–1.44) | 0.0003 | 1.65 (1.19–2.29) | 0.003 | 1.36 (1.18–1.56) | <0.001 | 0.94 (0.76–1.17) | 0.6 | 1.15 (0.93–1.42) | 0.2 |
| African American | 1.09 (1.03–1.15) | 0.002 | 1.05 (0.97–1.13) | 0.2 | 1.34 (1.09–1.64) | 0.005 | 1.17 (1.09–1.27) | <0.001 | 1.13 (1.01–1.27) | 0.03 | 1.15 (1.03–1.28) | 0.02 |
| Hispanic | 1.32 (1.24–1.40) | <0.001 | 1.36 (1.26–1.47) | <0.001 | 1.52 (1.21–1.90) | <0.001 | 1.33 (1.22–1.45) | <0.001 | 1.12 (0.97–1.28) | 0.1 | 1.37 (1.21–1.56) | <0.001 |
| Unknown/missing | 0.99 (0.93–1.06) | 0.8 | 1.07 (0.99–1.17) | 0.1 | 1.23 (0.94–1.61) | 0.1 | 0.96 (0.88–1.05) | 0.4 | 0.94 (0.81–1.08) | 0.4 | 1.02 (0.88–1.18) | 0.8 |
| Diabetes mellitus | 1.03 (0.99–1.07) | 0.09 | 1.06 (1.00–1.11) | 0.045 | 1.08 (0.92–1.28) | 0.4 | 1.01 (0.96–1.07) | 0.7 | 1.20 (1.10–1.31) | <0.001 | 0.99 (0.91–1.08) | 0.8 |
| COPD | 1.14 (1.09–1.20) | <0.001 | 1.07 (1.00–1.14) | 0.04 | 1.23 (1.01–1.49) | 0.04 | 1.14 (1.06–1.22) | <0.001 | 1.36 (1.23–1.51) | <0.001 | 1.14 (1.03–1.25) | 0.01 |
| PVD | 1.24 (1.16–1.31) | <0.001 | 1.1 (1.02–1.2) | 0.02 | 1.19 (0.93–1.51) | 0.2 | 1.28 (1.18–1.40) | <0.001 | 1.31 (1.15–1.48) | <0.001 | 1.26 (1.12–1.42) | <0.001 |
| Cerebrovascular disease | 1.11 (1.04–1.19) | 0.001 | 1.12 (1.03–1.22) | 0.01 | 1.21 (0.93–1.57) | 0.2 | 1.03 (0.94–1.14) | 0.5 | 1.26 (1.10–1.44) | <0.001 | 1.30 (1.15–1.47) | <0.001 |
| Cancer | 1.30 (1.22–1.39) | <0.001 | 1.16 (1.06–1.27) | 0.001 | 1.37 (1.05–1.79) | 0.02 | 1.40 (1.27–1.53) | <0.001 | 1.59 (1.39–1.82) | <0.001 | 1.32 (1.16–1.52) | <0.001 |
| Anticoagulant use | 1.14 (1.04–1.25) | 0.004 | 1.01 (0.89–1.14) | 0.9 | 1.48 (1.06–2.07) | 0.02 | 1.28 (1.14–1.45) | <0.001 | 1.39 (1.16–1.67) | <0.001 | 1.11 (0.92–1.33) | 0.3 |
| Prior fluoroquinolones | 1.27 (1.22–1.32) | <0.001 | 1.37 (1.31–1.45) | <0.001 | 1.55 (1.33–1.81) | <0.001 | 1.21 (1.14–1.28) | <0.001 | 1.25 (1.14–1.36) | <0.001 | 1.27 (1.16–1.38) | <0.001 |
* Models adjusted for year of biopsy and the features listed in the table.
CI = confidence interval; COPD = chronic obstructive pulmonary disease; PVD = peripheral vascular disease.
We conducted sensitivity analyses where the postintervention run-in period was varied from 2 to 4 mo (Supplementary Table 4). Results overall were similar, although effects were attenuated for the 2013 AUA guidelines on biopsy rates and the 2012 USPSTF recommendations on overall complication rates, likely due to fewer time points for analysis. In a separate sensitivity analysis, we examined bleeding complications excluding microhematuria from 2009 to 2014, following introduction of a separate code for microhematuria. Bleeding rates paralleled those of the primary bleeding end point (Supplementary Fig. 4), and results of both ITS and logistic regression analyses were similar (Supplementary Table 5 and 6). As an exploratory analysis, we examined PCa diagnosis rates within 60 d following biopsy (Supplementary Fig. 5). These appeared to increase steadily across the study period.
To our knowledge, this is the largest study to examine the impact of landmark PSA screening publications on rates of prostate biopsy and the first to examine their impact on postbiopsy complications. Overall, biopsy rates have declined by 33%, which appears attributable to the publication of the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines. Interestingly, the magnitude of reduction in biopsy rates exceeded previously reported decreases in PSA screening rates [6], [7], [8], [9], [10], [11], [12], [22], and [23], which suggests that either the threshold for referral to a urologist or the threshold for performing a biopsy has changed. More importantly, there was a concurrent 22% decrease in 30-d complication rates that appears largely attributable to the 2012 USPSTF recommendations. Still, the reduction in complication rates was less than the reduction in biopsy rates due to an increase in the relative morbidity of biopsy, driven by growth in infectious complications. This may in part be related to a shift to older patients with greater comorbidity undergoing biopsy, although similar results were obtained in analyses adjusted for case mix, which suggests additional factors may exist.
To this end, we evaluated predictors of postbiopsy complications to identify actionable targets for further risk reduction. Some of these features represent modifiable risk factors (eg, anticoagulant use), and others represent opportunities for altering periprocedural management. For instance, prior fluoroquinolone use may identify patients who would benefit from prebiopsy rectal swabs to guide periprocedural antibiotics [24] and [25]. Nonmodifiable risk factors, including specific comorbidities such as cancer and peripheral vascular disease, may encourage reconsideration of the risk–benefit ratio before proceeding with biopsy.
It is also important to consider the concurrent impact of landmark PSA screening publications on rates of PCa diagnosis to contextualize the observed changes in biopsy and its morbidity. An exploratory analysis suggested that cancer diagnosis rates have steadily increased, although further studies are necessary to examine this issue comprehensively.
A number of studies have examined the impact of the 2008 USPSTF recommendations [7], [8], [9], and [22], ERSPC/PLCO trials [6] and [9], and 2012 USPSTF recommendations [6], [10], [11], [12], and [23] on PSA screening rates, generally noting reductions ranging from 0% to 8%. However, much less data are available on the impact of these publications on biopsy rates. One single-center study noted a 31% reduction in the number of biopsies performed following the 2012 USPSTF recommendations [13]; another noted no change in the proportion of referred men who underwent biopsy [26]. A report from a Canadian academic health network noted a reduction in biopsy rates from 58.0 to 35.5 biopsies per month following the 2012 USPSTF recommendations [27]. Our findings therefore confirm a reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines.
Limited data exist regarding secular trends in postbiopsy complications, and prior studies have focused exclusively on the relative morbidity among men undergoing biopsy rather than absolute complication rates. A Canadian population-based study observed that the 30-d hospitalization rate increased from 1.0% in 1996 to 4.1% in 2005, driven by an increase in infectious complications [28]. A Surveillance, Epidemiology and End Results–Medicare study also noted an increase in hospitalizations related to infectious complications over time [14]. Although we observed a similar increase in 30-d overall complications among men undergoing biopsy, likewise driven by infectious complications, there was an absolute 22% reduction in 30-d complications over the study period. Taken together, these findings support a reduction in the morbidity of biopsy following publication of revised PSA screening guidelines but highlight the continued need to reduce the relative morbidity of biopsy.
A number of studies have described complication rates following prostate biopsy [15], [16], [29], and [30]. Hematuria has been reported in 10–84% of biopsies, rectal bleeding in 1–45%, infectious complications in up to 6.3%, urinary retention in up to 1.7%, and hospitalization in up to 6.9% [14] and [29]. The wide range in complication rates is due to differences in definitions, ascertainment of end points, and variation in clinical practice. Given the heterogeneity in postbiopsy complication rates, our findings have two advantages. First, they represent a contemporary, geographically diverse population across the entire United States. Second, use of claims-based ascertainment limited complications to those cases in which patients sought medical care, and although this may underestimate the comprehensive morbidity of biopsy, it more accurately reflects the economic burden of serious complications.
This study has several limitations. Most importantly, a control population of men who were not exposed to the PSA screening publications was not available. This study is retrospective, and additional, unrecognized changes in policy or clinical practice may possibly have confounded the impact of the interventions examined. However, the ITS approach has been described as the strongest quasi-experimental design to evaluate policy changes when a randomized trial is not possible [19]. Because we relied on administrative claims for ascertainment of complications, minor complications that did not generate a claim were not captured, thus underestimating these end points. Although we adjusted for case mix, we were unable to adjust for PSA. However, population shifts in PSA may be confounded by changes in screening practices, and PSA adjustment may therefore under- or overestimate the impact of interventions. We were also unable to adjust for other factors that may be associated with biopsy or complication rates, such as prostate magnetic resonance imaging (MRI), ancillary biomarker testing, biopsy approach (eg, MRI-fusion biopsy, transperineal biopsy), indication for anticoagulation, use of periprocedural bridging, or operator biopsy volume/experience.
We observed an immediate reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines, with a smaller reduction in absolute 30-d complication rates. Still, the relative morbidity of biopsy has increased over time, driven by infectious complications. Accordingly, if health policy aims to reduce the morbidity of PSA screening related to biopsy, further efforts should focus on improving patient selection and periprocedural processes of care.
Author contributions: Boris Gershman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Gershman, van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Acquisition of data: Van Houten, Shah.
Analysis and interpretation of data: Gershman, Van Houten, Shah, Karnes.
Drafting of the manuscript: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Critical revision of the manuscript for important intellectual content: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Statistical analysis: Van Houten.
Obtaining funding: Karnes, Shah.
Administrative, technical, or material support: Karnes, Shah.
Supervision: Karnes, Shah.
Other (specify): None.
Financial disclosures: Boris Gershman certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: This publication was made possible by funding from the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery.
Funding/Support and role of the sponsor: None.
In 2009, two landmark randomized trials—the European Randomized Study of Screening for Prostate Cancer [ERSPC] [1] and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial [PLCO] [2]—reported smaller than anticipated reductions in prostate cancer (PCa) mortality and prompted reexamination of prostate-specific antigen (PSA) screening recommendations. The US Preventive Services Task Force (USPSTF), which in 2008 had recommended against PSA screening in men aged >75 yr [3], updated its position in 2012 to recommend against systematic PSA screening in the general population [4]. The American Urological Association (AUA) likewise updated its PSA screening guidelines in May 2013, recommending shared decision making for men aged 55–69 yr [5]. Both the 2012 USPSTF and 2013 AUA guidelines cited prostate biopsy and its potential complications as a significant harm associated with PSA screening.
Although several studies have noted a reduction in PSA screening rates following these publications [6], [7], [8], [9], [10], [11], and [12], there are little data on whether this has translated into the intended reductions in screening-related morbidity in the form of prostate biopsy and postbiopsy complications [13]. We evaluated whether publication of the 2008 USPSTF recommendations, 2009 ERSPC/PLCO trials, 2012 USPSTF recommendations, or 2013 AUA guidelines was associated with reductions in prostate biopsy or postbiopsy complications. Biopsy-related complications also remain understudied both with regard to incidence as well as identification of risk factors [14], [15], and [16]. Consequently, we also sought to characterize postbiopsy complications and evaluate predictors using a contemporary population-based cohort.
We conducted a retrospective analysis of administrative claims from the Optum Labs Data Warehouse that contains data on >100 million individuals enrolled in private health plans and Medicare Advantage Plans, from geographically diverse regions across the United States, as described in Supplementary Methods [17]. Because this study involved analysis of preexisting deidentified data, it was exempt from institutional review board approval.
We identified all prostate biopsies performed in men aged ≥40 yr between January 1, 2005, and September 30, 2014, using Current Procedural Terminology-4 code 55700 alone (11%) or in combination with 76872 (84%) or 76942 (0%), or International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 60.11 (5%). The first biopsy was selected as the index biopsy when more than one biopsy was performed. Patients with a prior diagnosis of PCa (ICD-9 code 185) were excluded. We restricted the cohort to patients with continuous medical and pharmacy enrollment for at least 12 mo prior to biopsy and 30 d after biopsy to ascertain comorbidities and 30-d complications (Supplementary Fig. 1).
We recorded sociodemographic characteristics, comorbid conditions, anticoagulant use within 30 d of biopsy, and fluoroquinolone use in the 12 mo prior to biopsy as described in Supplement 1. The Charlson Comorbidity Index was calculated as previously described [18]. Prebiopsy PSA values within 60 d of biopsy were available for a subset of patients.
The primary outcomes were crude and adjusted rates of prostate biopsy and 30-d complications. Complications were identified using inpatient and outpatient administrative claims as summarized in Supplementary Table 1.
Baseline demographic features were compared across study periods using the Kruskal-Wallis test for continuous variables and the chi-square test for discrete variables. To evaluate the impact of each publication on rates of prostate biopsy and 30-d complications, we used an interrupted time series (ITS) approach as described in Supplementary Methods [19] and [20]. Rates were estimated using all men aged ≥40 yr with at least 12 mo of continuous medical and pharmacy enrollment as the denominator. As secondary analyses, we estimated rates of 30-d complications among men who underwent biopsy. Two separate ITS analyses were performed: (1) unadjusted analyses, using crude rates of biopsy and complications aggregated for each month, and (2) analyses standardized for case mix, to adjust for population shifts over time as described in Supplementary Methods [20] and [21].
Logistic regression was used to evaluate predictors of 30-d complications. Multivariable models were adjusted for age at biopsy, year of biopsy, race, anticoagulant use, history of fluoroquinolone use, and the presence of the five most prevalent Charlson comorbidities.
Statistical analyses were performed using SAS v.9.3 (SAS Institute Inc., Cary, NC, USA). All tests were two sided with p < 0.05 considered statistically significant.
Between January 2005 and September 2014, a total of 104 584 men underwent prostate biopsy (Table 1). Median age at biopsy was 63.0 yr (interquartile range [IQR]: 57.0–70.0). Overall, 18% of men had used fluoroquinolones in the 12 mo preceding biopsy, and 2.8% were on anticoagulants. Median prebiopsy PSA was 5.1 ng/dl (IQR: 4.0–7.1) for the 18 749 men (18%) for whom this was available. Over the time periods, men undergoing biopsy were older, had increased prevalence of anticoagulant use and comorbid conditions, and had a higher PSA (Table 1 and Supplementary Fig. 2).
Table 1
Patient demographics for men who underwent prostate biopsy (January 2005 to September 2014) stratified by pre/postintervention time periods (n = 104 584)
| Baseline Jan 2005–Jul 2008 (n = 37 217) |
2008 USPSTF Aug 2008–Feb 2009 (n = 6672) |
ERSPC/PLCO Mar 2009–Apr 2012 (n = 36 603) |
2012 USPSTF May 2012–Apr 2013 (n = 10 772) |
2013 AUA May 2013–Sep 2014 (n = 13 320) |
p value | Total (n = 104 584) | |
|---|---|---|---|---|---|---|---|
| Age at biopsy, yr | |||||||
| Median (IQR) | 62.0 (56.0–68.0) | 62.0 (56.0–68.0) | 63.0 (57.0–70.0) | 65.0 (59.0–71.0) | 66.0 (59.0–71.0) | <0.001 | 63.0 (57.0–70.0) |
| Age group, yr (%) | |||||||
| 40–49 | 2269 (6.1) | 416 (6.2) | 1737 (4.7) | 420 (3.9) | 467 (3.5) | <0.001 | 5309 (5.1) |
| 50–59 | 12 366 (33) | 2201 (33) | 10 342 (28) | 2644 (25) | 3098 (23) | 30 651 (29) | |
| 60–69 | 14 671 (39) | 2697 (40) | 15 105 (41) | 4426 (41) | 5546 (42) | 42 445 (41) | |
| ≥70 | 7911 (21) | 1358 (20) | 9419 (26) | 3282 (31) | 4209 (32) | 26 179 (25) | |
| Race (%) | |||||||
| White | 25 508 (69) | 4643 (70) | 26 187 (72) | 7795 (72) | 9407 (71) | <0.001 | 73 540 (70) |
| African American | 3695 (9.9) | 780 (12) | 4423 (12) | 1283 (12) | 1589 (12) | 11 770 (11) | |
| Hispanic | 2549 (6.8) | 538 (8.1) | 2911 (8.0) | 837 (7.8) | 1186 (8.9) | 8021 (7.7) | |
| Asian | 1133 (3.0) | 236 (3.5) | 1142 (3.1) | 306 (2.8) | 387 (2.9) | 3204 (3.1) | |
| Unknown/Missing | 4332 (12) | 475 (7.1) | 1940 (5.3) | 551 (5.1) | 751 (5.6) | 8049 (7.7) | |
| Census region (%) | |||||||
| Northeast | 7094 (19) | 1196 (18) | 6120 (17) | 1886 (18) | 2424 (18) | <0.001 | 18 720 (18) |
| Midwest | 10 094 (27) | 1562 (23) | 8837 (24) | 2742 (26) | 3627 (27) | 26 862 (26) | |
| South | 15 330 (41) | 2889 (43) | 16 371 (45) | 4698 (44) | 5538 (42) | 44 826 (43) | |
| West | 4660 (13) | 993 (15) | 5147 (14) | 1442 (13) | 1728 (13) | 13 970 (13) | |
| Other/Missing | 39 (0.1) | 32 (0.5) | 128 (0.3) | 4 (0) | 3 (0) | 206 (0.2) | |
| PSA, ng/d | n = 5174 | n = 548 | n = 7345 | n = 2211 | n = 3471 | n = 18 749 | |
| Median (IQR) | 5.1 (3.9–7.1) | 4.9 (3.8–6.8) | 5.0 (4.0–6.8) | 5.3 (4.2–7.4) | 5.4 (4.2–7.8) | <0.001 | 5.1 (4.0–7.1) |
| Anticoagulant use (%) | 928 (2.5) | 184 (2.8) | 1103 (3.0) | 299 (2.8) | 415 (3.1) | <0.001 | 2929 (2.8) |
| History of fluoroquinolone use in prior 12 mo (%) | 6797 (18) | 1267 (19) | 6557 (18) | 2017 (19) | 2504 (19) | 0.054 | 19 142 (18) |
| Charlson Comorbidity Index (%) | |||||||
| 0 | 22 354 (60) | 3887 (58) | 20 269 (55) | 5758 (54) | 7039 (53) | <0.001 | 59 307 (57) |
| 1 | 8199 (22) | 1453 (22) | 8252 (23) | 2503 (23) | 2999 (23) | 23 406 (22) | |
| 2 | 3232 (8.7) | 664 (10) | 3520 (9.6) | 1108 (10) | 1359 (10) | 9883 (9.4) | |
| ≥3 | 3432 (9.2) | 668 (10) | 4562 (13) | 1403 (13) | 1923 (14) | 11 988 (12) | |
| Comorbidities (%) | |||||||
| Myocardial infarction | 763 (2.1) | 144 (2.2) | 812 (2.2) | 227 (2.1) | 273 (2.0) | 0.6 | 2219 (2.1) |
| Congestive heart failure | 1443 (3.9) | 223 (3.3) | 1548 (4.2) | 467 (4.3) | 586 (4.4) | <0.001 | 4267 (4.1) |
| Peripheral vascular disease | 2054 (5.5) | 399 (6.0) | 2660 (7.3) | 783 (7.3) | 1071 (8.0) | <0.001 | 6967 (6.7) |
| Cerebrovascular disease | 2007 (5.4) | 379 (5.7) | 2358 (6.4) | 681 (6.3) | 947 (7.1) | <0.001 | 6372 (6.1) |
| Dementia | 370 (1.0) | 63 (0.9) | 442 (1.2) | 141 (1.3) | 198 (1.5) | <0.001 | 1214 (1.2) |
| Chronic obstructive pulmonary disease | 4251 (11) | 758 (11) | 4293 (12) | 1254 (12) | 1686 (13) | 0.004 | 12 242 (12) |
| Ulcer | 362 (1.0) | 60 (0.9) | 297 (0.8) | 84 (0.8) | 102 (0.8) | 0.07 | 905 (0.9) |
| Mild liver disease | 1182 (3.2) | 223 (3.3) | 1332 (3.6) | 413 (3.8) | 515 (3.9) | <0.001 | 3665 (3.5) |
| Diabetes mellitus | 6475 (17) | 1290 (19) | 7823 (21) | 2442 (23) | 2973 (22) | <0.001 | 21 003 (20) |
| Diabetes with organ damage | 1105 (3.0) | 225 (3.4) | 1600 (4.4) | 492 (4.6) | 734 (5.5) | <0.001 | 4156 (4.0) |
| Hemiplegia | 128 (0.3) | 25 (0.4) | 150 (0.4) | 51 (0.5) | 56 (0.4) | 0.3 | 410 (0.4) |
| Moderate/severe renal disease | 1133 (3.0) | 254 (3.8) | 1950 (5.3) | 674 (6.3) | 908 (6.8) | <0.001 | 4919 (4.7) |
| Moderate/severe liver disease | 27 (0.1) | 7 (0.1) | 31 (0.1) | 9 (0.1) | 12 (0.1) | 0.9 | 86 (0.1) |
| Metastatic solid tumor | 175 (0.5) | 37 (0.6) | 199 (0.5) | 58 (0.5) | 80 (0.6) | 0.4 | 549 (0.5) |
| AIDS | 82 (0.2) | 16 (0.2) | 101 (0.3) | 27 (0.3) | 46 (0.3) | 0.2 | 272 (0.3) |
| Rheumatologic disease | 449 (1.2) | 71 (1.1) | 522 (1.4) | 158 (1.5) | 196 (1.5) | 0.007 | 1396 (1.3) |
| Cancer | 1736 (4.7) | 337 (5.1) | 1796 (4.9) | 571 (5.3) | 701 (5.3) | 0.02 | 5141 (4.9) |
| Net worth range (%) | |||||||
| <$25 000 | 1831 (4.9) | 404 (6.1) | 2101 (5.7) | 561 (5.2) | 695 (5.2) | <0.001 | 5592 (5.3) |
| $25 000–149 000 | 5188 (14) | 1042 (16) | 5608 (15) | 1591 (15) | 2046 (15) | 15 475 (15) | |
| $150 000–249 000 | 4754 (13) | 837 (13) | 5182 (14) | 1499 (14) | 1862 (14) | 14 134 (14) | |
| $250 000–499 000 | 9694 (26) | 1779 (27) | 10 048 (28) | 3027 (28) | 3636 (27) | 28 184 (27) | |
| ≥$500 000 | 10 910 (29) | 2030 (30) | 10 860 (30) | 3208 (30) | 3852 (29) | 30 860 (30) | |
| Unknown/missing | 4840 (13) | 580 (8.7) | 2804 (7.7) | 886 (8.2) | 1229 (9.2) | 10 339 (9.9) | |
| 30-d complications (%) | |||||||
| Any | 5973 (16) | 1151 (17) | 6120 (17) | 1863 (17) | 2451 (18) | <0.001 | 17 558 (17) |
| Bleeding | 2547 (6.8) | 499 (7.5) | 2660 (7.3) | 860 (8.0) | 1083 (8.1) | <0.001 | 7649 (7.3) |
| Nonsepsis infection | 2978 (8.0) | 590 (8.8) | 2990 (8.2) | 875 (8.1) | 1236 (9.3) | <0.001 | 8669 (8.3) |
| Sepsis | 191 (0.5) | 50 (0.7) | 349 (1.0) | 105 (1.0) | 134 (1.0) | <0.001 | 829 (0.8) |
| Urinary retention | 980 (2.6) | 161 (2.4) | 1138 (3.1) | 341 (3.2) | 422 (3.2) | <0.001 | 3042 (2.9) |
| Hospitalization | 1019 (2.7) | 181 (2.7) | 1056 (2.9) | 320 (3.0) | 367 (2.8) | 0.6 | 2943 (2.8) |
AIDS = acquired immunodeficiency syndrome; AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; IQR = interquartile range; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer [screening trial]; PSA = prostate-specific antigen; USPTF = US Preventive Services Task Force.
Overall, 30-d complications occurred in 17% of biopsies. The most prevalent was nonsepsis genitourinary infections (8.3%), followed by bleeding complications (7.3%) and urinary retention (2.9%) (Table 1). The 30-d hospitalization rate was 2.8%.
We first examined unadjusted rates of prostate biopsy among men aged ≥40 yr (Fig. 1 and Table 2). From January 2005 to September 2014, biopsy rates fell 33% from 64.1 to 42.8 per 100 000 person-months. There was substantial seasonal variation in biopsy rates with reductions near the end of each calendar year. With the ITS approach, each intervention is potentially associated with an immediate change in event rates (ie, level shift or change in intercept) as well as a change in the rate of growth (ie, change in slope). At baseline, there was a statistically significant growth in prostate biopsy rates of 2.9 biopsies per 100 000 person-months per year (95% confidence interval [CI], 0.96–4.8; p < 0.001). Following the 2008 USPSTF recommendations, there was an immediate decrease (ie, level shift) in biopsy rates of −10.1 biopsies per 100 000 person-months (95% CI, −17.1 to −3.0; p < 0.001). Likewise, there were immediate decreases in biopsy rates following the 2012 USPSTF recommendations (−13.8 biopsies per 100 000 person-months; 95% CI, −21.0 to −6.7; p < 0.001) and 2013 AUA guidelines (−8.8 biopsies per 100 000 person-months; 95% CI, −16.7 to −0.92; p = 0.03). Analyses adjusted for case mix were similar, although the baseline growth rate was no longer significant, indicating this was related to population shifts over time (Table 2, model 2).
Fig. 1
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 2
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315)
| Prostate biopsy | Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Model 1* | ||||||||||||||
| Baseline rate | 57.8 (53.4–62.3) |
<0.001 | 8.6 (7.5–9.7) |
<0.001 | 4.0 (3.3–4.6) |
<0.001 | 0.17 (0.03–0.32) |
0.02 | 2.8 (2.3–3.3) |
<0.001 | 1.5 (1.3–1.6) |
<0.001 | 1.4 (1.1–1.7) |
<0.001 |
| Baseline trend in rate | 2.9 (0.96–4.8) |
<0.001 | 0.84 (0.48–1.3) |
<0.001 | 0.72 (0.48–1.1) |
<0.001 | 0.12 (0.12–0.12) |
<0.001 | 0.36 (0.12–0.48) |
<0.001 | 0.12 (0.00–0.12) |
<0.001 | 0.00 (−0.12 to 0.12) |
0.5 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.1 (−17.1 to −3.0) |
<0.001 | −0.61 (−3.1 to 1.9) |
0.6 | −1.4 (−2.8 to −0.03) |
0.053 | 0.04 (−0.35 to 0.42) |
0.8 | 0.56 (−0.70 to 1.8) |
0.4 | −0.06 (−0.69 to 0.57) |
0.9 | 0.16 (−0.58 to 0.90) |
0.7 |
| Trend change | 7.9 (−12.7 to 28.4) |
0.5 | −1.0 (−9.1 to 7.2) |
0.8 | −0.12 (−4.1 to 4.0) |
0.98 | −0.60 (−1.9 to 0.72) |
0.4 | −2.0 (−6.5 to 2.4) |
0.4 | −1.1 (−3.4 to 1.3) |
0.4 | −1.0 (−3.6 to 1.6) |
0.5 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.6 (−12 to 4.5) |
0.4 | −0.74 (−3.8 to 2.4) |
0.6 | −0.12 (−1.8 to 1.5) |
0.9 | 0.29 (−0.20 to 0.78) |
0.2 | −0.04 (−1.6 to 1.6) |
0.97 | 0.59 (−0.23 to 1.4) |
0.2 | 0.51 (−0.43 to 1.5) |
0.3 |
| Trend change | −11.2 (−32.0 to 9.6) |
0.3 | 0.0 (−8.2 to 8.0) |
0.99 | −0.96 (−5.0 to 3.1) |
0.7 | 0.36 (−0.96 to 1.8) |
0.5 | 1.7 (−2.8 to 6.1) |
0.5 | 1.0 (−1.4 to 3.4) |
0.4 | 0.84 (−1.7 to 3.5) |
0.5 |
| 2012 USPSTF | ||||||||||||||
| Level change | −13.8 (−21.0 to −6.7) |
<0.001 | −2.5 (−4.5 to −0.45) |
0.02 | −1.7 (−3.1 to −0.31) |
0.02 | −0.12 (−0.42 to 0.19) |
0.5 | −0.61 (−1.5 to 0.28) |
0.2 | −0.75 (−1.1 to −0.44) |
<0.001 | −0.25 (−0.77 to 0.27) |
0.4 |
| Trend change | 8.2 (−5.8 to 22.0) |
0.3 | 3.0 (−0.36 to 6.4) |
0.09 | 3.1 (0.72–5.5) |
0.01 | 0.12 (−0.36 to 0.60) |
0.7 | 0.84 (−0.72 to 2.3) |
0.3 | 0.48 (0.00 to 1.1) |
0.08 | 0.24 (−0.60 to 1.2) |
0.6 |
| 2013 AUA | ||||||||||||||
| Level change | −8.8 (−16.7 to −0.92) |
0.03 | −1.4 (−3.8 to 1.1) |
0.3 | −1.9 (−3.5 to −0.22) |
0.03 | −0.02 (−0.41 to 0.37) |
0.9 | −0.19 (−1.3 to 0.91) |
0.7 | −0.38 (−0.77 to 0.01) |
0.06 | −0.01 (−0.66 to 0.63) |
0.96 |
| Trend change | −10.4 (−30.0 to 9.2) |
0.3 | −4.1 (−7.9 to −0.24) |
0.04 | −2.9 (−6.0 to 0.12) |
0.07 | −0.24 (−0.72 to 0.36) |
0.5 | −1.8 (−3.5 to 0.0) |
0.046 | −0.60 (−1.2 to 0.12) |
0.09 | −0.60 (−1.7 to 0.36) |
0.2 |
| Model 2† | ||||||||||||||
| Baseline rate | 86.8 (82.1–91.4) |
<0.001 | 16.2 (14.9–17.6) |
<0.001 | 6.4 (4.9–7.9) |
<0.001 | 0.39 (−0.13 to 0.91) |
0.2 | 6.1 (4.6–7.7) |
<0.001 | 3.6 (2.8–4.4) |
<0.001 | 3.0 (2.2–3.8) |
<0.001 |
| Baseline trend in rate | 0.24 (−1.8 to 2.3) |
0.8 | −0.24 (−0.72 to 0.24) |
0.4 | 0.48 (−0.12 to 1.1) |
0.09 | 0.12 (−0.12 to 0.24) |
0.4 | −0.24 (−0.72 to 0.24) |
0.4 | −0.24 (−0.48 to 0.00) |
0.054 | −0.12 (−0.36 to 0.12) |
0.3 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.8 (−17.8 to −3.8) |
<0.001 | −5.4 (−9.6 to −1.2) |
0.01 | −2.5 (−5.6 to 0.71) |
0.1 | −0.61 (−1.9 to 0.71) |
0.4 | 0.14 (−3.8 to 4.1) |
0.9 | 1.1 (−0.98 to 3.1) |
0.3 | 0.14 (−1.9 to 2.1) |
0.9 |
| Trend change | 10.6 (−10.1 to 31.2) |
0.3 | 6.1 (−8.4 to 20.8) |
0.4 | −1.7 (−12.2 to 8.8) |
0.8 | 1.7 (−2.9 to 6.2) |
0.5 | 1.7 (−12.0 to 15.5) |
0.8 | −4.7 (−11.8 to 2.5) |
0.2 | −0.84 (−7.8 to 6.1) |
0.8 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.9 (−12 to 4.1) |
0.3 | −0.24 (−5.6 to 5.1) |
0.9 | 1.9 (−2.1 to 5.8) |
0.4 | −0.32 (−2.0 to 1.4) |
0.7 | −0.99 (−6.0 to 4.0) |
0.7 | 1.8 (−0.84 to 4.4) |
0.2 | 0.87 (−1.7 to 3.4) |
0.5 |
| Trend change | −16.2 (−37.1 to 4.7) |
0.1 | −6.8 (−21.4 to 7.8) |
0.4 | 0.36 (−10.2 to 10.9) |
0.95 | −1.9 (−6.5 to 2.8) |
0.4 | −1.6 (−15.2 to 12.2) |
0.8 | 4.7 (−2.4 to 11.9) |
0.2 | 0.48 (−6.4 to 7.4) |
0.9 |
| 2012 USPSTF | ||||||||||||||
| Level change | −11.0 (−18.1 to −4.0) |
0.001 | −2.8 (−6.0 to 0.38) |
0.09 | 0.53 (−1.9 to 3.0) |
0.7 | −0.02 (−0.95 to 0.91) |
0.97 | −1.8 (−4.6 to 0.96) |
0.2 | −1.5 (−3.1 to 0.17) |
0.08 | 0.16 (−1.3 to 1.6) |
0.8 |
| Trend change | 7.0 (−7.2 to 21.1) |
0.3 | 1.7 (−3.6 to 7.0) |
0.5 | 0.12 (−4.1 to 4.2) |
0.96 | 0.12 (−1.4 to 1.8) |
0.9 | −0.48 (−5.3 to 4.3) |
0.9 | 1.6 (−1.3 to 4.3) |
0.3 | 0.12 (−2.4 to 2.6) |
0.9 |
| 2013 AUA | ||||||||||||||
| Level change | −8.2 (−15.9 to −0.38) |
0.04 | −1.0 (−4.8 to 2.8) |
0.6 | −0.44 (−3.2 to 2.3) |
0.8 | −0.15 (−1.3 to 0.99) |
0.8 | 0.28 (−3.2 to 3.7) |
0.9 | −0.50 (−2.7 to 1.7) |
0.7 | 0.47 (−1.3 to 2.3) |
0.6 |
| Trend change | −4.2 (−24.5 to 16.1) |
0.7 | −0.7 (−7.6 to 6.2) |
0.9 | 1.1 (−4.2 to 6.2) |
0.7 | −0.12 (−1.8 to 1.7) |
0.9 | −0.60 (−5.9 to 4.8) |
0.8 | −1.8 (−4.9 to 1.2) |
0.3 | −0.48 (−3.1 to 2.2) |
0.7 |
* Model 1: Base model with constant, time trend, intervention indicator variables. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
† Model 2: Adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, and anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPTF = US Preventive Services Task Force.
Next, we examined 30-d postbiopsy complications among men aged ≥40 yr as a measure of the absolute morbidity of biopsy (Fig. 1, Supplementary Fig. 3, and Table 2). Across the study periods, overall complication rates decreased 10% from 8.7 to 7.8 per 100 000 person-months. In unadjusted ITS analyses, there was a baseline growth in the overall complication rate (0.84 per 100 000 person-months per year; 95% CI, 0.48–1.3; p < 0.001), driven by growth in the rate of nonsepsis infections (0.72 per 100 000 person-months per year; 95% CI, 0.48–1.1; p < 0.001). Only the 2012 USPSTF recommendations were associated with a significant change in complication rates, following which there was an immediate decrease in the overall complication rate of −2.5 per 100 000 person-months (95% CI, −4.5 to −0.45; p = 0.02), driven by reductions in nonsepsis infections (−1.7 per 100 000 person-months; 95% CI, −3.1 to −0.31; p = 0.02) and urinary retention (−0.75 per 100 000 person-months;95% CI, −1.1 to −0.44; p < 0.001). Following the 2013 AUA guidelines there was also a reduction in the overall complication baseline growth rate (−4.1 per 100 000 person-months per year; 95% CI, −7.9 to −0.24; p = 0.04). In analyses adjusted for case mix (Table 2, model 2), effects were overall attenuated, and only the 2008 USPSTF recommendations were associated with a reduction in overall complication rates (−5.4 per 100 000 person-months; 95% CI, −9.6 to −1.2; p = 0.01).
To examine whether there was a change in the relative morbidity of biopsy, we performed a secondary analysis examining complication rates among men undergoing biopsy (Fig. 2). The proportion of men undergoing biopsy who experienced a complication increased from 14% in January 2005 to 18% in September 2014, an increase of 36%. There was again a baseline growth in overall complication rates (0.72% per year; 95% CI, 0.36–1.1; p < 0.001), driven by an increase in nonsepsis infections and sepsis (Supplementary Table 2). None of the interventions affected this growth. Similar results were obtained after adjusting for case mix (Table 3).
Fig. 2
Interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 3
Adjusted* interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Baseline rate | 15.6 (14.3–16.8) |
<0.001 | 7.1 (6.5–7.7) |
<0.001 | 0.34 (0.11–0.57) |
<0.001 | 6.7 (5.9–7.5) |
<0.001 | 2.8 (2.5–3.2) |
<0.001 | 3.2 (2.8–3.6) |
<0.001 |
| Baseline trend in rate | 0.60 (0.12–1.1) |
0.02 | 0.72 (0.60–0.96) |
<0.001 | 0.12 (0.12–0.24) |
<0.001 | 0.12 (−0.12 to 0.36) |
0.3 | −0.12 (−0.24 to 0.00) |
0.3 | −0.12 (−0.24 to 0.00) |
0.0498 |
| 2008 USPSTF | ||||||||||||
| Level change | 1.1 (−2.0 to 4.1) |
0.5 | −1.3 (−3.2 to 0.58) |
0.2 | 0.13 (−0.45 to 0.72) |
0.7 | 1.9 (−0.11 to 3.9) |
0.07 | 0.46 (−0.50 to 1.4) |
0.4 | 0.27 (−0.69 to 1.2) |
0.6 |
| Trend change | −3.5 (−13.4 to 6.6) |
0.5 | −1.8 (−8.4 to 4.7) |
0.6 | −1.0 (−3.0 to 0.96) |
0.3 | −4.3 (−11.4 to 2.6) |
0.2 | −1.6 (−4.9 to 1.9) |
0.4 | −0.72 (−4.0 to 2.6) |
0.7 |
| ERSPC/PLCO | ||||||||||||
| Level change | −0.14 (−3.9 to 3.6) |
0.9 | 1.1 (−1.3 to 3.4) |
0.4 | 0.55 (−0.19 to 1.3) |
0.2 | 0.39 (−2.2 to 3.0) |
0.8 | 0.75 (−0.47 to 2.0) |
0.2 | 0.65 (−0.56 to 1.9) |
0.3 |
| Trend change | 2.9 (−7.1 to 13.0) |
0.6 | 0.72 (−5.8 to 7.2) |
0.8 | 0.84 (−1.2 to 2.9) |
0.4 | 4.2 (−2.8 to 11.3) |
0.2 | 1.7 (−1.8 to 5.0) |
0.4 | 0.72 (−2.6 to 4.0) |
0.7 |
| 2012 USPSTF | ||||||||||||
| Level change | −0.27 (−2.7 to 2.1) |
0.8 | −0.56 (−2.0 to 0.87) |
0.5 | 0.16 (−0.31 to 0.63) |
0.5 | 1.1 (−0.38 to 2.5) |
0.2 | −0.38 (−1.0 to 0.28) |
0.3 | 0.44 (−0.34 to 1.2) |
0.3 |
| Trend change | 1.3 (−2.6 to 5.3) |
0.5 | 2.9 (0.60–5.3) |
0.02 | −0.24 (−1.1 to 0.60) |
0.6 | −0.72 (−3.2 to 1.7) |
0.5 | 0.24 (−0.96 to 1.3) |
0.7 | −0.48 (−1.8 to 0.84) |
0.5 |
| 2013 AUA | ||||||||||||
| Level change | 0.25 (−2.6 to 3.1) |
0.9 | −0.64 (−2.3 to 1.1) |
0.5 | 0.22 (−0.40 to 0.83) |
0.5 | 1.1 (−0.68 to 2.8) |
0.2 | −0.29 (−1.1 to 0.48) |
0.5 | 0.47 (−0.55 to 1.5) |
0.4 |
| Trend change | −1.6 (−6.1 to 2.9) |
0.5 | −2.5 (−5.5 to 0.60) |
0.1 | 0.12 (−0.72 to 1.1) |
0.8 | −0.36 (−3.1 to 2.4) |
0.8 | 0.24 (−1.2 to 1.7) |
0.8 | 0.12 (−1.3 to 1.6) |
0.9 |
* Model adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 person-months. Baseline trend in rate and trend change per 100 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial]; USPTF = US Preventive Services Task Force.
Given the increase in the relative morbidity of biopsy, we performed logistic regression to identify predictors of each complication category (Supplementary Table 3 and Table 4). Due to the large sample size, we focused on those features associated with the greatest magnitude of effect. A prior diagnosis of cancer was associated with an increased risk of all complication categories including any complication (odds ratio [OR]: 1.30; 95% CI, 1.22–1.39; p < 0.001) and hospitalization (OR: 1.59; 95% CI, 1.39–1.82; p < 0.001). Similarly, prior fluoroquinolone use was associated with an increased risk of all complication categories including any complication (OR: 1.27; 95% CI, 1.22–1.32; p < 0.001) and sepsis (OR: 1.55; 95% CI, 1.33–1.81; p < 0.001). Age at biopsy ≥70 yr was associated with a nearly 4-fold increased risk of urinary retention. Although anticoagulant use was associated with increased risks of bleeding (OR: 1.28; 95% CI, 1.14–1.45; p < 0.001), sepsis (OR: 1.48; 95% CI, 1.06–2.07; p = 0.02), and hospitalization (OR: 1.39; 95% CI, 1.16–1.67; p < 0.001), the increase in the overall complication risk was modest (OR 1.14; 95% CI, 1.04–1.25; p = 0.004).
Table 4
Multivariable analysis* of 30-d postbiopsy complications among men aged ≥40 yr undergoing prostate biopsy (January 2005 to September 2014) (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Hospitalization | Urinary retention | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Age at biopsy, yr | ||||||||||||
| 40–49 | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| 50–59 | 1.01 (0.94–1.09) | 0.8 | 0.94 (0.85–1.04) | 0.2 | 1.16 (0.83–1.63) | 0.4 | 1.01 (0.90–1.14) | 0.8 | 1.12 (0.93–1.36) | 0.2 | 1.54 (1.20–1.97) | <0.001 |
| 60–69 | 1.05 (0.97–1.13) | 0.2 | 0.94 (0.85–1.03) | 0.2 | 0.98 (0.70–1.37) | 0.9 | 1.05 (0.94–1.18) | 0.4 | 1.16 (0.95–1.40) | 0.1 | 2.16 (1.68–2.77) | <0.001 |
| ≥70 | 1.25 (1.15–1.36) | <0.001 | 1.03 (0.93–1.15) | 0.6 | 1.05 (0.74–1.49) | 0.8 | 1.29 (1.15–1.46) | <0.001 | 1.31 (1.07–1.60) | 0.008 | 3.98 (3.08–5.15) | <0.001 |
| Race | ||||||||||||
| White | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| Asian | 1.27 (1.15–1.40) | <0.001 | 1.27 (1.11–1.44) | 0.0003 | 1.65 (1.19–2.29) | 0.003 | 1.36 (1.18–1.56) | <0.001 | 0.94 (0.76–1.17) | 0.6 | 1.15 (0.93–1.42) | 0.2 |
| African American | 1.09 (1.03–1.15) | 0.002 | 1.05 (0.97–1.13) | 0.2 | 1.34 (1.09–1.64) | 0.005 | 1.17 (1.09–1.27) | <0.001 | 1.13 (1.01–1.27) | 0.03 | 1.15 (1.03–1.28) | 0.02 |
| Hispanic | 1.32 (1.24–1.40) | <0.001 | 1.36 (1.26–1.47) | <0.001 | 1.52 (1.21–1.90) | <0.001 | 1.33 (1.22–1.45) | <0.001 | 1.12 (0.97–1.28) | 0.1 | 1.37 (1.21–1.56) | <0.001 |
| Unknown/missing | 0.99 (0.93–1.06) | 0.8 | 1.07 (0.99–1.17) | 0.1 | 1.23 (0.94–1.61) | 0.1 | 0.96 (0.88–1.05) | 0.4 | 0.94 (0.81–1.08) | 0.4 | 1.02 (0.88–1.18) | 0.8 |
| Diabetes mellitus | 1.03 (0.99–1.07) | 0.09 | 1.06 (1.00–1.11) | 0.045 | 1.08 (0.92–1.28) | 0.4 | 1.01 (0.96–1.07) | 0.7 | 1.20 (1.10–1.31) | <0.001 | 0.99 (0.91–1.08) | 0.8 |
| COPD | 1.14 (1.09–1.20) | <0.001 | 1.07 (1.00–1.14) | 0.04 | 1.23 (1.01–1.49) | 0.04 | 1.14 (1.06–1.22) | <0.001 | 1.36 (1.23–1.51) | <0.001 | 1.14 (1.03–1.25) | 0.01 |
| PVD | 1.24 (1.16–1.31) | <0.001 | 1.1 (1.02–1.2) | 0.02 | 1.19 (0.93–1.51) | 0.2 | 1.28 (1.18–1.40) | <0.001 | 1.31 (1.15–1.48) | <0.001 | 1.26 (1.12–1.42) | <0.001 |
| Cerebrovascular disease | 1.11 (1.04–1.19) | 0.001 | 1.12 (1.03–1.22) | 0.01 | 1.21 (0.93–1.57) | 0.2 | 1.03 (0.94–1.14) | 0.5 | 1.26 (1.10–1.44) | <0.001 | 1.30 (1.15–1.47) | <0.001 |
| Cancer | 1.30 (1.22–1.39) | <0.001 | 1.16 (1.06–1.27) | 0.001 | 1.37 (1.05–1.79) | 0.02 | 1.40 (1.27–1.53) | <0.001 | 1.59 (1.39–1.82) | <0.001 | 1.32 (1.16–1.52) | <0.001 |
| Anticoagulant use | 1.14 (1.04–1.25) | 0.004 | 1.01 (0.89–1.14) | 0.9 | 1.48 (1.06–2.07) | 0.02 | 1.28 (1.14–1.45) | <0.001 | 1.39 (1.16–1.67) | <0.001 | 1.11 (0.92–1.33) | 0.3 |
| Prior fluoroquinolones | 1.27 (1.22–1.32) | <0.001 | 1.37 (1.31–1.45) | <0.001 | 1.55 (1.33–1.81) | <0.001 | 1.21 (1.14–1.28) | <0.001 | 1.25 (1.14–1.36) | <0.001 | 1.27 (1.16–1.38) | <0.001 |
* Models adjusted for year of biopsy and the features listed in the table.
CI = confidence interval; COPD = chronic obstructive pulmonary disease; PVD = peripheral vascular disease.
We conducted sensitivity analyses where the postintervention run-in period was varied from 2 to 4 mo (Supplementary Table 4). Results overall were similar, although effects were attenuated for the 2013 AUA guidelines on biopsy rates and the 2012 USPSTF recommendations on overall complication rates, likely due to fewer time points for analysis. In a separate sensitivity analysis, we examined bleeding complications excluding microhematuria from 2009 to 2014, following introduction of a separate code for microhematuria. Bleeding rates paralleled those of the primary bleeding end point (Supplementary Fig. 4), and results of both ITS and logistic regression analyses were similar (Supplementary Table 5 and 6). As an exploratory analysis, we examined PCa diagnosis rates within 60 d following biopsy (Supplementary Fig. 5). These appeared to increase steadily across the study period.
To our knowledge, this is the largest study to examine the impact of landmark PSA screening publications on rates of prostate biopsy and the first to examine their impact on postbiopsy complications. Overall, biopsy rates have declined by 33%, which appears attributable to the publication of the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines. Interestingly, the magnitude of reduction in biopsy rates exceeded previously reported decreases in PSA screening rates [6], [7], [8], [9], [10], [11], [12], [22], and [23], which suggests that either the threshold for referral to a urologist or the threshold for performing a biopsy has changed. More importantly, there was a concurrent 22% decrease in 30-d complication rates that appears largely attributable to the 2012 USPSTF recommendations. Still, the reduction in complication rates was less than the reduction in biopsy rates due to an increase in the relative morbidity of biopsy, driven by growth in infectious complications. This may in part be related to a shift to older patients with greater comorbidity undergoing biopsy, although similar results were obtained in analyses adjusted for case mix, which suggests additional factors may exist.
To this end, we evaluated predictors of postbiopsy complications to identify actionable targets for further risk reduction. Some of these features represent modifiable risk factors (eg, anticoagulant use), and others represent opportunities for altering periprocedural management. For instance, prior fluoroquinolone use may identify patients who would benefit from prebiopsy rectal swabs to guide periprocedural antibiotics [24] and [25]. Nonmodifiable risk factors, including specific comorbidities such as cancer and peripheral vascular disease, may encourage reconsideration of the risk–benefit ratio before proceeding with biopsy.
It is also important to consider the concurrent impact of landmark PSA screening publications on rates of PCa diagnosis to contextualize the observed changes in biopsy and its morbidity. An exploratory analysis suggested that cancer diagnosis rates have steadily increased, although further studies are necessary to examine this issue comprehensively.
A number of studies have examined the impact of the 2008 USPSTF recommendations [7], [8], [9], and [22], ERSPC/PLCO trials [6] and [9], and 2012 USPSTF recommendations [6], [10], [11], [12], and [23] on PSA screening rates, generally noting reductions ranging from 0% to 8%. However, much less data are available on the impact of these publications on biopsy rates. One single-center study noted a 31% reduction in the number of biopsies performed following the 2012 USPSTF recommendations [13]; another noted no change in the proportion of referred men who underwent biopsy [26]. A report from a Canadian academic health network noted a reduction in biopsy rates from 58.0 to 35.5 biopsies per month following the 2012 USPSTF recommendations [27]. Our findings therefore confirm a reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines.
Limited data exist regarding secular trends in postbiopsy complications, and prior studies have focused exclusively on the relative morbidity among men undergoing biopsy rather than absolute complication rates. A Canadian population-based study observed that the 30-d hospitalization rate increased from 1.0% in 1996 to 4.1% in 2005, driven by an increase in infectious complications [28]. A Surveillance, Epidemiology and End Results–Medicare study also noted an increase in hospitalizations related to infectious complications over time [14]. Although we observed a similar increase in 30-d overall complications among men undergoing biopsy, likewise driven by infectious complications, there was an absolute 22% reduction in 30-d complications over the study period. Taken together, these findings support a reduction in the morbidity of biopsy following publication of revised PSA screening guidelines but highlight the continued need to reduce the relative morbidity of biopsy.
A number of studies have described complication rates following prostate biopsy [15], [16], [29], and [30]. Hematuria has been reported in 10–84% of biopsies, rectal bleeding in 1–45%, infectious complications in up to 6.3%, urinary retention in up to 1.7%, and hospitalization in up to 6.9% [14] and [29]. The wide range in complication rates is due to differences in definitions, ascertainment of end points, and variation in clinical practice. Given the heterogeneity in postbiopsy complication rates, our findings have two advantages. First, they represent a contemporary, geographically diverse population across the entire United States. Second, use of claims-based ascertainment limited complications to those cases in which patients sought medical care, and although this may underestimate the comprehensive morbidity of biopsy, it more accurately reflects the economic burden of serious complications.
This study has several limitations. Most importantly, a control population of men who were not exposed to the PSA screening publications was not available. This study is retrospective, and additional, unrecognized changes in policy or clinical practice may possibly have confounded the impact of the interventions examined. However, the ITS approach has been described as the strongest quasi-experimental design to evaluate policy changes when a randomized trial is not possible [19]. Because we relied on administrative claims for ascertainment of complications, minor complications that did not generate a claim were not captured, thus underestimating these end points. Although we adjusted for case mix, we were unable to adjust for PSA. However, population shifts in PSA may be confounded by changes in screening practices, and PSA adjustment may therefore under- or overestimate the impact of interventions. We were also unable to adjust for other factors that may be associated with biopsy or complication rates, such as prostate magnetic resonance imaging (MRI), ancillary biomarker testing, biopsy approach (eg, MRI-fusion biopsy, transperineal biopsy), indication for anticoagulation, use of periprocedural bridging, or operator biopsy volume/experience.
We observed an immediate reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines, with a smaller reduction in absolute 30-d complication rates. Still, the relative morbidity of biopsy has increased over time, driven by infectious complications. Accordingly, if health policy aims to reduce the morbidity of PSA screening related to biopsy, further efforts should focus on improving patient selection and periprocedural processes of care.
Author contributions: Boris Gershman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Gershman, van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Acquisition of data: Van Houten, Shah.
Analysis and interpretation of data: Gershman, Van Houten, Shah, Karnes.
Drafting of the manuscript: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Critical revision of the manuscript for important intellectual content: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Statistical analysis: Van Houten.
Obtaining funding: Karnes, Shah.
Administrative, technical, or material support: Karnes, Shah.
Supervision: Karnes, Shah.
Other (specify): None.
Financial disclosures: Boris Gershman certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: This publication was made possible by funding from the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery.
Funding/Support and role of the sponsor: None.
In 2009, two landmark randomized trials—the European Randomized Study of Screening for Prostate Cancer [ERSPC] [1] and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial [PLCO] [2]—reported smaller than anticipated reductions in prostate cancer (PCa) mortality and prompted reexamination of prostate-specific antigen (PSA) screening recommendations. The US Preventive Services Task Force (USPSTF), which in 2008 had recommended against PSA screening in men aged >75 yr [3], updated its position in 2012 to recommend against systematic PSA screening in the general population [4]. The American Urological Association (AUA) likewise updated its PSA screening guidelines in May 2013, recommending shared decision making for men aged 55–69 yr [5]. Both the 2012 USPSTF and 2013 AUA guidelines cited prostate biopsy and its potential complications as a significant harm associated with PSA screening.
Although several studies have noted a reduction in PSA screening rates following these publications [6], [7], [8], [9], [10], [11], and [12], there are little data on whether this has translated into the intended reductions in screening-related morbidity in the form of prostate biopsy and postbiopsy complications [13]. We evaluated whether publication of the 2008 USPSTF recommendations, 2009 ERSPC/PLCO trials, 2012 USPSTF recommendations, or 2013 AUA guidelines was associated with reductions in prostate biopsy or postbiopsy complications. Biopsy-related complications also remain understudied both with regard to incidence as well as identification of risk factors [14], [15], and [16]. Consequently, we also sought to characterize postbiopsy complications and evaluate predictors using a contemporary population-based cohort.
We conducted a retrospective analysis of administrative claims from the Optum Labs Data Warehouse that contains data on >100 million individuals enrolled in private health plans and Medicare Advantage Plans, from geographically diverse regions across the United States, as described in Supplementary Methods [17]. Because this study involved analysis of preexisting deidentified data, it was exempt from institutional review board approval.
We identified all prostate biopsies performed in men aged ≥40 yr between January 1, 2005, and September 30, 2014, using Current Procedural Terminology-4 code 55700 alone (11%) or in combination with 76872 (84%) or 76942 (0%), or International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 60.11 (5%). The first biopsy was selected as the index biopsy when more than one biopsy was performed. Patients with a prior diagnosis of PCa (ICD-9 code 185) were excluded. We restricted the cohort to patients with continuous medical and pharmacy enrollment for at least 12 mo prior to biopsy and 30 d after biopsy to ascertain comorbidities and 30-d complications (Supplementary Fig. 1).
We recorded sociodemographic characteristics, comorbid conditions, anticoagulant use within 30 d of biopsy, and fluoroquinolone use in the 12 mo prior to biopsy as described in Supplement 1. The Charlson Comorbidity Index was calculated as previously described [18]. Prebiopsy PSA values within 60 d of biopsy were available for a subset of patients.
The primary outcomes were crude and adjusted rates of prostate biopsy and 30-d complications. Complications were identified using inpatient and outpatient administrative claims as summarized in Supplementary Table 1.
Baseline demographic features were compared across study periods using the Kruskal-Wallis test for continuous variables and the chi-square test for discrete variables. To evaluate the impact of each publication on rates of prostate biopsy and 30-d complications, we used an interrupted time series (ITS) approach as described in Supplementary Methods [19] and [20]. Rates were estimated using all men aged ≥40 yr with at least 12 mo of continuous medical and pharmacy enrollment as the denominator. As secondary analyses, we estimated rates of 30-d complications among men who underwent biopsy. Two separate ITS analyses were performed: (1) unadjusted analyses, using crude rates of biopsy and complications aggregated for each month, and (2) analyses standardized for case mix, to adjust for population shifts over time as described in Supplementary Methods [20] and [21].
Logistic regression was used to evaluate predictors of 30-d complications. Multivariable models were adjusted for age at biopsy, year of biopsy, race, anticoagulant use, history of fluoroquinolone use, and the presence of the five most prevalent Charlson comorbidities.
Statistical analyses were performed using SAS v.9.3 (SAS Institute Inc., Cary, NC, USA). All tests were two sided with p < 0.05 considered statistically significant.
Between January 2005 and September 2014, a total of 104 584 men underwent prostate biopsy (Table 1). Median age at biopsy was 63.0 yr (interquartile range [IQR]: 57.0–70.0). Overall, 18% of men had used fluoroquinolones in the 12 mo preceding biopsy, and 2.8% were on anticoagulants. Median prebiopsy PSA was 5.1 ng/dl (IQR: 4.0–7.1) for the 18 749 men (18%) for whom this was available. Over the time periods, men undergoing biopsy were older, had increased prevalence of anticoagulant use and comorbid conditions, and had a higher PSA (Table 1 and Supplementary Fig. 2).
Table 1
Patient demographics for men who underwent prostate biopsy (January 2005 to September 2014) stratified by pre/postintervention time periods (n = 104 584)
| Baseline Jan 2005–Jul 2008 (n = 37 217) |
2008 USPSTF Aug 2008–Feb 2009 (n = 6672) |
ERSPC/PLCO Mar 2009–Apr 2012 (n = 36 603) |
2012 USPSTF May 2012–Apr 2013 (n = 10 772) |
2013 AUA May 2013–Sep 2014 (n = 13 320) |
p value | Total (n = 104 584) | |
|---|---|---|---|---|---|---|---|
| Age at biopsy, yr | |||||||
| Median (IQR) | 62.0 (56.0–68.0) | 62.0 (56.0–68.0) | 63.0 (57.0–70.0) | 65.0 (59.0–71.0) | 66.0 (59.0–71.0) | <0.001 | 63.0 (57.0–70.0) |
| Age group, yr (%) | |||||||
| 40–49 | 2269 (6.1) | 416 (6.2) | 1737 (4.7) | 420 (3.9) | 467 (3.5) | <0.001 | 5309 (5.1) |
| 50–59 | 12 366 (33) | 2201 (33) | 10 342 (28) | 2644 (25) | 3098 (23) | 30 651 (29) | |
| 60–69 | 14 671 (39) | 2697 (40) | 15 105 (41) | 4426 (41) | 5546 (42) | 42 445 (41) | |
| ≥70 | 7911 (21) | 1358 (20) | 9419 (26) | 3282 (31) | 4209 (32) | 26 179 (25) | |
| Race (%) | |||||||
| White | 25 508 (69) | 4643 (70) | 26 187 (72) | 7795 (72) | 9407 (71) | <0.001 | 73 540 (70) |
| African American | 3695 (9.9) | 780 (12) | 4423 (12) | 1283 (12) | 1589 (12) | 11 770 (11) | |
| Hispanic | 2549 (6.8) | 538 (8.1) | 2911 (8.0) | 837 (7.8) | 1186 (8.9) | 8021 (7.7) | |
| Asian | 1133 (3.0) | 236 (3.5) | 1142 (3.1) | 306 (2.8) | 387 (2.9) | 3204 (3.1) | |
| Unknown/Missing | 4332 (12) | 475 (7.1) | 1940 (5.3) | 551 (5.1) | 751 (5.6) | 8049 (7.7) | |
| Census region (%) | |||||||
| Northeast | 7094 (19) | 1196 (18) | 6120 (17) | 1886 (18) | 2424 (18) | <0.001 | 18 720 (18) |
| Midwest | 10 094 (27) | 1562 (23) | 8837 (24) | 2742 (26) | 3627 (27) | 26 862 (26) | |
| South | 15 330 (41) | 2889 (43) | 16 371 (45) | 4698 (44) | 5538 (42) | 44 826 (43) | |
| West | 4660 (13) | 993 (15) | 5147 (14) | 1442 (13) | 1728 (13) | 13 970 (13) | |
| Other/Missing | 39 (0.1) | 32 (0.5) | 128 (0.3) | 4 (0) | 3 (0) | 206 (0.2) | |
| PSA, ng/d | n = 5174 | n = 548 | n = 7345 | n = 2211 | n = 3471 | n = 18 749 | |
| Median (IQR) | 5.1 (3.9–7.1) | 4.9 (3.8–6.8) | 5.0 (4.0–6.8) | 5.3 (4.2–7.4) | 5.4 (4.2–7.8) | <0.001 | 5.1 (4.0–7.1) |
| Anticoagulant use (%) | 928 (2.5) | 184 (2.8) | 1103 (3.0) | 299 (2.8) | 415 (3.1) | <0.001 | 2929 (2.8) |
| History of fluoroquinolone use in prior 12 mo (%) | 6797 (18) | 1267 (19) | 6557 (18) | 2017 (19) | 2504 (19) | 0.054 | 19 142 (18) |
| Charlson Comorbidity Index (%) | |||||||
| 0 | 22 354 (60) | 3887 (58) | 20 269 (55) | 5758 (54) | 7039 (53) | <0.001 | 59 307 (57) |
| 1 | 8199 (22) | 1453 (22) | 8252 (23) | 2503 (23) | 2999 (23) | 23 406 (22) | |
| 2 | 3232 (8.7) | 664 (10) | 3520 (9.6) | 1108 (10) | 1359 (10) | 9883 (9.4) | |
| ≥3 | 3432 (9.2) | 668 (10) | 4562 (13) | 1403 (13) | 1923 (14) | 11 988 (12) | |
| Comorbidities (%) | |||||||
| Myocardial infarction | 763 (2.1) | 144 (2.2) | 812 (2.2) | 227 (2.1) | 273 (2.0) | 0.6 | 2219 (2.1) |
| Congestive heart failure | 1443 (3.9) | 223 (3.3) | 1548 (4.2) | 467 (4.3) | 586 (4.4) | <0.001 | 4267 (4.1) |
| Peripheral vascular disease | 2054 (5.5) | 399 (6.0) | 2660 (7.3) | 783 (7.3) | 1071 (8.0) | <0.001 | 6967 (6.7) |
| Cerebrovascular disease | 2007 (5.4) | 379 (5.7) | 2358 (6.4) | 681 (6.3) | 947 (7.1) | <0.001 | 6372 (6.1) |
| Dementia | 370 (1.0) | 63 (0.9) | 442 (1.2) | 141 (1.3) | 198 (1.5) | <0.001 | 1214 (1.2) |
| Chronic obstructive pulmonary disease | 4251 (11) | 758 (11) | 4293 (12) | 1254 (12) | 1686 (13) | 0.004 | 12 242 (12) |
| Ulcer | 362 (1.0) | 60 (0.9) | 297 (0.8) | 84 (0.8) | 102 (0.8) | 0.07 | 905 (0.9) |
| Mild liver disease | 1182 (3.2) | 223 (3.3) | 1332 (3.6) | 413 (3.8) | 515 (3.9) | <0.001 | 3665 (3.5) |
| Diabetes mellitus | 6475 (17) | 1290 (19) | 7823 (21) | 2442 (23) | 2973 (22) | <0.001 | 21 003 (20) |
| Diabetes with organ damage | 1105 (3.0) | 225 (3.4) | 1600 (4.4) | 492 (4.6) | 734 (5.5) | <0.001 | 4156 (4.0) |
| Hemiplegia | 128 (0.3) | 25 (0.4) | 150 (0.4) | 51 (0.5) | 56 (0.4) | 0.3 | 410 (0.4) |
| Moderate/severe renal disease | 1133 (3.0) | 254 (3.8) | 1950 (5.3) | 674 (6.3) | 908 (6.8) | <0.001 | 4919 (4.7) |
| Moderate/severe liver disease | 27 (0.1) | 7 (0.1) | 31 (0.1) | 9 (0.1) | 12 (0.1) | 0.9 | 86 (0.1) |
| Metastatic solid tumor | 175 (0.5) | 37 (0.6) | 199 (0.5) | 58 (0.5) | 80 (0.6) | 0.4 | 549 (0.5) |
| AIDS | 82 (0.2) | 16 (0.2) | 101 (0.3) | 27 (0.3) | 46 (0.3) | 0.2 | 272 (0.3) |
| Rheumatologic disease | 449 (1.2) | 71 (1.1) | 522 (1.4) | 158 (1.5) | 196 (1.5) | 0.007 | 1396 (1.3) |
| Cancer | 1736 (4.7) | 337 (5.1) | 1796 (4.9) | 571 (5.3) | 701 (5.3) | 0.02 | 5141 (4.9) |
| Net worth range (%) | |||||||
| <$25 000 | 1831 (4.9) | 404 (6.1) | 2101 (5.7) | 561 (5.2) | 695 (5.2) | <0.001 | 5592 (5.3) |
| $25 000–149 000 | 5188 (14) | 1042 (16) | 5608 (15) | 1591 (15) | 2046 (15) | 15 475 (15) | |
| $150 000–249 000 | 4754 (13) | 837 (13) | 5182 (14) | 1499 (14) | 1862 (14) | 14 134 (14) | |
| $250 000–499 000 | 9694 (26) | 1779 (27) | 10 048 (28) | 3027 (28) | 3636 (27) | 28 184 (27) | |
| ≥$500 000 | 10 910 (29) | 2030 (30) | 10 860 (30) | 3208 (30) | 3852 (29) | 30 860 (30) | |
| Unknown/missing | 4840 (13) | 580 (8.7) | 2804 (7.7) | 886 (8.2) | 1229 (9.2) | 10 339 (9.9) | |
| 30-d complications (%) | |||||||
| Any | 5973 (16) | 1151 (17) | 6120 (17) | 1863 (17) | 2451 (18) | <0.001 | 17 558 (17) |
| Bleeding | 2547 (6.8) | 499 (7.5) | 2660 (7.3) | 860 (8.0) | 1083 (8.1) | <0.001 | 7649 (7.3) |
| Nonsepsis infection | 2978 (8.0) | 590 (8.8) | 2990 (8.2) | 875 (8.1) | 1236 (9.3) | <0.001 | 8669 (8.3) |
| Sepsis | 191 (0.5) | 50 (0.7) | 349 (1.0) | 105 (1.0) | 134 (1.0) | <0.001 | 829 (0.8) |
| Urinary retention | 980 (2.6) | 161 (2.4) | 1138 (3.1) | 341 (3.2) | 422 (3.2) | <0.001 | 3042 (2.9) |
| Hospitalization | 1019 (2.7) | 181 (2.7) | 1056 (2.9) | 320 (3.0) | 367 (2.8) | 0.6 | 2943 (2.8) |
AIDS = acquired immunodeficiency syndrome; AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; IQR = interquartile range; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer [screening trial]; PSA = prostate-specific antigen; USPTF = US Preventive Services Task Force.
Overall, 30-d complications occurred in 17% of biopsies. The most prevalent was nonsepsis genitourinary infections (8.3%), followed by bleeding complications (7.3%) and urinary retention (2.9%) (Table 1). The 30-d hospitalization rate was 2.8%.
We first examined unadjusted rates of prostate biopsy among men aged ≥40 yr (Fig. 1 and Table 2). From January 2005 to September 2014, biopsy rates fell 33% from 64.1 to 42.8 per 100 000 person-months. There was substantial seasonal variation in biopsy rates with reductions near the end of each calendar year. With the ITS approach, each intervention is potentially associated with an immediate change in event rates (ie, level shift or change in intercept) as well as a change in the rate of growth (ie, change in slope). At baseline, there was a statistically significant growth in prostate biopsy rates of 2.9 biopsies per 100 000 person-months per year (95% confidence interval [CI], 0.96–4.8; p < 0.001). Following the 2008 USPSTF recommendations, there was an immediate decrease (ie, level shift) in biopsy rates of −10.1 biopsies per 100 000 person-months (95% CI, −17.1 to −3.0; p < 0.001). Likewise, there were immediate decreases in biopsy rates following the 2012 USPSTF recommendations (−13.8 biopsies per 100 000 person-months; 95% CI, −21.0 to −6.7; p < 0.001) and 2013 AUA guidelines (−8.8 biopsies per 100 000 person-months; 95% CI, −16.7 to −0.92; p = 0.03). Analyses adjusted for case mix were similar, although the baseline growth rate was no longer significant, indicating this was related to population shifts over time (Table 2, model 2).
Fig. 1
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 2
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315)
| Prostate biopsy | Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Model 1* | ||||||||||||||
| Baseline rate | 57.8 (53.4–62.3) |
<0.001 | 8.6 (7.5–9.7) |
<0.001 | 4.0 (3.3–4.6) |
<0.001 | 0.17 (0.03–0.32) |
0.02 | 2.8 (2.3–3.3) |
<0.001 | 1.5 (1.3–1.6) |
<0.001 | 1.4 (1.1–1.7) |
<0.001 |
| Baseline trend in rate | 2.9 (0.96–4.8) |
<0.001 | 0.84 (0.48–1.3) |
<0.001 | 0.72 (0.48–1.1) |
<0.001 | 0.12 (0.12–0.12) |
<0.001 | 0.36 (0.12–0.48) |
<0.001 | 0.12 (0.00–0.12) |
<0.001 | 0.00 (−0.12 to 0.12) |
0.5 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.1 (−17.1 to −3.0) |
<0.001 | −0.61 (−3.1 to 1.9) |
0.6 | −1.4 (−2.8 to −0.03) |
0.053 | 0.04 (−0.35 to 0.42) |
0.8 | 0.56 (−0.70 to 1.8) |
0.4 | −0.06 (−0.69 to 0.57) |
0.9 | 0.16 (−0.58 to 0.90) |
0.7 |
| Trend change | 7.9 (−12.7 to 28.4) |
0.5 | −1.0 (−9.1 to 7.2) |
0.8 | −0.12 (−4.1 to 4.0) |
0.98 | −0.60 (−1.9 to 0.72) |
0.4 | −2.0 (−6.5 to 2.4) |
0.4 | −1.1 (−3.4 to 1.3) |
0.4 | −1.0 (−3.6 to 1.6) |
0.5 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.6 (−12 to 4.5) |
0.4 | −0.74 (−3.8 to 2.4) |
0.6 | −0.12 (−1.8 to 1.5) |
0.9 | 0.29 (−0.20 to 0.78) |
0.2 | −0.04 (−1.6 to 1.6) |
0.97 | 0.59 (−0.23 to 1.4) |
0.2 | 0.51 (−0.43 to 1.5) |
0.3 |
| Trend change | −11.2 (−32.0 to 9.6) |
0.3 | 0.0 (−8.2 to 8.0) |
0.99 | −0.96 (−5.0 to 3.1) |
0.7 | 0.36 (−0.96 to 1.8) |
0.5 | 1.7 (−2.8 to 6.1) |
0.5 | 1.0 (−1.4 to 3.4) |
0.4 | 0.84 (−1.7 to 3.5) |
0.5 |
| 2012 USPSTF | ||||||||||||||
| Level change | −13.8 (−21.0 to −6.7) |
<0.001 | −2.5 (−4.5 to −0.45) |
0.02 | −1.7 (−3.1 to −0.31) |
0.02 | −0.12 (−0.42 to 0.19) |
0.5 | −0.61 (−1.5 to 0.28) |
0.2 | −0.75 (−1.1 to −0.44) |
<0.001 | −0.25 (−0.77 to 0.27) |
0.4 |
| Trend change | 8.2 (−5.8 to 22.0) |
0.3 | 3.0 (−0.36 to 6.4) |
0.09 | 3.1 (0.72–5.5) |
0.01 | 0.12 (−0.36 to 0.60) |
0.7 | 0.84 (−0.72 to 2.3) |
0.3 | 0.48 (0.00 to 1.1) |
0.08 | 0.24 (−0.60 to 1.2) |
0.6 |
| 2013 AUA | ||||||||||||||
| Level change | −8.8 (−16.7 to −0.92) |
0.03 | −1.4 (−3.8 to 1.1) |
0.3 | −1.9 (−3.5 to −0.22) |
0.03 | −0.02 (−0.41 to 0.37) |
0.9 | −0.19 (−1.3 to 0.91) |
0.7 | −0.38 (−0.77 to 0.01) |
0.06 | −0.01 (−0.66 to 0.63) |
0.96 |
| Trend change | −10.4 (−30.0 to 9.2) |
0.3 | −4.1 (−7.9 to −0.24) |
0.04 | −2.9 (−6.0 to 0.12) |
0.07 | −0.24 (−0.72 to 0.36) |
0.5 | −1.8 (−3.5 to 0.0) |
0.046 | −0.60 (−1.2 to 0.12) |
0.09 | −0.60 (−1.7 to 0.36) |
0.2 |
| Model 2† | ||||||||||||||
| Baseline rate | 86.8 (82.1–91.4) |
<0.001 | 16.2 (14.9–17.6) |
<0.001 | 6.4 (4.9–7.9) |
<0.001 | 0.39 (−0.13 to 0.91) |
0.2 | 6.1 (4.6–7.7) |
<0.001 | 3.6 (2.8–4.4) |
<0.001 | 3.0 (2.2–3.8) |
<0.001 |
| Baseline trend in rate | 0.24 (−1.8 to 2.3) |
0.8 | −0.24 (−0.72 to 0.24) |
0.4 | 0.48 (−0.12 to 1.1) |
0.09 | 0.12 (−0.12 to 0.24) |
0.4 | −0.24 (−0.72 to 0.24) |
0.4 | −0.24 (−0.48 to 0.00) |
0.054 | −0.12 (−0.36 to 0.12) |
0.3 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.8 (−17.8 to −3.8) |
<0.001 | −5.4 (−9.6 to −1.2) |
0.01 | −2.5 (−5.6 to 0.71) |
0.1 | −0.61 (−1.9 to 0.71) |
0.4 | 0.14 (−3.8 to 4.1) |
0.9 | 1.1 (−0.98 to 3.1) |
0.3 | 0.14 (−1.9 to 2.1) |
0.9 |
| Trend change | 10.6 (−10.1 to 31.2) |
0.3 | 6.1 (−8.4 to 20.8) |
0.4 | −1.7 (−12.2 to 8.8) |
0.8 | 1.7 (−2.9 to 6.2) |
0.5 | 1.7 (−12.0 to 15.5) |
0.8 | −4.7 (−11.8 to 2.5) |
0.2 | −0.84 (−7.8 to 6.1) |
0.8 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.9 (−12 to 4.1) |
0.3 | −0.24 (−5.6 to 5.1) |
0.9 | 1.9 (−2.1 to 5.8) |
0.4 | −0.32 (−2.0 to 1.4) |
0.7 | −0.99 (−6.0 to 4.0) |
0.7 | 1.8 (−0.84 to 4.4) |
0.2 | 0.87 (−1.7 to 3.4) |
0.5 |
| Trend change | −16.2 (−37.1 to 4.7) |
0.1 | −6.8 (−21.4 to 7.8) |
0.4 | 0.36 (−10.2 to 10.9) |
0.95 | −1.9 (−6.5 to 2.8) |
0.4 | −1.6 (−15.2 to 12.2) |
0.8 | 4.7 (−2.4 to 11.9) |
0.2 | 0.48 (−6.4 to 7.4) |
0.9 |
| 2012 USPSTF | ||||||||||||||
| Level change | −11.0 (−18.1 to −4.0) |
0.001 | −2.8 (−6.0 to 0.38) |
0.09 | 0.53 (−1.9 to 3.0) |
0.7 | −0.02 (−0.95 to 0.91) |
0.97 | −1.8 (−4.6 to 0.96) |
0.2 | −1.5 (−3.1 to 0.17) |
0.08 | 0.16 (−1.3 to 1.6) |
0.8 |
| Trend change | 7.0 (−7.2 to 21.1) |
0.3 | 1.7 (−3.6 to 7.0) |
0.5 | 0.12 (−4.1 to 4.2) |
0.96 | 0.12 (−1.4 to 1.8) |
0.9 | −0.48 (−5.3 to 4.3) |
0.9 | 1.6 (−1.3 to 4.3) |
0.3 | 0.12 (−2.4 to 2.6) |
0.9 |
| 2013 AUA | ||||||||||||||
| Level change | −8.2 (−15.9 to −0.38) |
0.04 | −1.0 (−4.8 to 2.8) |
0.6 | −0.44 (−3.2 to 2.3) |
0.8 | −0.15 (−1.3 to 0.99) |
0.8 | 0.28 (−3.2 to 3.7) |
0.9 | −0.50 (−2.7 to 1.7) |
0.7 | 0.47 (−1.3 to 2.3) |
0.6 |
| Trend change | −4.2 (−24.5 to 16.1) |
0.7 | −0.7 (−7.6 to 6.2) |
0.9 | 1.1 (−4.2 to 6.2) |
0.7 | −0.12 (−1.8 to 1.7) |
0.9 | −0.60 (−5.9 to 4.8) |
0.8 | −1.8 (−4.9 to 1.2) |
0.3 | −0.48 (−3.1 to 2.2) |
0.7 |
* Model 1: Base model with constant, time trend, intervention indicator variables. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
† Model 2: Adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, and anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPTF = US Preventive Services Task Force.
Next, we examined 30-d postbiopsy complications among men aged ≥40 yr as a measure of the absolute morbidity of biopsy (Fig. 1, Supplementary Fig. 3, and Table 2). Across the study periods, overall complication rates decreased 10% from 8.7 to 7.8 per 100 000 person-months. In unadjusted ITS analyses, there was a baseline growth in the overall complication rate (0.84 per 100 000 person-months per year; 95% CI, 0.48–1.3; p < 0.001), driven by growth in the rate of nonsepsis infections (0.72 per 100 000 person-months per year; 95% CI, 0.48–1.1; p < 0.001). Only the 2012 USPSTF recommendations were associated with a significant change in complication rates, following which there was an immediate decrease in the overall complication rate of −2.5 per 100 000 person-months (95% CI, −4.5 to −0.45; p = 0.02), driven by reductions in nonsepsis infections (−1.7 per 100 000 person-months; 95% CI, −3.1 to −0.31; p = 0.02) and urinary retention (−0.75 per 100 000 person-months;95% CI, −1.1 to −0.44; p < 0.001). Following the 2013 AUA guidelines there was also a reduction in the overall complication baseline growth rate (−4.1 per 100 000 person-months per year; 95% CI, −7.9 to −0.24; p = 0.04). In analyses adjusted for case mix (Table 2, model 2), effects were overall attenuated, and only the 2008 USPSTF recommendations were associated with a reduction in overall complication rates (−5.4 per 100 000 person-months; 95% CI, −9.6 to −1.2; p = 0.01).
To examine whether there was a change in the relative morbidity of biopsy, we performed a secondary analysis examining complication rates among men undergoing biopsy (Fig. 2). The proportion of men undergoing biopsy who experienced a complication increased from 14% in January 2005 to 18% in September 2014, an increase of 36%. There was again a baseline growth in overall complication rates (0.72% per year; 95% CI, 0.36–1.1; p < 0.001), driven by an increase in nonsepsis infections and sepsis (Supplementary Table 2). None of the interventions affected this growth. Similar results were obtained after adjusting for case mix (Table 3).
Fig. 2
Interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 3
Adjusted* interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Baseline rate | 15.6 (14.3–16.8) |
<0.001 | 7.1 (6.5–7.7) |
<0.001 | 0.34 (0.11–0.57) |
<0.001 | 6.7 (5.9–7.5) |
<0.001 | 2.8 (2.5–3.2) |
<0.001 | 3.2 (2.8–3.6) |
<0.001 |
| Baseline trend in rate | 0.60 (0.12–1.1) |
0.02 | 0.72 (0.60–0.96) |
<0.001 | 0.12 (0.12–0.24) |
<0.001 | 0.12 (−0.12 to 0.36) |
0.3 | −0.12 (−0.24 to 0.00) |
0.3 | −0.12 (−0.24 to 0.00) |
0.0498 |
| 2008 USPSTF | ||||||||||||
| Level change | 1.1 (−2.0 to 4.1) |
0.5 | −1.3 (−3.2 to 0.58) |
0.2 | 0.13 (−0.45 to 0.72) |
0.7 | 1.9 (−0.11 to 3.9) |
0.07 | 0.46 (−0.50 to 1.4) |
0.4 | 0.27 (−0.69 to 1.2) |
0.6 |
| Trend change | −3.5 (−13.4 to 6.6) |
0.5 | −1.8 (−8.4 to 4.7) |
0.6 | −1.0 (−3.0 to 0.96) |
0.3 | −4.3 (−11.4 to 2.6) |
0.2 | −1.6 (−4.9 to 1.9) |
0.4 | −0.72 (−4.0 to 2.6) |
0.7 |
| ERSPC/PLCO | ||||||||||||
| Level change | −0.14 (−3.9 to 3.6) |
0.9 | 1.1 (−1.3 to 3.4) |
0.4 | 0.55 (−0.19 to 1.3) |
0.2 | 0.39 (−2.2 to 3.0) |
0.8 | 0.75 (−0.47 to 2.0) |
0.2 | 0.65 (−0.56 to 1.9) |
0.3 |
| Trend change | 2.9 (−7.1 to 13.0) |
0.6 | 0.72 (−5.8 to 7.2) |
0.8 | 0.84 (−1.2 to 2.9) |
0.4 | 4.2 (−2.8 to 11.3) |
0.2 | 1.7 (−1.8 to 5.0) |
0.4 | 0.72 (−2.6 to 4.0) |
0.7 |
| 2012 USPSTF | ||||||||||||
| Level change | −0.27 (−2.7 to 2.1) |
0.8 | −0.56 (−2.0 to 0.87) |
0.5 | 0.16 (−0.31 to 0.63) |
0.5 | 1.1 (−0.38 to 2.5) |
0.2 | −0.38 (−1.0 to 0.28) |
0.3 | 0.44 (−0.34 to 1.2) |
0.3 |
| Trend change | 1.3 (−2.6 to 5.3) |
0.5 | 2.9 (0.60–5.3) |
0.02 | −0.24 (−1.1 to 0.60) |
0.6 | −0.72 (−3.2 to 1.7) |
0.5 | 0.24 (−0.96 to 1.3) |
0.7 | −0.48 (−1.8 to 0.84) |
0.5 |
| 2013 AUA | ||||||||||||
| Level change | 0.25 (−2.6 to 3.1) |
0.9 | −0.64 (−2.3 to 1.1) |
0.5 | 0.22 (−0.40 to 0.83) |
0.5 | 1.1 (−0.68 to 2.8) |
0.2 | −0.29 (−1.1 to 0.48) |
0.5 | 0.47 (−0.55 to 1.5) |
0.4 |
| Trend change | −1.6 (−6.1 to 2.9) |
0.5 | −2.5 (−5.5 to 0.60) |
0.1 | 0.12 (−0.72 to 1.1) |
0.8 | −0.36 (−3.1 to 2.4) |
0.8 | 0.24 (−1.2 to 1.7) |
0.8 | 0.12 (−1.3 to 1.6) |
0.9 |
* Model adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 person-months. Baseline trend in rate and trend change per 100 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial]; USPTF = US Preventive Services Task Force.
Given the increase in the relative morbidity of biopsy, we performed logistic regression to identify predictors of each complication category (Supplementary Table 3 and Table 4). Due to the large sample size, we focused on those features associated with the greatest magnitude of effect. A prior diagnosis of cancer was associated with an increased risk of all complication categories including any complication (odds ratio [OR]: 1.30; 95% CI, 1.22–1.39; p < 0.001) and hospitalization (OR: 1.59; 95% CI, 1.39–1.82; p < 0.001). Similarly, prior fluoroquinolone use was associated with an increased risk of all complication categories including any complication (OR: 1.27; 95% CI, 1.22–1.32; p < 0.001) and sepsis (OR: 1.55; 95% CI, 1.33–1.81; p < 0.001). Age at biopsy ≥70 yr was associated with a nearly 4-fold increased risk of urinary retention. Although anticoagulant use was associated with increased risks of bleeding (OR: 1.28; 95% CI, 1.14–1.45; p < 0.001), sepsis (OR: 1.48; 95% CI, 1.06–2.07; p = 0.02), and hospitalization (OR: 1.39; 95% CI, 1.16–1.67; p < 0.001), the increase in the overall complication risk was modest (OR 1.14; 95% CI, 1.04–1.25; p = 0.004).
Table 4
Multivariable analysis* of 30-d postbiopsy complications among men aged ≥40 yr undergoing prostate biopsy (January 2005 to September 2014) (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Hospitalization | Urinary retention | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Age at biopsy, yr | ||||||||||||
| 40–49 | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| 50–59 | 1.01 (0.94–1.09) | 0.8 | 0.94 (0.85–1.04) | 0.2 | 1.16 (0.83–1.63) | 0.4 | 1.01 (0.90–1.14) | 0.8 | 1.12 (0.93–1.36) | 0.2 | 1.54 (1.20–1.97) | <0.001 |
| 60–69 | 1.05 (0.97–1.13) | 0.2 | 0.94 (0.85–1.03) | 0.2 | 0.98 (0.70–1.37) | 0.9 | 1.05 (0.94–1.18) | 0.4 | 1.16 (0.95–1.40) | 0.1 | 2.16 (1.68–2.77) | <0.001 |
| ≥70 | 1.25 (1.15–1.36) | <0.001 | 1.03 (0.93–1.15) | 0.6 | 1.05 (0.74–1.49) | 0.8 | 1.29 (1.15–1.46) | <0.001 | 1.31 (1.07–1.60) | 0.008 | 3.98 (3.08–5.15) | <0.001 |
| Race | ||||||||||||
| White | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| Asian | 1.27 (1.15–1.40) | <0.001 | 1.27 (1.11–1.44) | 0.0003 | 1.65 (1.19–2.29) | 0.003 | 1.36 (1.18–1.56) | <0.001 | 0.94 (0.76–1.17) | 0.6 | 1.15 (0.93–1.42) | 0.2 |
| African American | 1.09 (1.03–1.15) | 0.002 | 1.05 (0.97–1.13) | 0.2 | 1.34 (1.09–1.64) | 0.005 | 1.17 (1.09–1.27) | <0.001 | 1.13 (1.01–1.27) | 0.03 | 1.15 (1.03–1.28) | 0.02 |
| Hispanic | 1.32 (1.24–1.40) | <0.001 | 1.36 (1.26–1.47) | <0.001 | 1.52 (1.21–1.90) | <0.001 | 1.33 (1.22–1.45) | <0.001 | 1.12 (0.97–1.28) | 0.1 | 1.37 (1.21–1.56) | <0.001 |
| Unknown/missing | 0.99 (0.93–1.06) | 0.8 | 1.07 (0.99–1.17) | 0.1 | 1.23 (0.94–1.61) | 0.1 | 0.96 (0.88–1.05) | 0.4 | 0.94 (0.81–1.08) | 0.4 | 1.02 (0.88–1.18) | 0.8 |
| Diabetes mellitus | 1.03 (0.99–1.07) | 0.09 | 1.06 (1.00–1.11) | 0.045 | 1.08 (0.92–1.28) | 0.4 | 1.01 (0.96–1.07) | 0.7 | 1.20 (1.10–1.31) | <0.001 | 0.99 (0.91–1.08) | 0.8 |
| COPD | 1.14 (1.09–1.20) | <0.001 | 1.07 (1.00–1.14) | 0.04 | 1.23 (1.01–1.49) | 0.04 | 1.14 (1.06–1.22) | <0.001 | 1.36 (1.23–1.51) | <0.001 | 1.14 (1.03–1.25) | 0.01 |
| PVD | 1.24 (1.16–1.31) | <0.001 | 1.1 (1.02–1.2) | 0.02 | 1.19 (0.93–1.51) | 0.2 | 1.28 (1.18–1.40) | <0.001 | 1.31 (1.15–1.48) | <0.001 | 1.26 (1.12–1.42) | <0.001 |
| Cerebrovascular disease | 1.11 (1.04–1.19) | 0.001 | 1.12 (1.03–1.22) | 0.01 | 1.21 (0.93–1.57) | 0.2 | 1.03 (0.94–1.14) | 0.5 | 1.26 (1.10–1.44) | <0.001 | 1.30 (1.15–1.47) | <0.001 |
| Cancer | 1.30 (1.22–1.39) | <0.001 | 1.16 (1.06–1.27) | 0.001 | 1.37 (1.05–1.79) | 0.02 | 1.40 (1.27–1.53) | <0.001 | 1.59 (1.39–1.82) | <0.001 | 1.32 (1.16–1.52) | <0.001 |
| Anticoagulant use | 1.14 (1.04–1.25) | 0.004 | 1.01 (0.89–1.14) | 0.9 | 1.48 (1.06–2.07) | 0.02 | 1.28 (1.14–1.45) | <0.001 | 1.39 (1.16–1.67) | <0.001 | 1.11 (0.92–1.33) | 0.3 |
| Prior fluoroquinolones | 1.27 (1.22–1.32) | <0.001 | 1.37 (1.31–1.45) | <0.001 | 1.55 (1.33–1.81) | <0.001 | 1.21 (1.14–1.28) | <0.001 | 1.25 (1.14–1.36) | <0.001 | 1.27 (1.16–1.38) | <0.001 |
* Models adjusted for year of biopsy and the features listed in the table.
CI = confidence interval; COPD = chronic obstructive pulmonary disease; PVD = peripheral vascular disease.
We conducted sensitivity analyses where the postintervention run-in period was varied from 2 to 4 mo (Supplementary Table 4). Results overall were similar, although effects were attenuated for the 2013 AUA guidelines on biopsy rates and the 2012 USPSTF recommendations on overall complication rates, likely due to fewer time points for analysis. In a separate sensitivity analysis, we examined bleeding complications excluding microhematuria from 2009 to 2014, following introduction of a separate code for microhematuria. Bleeding rates paralleled those of the primary bleeding end point (Supplementary Fig. 4), and results of both ITS and logistic regression analyses were similar (Supplementary Table 5 and 6). As an exploratory analysis, we examined PCa diagnosis rates within 60 d following biopsy (Supplementary Fig. 5). These appeared to increase steadily across the study period.
To our knowledge, this is the largest study to examine the impact of landmark PSA screening publications on rates of prostate biopsy and the first to examine their impact on postbiopsy complications. Overall, biopsy rates have declined by 33%, which appears attributable to the publication of the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines. Interestingly, the magnitude of reduction in biopsy rates exceeded previously reported decreases in PSA screening rates [6], [7], [8], [9], [10], [11], [12], [22], and [23], which suggests that either the threshold for referral to a urologist or the threshold for performing a biopsy has changed. More importantly, there was a concurrent 22% decrease in 30-d complication rates that appears largely attributable to the 2012 USPSTF recommendations. Still, the reduction in complication rates was less than the reduction in biopsy rates due to an increase in the relative morbidity of biopsy, driven by growth in infectious complications. This may in part be related to a shift to older patients with greater comorbidity undergoing biopsy, although similar results were obtained in analyses adjusted for case mix, which suggests additional factors may exist.
To this end, we evaluated predictors of postbiopsy complications to identify actionable targets for further risk reduction. Some of these features represent modifiable risk factors (eg, anticoagulant use), and others represent opportunities for altering periprocedural management. For instance, prior fluoroquinolone use may identify patients who would benefit from prebiopsy rectal swabs to guide periprocedural antibiotics [24] and [25]. Nonmodifiable risk factors, including specific comorbidities such as cancer and peripheral vascular disease, may encourage reconsideration of the risk–benefit ratio before proceeding with biopsy.
It is also important to consider the concurrent impact of landmark PSA screening publications on rates of PCa diagnosis to contextualize the observed changes in biopsy and its morbidity. An exploratory analysis suggested that cancer diagnosis rates have steadily increased, although further studies are necessary to examine this issue comprehensively.
A number of studies have examined the impact of the 2008 USPSTF recommendations [7], [8], [9], and [22], ERSPC/PLCO trials [6] and [9], and 2012 USPSTF recommendations [6], [10], [11], [12], and [23] on PSA screening rates, generally noting reductions ranging from 0% to 8%. However, much less data are available on the impact of these publications on biopsy rates. One single-center study noted a 31% reduction in the number of biopsies performed following the 2012 USPSTF recommendations [13]; another noted no change in the proportion of referred men who underwent biopsy [26]. A report from a Canadian academic health network noted a reduction in biopsy rates from 58.0 to 35.5 biopsies per month following the 2012 USPSTF recommendations [27]. Our findings therefore confirm a reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines.
Limited data exist regarding secular trends in postbiopsy complications, and prior studies have focused exclusively on the relative morbidity among men undergoing biopsy rather than absolute complication rates. A Canadian population-based study observed that the 30-d hospitalization rate increased from 1.0% in 1996 to 4.1% in 2005, driven by an increase in infectious complications [28]. A Surveillance, Epidemiology and End Results–Medicare study also noted an increase in hospitalizations related to infectious complications over time [14]. Although we observed a similar increase in 30-d overall complications among men undergoing biopsy, likewise driven by infectious complications, there was an absolute 22% reduction in 30-d complications over the study period. Taken together, these findings support a reduction in the morbidity of biopsy following publication of revised PSA screening guidelines but highlight the continued need to reduce the relative morbidity of biopsy.
A number of studies have described complication rates following prostate biopsy [15], [16], [29], and [30]. Hematuria has been reported in 10–84% of biopsies, rectal bleeding in 1–45%, infectious complications in up to 6.3%, urinary retention in up to 1.7%, and hospitalization in up to 6.9% [14] and [29]. The wide range in complication rates is due to differences in definitions, ascertainment of end points, and variation in clinical practice. Given the heterogeneity in postbiopsy complication rates, our findings have two advantages. First, they represent a contemporary, geographically diverse population across the entire United States. Second, use of claims-based ascertainment limited complications to those cases in which patients sought medical care, and although this may underestimate the comprehensive morbidity of biopsy, it more accurately reflects the economic burden of serious complications.
This study has several limitations. Most importantly, a control population of men who were not exposed to the PSA screening publications was not available. This study is retrospective, and additional, unrecognized changes in policy or clinical practice may possibly have confounded the impact of the interventions examined. However, the ITS approach has been described as the strongest quasi-experimental design to evaluate policy changes when a randomized trial is not possible [19]. Because we relied on administrative claims for ascertainment of complications, minor complications that did not generate a claim were not captured, thus underestimating these end points. Although we adjusted for case mix, we were unable to adjust for PSA. However, population shifts in PSA may be confounded by changes in screening practices, and PSA adjustment may therefore under- or overestimate the impact of interventions. We were also unable to adjust for other factors that may be associated with biopsy or complication rates, such as prostate magnetic resonance imaging (MRI), ancillary biomarker testing, biopsy approach (eg, MRI-fusion biopsy, transperineal biopsy), indication for anticoagulation, use of periprocedural bridging, or operator biopsy volume/experience.
We observed an immediate reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines, with a smaller reduction in absolute 30-d complication rates. Still, the relative morbidity of biopsy has increased over time, driven by infectious complications. Accordingly, if health policy aims to reduce the morbidity of PSA screening related to biopsy, further efforts should focus on improving patient selection and periprocedural processes of care.
Author contributions: Boris Gershman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Gershman, van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Acquisition of data: Van Houten, Shah.
Analysis and interpretation of data: Gershman, Van Houten, Shah, Karnes.
Drafting of the manuscript: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Critical revision of the manuscript for important intellectual content: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Statistical analysis: Van Houten.
Obtaining funding: Karnes, Shah.
Administrative, technical, or material support: Karnes, Shah.
Supervision: Karnes, Shah.
Other (specify): None.
Financial disclosures: Boris Gershman certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: This publication was made possible by funding from the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery.
Funding/Support and role of the sponsor: None.
In 2009, two landmark randomized trials—the European Randomized Study of Screening for Prostate Cancer [ERSPC] [1] and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial [PLCO] [2]—reported smaller than anticipated reductions in prostate cancer (PCa) mortality and prompted reexamination of prostate-specific antigen (PSA) screening recommendations. The US Preventive Services Task Force (USPSTF), which in 2008 had recommended against PSA screening in men aged >75 yr [3], updated its position in 2012 to recommend against systematic PSA screening in the general population [4]. The American Urological Association (AUA) likewise updated its PSA screening guidelines in May 2013, recommending shared decision making for men aged 55–69 yr [5]. Both the 2012 USPSTF and 2013 AUA guidelines cited prostate biopsy and its potential complications as a significant harm associated with PSA screening.
Although several studies have noted a reduction in PSA screening rates following these publications [6], [7], [8], [9], [10], [11], and [12], there are little data on whether this has translated into the intended reductions in screening-related morbidity in the form of prostate biopsy and postbiopsy complications [13]. We evaluated whether publication of the 2008 USPSTF recommendations, 2009 ERSPC/PLCO trials, 2012 USPSTF recommendations, or 2013 AUA guidelines was associated with reductions in prostate biopsy or postbiopsy complications. Biopsy-related complications also remain understudied both with regard to incidence as well as identification of risk factors [14], [15], and [16]. Consequently, we also sought to characterize postbiopsy complications and evaluate predictors using a contemporary population-based cohort.
We conducted a retrospective analysis of administrative claims from the Optum Labs Data Warehouse that contains data on >100 million individuals enrolled in private health plans and Medicare Advantage Plans, from geographically diverse regions across the United States, as described in Supplementary Methods [17]. Because this study involved analysis of preexisting deidentified data, it was exempt from institutional review board approval.
We identified all prostate biopsies performed in men aged ≥40 yr between January 1, 2005, and September 30, 2014, using Current Procedural Terminology-4 code 55700 alone (11%) or in combination with 76872 (84%) or 76942 (0%), or International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 60.11 (5%). The first biopsy was selected as the index biopsy when more than one biopsy was performed. Patients with a prior diagnosis of PCa (ICD-9 code 185) were excluded. We restricted the cohort to patients with continuous medical and pharmacy enrollment for at least 12 mo prior to biopsy and 30 d after biopsy to ascertain comorbidities and 30-d complications (Supplementary Fig. 1).
We recorded sociodemographic characteristics, comorbid conditions, anticoagulant use within 30 d of biopsy, and fluoroquinolone use in the 12 mo prior to biopsy as described in Supplement 1. The Charlson Comorbidity Index was calculated as previously described [18]. Prebiopsy PSA values within 60 d of biopsy were available for a subset of patients.
The primary outcomes were crude and adjusted rates of prostate biopsy and 30-d complications. Complications were identified using inpatient and outpatient administrative claims as summarized in Supplementary Table 1.
Baseline demographic features were compared across study periods using the Kruskal-Wallis test for continuous variables and the chi-square test for discrete variables. To evaluate the impact of each publication on rates of prostate biopsy and 30-d complications, we used an interrupted time series (ITS) approach as described in Supplementary Methods [19] and [20]. Rates were estimated using all men aged ≥40 yr with at least 12 mo of continuous medical and pharmacy enrollment as the denominator. As secondary analyses, we estimated rates of 30-d complications among men who underwent biopsy. Two separate ITS analyses were performed: (1) unadjusted analyses, using crude rates of biopsy and complications aggregated for each month, and (2) analyses standardized for case mix, to adjust for population shifts over time as described in Supplementary Methods [20] and [21].
Logistic regression was used to evaluate predictors of 30-d complications. Multivariable models were adjusted for age at biopsy, year of biopsy, race, anticoagulant use, history of fluoroquinolone use, and the presence of the five most prevalent Charlson comorbidities.
Statistical analyses were performed using SAS v.9.3 (SAS Institute Inc., Cary, NC, USA). All tests were two sided with p < 0.05 considered statistically significant.
Between January 2005 and September 2014, a total of 104 584 men underwent prostate biopsy (Table 1). Median age at biopsy was 63.0 yr (interquartile range [IQR]: 57.0–70.0). Overall, 18% of men had used fluoroquinolones in the 12 mo preceding biopsy, and 2.8% were on anticoagulants. Median prebiopsy PSA was 5.1 ng/dl (IQR: 4.0–7.1) for the 18 749 men (18%) for whom this was available. Over the time periods, men undergoing biopsy were older, had increased prevalence of anticoagulant use and comorbid conditions, and had a higher PSA (Table 1 and Supplementary Fig. 2).
Table 1
Patient demographics for men who underwent prostate biopsy (January 2005 to September 2014) stratified by pre/postintervention time periods (n = 104 584)
| Baseline Jan 2005–Jul 2008 (n = 37 217) |
2008 USPSTF Aug 2008–Feb 2009 (n = 6672) |
ERSPC/PLCO Mar 2009–Apr 2012 (n = 36 603) |
2012 USPSTF May 2012–Apr 2013 (n = 10 772) |
2013 AUA May 2013–Sep 2014 (n = 13 320) |
p value | Total (n = 104 584) | |
|---|---|---|---|---|---|---|---|
| Age at biopsy, yr | |||||||
| Median (IQR) | 62.0 (56.0–68.0) | 62.0 (56.0–68.0) | 63.0 (57.0–70.0) | 65.0 (59.0–71.0) | 66.0 (59.0–71.0) | <0.001 | 63.0 (57.0–70.0) |
| Age group, yr (%) | |||||||
| 40–49 | 2269 (6.1) | 416 (6.2) | 1737 (4.7) | 420 (3.9) | 467 (3.5) | <0.001 | 5309 (5.1) |
| 50–59 | 12 366 (33) | 2201 (33) | 10 342 (28) | 2644 (25) | 3098 (23) | 30 651 (29) | |
| 60–69 | 14 671 (39) | 2697 (40) | 15 105 (41) | 4426 (41) | 5546 (42) | 42 445 (41) | |
| ≥70 | 7911 (21) | 1358 (20) | 9419 (26) | 3282 (31) | 4209 (32) | 26 179 (25) | |
| Race (%) | |||||||
| White | 25 508 (69) | 4643 (70) | 26 187 (72) | 7795 (72) | 9407 (71) | <0.001 | 73 540 (70) |
| African American | 3695 (9.9) | 780 (12) | 4423 (12) | 1283 (12) | 1589 (12) | 11 770 (11) | |
| Hispanic | 2549 (6.8) | 538 (8.1) | 2911 (8.0) | 837 (7.8) | 1186 (8.9) | 8021 (7.7) | |
| Asian | 1133 (3.0) | 236 (3.5) | 1142 (3.1) | 306 (2.8) | 387 (2.9) | 3204 (3.1) | |
| Unknown/Missing | 4332 (12) | 475 (7.1) | 1940 (5.3) | 551 (5.1) | 751 (5.6) | 8049 (7.7) | |
| Census region (%) | |||||||
| Northeast | 7094 (19) | 1196 (18) | 6120 (17) | 1886 (18) | 2424 (18) | <0.001 | 18 720 (18) |
| Midwest | 10 094 (27) | 1562 (23) | 8837 (24) | 2742 (26) | 3627 (27) | 26 862 (26) | |
| South | 15 330 (41) | 2889 (43) | 16 371 (45) | 4698 (44) | 5538 (42) | 44 826 (43) | |
| West | 4660 (13) | 993 (15) | 5147 (14) | 1442 (13) | 1728 (13) | 13 970 (13) | |
| Other/Missing | 39 (0.1) | 32 (0.5) | 128 (0.3) | 4 (0) | 3 (0) | 206 (0.2) | |
| PSA, ng/d | n = 5174 | n = 548 | n = 7345 | n = 2211 | n = 3471 | n = 18 749 | |
| Median (IQR) | 5.1 (3.9–7.1) | 4.9 (3.8–6.8) | 5.0 (4.0–6.8) | 5.3 (4.2–7.4) | 5.4 (4.2–7.8) | <0.001 | 5.1 (4.0–7.1) |
| Anticoagulant use (%) | 928 (2.5) | 184 (2.8) | 1103 (3.0) | 299 (2.8) | 415 (3.1) | <0.001 | 2929 (2.8) |
| History of fluoroquinolone use in prior 12 mo (%) | 6797 (18) | 1267 (19) | 6557 (18) | 2017 (19) | 2504 (19) | 0.054 | 19 142 (18) |
| Charlson Comorbidity Index (%) | |||||||
| 0 | 22 354 (60) | 3887 (58) | 20 269 (55) | 5758 (54) | 7039 (53) | <0.001 | 59 307 (57) |
| 1 | 8199 (22) | 1453 (22) | 8252 (23) | 2503 (23) | 2999 (23) | 23 406 (22) | |
| 2 | 3232 (8.7) | 664 (10) | 3520 (9.6) | 1108 (10) | 1359 (10) | 9883 (9.4) | |
| ≥3 | 3432 (9.2) | 668 (10) | 4562 (13) | 1403 (13) | 1923 (14) | 11 988 (12) | |
| Comorbidities (%) | |||||||
| Myocardial infarction | 763 (2.1) | 144 (2.2) | 812 (2.2) | 227 (2.1) | 273 (2.0) | 0.6 | 2219 (2.1) |
| Congestive heart failure | 1443 (3.9) | 223 (3.3) | 1548 (4.2) | 467 (4.3) | 586 (4.4) | <0.001 | 4267 (4.1) |
| Peripheral vascular disease | 2054 (5.5) | 399 (6.0) | 2660 (7.3) | 783 (7.3) | 1071 (8.0) | <0.001 | 6967 (6.7) |
| Cerebrovascular disease | 2007 (5.4) | 379 (5.7) | 2358 (6.4) | 681 (6.3) | 947 (7.1) | <0.001 | 6372 (6.1) |
| Dementia | 370 (1.0) | 63 (0.9) | 442 (1.2) | 141 (1.3) | 198 (1.5) | <0.001 | 1214 (1.2) |
| Chronic obstructive pulmonary disease | 4251 (11) | 758 (11) | 4293 (12) | 1254 (12) | 1686 (13) | 0.004 | 12 242 (12) |
| Ulcer | 362 (1.0) | 60 (0.9) | 297 (0.8) | 84 (0.8) | 102 (0.8) | 0.07 | 905 (0.9) |
| Mild liver disease | 1182 (3.2) | 223 (3.3) | 1332 (3.6) | 413 (3.8) | 515 (3.9) | <0.001 | 3665 (3.5) |
| Diabetes mellitus | 6475 (17) | 1290 (19) | 7823 (21) | 2442 (23) | 2973 (22) | <0.001 | 21 003 (20) |
| Diabetes with organ damage | 1105 (3.0) | 225 (3.4) | 1600 (4.4) | 492 (4.6) | 734 (5.5) | <0.001 | 4156 (4.0) |
| Hemiplegia | 128 (0.3) | 25 (0.4) | 150 (0.4) | 51 (0.5) | 56 (0.4) | 0.3 | 410 (0.4) |
| Moderate/severe renal disease | 1133 (3.0) | 254 (3.8) | 1950 (5.3) | 674 (6.3) | 908 (6.8) | <0.001 | 4919 (4.7) |
| Moderate/severe liver disease | 27 (0.1) | 7 (0.1) | 31 (0.1) | 9 (0.1) | 12 (0.1) | 0.9 | 86 (0.1) |
| Metastatic solid tumor | 175 (0.5) | 37 (0.6) | 199 (0.5) | 58 (0.5) | 80 (0.6) | 0.4 | 549 (0.5) |
| AIDS | 82 (0.2) | 16 (0.2) | 101 (0.3) | 27 (0.3) | 46 (0.3) | 0.2 | 272 (0.3) |
| Rheumatologic disease | 449 (1.2) | 71 (1.1) | 522 (1.4) | 158 (1.5) | 196 (1.5) | 0.007 | 1396 (1.3) |
| Cancer | 1736 (4.7) | 337 (5.1) | 1796 (4.9) | 571 (5.3) | 701 (5.3) | 0.02 | 5141 (4.9) |
| Net worth range (%) | |||||||
| <$25 000 | 1831 (4.9) | 404 (6.1) | 2101 (5.7) | 561 (5.2) | 695 (5.2) | <0.001 | 5592 (5.3) |
| $25 000–149 000 | 5188 (14) | 1042 (16) | 5608 (15) | 1591 (15) | 2046 (15) | 15 475 (15) | |
| $150 000–249 000 | 4754 (13) | 837 (13) | 5182 (14) | 1499 (14) | 1862 (14) | 14 134 (14) | |
| $250 000–499 000 | 9694 (26) | 1779 (27) | 10 048 (28) | 3027 (28) | 3636 (27) | 28 184 (27) | |
| ≥$500 000 | 10 910 (29) | 2030 (30) | 10 860 (30) | 3208 (30) | 3852 (29) | 30 860 (30) | |
| Unknown/missing | 4840 (13) | 580 (8.7) | 2804 (7.7) | 886 (8.2) | 1229 (9.2) | 10 339 (9.9) | |
| 30-d complications (%) | |||||||
| Any | 5973 (16) | 1151 (17) | 6120 (17) | 1863 (17) | 2451 (18) | <0.001 | 17 558 (17) |
| Bleeding | 2547 (6.8) | 499 (7.5) | 2660 (7.3) | 860 (8.0) | 1083 (8.1) | <0.001 | 7649 (7.3) |
| Nonsepsis infection | 2978 (8.0) | 590 (8.8) | 2990 (8.2) | 875 (8.1) | 1236 (9.3) | <0.001 | 8669 (8.3) |
| Sepsis | 191 (0.5) | 50 (0.7) | 349 (1.0) | 105 (1.0) | 134 (1.0) | <0.001 | 829 (0.8) |
| Urinary retention | 980 (2.6) | 161 (2.4) | 1138 (3.1) | 341 (3.2) | 422 (3.2) | <0.001 | 3042 (2.9) |
| Hospitalization | 1019 (2.7) | 181 (2.7) | 1056 (2.9) | 320 (3.0) | 367 (2.8) | 0.6 | 2943 (2.8) |
AIDS = acquired immunodeficiency syndrome; AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; IQR = interquartile range; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer [screening trial]; PSA = prostate-specific antigen; USPTF = US Preventive Services Task Force.
Overall, 30-d complications occurred in 17% of biopsies. The most prevalent was nonsepsis genitourinary infections (8.3%), followed by bleeding complications (7.3%) and urinary retention (2.9%) (Table 1). The 30-d hospitalization rate was 2.8%.
We first examined unadjusted rates of prostate biopsy among men aged ≥40 yr (Fig. 1 and Table 2). From January 2005 to September 2014, biopsy rates fell 33% from 64.1 to 42.8 per 100 000 person-months. There was substantial seasonal variation in biopsy rates with reductions near the end of each calendar year. With the ITS approach, each intervention is potentially associated with an immediate change in event rates (ie, level shift or change in intercept) as well as a change in the rate of growth (ie, change in slope). At baseline, there was a statistically significant growth in prostate biopsy rates of 2.9 biopsies per 100 000 person-months per year (95% confidence interval [CI], 0.96–4.8; p < 0.001). Following the 2008 USPSTF recommendations, there was an immediate decrease (ie, level shift) in biopsy rates of −10.1 biopsies per 100 000 person-months (95% CI, −17.1 to −3.0; p < 0.001). Likewise, there were immediate decreases in biopsy rates following the 2012 USPSTF recommendations (−13.8 biopsies per 100 000 person-months; 95% CI, −21.0 to −6.7; p < 0.001) and 2013 AUA guidelines (−8.8 biopsies per 100 000 person-months; 95% CI, −16.7 to −0.92; p = 0.03). Analyses adjusted for case mix were similar, although the baseline growth rate was no longer significant, indicating this was related to population shifts over time (Table 2, model 2).
Fig. 1
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 2
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315)
| Prostate biopsy | Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Model 1* | ||||||||||||||
| Baseline rate | 57.8 (53.4–62.3) |
<0.001 | 8.6 (7.5–9.7) |
<0.001 | 4.0 (3.3–4.6) |
<0.001 | 0.17 (0.03–0.32) |
0.02 | 2.8 (2.3–3.3) |
<0.001 | 1.5 (1.3–1.6) |
<0.001 | 1.4 (1.1–1.7) |
<0.001 |
| Baseline trend in rate | 2.9 (0.96–4.8) |
<0.001 | 0.84 (0.48–1.3) |
<0.001 | 0.72 (0.48–1.1) |
<0.001 | 0.12 (0.12–0.12) |
<0.001 | 0.36 (0.12–0.48) |
<0.001 | 0.12 (0.00–0.12) |
<0.001 | 0.00 (−0.12 to 0.12) |
0.5 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.1 (−17.1 to −3.0) |
<0.001 | −0.61 (−3.1 to 1.9) |
0.6 | −1.4 (−2.8 to −0.03) |
0.053 | 0.04 (−0.35 to 0.42) |
0.8 | 0.56 (−0.70 to 1.8) |
0.4 | −0.06 (−0.69 to 0.57) |
0.9 | 0.16 (−0.58 to 0.90) |
0.7 |
| Trend change | 7.9 (−12.7 to 28.4) |
0.5 | −1.0 (−9.1 to 7.2) |
0.8 | −0.12 (−4.1 to 4.0) |
0.98 | −0.60 (−1.9 to 0.72) |
0.4 | −2.0 (−6.5 to 2.4) |
0.4 | −1.1 (−3.4 to 1.3) |
0.4 | −1.0 (−3.6 to 1.6) |
0.5 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.6 (−12 to 4.5) |
0.4 | −0.74 (−3.8 to 2.4) |
0.6 | −0.12 (−1.8 to 1.5) |
0.9 | 0.29 (−0.20 to 0.78) |
0.2 | −0.04 (−1.6 to 1.6) |
0.97 | 0.59 (−0.23 to 1.4) |
0.2 | 0.51 (−0.43 to 1.5) |
0.3 |
| Trend change | −11.2 (−32.0 to 9.6) |
0.3 | 0.0 (−8.2 to 8.0) |
0.99 | −0.96 (−5.0 to 3.1) |
0.7 | 0.36 (−0.96 to 1.8) |
0.5 | 1.7 (−2.8 to 6.1) |
0.5 | 1.0 (−1.4 to 3.4) |
0.4 | 0.84 (−1.7 to 3.5) |
0.5 |
| 2012 USPSTF | ||||||||||||||
| Level change | −13.8 (−21.0 to −6.7) |
<0.001 | −2.5 (−4.5 to −0.45) |
0.02 | −1.7 (−3.1 to −0.31) |
0.02 | −0.12 (−0.42 to 0.19) |
0.5 | −0.61 (−1.5 to 0.28) |
0.2 | −0.75 (−1.1 to −0.44) |
<0.001 | −0.25 (−0.77 to 0.27) |
0.4 |
| Trend change | 8.2 (−5.8 to 22.0) |
0.3 | 3.0 (−0.36 to 6.4) |
0.09 | 3.1 (0.72–5.5) |
0.01 | 0.12 (−0.36 to 0.60) |
0.7 | 0.84 (−0.72 to 2.3) |
0.3 | 0.48 (0.00 to 1.1) |
0.08 | 0.24 (−0.60 to 1.2) |
0.6 |
| 2013 AUA | ||||||||||||||
| Level change | −8.8 (−16.7 to −0.92) |
0.03 | −1.4 (−3.8 to 1.1) |
0.3 | −1.9 (−3.5 to −0.22) |
0.03 | −0.02 (−0.41 to 0.37) |
0.9 | −0.19 (−1.3 to 0.91) |
0.7 | −0.38 (−0.77 to 0.01) |
0.06 | −0.01 (−0.66 to 0.63) |
0.96 |
| Trend change | −10.4 (−30.0 to 9.2) |
0.3 | −4.1 (−7.9 to −0.24) |
0.04 | −2.9 (−6.0 to 0.12) |
0.07 | −0.24 (−0.72 to 0.36) |
0.5 | −1.8 (−3.5 to 0.0) |
0.046 | −0.60 (−1.2 to 0.12) |
0.09 | −0.60 (−1.7 to 0.36) |
0.2 |
| Model 2† | ||||||||||||||
| Baseline rate | 86.8 (82.1–91.4) |
<0.001 | 16.2 (14.9–17.6) |
<0.001 | 6.4 (4.9–7.9) |
<0.001 | 0.39 (−0.13 to 0.91) |
0.2 | 6.1 (4.6–7.7) |
<0.001 | 3.6 (2.8–4.4) |
<0.001 | 3.0 (2.2–3.8) |
<0.001 |
| Baseline trend in rate | 0.24 (−1.8 to 2.3) |
0.8 | −0.24 (−0.72 to 0.24) |
0.4 | 0.48 (−0.12 to 1.1) |
0.09 | 0.12 (−0.12 to 0.24) |
0.4 | −0.24 (−0.72 to 0.24) |
0.4 | −0.24 (−0.48 to 0.00) |
0.054 | −0.12 (−0.36 to 0.12) |
0.3 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.8 (−17.8 to −3.8) |
<0.001 | −5.4 (−9.6 to −1.2) |
0.01 | −2.5 (−5.6 to 0.71) |
0.1 | −0.61 (−1.9 to 0.71) |
0.4 | 0.14 (−3.8 to 4.1) |
0.9 | 1.1 (−0.98 to 3.1) |
0.3 | 0.14 (−1.9 to 2.1) |
0.9 |
| Trend change | 10.6 (−10.1 to 31.2) |
0.3 | 6.1 (−8.4 to 20.8) |
0.4 | −1.7 (−12.2 to 8.8) |
0.8 | 1.7 (−2.9 to 6.2) |
0.5 | 1.7 (−12.0 to 15.5) |
0.8 | −4.7 (−11.8 to 2.5) |
0.2 | −0.84 (−7.8 to 6.1) |
0.8 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.9 (−12 to 4.1) |
0.3 | −0.24 (−5.6 to 5.1) |
0.9 | 1.9 (−2.1 to 5.8) |
0.4 | −0.32 (−2.0 to 1.4) |
0.7 | −0.99 (−6.0 to 4.0) |
0.7 | 1.8 (−0.84 to 4.4) |
0.2 | 0.87 (−1.7 to 3.4) |
0.5 |
| Trend change | −16.2 (−37.1 to 4.7) |
0.1 | −6.8 (−21.4 to 7.8) |
0.4 | 0.36 (−10.2 to 10.9) |
0.95 | −1.9 (−6.5 to 2.8) |
0.4 | −1.6 (−15.2 to 12.2) |
0.8 | 4.7 (−2.4 to 11.9) |
0.2 | 0.48 (−6.4 to 7.4) |
0.9 |
| 2012 USPSTF | ||||||||||||||
| Level change | −11.0 (−18.1 to −4.0) |
0.001 | −2.8 (−6.0 to 0.38) |
0.09 | 0.53 (−1.9 to 3.0) |
0.7 | −0.02 (−0.95 to 0.91) |
0.97 | −1.8 (−4.6 to 0.96) |
0.2 | −1.5 (−3.1 to 0.17) |
0.08 | 0.16 (−1.3 to 1.6) |
0.8 |
| Trend change | 7.0 (−7.2 to 21.1) |
0.3 | 1.7 (−3.6 to 7.0) |
0.5 | 0.12 (−4.1 to 4.2) |
0.96 | 0.12 (−1.4 to 1.8) |
0.9 | −0.48 (−5.3 to 4.3) |
0.9 | 1.6 (−1.3 to 4.3) |
0.3 | 0.12 (−2.4 to 2.6) |
0.9 |
| 2013 AUA | ||||||||||||||
| Level change | −8.2 (−15.9 to −0.38) |
0.04 | −1.0 (−4.8 to 2.8) |
0.6 | −0.44 (−3.2 to 2.3) |
0.8 | −0.15 (−1.3 to 0.99) |
0.8 | 0.28 (−3.2 to 3.7) |
0.9 | −0.50 (−2.7 to 1.7) |
0.7 | 0.47 (−1.3 to 2.3) |
0.6 |
| Trend change | −4.2 (−24.5 to 16.1) |
0.7 | −0.7 (−7.6 to 6.2) |
0.9 | 1.1 (−4.2 to 6.2) |
0.7 | −0.12 (−1.8 to 1.7) |
0.9 | −0.60 (−5.9 to 4.8) |
0.8 | −1.8 (−4.9 to 1.2) |
0.3 | −0.48 (−3.1 to 2.2) |
0.7 |
* Model 1: Base model with constant, time trend, intervention indicator variables. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
† Model 2: Adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, and anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPTF = US Preventive Services Task Force.
Next, we examined 30-d postbiopsy complications among men aged ≥40 yr as a measure of the absolute morbidity of biopsy (Fig. 1, Supplementary Fig. 3, and Table 2). Across the study periods, overall complication rates decreased 10% from 8.7 to 7.8 per 100 000 person-months. In unadjusted ITS analyses, there was a baseline growth in the overall complication rate (0.84 per 100 000 person-months per year; 95% CI, 0.48–1.3; p < 0.001), driven by growth in the rate of nonsepsis infections (0.72 per 100 000 person-months per year; 95% CI, 0.48–1.1; p < 0.001). Only the 2012 USPSTF recommendations were associated with a significant change in complication rates, following which there was an immediate decrease in the overall complication rate of −2.5 per 100 000 person-months (95% CI, −4.5 to −0.45; p = 0.02), driven by reductions in nonsepsis infections (−1.7 per 100 000 person-months; 95% CI, −3.1 to −0.31; p = 0.02) and urinary retention (−0.75 per 100 000 person-months;95% CI, −1.1 to −0.44; p < 0.001). Following the 2013 AUA guidelines there was also a reduction in the overall complication baseline growth rate (−4.1 per 100 000 person-months per year; 95% CI, −7.9 to −0.24; p = 0.04). In analyses adjusted for case mix (Table 2, model 2), effects were overall attenuated, and only the 2008 USPSTF recommendations were associated with a reduction in overall complication rates (−5.4 per 100 000 person-months; 95% CI, −9.6 to −1.2; p = 0.01).
To examine whether there was a change in the relative morbidity of biopsy, we performed a secondary analysis examining complication rates among men undergoing biopsy (Fig. 2). The proportion of men undergoing biopsy who experienced a complication increased from 14% in January 2005 to 18% in September 2014, an increase of 36%. There was again a baseline growth in overall complication rates (0.72% per year; 95% CI, 0.36–1.1; p < 0.001), driven by an increase in nonsepsis infections and sepsis (Supplementary Table 2). None of the interventions affected this growth. Similar results were obtained after adjusting for case mix (Table 3).
Fig. 2
Interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 3
Adjusted* interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Baseline rate | 15.6 (14.3–16.8) |
<0.001 | 7.1 (6.5–7.7) |
<0.001 | 0.34 (0.11–0.57) |
<0.001 | 6.7 (5.9–7.5) |
<0.001 | 2.8 (2.5–3.2) |
<0.001 | 3.2 (2.8–3.6) |
<0.001 |
| Baseline trend in rate | 0.60 (0.12–1.1) |
0.02 | 0.72 (0.60–0.96) |
<0.001 | 0.12 (0.12–0.24) |
<0.001 | 0.12 (−0.12 to 0.36) |
0.3 | −0.12 (−0.24 to 0.00) |
0.3 | −0.12 (−0.24 to 0.00) |
0.0498 |
| 2008 USPSTF | ||||||||||||
| Level change | 1.1 (−2.0 to 4.1) |
0.5 | −1.3 (−3.2 to 0.58) |
0.2 | 0.13 (−0.45 to 0.72) |
0.7 | 1.9 (−0.11 to 3.9) |
0.07 | 0.46 (−0.50 to 1.4) |
0.4 | 0.27 (−0.69 to 1.2) |
0.6 |
| Trend change | −3.5 (−13.4 to 6.6) |
0.5 | −1.8 (−8.4 to 4.7) |
0.6 | −1.0 (−3.0 to 0.96) |
0.3 | −4.3 (−11.4 to 2.6) |
0.2 | −1.6 (−4.9 to 1.9) |
0.4 | −0.72 (−4.0 to 2.6) |
0.7 |
| ERSPC/PLCO | ||||||||||||
| Level change | −0.14 (−3.9 to 3.6) |
0.9 | 1.1 (−1.3 to 3.4) |
0.4 | 0.55 (−0.19 to 1.3) |
0.2 | 0.39 (−2.2 to 3.0) |
0.8 | 0.75 (−0.47 to 2.0) |
0.2 | 0.65 (−0.56 to 1.9) |
0.3 |
| Trend change | 2.9 (−7.1 to 13.0) |
0.6 | 0.72 (−5.8 to 7.2) |
0.8 | 0.84 (−1.2 to 2.9) |
0.4 | 4.2 (−2.8 to 11.3) |
0.2 | 1.7 (−1.8 to 5.0) |
0.4 | 0.72 (−2.6 to 4.0) |
0.7 |
| 2012 USPSTF | ||||||||||||
| Level change | −0.27 (−2.7 to 2.1) |
0.8 | −0.56 (−2.0 to 0.87) |
0.5 | 0.16 (−0.31 to 0.63) |
0.5 | 1.1 (−0.38 to 2.5) |
0.2 | −0.38 (−1.0 to 0.28) |
0.3 | 0.44 (−0.34 to 1.2) |
0.3 |
| Trend change | 1.3 (−2.6 to 5.3) |
0.5 | 2.9 (0.60–5.3) |
0.02 | −0.24 (−1.1 to 0.60) |
0.6 | −0.72 (−3.2 to 1.7) |
0.5 | 0.24 (−0.96 to 1.3) |
0.7 | −0.48 (−1.8 to 0.84) |
0.5 |
| 2013 AUA | ||||||||||||
| Level change | 0.25 (−2.6 to 3.1) |
0.9 | −0.64 (−2.3 to 1.1) |
0.5 | 0.22 (−0.40 to 0.83) |
0.5 | 1.1 (−0.68 to 2.8) |
0.2 | −0.29 (−1.1 to 0.48) |
0.5 | 0.47 (−0.55 to 1.5) |
0.4 |
| Trend change | −1.6 (−6.1 to 2.9) |
0.5 | −2.5 (−5.5 to 0.60) |
0.1 | 0.12 (−0.72 to 1.1) |
0.8 | −0.36 (−3.1 to 2.4) |
0.8 | 0.24 (−1.2 to 1.7) |
0.8 | 0.12 (−1.3 to 1.6) |
0.9 |
* Model adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 person-months. Baseline trend in rate and trend change per 100 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial]; USPTF = US Preventive Services Task Force.
Given the increase in the relative morbidity of biopsy, we performed logistic regression to identify predictors of each complication category (Supplementary Table 3 and Table 4). Due to the large sample size, we focused on those features associated with the greatest magnitude of effect. A prior diagnosis of cancer was associated with an increased risk of all complication categories including any complication (odds ratio [OR]: 1.30; 95% CI, 1.22–1.39; p < 0.001) and hospitalization (OR: 1.59; 95% CI, 1.39–1.82; p < 0.001). Similarly, prior fluoroquinolone use was associated with an increased risk of all complication categories including any complication (OR: 1.27; 95% CI, 1.22–1.32; p < 0.001) and sepsis (OR: 1.55; 95% CI, 1.33–1.81; p < 0.001). Age at biopsy ≥70 yr was associated with a nearly 4-fold increased risk of urinary retention. Although anticoagulant use was associated with increased risks of bleeding (OR: 1.28; 95% CI, 1.14–1.45; p < 0.001), sepsis (OR: 1.48; 95% CI, 1.06–2.07; p = 0.02), and hospitalization (OR: 1.39; 95% CI, 1.16–1.67; p < 0.001), the increase in the overall complication risk was modest (OR 1.14; 95% CI, 1.04–1.25; p = 0.004).
Table 4
Multivariable analysis* of 30-d postbiopsy complications among men aged ≥40 yr undergoing prostate biopsy (January 2005 to September 2014) (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Hospitalization | Urinary retention | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Age at biopsy, yr | ||||||||||||
| 40–49 | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| 50–59 | 1.01 (0.94–1.09) | 0.8 | 0.94 (0.85–1.04) | 0.2 | 1.16 (0.83–1.63) | 0.4 | 1.01 (0.90–1.14) | 0.8 | 1.12 (0.93–1.36) | 0.2 | 1.54 (1.20–1.97) | <0.001 |
| 60–69 | 1.05 (0.97–1.13) | 0.2 | 0.94 (0.85–1.03) | 0.2 | 0.98 (0.70–1.37) | 0.9 | 1.05 (0.94–1.18) | 0.4 | 1.16 (0.95–1.40) | 0.1 | 2.16 (1.68–2.77) | <0.001 |
| ≥70 | 1.25 (1.15–1.36) | <0.001 | 1.03 (0.93–1.15) | 0.6 | 1.05 (0.74–1.49) | 0.8 | 1.29 (1.15–1.46) | <0.001 | 1.31 (1.07–1.60) | 0.008 | 3.98 (3.08–5.15) | <0.001 |
| Race | ||||||||||||
| White | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| Asian | 1.27 (1.15–1.40) | <0.001 | 1.27 (1.11–1.44) | 0.0003 | 1.65 (1.19–2.29) | 0.003 | 1.36 (1.18–1.56) | <0.001 | 0.94 (0.76–1.17) | 0.6 | 1.15 (0.93–1.42) | 0.2 |
| African American | 1.09 (1.03–1.15) | 0.002 | 1.05 (0.97–1.13) | 0.2 | 1.34 (1.09–1.64) | 0.005 | 1.17 (1.09–1.27) | <0.001 | 1.13 (1.01–1.27) | 0.03 | 1.15 (1.03–1.28) | 0.02 |
| Hispanic | 1.32 (1.24–1.40) | <0.001 | 1.36 (1.26–1.47) | <0.001 | 1.52 (1.21–1.90) | <0.001 | 1.33 (1.22–1.45) | <0.001 | 1.12 (0.97–1.28) | 0.1 | 1.37 (1.21–1.56) | <0.001 |
| Unknown/missing | 0.99 (0.93–1.06) | 0.8 | 1.07 (0.99–1.17) | 0.1 | 1.23 (0.94–1.61) | 0.1 | 0.96 (0.88–1.05) | 0.4 | 0.94 (0.81–1.08) | 0.4 | 1.02 (0.88–1.18) | 0.8 |
| Diabetes mellitus | 1.03 (0.99–1.07) | 0.09 | 1.06 (1.00–1.11) | 0.045 | 1.08 (0.92–1.28) | 0.4 | 1.01 (0.96–1.07) | 0.7 | 1.20 (1.10–1.31) | <0.001 | 0.99 (0.91–1.08) | 0.8 |
| COPD | 1.14 (1.09–1.20) | <0.001 | 1.07 (1.00–1.14) | 0.04 | 1.23 (1.01–1.49) | 0.04 | 1.14 (1.06–1.22) | <0.001 | 1.36 (1.23–1.51) | <0.001 | 1.14 (1.03–1.25) | 0.01 |
| PVD | 1.24 (1.16–1.31) | <0.001 | 1.1 (1.02–1.2) | 0.02 | 1.19 (0.93–1.51) | 0.2 | 1.28 (1.18–1.40) | <0.001 | 1.31 (1.15–1.48) | <0.001 | 1.26 (1.12–1.42) | <0.001 |
| Cerebrovascular disease | 1.11 (1.04–1.19) | 0.001 | 1.12 (1.03–1.22) | 0.01 | 1.21 (0.93–1.57) | 0.2 | 1.03 (0.94–1.14) | 0.5 | 1.26 (1.10–1.44) | <0.001 | 1.30 (1.15–1.47) | <0.001 |
| Cancer | 1.30 (1.22–1.39) | <0.001 | 1.16 (1.06–1.27) | 0.001 | 1.37 (1.05–1.79) | 0.02 | 1.40 (1.27–1.53) | <0.001 | 1.59 (1.39–1.82) | <0.001 | 1.32 (1.16–1.52) | <0.001 |
| Anticoagulant use | 1.14 (1.04–1.25) | 0.004 | 1.01 (0.89–1.14) | 0.9 | 1.48 (1.06–2.07) | 0.02 | 1.28 (1.14–1.45) | <0.001 | 1.39 (1.16–1.67) | <0.001 | 1.11 (0.92–1.33) | 0.3 |
| Prior fluoroquinolones | 1.27 (1.22–1.32) | <0.001 | 1.37 (1.31–1.45) | <0.001 | 1.55 (1.33–1.81) | <0.001 | 1.21 (1.14–1.28) | <0.001 | 1.25 (1.14–1.36) | <0.001 | 1.27 (1.16–1.38) | <0.001 |
* Models adjusted for year of biopsy and the features listed in the table.
CI = confidence interval; COPD = chronic obstructive pulmonary disease; PVD = peripheral vascular disease.
We conducted sensitivity analyses where the postintervention run-in period was varied from 2 to 4 mo (Supplementary Table 4). Results overall were similar, although effects were attenuated for the 2013 AUA guidelines on biopsy rates and the 2012 USPSTF recommendations on overall complication rates, likely due to fewer time points for analysis. In a separate sensitivity analysis, we examined bleeding complications excluding microhematuria from 2009 to 2014, following introduction of a separate code for microhematuria. Bleeding rates paralleled those of the primary bleeding end point (Supplementary Fig. 4), and results of both ITS and logistic regression analyses were similar (Supplementary Table 5 and 6). As an exploratory analysis, we examined PCa diagnosis rates within 60 d following biopsy (Supplementary Fig. 5). These appeared to increase steadily across the study period.
To our knowledge, this is the largest study to examine the impact of landmark PSA screening publications on rates of prostate biopsy and the first to examine their impact on postbiopsy complications. Overall, biopsy rates have declined by 33%, which appears attributable to the publication of the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines. Interestingly, the magnitude of reduction in biopsy rates exceeded previously reported decreases in PSA screening rates [6], [7], [8], [9], [10], [11], [12], [22], and [23], which suggests that either the threshold for referral to a urologist or the threshold for performing a biopsy has changed. More importantly, there was a concurrent 22% decrease in 30-d complication rates that appears largely attributable to the 2012 USPSTF recommendations. Still, the reduction in complication rates was less than the reduction in biopsy rates due to an increase in the relative morbidity of biopsy, driven by growth in infectious complications. This may in part be related to a shift to older patients with greater comorbidity undergoing biopsy, although similar results were obtained in analyses adjusted for case mix, which suggests additional factors may exist.
To this end, we evaluated predictors of postbiopsy complications to identify actionable targets for further risk reduction. Some of these features represent modifiable risk factors (eg, anticoagulant use), and others represent opportunities for altering periprocedural management. For instance, prior fluoroquinolone use may identify patients who would benefit from prebiopsy rectal swabs to guide periprocedural antibiotics [24] and [25]. Nonmodifiable risk factors, including specific comorbidities such as cancer and peripheral vascular disease, may encourage reconsideration of the risk–benefit ratio before proceeding with biopsy.
It is also important to consider the concurrent impact of landmark PSA screening publications on rates of PCa diagnosis to contextualize the observed changes in biopsy and its morbidity. An exploratory analysis suggested that cancer diagnosis rates have steadily increased, although further studies are necessary to examine this issue comprehensively.
A number of studies have examined the impact of the 2008 USPSTF recommendations [7], [8], [9], and [22], ERSPC/PLCO trials [6] and [9], and 2012 USPSTF recommendations [6], [10], [11], [12], and [23] on PSA screening rates, generally noting reductions ranging from 0% to 8%. However, much less data are available on the impact of these publications on biopsy rates. One single-center study noted a 31% reduction in the number of biopsies performed following the 2012 USPSTF recommendations [13]; another noted no change in the proportion of referred men who underwent biopsy [26]. A report from a Canadian academic health network noted a reduction in biopsy rates from 58.0 to 35.5 biopsies per month following the 2012 USPSTF recommendations [27]. Our findings therefore confirm a reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines.
Limited data exist regarding secular trends in postbiopsy complications, and prior studies have focused exclusively on the relative morbidity among men undergoing biopsy rather than absolute complication rates. A Canadian population-based study observed that the 30-d hospitalization rate increased from 1.0% in 1996 to 4.1% in 2005, driven by an increase in infectious complications [28]. A Surveillance, Epidemiology and End Results–Medicare study also noted an increase in hospitalizations related to infectious complications over time [14]. Although we observed a similar increase in 30-d overall complications among men undergoing biopsy, likewise driven by infectious complications, there was an absolute 22% reduction in 30-d complications over the study period. Taken together, these findings support a reduction in the morbidity of biopsy following publication of revised PSA screening guidelines but highlight the continued need to reduce the relative morbidity of biopsy.
A number of studies have described complication rates following prostate biopsy [15], [16], [29], and [30]. Hematuria has been reported in 10–84% of biopsies, rectal bleeding in 1–45%, infectious complications in up to 6.3%, urinary retention in up to 1.7%, and hospitalization in up to 6.9% [14] and [29]. The wide range in complication rates is due to differences in definitions, ascertainment of end points, and variation in clinical practice. Given the heterogeneity in postbiopsy complication rates, our findings have two advantages. First, they represent a contemporary, geographically diverse population across the entire United States. Second, use of claims-based ascertainment limited complications to those cases in which patients sought medical care, and although this may underestimate the comprehensive morbidity of biopsy, it more accurately reflects the economic burden of serious complications.
This study has several limitations. Most importantly, a control population of men who were not exposed to the PSA screening publications was not available. This study is retrospective, and additional, unrecognized changes in policy or clinical practice may possibly have confounded the impact of the interventions examined. However, the ITS approach has been described as the strongest quasi-experimental design to evaluate policy changes when a randomized trial is not possible [19]. Because we relied on administrative claims for ascertainment of complications, minor complications that did not generate a claim were not captured, thus underestimating these end points. Although we adjusted for case mix, we were unable to adjust for PSA. However, population shifts in PSA may be confounded by changes in screening practices, and PSA adjustment may therefore under- or overestimate the impact of interventions. We were also unable to adjust for other factors that may be associated with biopsy or complication rates, such as prostate magnetic resonance imaging (MRI), ancillary biomarker testing, biopsy approach (eg, MRI-fusion biopsy, transperineal biopsy), indication for anticoagulation, use of periprocedural bridging, or operator biopsy volume/experience.
We observed an immediate reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines, with a smaller reduction in absolute 30-d complication rates. Still, the relative morbidity of biopsy has increased over time, driven by infectious complications. Accordingly, if health policy aims to reduce the morbidity of PSA screening related to biopsy, further efforts should focus on improving patient selection and periprocedural processes of care.
Author contributions: Boris Gershman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Gershman, van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Acquisition of data: Van Houten, Shah.
Analysis and interpretation of data: Gershman, Van Houten, Shah, Karnes.
Drafting of the manuscript: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Critical revision of the manuscript for important intellectual content: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Statistical analysis: Van Houten.
Obtaining funding: Karnes, Shah.
Administrative, technical, or material support: Karnes, Shah.
Supervision: Karnes, Shah.
Other (specify): None.
Financial disclosures: Boris Gershman certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: This publication was made possible by funding from the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery.
Funding/Support and role of the sponsor: None.
In 2009, two landmark randomized trials—the European Randomized Study of Screening for Prostate Cancer [ERSPC] [1] and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial [PLCO] [2]—reported smaller than anticipated reductions in prostate cancer (PCa) mortality and prompted reexamination of prostate-specific antigen (PSA) screening recommendations. The US Preventive Services Task Force (USPSTF), which in 2008 had recommended against PSA screening in men aged >75 yr [3], updated its position in 2012 to recommend against systematic PSA screening in the general population [4]. The American Urological Association (AUA) likewise updated its PSA screening guidelines in May 2013, recommending shared decision making for men aged 55–69 yr [5]. Both the 2012 USPSTF and 2013 AUA guidelines cited prostate biopsy and its potential complications as a significant harm associated with PSA screening.
Although several studies have noted a reduction in PSA screening rates following these publications [6], [7], [8], [9], [10], [11], and [12], there are little data on whether this has translated into the intended reductions in screening-related morbidity in the form of prostate biopsy and postbiopsy complications [13]. We evaluated whether publication of the 2008 USPSTF recommendations, 2009 ERSPC/PLCO trials, 2012 USPSTF recommendations, or 2013 AUA guidelines was associated with reductions in prostate biopsy or postbiopsy complications. Biopsy-related complications also remain understudied both with regard to incidence as well as identification of risk factors [14], [15], and [16]. Consequently, we also sought to characterize postbiopsy complications and evaluate predictors using a contemporary population-based cohort.
We conducted a retrospective analysis of administrative claims from the Optum Labs Data Warehouse that contains data on >100 million individuals enrolled in private health plans and Medicare Advantage Plans, from geographically diverse regions across the United States, as described in Supplementary Methods [17]. Because this study involved analysis of preexisting deidentified data, it was exempt from institutional review board approval.
We identified all prostate biopsies performed in men aged ≥40 yr between January 1, 2005, and September 30, 2014, using Current Procedural Terminology-4 code 55700 alone (11%) or in combination with 76872 (84%) or 76942 (0%), or International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 60.11 (5%). The first biopsy was selected as the index biopsy when more than one biopsy was performed. Patients with a prior diagnosis of PCa (ICD-9 code 185) were excluded. We restricted the cohort to patients with continuous medical and pharmacy enrollment for at least 12 mo prior to biopsy and 30 d after biopsy to ascertain comorbidities and 30-d complications (Supplementary Fig. 1).
We recorded sociodemographic characteristics, comorbid conditions, anticoagulant use within 30 d of biopsy, and fluoroquinolone use in the 12 mo prior to biopsy as described in Supplement 1. The Charlson Comorbidity Index was calculated as previously described [18]. Prebiopsy PSA values within 60 d of biopsy were available for a subset of patients.
The primary outcomes were crude and adjusted rates of prostate biopsy and 30-d complications. Complications were identified using inpatient and outpatient administrative claims as summarized in Supplementary Table 1.
Baseline demographic features were compared across study periods using the Kruskal-Wallis test for continuous variables and the chi-square test for discrete variables. To evaluate the impact of each publication on rates of prostate biopsy and 30-d complications, we used an interrupted time series (ITS) approach as described in Supplementary Methods [19] and [20]. Rates were estimated using all men aged ≥40 yr with at least 12 mo of continuous medical and pharmacy enrollment as the denominator. As secondary analyses, we estimated rates of 30-d complications among men who underwent biopsy. Two separate ITS analyses were performed: (1) unadjusted analyses, using crude rates of biopsy and complications aggregated for each month, and (2) analyses standardized for case mix, to adjust for population shifts over time as described in Supplementary Methods [20] and [21].
Logistic regression was used to evaluate predictors of 30-d complications. Multivariable models were adjusted for age at biopsy, year of biopsy, race, anticoagulant use, history of fluoroquinolone use, and the presence of the five most prevalent Charlson comorbidities.
Statistical analyses were performed using SAS v.9.3 (SAS Institute Inc., Cary, NC, USA). All tests were two sided with p < 0.05 considered statistically significant.
Between January 2005 and September 2014, a total of 104 584 men underwent prostate biopsy (Table 1). Median age at biopsy was 63.0 yr (interquartile range [IQR]: 57.0–70.0). Overall, 18% of men had used fluoroquinolones in the 12 mo preceding biopsy, and 2.8% were on anticoagulants. Median prebiopsy PSA was 5.1 ng/dl (IQR: 4.0–7.1) for the 18 749 men (18%) for whom this was available. Over the time periods, men undergoing biopsy were older, had increased prevalence of anticoagulant use and comorbid conditions, and had a higher PSA (Table 1 and Supplementary Fig. 2).
Table 1
Patient demographics for men who underwent prostate biopsy (January 2005 to September 2014) stratified by pre/postintervention time periods (n = 104 584)
| Baseline Jan 2005–Jul 2008 (n = 37 217) |
2008 USPSTF Aug 2008–Feb 2009 (n = 6672) |
ERSPC/PLCO Mar 2009–Apr 2012 (n = 36 603) |
2012 USPSTF May 2012–Apr 2013 (n = 10 772) |
2013 AUA May 2013–Sep 2014 (n = 13 320) |
p value | Total (n = 104 584) | |
|---|---|---|---|---|---|---|---|
| Age at biopsy, yr | |||||||
| Median (IQR) | 62.0 (56.0–68.0) | 62.0 (56.0–68.0) | 63.0 (57.0–70.0) | 65.0 (59.0–71.0) | 66.0 (59.0–71.0) | <0.001 | 63.0 (57.0–70.0) |
| Age group, yr (%) | |||||||
| 40–49 | 2269 (6.1) | 416 (6.2) | 1737 (4.7) | 420 (3.9) | 467 (3.5) | <0.001 | 5309 (5.1) |
| 50–59 | 12 366 (33) | 2201 (33) | 10 342 (28) | 2644 (25) | 3098 (23) | 30 651 (29) | |
| 60–69 | 14 671 (39) | 2697 (40) | 15 105 (41) | 4426 (41) | 5546 (42) | 42 445 (41) | |
| ≥70 | 7911 (21) | 1358 (20) | 9419 (26) | 3282 (31) | 4209 (32) | 26 179 (25) | |
| Race (%) | |||||||
| White | 25 508 (69) | 4643 (70) | 26 187 (72) | 7795 (72) | 9407 (71) | <0.001 | 73 540 (70) |
| African American | 3695 (9.9) | 780 (12) | 4423 (12) | 1283 (12) | 1589 (12) | 11 770 (11) | |
| Hispanic | 2549 (6.8) | 538 (8.1) | 2911 (8.0) | 837 (7.8) | 1186 (8.9) | 8021 (7.7) | |
| Asian | 1133 (3.0) | 236 (3.5) | 1142 (3.1) | 306 (2.8) | 387 (2.9) | 3204 (3.1) | |
| Unknown/Missing | 4332 (12) | 475 (7.1) | 1940 (5.3) | 551 (5.1) | 751 (5.6) | 8049 (7.7) | |
| Census region (%) | |||||||
| Northeast | 7094 (19) | 1196 (18) | 6120 (17) | 1886 (18) | 2424 (18) | <0.001 | 18 720 (18) |
| Midwest | 10 094 (27) | 1562 (23) | 8837 (24) | 2742 (26) | 3627 (27) | 26 862 (26) | |
| South | 15 330 (41) | 2889 (43) | 16 371 (45) | 4698 (44) | 5538 (42) | 44 826 (43) | |
| West | 4660 (13) | 993 (15) | 5147 (14) | 1442 (13) | 1728 (13) | 13 970 (13) | |
| Other/Missing | 39 (0.1) | 32 (0.5) | 128 (0.3) | 4 (0) | 3 (0) | 206 (0.2) | |
| PSA, ng/d | n = 5174 | n = 548 | n = 7345 | n = 2211 | n = 3471 | n = 18 749 | |
| Median (IQR) | 5.1 (3.9–7.1) | 4.9 (3.8–6.8) | 5.0 (4.0–6.8) | 5.3 (4.2–7.4) | 5.4 (4.2–7.8) | <0.001 | 5.1 (4.0–7.1) |
| Anticoagulant use (%) | 928 (2.5) | 184 (2.8) | 1103 (3.0) | 299 (2.8) | 415 (3.1) | <0.001 | 2929 (2.8) |
| History of fluoroquinolone use in prior 12 mo (%) | 6797 (18) | 1267 (19) | 6557 (18) | 2017 (19) | 2504 (19) | 0.054 | 19 142 (18) |
| Charlson Comorbidity Index (%) | |||||||
| 0 | 22 354 (60) | 3887 (58) | 20 269 (55) | 5758 (54) | 7039 (53) | <0.001 | 59 307 (57) |
| 1 | 8199 (22) | 1453 (22) | 8252 (23) | 2503 (23) | 2999 (23) | 23 406 (22) | |
| 2 | 3232 (8.7) | 664 (10) | 3520 (9.6) | 1108 (10) | 1359 (10) | 9883 (9.4) | |
| ≥3 | 3432 (9.2) | 668 (10) | 4562 (13) | 1403 (13) | 1923 (14) | 11 988 (12) | |
| Comorbidities (%) | |||||||
| Myocardial infarction | 763 (2.1) | 144 (2.2) | 812 (2.2) | 227 (2.1) | 273 (2.0) | 0.6 | 2219 (2.1) |
| Congestive heart failure | 1443 (3.9) | 223 (3.3) | 1548 (4.2) | 467 (4.3) | 586 (4.4) | <0.001 | 4267 (4.1) |
| Peripheral vascular disease | 2054 (5.5) | 399 (6.0) | 2660 (7.3) | 783 (7.3) | 1071 (8.0) | <0.001 | 6967 (6.7) |
| Cerebrovascular disease | 2007 (5.4) | 379 (5.7) | 2358 (6.4) | 681 (6.3) | 947 (7.1) | <0.001 | 6372 (6.1) |
| Dementia | 370 (1.0) | 63 (0.9) | 442 (1.2) | 141 (1.3) | 198 (1.5) | <0.001 | 1214 (1.2) |
| Chronic obstructive pulmonary disease | 4251 (11) | 758 (11) | 4293 (12) | 1254 (12) | 1686 (13) | 0.004 | 12 242 (12) |
| Ulcer | 362 (1.0) | 60 (0.9) | 297 (0.8) | 84 (0.8) | 102 (0.8) | 0.07 | 905 (0.9) |
| Mild liver disease | 1182 (3.2) | 223 (3.3) | 1332 (3.6) | 413 (3.8) | 515 (3.9) | <0.001 | 3665 (3.5) |
| Diabetes mellitus | 6475 (17) | 1290 (19) | 7823 (21) | 2442 (23) | 2973 (22) | <0.001 | 21 003 (20) |
| Diabetes with organ damage | 1105 (3.0) | 225 (3.4) | 1600 (4.4) | 492 (4.6) | 734 (5.5) | <0.001 | 4156 (4.0) |
| Hemiplegia | 128 (0.3) | 25 (0.4) | 150 (0.4) | 51 (0.5) | 56 (0.4) | 0.3 | 410 (0.4) |
| Moderate/severe renal disease | 1133 (3.0) | 254 (3.8) | 1950 (5.3) | 674 (6.3) | 908 (6.8) | <0.001 | 4919 (4.7) |
| Moderate/severe liver disease | 27 (0.1) | 7 (0.1) | 31 (0.1) | 9 (0.1) | 12 (0.1) | 0.9 | 86 (0.1) |
| Metastatic solid tumor | 175 (0.5) | 37 (0.6) | 199 (0.5) | 58 (0.5) | 80 (0.6) | 0.4 | 549 (0.5) |
| AIDS | 82 (0.2) | 16 (0.2) | 101 (0.3) | 27 (0.3) | 46 (0.3) | 0.2 | 272 (0.3) |
| Rheumatologic disease | 449 (1.2) | 71 (1.1) | 522 (1.4) | 158 (1.5) | 196 (1.5) | 0.007 | 1396 (1.3) |
| Cancer | 1736 (4.7) | 337 (5.1) | 1796 (4.9) | 571 (5.3) | 701 (5.3) | 0.02 | 5141 (4.9) |
| Net worth range (%) | |||||||
| <$25 000 | 1831 (4.9) | 404 (6.1) | 2101 (5.7) | 561 (5.2) | 695 (5.2) | <0.001 | 5592 (5.3) |
| $25 000–149 000 | 5188 (14) | 1042 (16) | 5608 (15) | 1591 (15) | 2046 (15) | 15 475 (15) | |
| $150 000–249 000 | 4754 (13) | 837 (13) | 5182 (14) | 1499 (14) | 1862 (14) | 14 134 (14) | |
| $250 000–499 000 | 9694 (26) | 1779 (27) | 10 048 (28) | 3027 (28) | 3636 (27) | 28 184 (27) | |
| ≥$500 000 | 10 910 (29) | 2030 (30) | 10 860 (30) | 3208 (30) | 3852 (29) | 30 860 (30) | |
| Unknown/missing | 4840 (13) | 580 (8.7) | 2804 (7.7) | 886 (8.2) | 1229 (9.2) | 10 339 (9.9) | |
| 30-d complications (%) | |||||||
| Any | 5973 (16) | 1151 (17) | 6120 (17) | 1863 (17) | 2451 (18) | <0.001 | 17 558 (17) |
| Bleeding | 2547 (6.8) | 499 (7.5) | 2660 (7.3) | 860 (8.0) | 1083 (8.1) | <0.001 | 7649 (7.3) |
| Nonsepsis infection | 2978 (8.0) | 590 (8.8) | 2990 (8.2) | 875 (8.1) | 1236 (9.3) | <0.001 | 8669 (8.3) |
| Sepsis | 191 (0.5) | 50 (0.7) | 349 (1.0) | 105 (1.0) | 134 (1.0) | <0.001 | 829 (0.8) |
| Urinary retention | 980 (2.6) | 161 (2.4) | 1138 (3.1) | 341 (3.2) | 422 (3.2) | <0.001 | 3042 (2.9) |
| Hospitalization | 1019 (2.7) | 181 (2.7) | 1056 (2.9) | 320 (3.0) | 367 (2.8) | 0.6 | 2943 (2.8) |
AIDS = acquired immunodeficiency syndrome; AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; IQR = interquartile range; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer [screening trial]; PSA = prostate-specific antigen; USPTF = US Preventive Services Task Force.
Overall, 30-d complications occurred in 17% of biopsies. The most prevalent was nonsepsis genitourinary infections (8.3%), followed by bleeding complications (7.3%) and urinary retention (2.9%) (Table 1). The 30-d hospitalization rate was 2.8%.
We first examined unadjusted rates of prostate biopsy among men aged ≥40 yr (Fig. 1 and Table 2). From January 2005 to September 2014, biopsy rates fell 33% from 64.1 to 42.8 per 100 000 person-months. There was substantial seasonal variation in biopsy rates with reductions near the end of each calendar year. With the ITS approach, each intervention is potentially associated with an immediate change in event rates (ie, level shift or change in intercept) as well as a change in the rate of growth (ie, change in slope). At baseline, there was a statistically significant growth in prostate biopsy rates of 2.9 biopsies per 100 000 person-months per year (95% confidence interval [CI], 0.96–4.8; p < 0.001). Following the 2008 USPSTF recommendations, there was an immediate decrease (ie, level shift) in biopsy rates of −10.1 biopsies per 100 000 person-months (95% CI, −17.1 to −3.0; p < 0.001). Likewise, there were immediate decreases in biopsy rates following the 2012 USPSTF recommendations (−13.8 biopsies per 100 000 person-months; 95% CI, −21.0 to −6.7; p < 0.001) and 2013 AUA guidelines (−8.8 biopsies per 100 000 person-months; 95% CI, −16.7 to −0.92; p = 0.03). Analyses adjusted for case mix were similar, although the baseline growth rate was no longer significant, indicating this was related to population shifts over time (Table 2, model 2).
Fig. 1
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 2
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315)
| Prostate biopsy | Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Model 1* | ||||||||||||||
| Baseline rate | 57.8 (53.4–62.3) |
<0.001 | 8.6 (7.5–9.7) |
<0.001 | 4.0 (3.3–4.6) |
<0.001 | 0.17 (0.03–0.32) |
0.02 | 2.8 (2.3–3.3) |
<0.001 | 1.5 (1.3–1.6) |
<0.001 | 1.4 (1.1–1.7) |
<0.001 |
| Baseline trend in rate | 2.9 (0.96–4.8) |
<0.001 | 0.84 (0.48–1.3) |
<0.001 | 0.72 (0.48–1.1) |
<0.001 | 0.12 (0.12–0.12) |
<0.001 | 0.36 (0.12–0.48) |
<0.001 | 0.12 (0.00–0.12) |
<0.001 | 0.00 (−0.12 to 0.12) |
0.5 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.1 (−17.1 to −3.0) |
<0.001 | −0.61 (−3.1 to 1.9) |
0.6 | −1.4 (−2.8 to −0.03) |
0.053 | 0.04 (−0.35 to 0.42) |
0.8 | 0.56 (−0.70 to 1.8) |
0.4 | −0.06 (−0.69 to 0.57) |
0.9 | 0.16 (−0.58 to 0.90) |
0.7 |
| Trend change | 7.9 (−12.7 to 28.4) |
0.5 | −1.0 (−9.1 to 7.2) |
0.8 | −0.12 (−4.1 to 4.0) |
0.98 | −0.60 (−1.9 to 0.72) |
0.4 | −2.0 (−6.5 to 2.4) |
0.4 | −1.1 (−3.4 to 1.3) |
0.4 | −1.0 (−3.6 to 1.6) |
0.5 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.6 (−12 to 4.5) |
0.4 | −0.74 (−3.8 to 2.4) |
0.6 | −0.12 (−1.8 to 1.5) |
0.9 | 0.29 (−0.20 to 0.78) |
0.2 | −0.04 (−1.6 to 1.6) |
0.97 | 0.59 (−0.23 to 1.4) |
0.2 | 0.51 (−0.43 to 1.5) |
0.3 |
| Trend change | −11.2 (−32.0 to 9.6) |
0.3 | 0.0 (−8.2 to 8.0) |
0.99 | −0.96 (−5.0 to 3.1) |
0.7 | 0.36 (−0.96 to 1.8) |
0.5 | 1.7 (−2.8 to 6.1) |
0.5 | 1.0 (−1.4 to 3.4) |
0.4 | 0.84 (−1.7 to 3.5) |
0.5 |
| 2012 USPSTF | ||||||||||||||
| Level change | −13.8 (−21.0 to −6.7) |
<0.001 | −2.5 (−4.5 to −0.45) |
0.02 | −1.7 (−3.1 to −0.31) |
0.02 | −0.12 (−0.42 to 0.19) |
0.5 | −0.61 (−1.5 to 0.28) |
0.2 | −0.75 (−1.1 to −0.44) |
<0.001 | −0.25 (−0.77 to 0.27) |
0.4 |
| Trend change | 8.2 (−5.8 to 22.0) |
0.3 | 3.0 (−0.36 to 6.4) |
0.09 | 3.1 (0.72–5.5) |
0.01 | 0.12 (−0.36 to 0.60) |
0.7 | 0.84 (−0.72 to 2.3) |
0.3 | 0.48 (0.00 to 1.1) |
0.08 | 0.24 (−0.60 to 1.2) |
0.6 |
| 2013 AUA | ||||||||||||||
| Level change | −8.8 (−16.7 to −0.92) |
0.03 | −1.4 (−3.8 to 1.1) |
0.3 | −1.9 (−3.5 to −0.22) |
0.03 | −0.02 (−0.41 to 0.37) |
0.9 | −0.19 (−1.3 to 0.91) |
0.7 | −0.38 (−0.77 to 0.01) |
0.06 | −0.01 (−0.66 to 0.63) |
0.96 |
| Trend change | −10.4 (−30.0 to 9.2) |
0.3 | −4.1 (−7.9 to −0.24) |
0.04 | −2.9 (−6.0 to 0.12) |
0.07 | −0.24 (−0.72 to 0.36) |
0.5 | −1.8 (−3.5 to 0.0) |
0.046 | −0.60 (−1.2 to 0.12) |
0.09 | −0.60 (−1.7 to 0.36) |
0.2 |
| Model 2† | ||||||||||||||
| Baseline rate | 86.8 (82.1–91.4) |
<0.001 | 16.2 (14.9–17.6) |
<0.001 | 6.4 (4.9–7.9) |
<0.001 | 0.39 (−0.13 to 0.91) |
0.2 | 6.1 (4.6–7.7) |
<0.001 | 3.6 (2.8–4.4) |
<0.001 | 3.0 (2.2–3.8) |
<0.001 |
| Baseline trend in rate | 0.24 (−1.8 to 2.3) |
0.8 | −0.24 (−0.72 to 0.24) |
0.4 | 0.48 (−0.12 to 1.1) |
0.09 | 0.12 (−0.12 to 0.24) |
0.4 | −0.24 (−0.72 to 0.24) |
0.4 | −0.24 (−0.48 to 0.00) |
0.054 | −0.12 (−0.36 to 0.12) |
0.3 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.8 (−17.8 to −3.8) |
<0.001 | −5.4 (−9.6 to −1.2) |
0.01 | −2.5 (−5.6 to 0.71) |
0.1 | −0.61 (−1.9 to 0.71) |
0.4 | 0.14 (−3.8 to 4.1) |
0.9 | 1.1 (−0.98 to 3.1) |
0.3 | 0.14 (−1.9 to 2.1) |
0.9 |
| Trend change | 10.6 (−10.1 to 31.2) |
0.3 | 6.1 (−8.4 to 20.8) |
0.4 | −1.7 (−12.2 to 8.8) |
0.8 | 1.7 (−2.9 to 6.2) |
0.5 | 1.7 (−12.0 to 15.5) |
0.8 | −4.7 (−11.8 to 2.5) |
0.2 | −0.84 (−7.8 to 6.1) |
0.8 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.9 (−12 to 4.1) |
0.3 | −0.24 (−5.6 to 5.1) |
0.9 | 1.9 (−2.1 to 5.8) |
0.4 | −0.32 (−2.0 to 1.4) |
0.7 | −0.99 (−6.0 to 4.0) |
0.7 | 1.8 (−0.84 to 4.4) |
0.2 | 0.87 (−1.7 to 3.4) |
0.5 |
| Trend change | −16.2 (−37.1 to 4.7) |
0.1 | −6.8 (−21.4 to 7.8) |
0.4 | 0.36 (−10.2 to 10.9) |
0.95 | −1.9 (−6.5 to 2.8) |
0.4 | −1.6 (−15.2 to 12.2) |
0.8 | 4.7 (−2.4 to 11.9) |
0.2 | 0.48 (−6.4 to 7.4) |
0.9 |
| 2012 USPSTF | ||||||||||||||
| Level change | −11.0 (−18.1 to −4.0) |
0.001 | −2.8 (−6.0 to 0.38) |
0.09 | 0.53 (−1.9 to 3.0) |
0.7 | −0.02 (−0.95 to 0.91) |
0.97 | −1.8 (−4.6 to 0.96) |
0.2 | −1.5 (−3.1 to 0.17) |
0.08 | 0.16 (−1.3 to 1.6) |
0.8 |
| Trend change | 7.0 (−7.2 to 21.1) |
0.3 | 1.7 (−3.6 to 7.0) |
0.5 | 0.12 (−4.1 to 4.2) |
0.96 | 0.12 (−1.4 to 1.8) |
0.9 | −0.48 (−5.3 to 4.3) |
0.9 | 1.6 (−1.3 to 4.3) |
0.3 | 0.12 (−2.4 to 2.6) |
0.9 |
| 2013 AUA | ||||||||||||||
| Level change | −8.2 (−15.9 to −0.38) |
0.04 | −1.0 (−4.8 to 2.8) |
0.6 | −0.44 (−3.2 to 2.3) |
0.8 | −0.15 (−1.3 to 0.99) |
0.8 | 0.28 (−3.2 to 3.7) |
0.9 | −0.50 (−2.7 to 1.7) |
0.7 | 0.47 (−1.3 to 2.3) |
0.6 |
| Trend change | −4.2 (−24.5 to 16.1) |
0.7 | −0.7 (−7.6 to 6.2) |
0.9 | 1.1 (−4.2 to 6.2) |
0.7 | −0.12 (−1.8 to 1.7) |
0.9 | −0.60 (−5.9 to 4.8) |
0.8 | −1.8 (−4.9 to 1.2) |
0.3 | −0.48 (−3.1 to 2.2) |
0.7 |
* Model 1: Base model with constant, time trend, intervention indicator variables. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
† Model 2: Adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, and anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPTF = US Preventive Services Task Force.
Next, we examined 30-d postbiopsy complications among men aged ≥40 yr as a measure of the absolute morbidity of biopsy (Fig. 1, Supplementary Fig. 3, and Table 2). Across the study periods, overall complication rates decreased 10% from 8.7 to 7.8 per 100 000 person-months. In unadjusted ITS analyses, there was a baseline growth in the overall complication rate (0.84 per 100 000 person-months per year; 95% CI, 0.48–1.3; p < 0.001), driven by growth in the rate of nonsepsis infections (0.72 per 100 000 person-months per year; 95% CI, 0.48–1.1; p < 0.001). Only the 2012 USPSTF recommendations were associated with a significant change in complication rates, following which there was an immediate decrease in the overall complication rate of −2.5 per 100 000 person-months (95% CI, −4.5 to −0.45; p = 0.02), driven by reductions in nonsepsis infections (−1.7 per 100 000 person-months; 95% CI, −3.1 to −0.31; p = 0.02) and urinary retention (−0.75 per 100 000 person-months;95% CI, −1.1 to −0.44; p < 0.001). Following the 2013 AUA guidelines there was also a reduction in the overall complication baseline growth rate (−4.1 per 100 000 person-months per year; 95% CI, −7.9 to −0.24; p = 0.04). In analyses adjusted for case mix (Table 2, model 2), effects were overall attenuated, and only the 2008 USPSTF recommendations were associated with a reduction in overall complication rates (−5.4 per 100 000 person-months; 95% CI, −9.6 to −1.2; p = 0.01).
To examine whether there was a change in the relative morbidity of biopsy, we performed a secondary analysis examining complication rates among men undergoing biopsy (Fig. 2). The proportion of men undergoing biopsy who experienced a complication increased from 14% in January 2005 to 18% in September 2014, an increase of 36%. There was again a baseline growth in overall complication rates (0.72% per year; 95% CI, 0.36–1.1; p < 0.001), driven by an increase in nonsepsis infections and sepsis (Supplementary Table 2). None of the interventions affected this growth. Similar results were obtained after adjusting for case mix (Table 3).
Fig. 2
Interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 3
Adjusted* interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Baseline rate | 15.6 (14.3–16.8) |
<0.001 | 7.1 (6.5–7.7) |
<0.001 | 0.34 (0.11–0.57) |
<0.001 | 6.7 (5.9–7.5) |
<0.001 | 2.8 (2.5–3.2) |
<0.001 | 3.2 (2.8–3.6) |
<0.001 |
| Baseline trend in rate | 0.60 (0.12–1.1) |
0.02 | 0.72 (0.60–0.96) |
<0.001 | 0.12 (0.12–0.24) |
<0.001 | 0.12 (−0.12 to 0.36) |
0.3 | −0.12 (−0.24 to 0.00) |
0.3 | −0.12 (−0.24 to 0.00) |
0.0498 |
| 2008 USPSTF | ||||||||||||
| Level change | 1.1 (−2.0 to 4.1) |
0.5 | −1.3 (−3.2 to 0.58) |
0.2 | 0.13 (−0.45 to 0.72) |
0.7 | 1.9 (−0.11 to 3.9) |
0.07 | 0.46 (−0.50 to 1.4) |
0.4 | 0.27 (−0.69 to 1.2) |
0.6 |
| Trend change | −3.5 (−13.4 to 6.6) |
0.5 | −1.8 (−8.4 to 4.7) |
0.6 | −1.0 (−3.0 to 0.96) |
0.3 | −4.3 (−11.4 to 2.6) |
0.2 | −1.6 (−4.9 to 1.9) |
0.4 | −0.72 (−4.0 to 2.6) |
0.7 |
| ERSPC/PLCO | ||||||||||||
| Level change | −0.14 (−3.9 to 3.6) |
0.9 | 1.1 (−1.3 to 3.4) |
0.4 | 0.55 (−0.19 to 1.3) |
0.2 | 0.39 (−2.2 to 3.0) |
0.8 | 0.75 (−0.47 to 2.0) |
0.2 | 0.65 (−0.56 to 1.9) |
0.3 |
| Trend change | 2.9 (−7.1 to 13.0) |
0.6 | 0.72 (−5.8 to 7.2) |
0.8 | 0.84 (−1.2 to 2.9) |
0.4 | 4.2 (−2.8 to 11.3) |
0.2 | 1.7 (−1.8 to 5.0) |
0.4 | 0.72 (−2.6 to 4.0) |
0.7 |
| 2012 USPSTF | ||||||||||||
| Level change | −0.27 (−2.7 to 2.1) |
0.8 | −0.56 (−2.0 to 0.87) |
0.5 | 0.16 (−0.31 to 0.63) |
0.5 | 1.1 (−0.38 to 2.5) |
0.2 | −0.38 (−1.0 to 0.28) |
0.3 | 0.44 (−0.34 to 1.2) |
0.3 |
| Trend change | 1.3 (−2.6 to 5.3) |
0.5 | 2.9 (0.60–5.3) |
0.02 | −0.24 (−1.1 to 0.60) |
0.6 | −0.72 (−3.2 to 1.7) |
0.5 | 0.24 (−0.96 to 1.3) |
0.7 | −0.48 (−1.8 to 0.84) |
0.5 |
| 2013 AUA | ||||||||||||
| Level change | 0.25 (−2.6 to 3.1) |
0.9 | −0.64 (−2.3 to 1.1) |
0.5 | 0.22 (−0.40 to 0.83) |
0.5 | 1.1 (−0.68 to 2.8) |
0.2 | −0.29 (−1.1 to 0.48) |
0.5 | 0.47 (−0.55 to 1.5) |
0.4 |
| Trend change | −1.6 (−6.1 to 2.9) |
0.5 | −2.5 (−5.5 to 0.60) |
0.1 | 0.12 (−0.72 to 1.1) |
0.8 | −0.36 (−3.1 to 2.4) |
0.8 | 0.24 (−1.2 to 1.7) |
0.8 | 0.12 (−1.3 to 1.6) |
0.9 |
* Model adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 person-months. Baseline trend in rate and trend change per 100 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial]; USPTF = US Preventive Services Task Force.
Given the increase in the relative morbidity of biopsy, we performed logistic regression to identify predictors of each complication category (Supplementary Table 3 and Table 4). Due to the large sample size, we focused on those features associated with the greatest magnitude of effect. A prior diagnosis of cancer was associated with an increased risk of all complication categories including any complication (odds ratio [OR]: 1.30; 95% CI, 1.22–1.39; p < 0.001) and hospitalization (OR: 1.59; 95% CI, 1.39–1.82; p < 0.001). Similarly, prior fluoroquinolone use was associated with an increased risk of all complication categories including any complication (OR: 1.27; 95% CI, 1.22–1.32; p < 0.001) and sepsis (OR: 1.55; 95% CI, 1.33–1.81; p < 0.001). Age at biopsy ≥70 yr was associated with a nearly 4-fold increased risk of urinary retention. Although anticoagulant use was associated with increased risks of bleeding (OR: 1.28; 95% CI, 1.14–1.45; p < 0.001), sepsis (OR: 1.48; 95% CI, 1.06–2.07; p = 0.02), and hospitalization (OR: 1.39; 95% CI, 1.16–1.67; p < 0.001), the increase in the overall complication risk was modest (OR 1.14; 95% CI, 1.04–1.25; p = 0.004).
Table 4
Multivariable analysis* of 30-d postbiopsy complications among men aged ≥40 yr undergoing prostate biopsy (January 2005 to September 2014) (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Hospitalization | Urinary retention | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Age at biopsy, yr | ||||||||||||
| 40–49 | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| 50–59 | 1.01 (0.94–1.09) | 0.8 | 0.94 (0.85–1.04) | 0.2 | 1.16 (0.83–1.63) | 0.4 | 1.01 (0.90–1.14) | 0.8 | 1.12 (0.93–1.36) | 0.2 | 1.54 (1.20–1.97) | <0.001 |
| 60–69 | 1.05 (0.97–1.13) | 0.2 | 0.94 (0.85–1.03) | 0.2 | 0.98 (0.70–1.37) | 0.9 | 1.05 (0.94–1.18) | 0.4 | 1.16 (0.95–1.40) | 0.1 | 2.16 (1.68–2.77) | <0.001 |
| ≥70 | 1.25 (1.15–1.36) | <0.001 | 1.03 (0.93–1.15) | 0.6 | 1.05 (0.74–1.49) | 0.8 | 1.29 (1.15–1.46) | <0.001 | 1.31 (1.07–1.60) | 0.008 | 3.98 (3.08–5.15) | <0.001 |
| Race | ||||||||||||
| White | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| Asian | 1.27 (1.15–1.40) | <0.001 | 1.27 (1.11–1.44) | 0.0003 | 1.65 (1.19–2.29) | 0.003 | 1.36 (1.18–1.56) | <0.001 | 0.94 (0.76–1.17) | 0.6 | 1.15 (0.93–1.42) | 0.2 |
| African American | 1.09 (1.03–1.15) | 0.002 | 1.05 (0.97–1.13) | 0.2 | 1.34 (1.09–1.64) | 0.005 | 1.17 (1.09–1.27) | <0.001 | 1.13 (1.01–1.27) | 0.03 | 1.15 (1.03–1.28) | 0.02 |
| Hispanic | 1.32 (1.24–1.40) | <0.001 | 1.36 (1.26–1.47) | <0.001 | 1.52 (1.21–1.90) | <0.001 | 1.33 (1.22–1.45) | <0.001 | 1.12 (0.97–1.28) | 0.1 | 1.37 (1.21–1.56) | <0.001 |
| Unknown/missing | 0.99 (0.93–1.06) | 0.8 | 1.07 (0.99–1.17) | 0.1 | 1.23 (0.94–1.61) | 0.1 | 0.96 (0.88–1.05) | 0.4 | 0.94 (0.81–1.08) | 0.4 | 1.02 (0.88–1.18) | 0.8 |
| Diabetes mellitus | 1.03 (0.99–1.07) | 0.09 | 1.06 (1.00–1.11) | 0.045 | 1.08 (0.92–1.28) | 0.4 | 1.01 (0.96–1.07) | 0.7 | 1.20 (1.10–1.31) | <0.001 | 0.99 (0.91–1.08) | 0.8 |
| COPD | 1.14 (1.09–1.20) | <0.001 | 1.07 (1.00–1.14) | 0.04 | 1.23 (1.01–1.49) | 0.04 | 1.14 (1.06–1.22) | <0.001 | 1.36 (1.23–1.51) | <0.001 | 1.14 (1.03–1.25) | 0.01 |
| PVD | 1.24 (1.16–1.31) | <0.001 | 1.1 (1.02–1.2) | 0.02 | 1.19 (0.93–1.51) | 0.2 | 1.28 (1.18–1.40) | <0.001 | 1.31 (1.15–1.48) | <0.001 | 1.26 (1.12–1.42) | <0.001 |
| Cerebrovascular disease | 1.11 (1.04–1.19) | 0.001 | 1.12 (1.03–1.22) | 0.01 | 1.21 (0.93–1.57) | 0.2 | 1.03 (0.94–1.14) | 0.5 | 1.26 (1.10–1.44) | <0.001 | 1.30 (1.15–1.47) | <0.001 |
| Cancer | 1.30 (1.22–1.39) | <0.001 | 1.16 (1.06–1.27) | 0.001 | 1.37 (1.05–1.79) | 0.02 | 1.40 (1.27–1.53) | <0.001 | 1.59 (1.39–1.82) | <0.001 | 1.32 (1.16–1.52) | <0.001 |
| Anticoagulant use | 1.14 (1.04–1.25) | 0.004 | 1.01 (0.89–1.14) | 0.9 | 1.48 (1.06–2.07) | 0.02 | 1.28 (1.14–1.45) | <0.001 | 1.39 (1.16–1.67) | <0.001 | 1.11 (0.92–1.33) | 0.3 |
| Prior fluoroquinolones | 1.27 (1.22–1.32) | <0.001 | 1.37 (1.31–1.45) | <0.001 | 1.55 (1.33–1.81) | <0.001 | 1.21 (1.14–1.28) | <0.001 | 1.25 (1.14–1.36) | <0.001 | 1.27 (1.16–1.38) | <0.001 |
* Models adjusted for year of biopsy and the features listed in the table.
CI = confidence interval; COPD = chronic obstructive pulmonary disease; PVD = peripheral vascular disease.
We conducted sensitivity analyses where the postintervention run-in period was varied from 2 to 4 mo (Supplementary Table 4). Results overall were similar, although effects were attenuated for the 2013 AUA guidelines on biopsy rates and the 2012 USPSTF recommendations on overall complication rates, likely due to fewer time points for analysis. In a separate sensitivity analysis, we examined bleeding complications excluding microhematuria from 2009 to 2014, following introduction of a separate code for microhematuria. Bleeding rates paralleled those of the primary bleeding end point (Supplementary Fig. 4), and results of both ITS and logistic regression analyses were similar (Supplementary Table 5 and 6). As an exploratory analysis, we examined PCa diagnosis rates within 60 d following biopsy (Supplementary Fig. 5). These appeared to increase steadily across the study period.
To our knowledge, this is the largest study to examine the impact of landmark PSA screening publications on rates of prostate biopsy and the first to examine their impact on postbiopsy complications. Overall, biopsy rates have declined by 33%, which appears attributable to the publication of the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines. Interestingly, the magnitude of reduction in biopsy rates exceeded previously reported decreases in PSA screening rates [6], [7], [8], [9], [10], [11], [12], [22], and [23], which suggests that either the threshold for referral to a urologist or the threshold for performing a biopsy has changed. More importantly, there was a concurrent 22% decrease in 30-d complication rates that appears largely attributable to the 2012 USPSTF recommendations. Still, the reduction in complication rates was less than the reduction in biopsy rates due to an increase in the relative morbidity of biopsy, driven by growth in infectious complications. This may in part be related to a shift to older patients with greater comorbidity undergoing biopsy, although similar results were obtained in analyses adjusted for case mix, which suggests additional factors may exist.
To this end, we evaluated predictors of postbiopsy complications to identify actionable targets for further risk reduction. Some of these features represent modifiable risk factors (eg, anticoagulant use), and others represent opportunities for altering periprocedural management. For instance, prior fluoroquinolone use may identify patients who would benefit from prebiopsy rectal swabs to guide periprocedural antibiotics [24] and [25]. Nonmodifiable risk factors, including specific comorbidities such as cancer and peripheral vascular disease, may encourage reconsideration of the risk–benefit ratio before proceeding with biopsy.
It is also important to consider the concurrent impact of landmark PSA screening publications on rates of PCa diagnosis to contextualize the observed changes in biopsy and its morbidity. An exploratory analysis suggested that cancer diagnosis rates have steadily increased, although further studies are necessary to examine this issue comprehensively.
A number of studies have examined the impact of the 2008 USPSTF recommendations [7], [8], [9], and [22], ERSPC/PLCO trials [6] and [9], and 2012 USPSTF recommendations [6], [10], [11], [12], and [23] on PSA screening rates, generally noting reductions ranging from 0% to 8%. However, much less data are available on the impact of these publications on biopsy rates. One single-center study noted a 31% reduction in the number of biopsies performed following the 2012 USPSTF recommendations [13]; another noted no change in the proportion of referred men who underwent biopsy [26]. A report from a Canadian academic health network noted a reduction in biopsy rates from 58.0 to 35.5 biopsies per month following the 2012 USPSTF recommendations [27]. Our findings therefore confirm a reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines.
Limited data exist regarding secular trends in postbiopsy complications, and prior studies have focused exclusively on the relative morbidity among men undergoing biopsy rather than absolute complication rates. A Canadian population-based study observed that the 30-d hospitalization rate increased from 1.0% in 1996 to 4.1% in 2005, driven by an increase in infectious complications [28]. A Surveillance, Epidemiology and End Results–Medicare study also noted an increase in hospitalizations related to infectious complications over time [14]. Although we observed a similar increase in 30-d overall complications among men undergoing biopsy, likewise driven by infectious complications, there was an absolute 22% reduction in 30-d complications over the study period. Taken together, these findings support a reduction in the morbidity of biopsy following publication of revised PSA screening guidelines but highlight the continued need to reduce the relative morbidity of biopsy.
A number of studies have described complication rates following prostate biopsy [15], [16], [29], and [30]. Hematuria has been reported in 10–84% of biopsies, rectal bleeding in 1–45%, infectious complications in up to 6.3%, urinary retention in up to 1.7%, and hospitalization in up to 6.9% [14] and [29]. The wide range in complication rates is due to differences in definitions, ascertainment of end points, and variation in clinical practice. Given the heterogeneity in postbiopsy complication rates, our findings have two advantages. First, they represent a contemporary, geographically diverse population across the entire United States. Second, use of claims-based ascertainment limited complications to those cases in which patients sought medical care, and although this may underestimate the comprehensive morbidity of biopsy, it more accurately reflects the economic burden of serious complications.
This study has several limitations. Most importantly, a control population of men who were not exposed to the PSA screening publications was not available. This study is retrospective, and additional, unrecognized changes in policy or clinical practice may possibly have confounded the impact of the interventions examined. However, the ITS approach has been described as the strongest quasi-experimental design to evaluate policy changes when a randomized trial is not possible [19]. Because we relied on administrative claims for ascertainment of complications, minor complications that did not generate a claim were not captured, thus underestimating these end points. Although we adjusted for case mix, we were unable to adjust for PSA. However, population shifts in PSA may be confounded by changes in screening practices, and PSA adjustment may therefore under- or overestimate the impact of interventions. We were also unable to adjust for other factors that may be associated with biopsy or complication rates, such as prostate magnetic resonance imaging (MRI), ancillary biomarker testing, biopsy approach (eg, MRI-fusion biopsy, transperineal biopsy), indication for anticoagulation, use of periprocedural bridging, or operator biopsy volume/experience.
We observed an immediate reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines, with a smaller reduction in absolute 30-d complication rates. Still, the relative morbidity of biopsy has increased over time, driven by infectious complications. Accordingly, if health policy aims to reduce the morbidity of PSA screening related to biopsy, further efforts should focus on improving patient selection and periprocedural processes of care.
Author contributions: Boris Gershman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Gershman, van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Acquisition of data: Van Houten, Shah.
Analysis and interpretation of data: Gershman, Van Houten, Shah, Karnes.
Drafting of the manuscript: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Critical revision of the manuscript for important intellectual content: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Statistical analysis: Van Houten.
Obtaining funding: Karnes, Shah.
Administrative, technical, or material support: Karnes, Shah.
Supervision: Karnes, Shah.
Other (specify): None.
Financial disclosures: Boris Gershman certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: This publication was made possible by funding from the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery.
Funding/Support and role of the sponsor: None.
In 2009, two landmark randomized trials—the European Randomized Study of Screening for Prostate Cancer [ERSPC] [1] and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial [PLCO] [2]—reported smaller than anticipated reductions in prostate cancer (PCa) mortality and prompted reexamination of prostate-specific antigen (PSA) screening recommendations. The US Preventive Services Task Force (USPSTF), which in 2008 had recommended against PSA screening in men aged >75 yr [3], updated its position in 2012 to recommend against systematic PSA screening in the general population [4]. The American Urological Association (AUA) likewise updated its PSA screening guidelines in May 2013, recommending shared decision making for men aged 55–69 yr [5]. Both the 2012 USPSTF and 2013 AUA guidelines cited prostate biopsy and its potential complications as a significant harm associated with PSA screening.
Although several studies have noted a reduction in PSA screening rates following these publications [6], [7], [8], [9], [10], [11], and [12], there are little data on whether this has translated into the intended reductions in screening-related morbidity in the form of prostate biopsy and postbiopsy complications [13]. We evaluated whether publication of the 2008 USPSTF recommendations, 2009 ERSPC/PLCO trials, 2012 USPSTF recommendations, or 2013 AUA guidelines was associated with reductions in prostate biopsy or postbiopsy complications. Biopsy-related complications also remain understudied both with regard to incidence as well as identification of risk factors [14], [15], and [16]. Consequently, we also sought to characterize postbiopsy complications and evaluate predictors using a contemporary population-based cohort.
We conducted a retrospective analysis of administrative claims from the Optum Labs Data Warehouse that contains data on >100 million individuals enrolled in private health plans and Medicare Advantage Plans, from geographically diverse regions across the United States, as described in Supplementary Methods [17]. Because this study involved analysis of preexisting deidentified data, it was exempt from institutional review board approval.
We identified all prostate biopsies performed in men aged ≥40 yr between January 1, 2005, and September 30, 2014, using Current Procedural Terminology-4 code 55700 alone (11%) or in combination with 76872 (84%) or 76942 (0%), or International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 60.11 (5%). The first biopsy was selected as the index biopsy when more than one biopsy was performed. Patients with a prior diagnosis of PCa (ICD-9 code 185) were excluded. We restricted the cohort to patients with continuous medical and pharmacy enrollment for at least 12 mo prior to biopsy and 30 d after biopsy to ascertain comorbidities and 30-d complications (Supplementary Fig. 1).
We recorded sociodemographic characteristics, comorbid conditions, anticoagulant use within 30 d of biopsy, and fluoroquinolone use in the 12 mo prior to biopsy as described in Supplement 1. The Charlson Comorbidity Index was calculated as previously described [18]. Prebiopsy PSA values within 60 d of biopsy were available for a subset of patients.
The primary outcomes were crude and adjusted rates of prostate biopsy and 30-d complications. Complications were identified using inpatient and outpatient administrative claims as summarized in Supplementary Table 1.
Baseline demographic features were compared across study periods using the Kruskal-Wallis test for continuous variables and the chi-square test for discrete variables. To evaluate the impact of each publication on rates of prostate biopsy and 30-d complications, we used an interrupted time series (ITS) approach as described in Supplementary Methods [19] and [20]. Rates were estimated using all men aged ≥40 yr with at least 12 mo of continuous medical and pharmacy enrollment as the denominator. As secondary analyses, we estimated rates of 30-d complications among men who underwent biopsy. Two separate ITS analyses were performed: (1) unadjusted analyses, using crude rates of biopsy and complications aggregated for each month, and (2) analyses standardized for case mix, to adjust for population shifts over time as described in Supplementary Methods [20] and [21].
Logistic regression was used to evaluate predictors of 30-d complications. Multivariable models were adjusted for age at biopsy, year of biopsy, race, anticoagulant use, history of fluoroquinolone use, and the presence of the five most prevalent Charlson comorbidities.
Statistical analyses were performed using SAS v.9.3 (SAS Institute Inc., Cary, NC, USA). All tests were two sided with p < 0.05 considered statistically significant.
Between January 2005 and September 2014, a total of 104 584 men underwent prostate biopsy (Table 1). Median age at biopsy was 63.0 yr (interquartile range [IQR]: 57.0–70.0). Overall, 18% of men had used fluoroquinolones in the 12 mo preceding biopsy, and 2.8% were on anticoagulants. Median prebiopsy PSA was 5.1 ng/dl (IQR: 4.0–7.1) for the 18 749 men (18%) for whom this was available. Over the time periods, men undergoing biopsy were older, had increased prevalence of anticoagulant use and comorbid conditions, and had a higher PSA (Table 1 and Supplementary Fig. 2).
Table 1
Patient demographics for men who underwent prostate biopsy (January 2005 to September 2014) stratified by pre/postintervention time periods (n = 104 584)
| Baseline Jan 2005–Jul 2008 (n = 37 217) |
2008 USPSTF Aug 2008–Feb 2009 (n = 6672) |
ERSPC/PLCO Mar 2009–Apr 2012 (n = 36 603) |
2012 USPSTF May 2012–Apr 2013 (n = 10 772) |
2013 AUA May 2013–Sep 2014 (n = 13 320) |
p value | Total (n = 104 584) | |
|---|---|---|---|---|---|---|---|
| Age at biopsy, yr | |||||||
| Median (IQR) | 62.0 (56.0–68.0) | 62.0 (56.0–68.0) | 63.0 (57.0–70.0) | 65.0 (59.0–71.0) | 66.0 (59.0–71.0) | <0.001 | 63.0 (57.0–70.0) |
| Age group, yr (%) | |||||||
| 40–49 | 2269 (6.1) | 416 (6.2) | 1737 (4.7) | 420 (3.9) | 467 (3.5) | <0.001 | 5309 (5.1) |
| 50–59 | 12 366 (33) | 2201 (33) | 10 342 (28) | 2644 (25) | 3098 (23) | 30 651 (29) | |
| 60–69 | 14 671 (39) | 2697 (40) | 15 105 (41) | 4426 (41) | 5546 (42) | 42 445 (41) | |
| ≥70 | 7911 (21) | 1358 (20) | 9419 (26) | 3282 (31) | 4209 (32) | 26 179 (25) | |
| Race (%) | |||||||
| White | 25 508 (69) | 4643 (70) | 26 187 (72) | 7795 (72) | 9407 (71) | <0.001 | 73 540 (70) |
| African American | 3695 (9.9) | 780 (12) | 4423 (12) | 1283 (12) | 1589 (12) | 11 770 (11) | |
| Hispanic | 2549 (6.8) | 538 (8.1) | 2911 (8.0) | 837 (7.8) | 1186 (8.9) | 8021 (7.7) | |
| Asian | 1133 (3.0) | 236 (3.5) | 1142 (3.1) | 306 (2.8) | 387 (2.9) | 3204 (3.1) | |
| Unknown/Missing | 4332 (12) | 475 (7.1) | 1940 (5.3) | 551 (5.1) | 751 (5.6) | 8049 (7.7) | |
| Census region (%) | |||||||
| Northeast | 7094 (19) | 1196 (18) | 6120 (17) | 1886 (18) | 2424 (18) | <0.001 | 18 720 (18) |
| Midwest | 10 094 (27) | 1562 (23) | 8837 (24) | 2742 (26) | 3627 (27) | 26 862 (26) | |
| South | 15 330 (41) | 2889 (43) | 16 371 (45) | 4698 (44) | 5538 (42) | 44 826 (43) | |
| West | 4660 (13) | 993 (15) | 5147 (14) | 1442 (13) | 1728 (13) | 13 970 (13) | |
| Other/Missing | 39 (0.1) | 32 (0.5) | 128 (0.3) | 4 (0) | 3 (0) | 206 (0.2) | |
| PSA, ng/d | n = 5174 | n = 548 | n = 7345 | n = 2211 | n = 3471 | n = 18 749 | |
| Median (IQR) | 5.1 (3.9–7.1) | 4.9 (3.8–6.8) | 5.0 (4.0–6.8) | 5.3 (4.2–7.4) | 5.4 (4.2–7.8) | <0.001 | 5.1 (4.0–7.1) |
| Anticoagulant use (%) | 928 (2.5) | 184 (2.8) | 1103 (3.0) | 299 (2.8) | 415 (3.1) | <0.001 | 2929 (2.8) |
| History of fluoroquinolone use in prior 12 mo (%) | 6797 (18) | 1267 (19) | 6557 (18) | 2017 (19) | 2504 (19) | 0.054 | 19 142 (18) |
| Charlson Comorbidity Index (%) | |||||||
| 0 | 22 354 (60) | 3887 (58) | 20 269 (55) | 5758 (54) | 7039 (53) | <0.001 | 59 307 (57) |
| 1 | 8199 (22) | 1453 (22) | 8252 (23) | 2503 (23) | 2999 (23) | 23 406 (22) | |
| 2 | 3232 (8.7) | 664 (10) | 3520 (9.6) | 1108 (10) | 1359 (10) | 9883 (9.4) | |
| ≥3 | 3432 (9.2) | 668 (10) | 4562 (13) | 1403 (13) | 1923 (14) | 11 988 (12) | |
| Comorbidities (%) | |||||||
| Myocardial infarction | 763 (2.1) | 144 (2.2) | 812 (2.2) | 227 (2.1) | 273 (2.0) | 0.6 | 2219 (2.1) |
| Congestive heart failure | 1443 (3.9) | 223 (3.3) | 1548 (4.2) | 467 (4.3) | 586 (4.4) | <0.001 | 4267 (4.1) |
| Peripheral vascular disease | 2054 (5.5) | 399 (6.0) | 2660 (7.3) | 783 (7.3) | 1071 (8.0) | <0.001 | 6967 (6.7) |
| Cerebrovascular disease | 2007 (5.4) | 379 (5.7) | 2358 (6.4) | 681 (6.3) | 947 (7.1) | <0.001 | 6372 (6.1) |
| Dementia | 370 (1.0) | 63 (0.9) | 442 (1.2) | 141 (1.3) | 198 (1.5) | <0.001 | 1214 (1.2) |
| Chronic obstructive pulmonary disease | 4251 (11) | 758 (11) | 4293 (12) | 1254 (12) | 1686 (13) | 0.004 | 12 242 (12) |
| Ulcer | 362 (1.0) | 60 (0.9) | 297 (0.8) | 84 (0.8) | 102 (0.8) | 0.07 | 905 (0.9) |
| Mild liver disease | 1182 (3.2) | 223 (3.3) | 1332 (3.6) | 413 (3.8) | 515 (3.9) | <0.001 | 3665 (3.5) |
| Diabetes mellitus | 6475 (17) | 1290 (19) | 7823 (21) | 2442 (23) | 2973 (22) | <0.001 | 21 003 (20) |
| Diabetes with organ damage | 1105 (3.0) | 225 (3.4) | 1600 (4.4) | 492 (4.6) | 734 (5.5) | <0.001 | 4156 (4.0) |
| Hemiplegia | 128 (0.3) | 25 (0.4) | 150 (0.4) | 51 (0.5) | 56 (0.4) | 0.3 | 410 (0.4) |
| Moderate/severe renal disease | 1133 (3.0) | 254 (3.8) | 1950 (5.3) | 674 (6.3) | 908 (6.8) | <0.001 | 4919 (4.7) |
| Moderate/severe liver disease | 27 (0.1) | 7 (0.1) | 31 (0.1) | 9 (0.1) | 12 (0.1) | 0.9 | 86 (0.1) |
| Metastatic solid tumor | 175 (0.5) | 37 (0.6) | 199 (0.5) | 58 (0.5) | 80 (0.6) | 0.4 | 549 (0.5) |
| AIDS | 82 (0.2) | 16 (0.2) | 101 (0.3) | 27 (0.3) | 46 (0.3) | 0.2 | 272 (0.3) |
| Rheumatologic disease | 449 (1.2) | 71 (1.1) | 522 (1.4) | 158 (1.5) | 196 (1.5) | 0.007 | 1396 (1.3) |
| Cancer | 1736 (4.7) | 337 (5.1) | 1796 (4.9) | 571 (5.3) | 701 (5.3) | 0.02 | 5141 (4.9) |
| Net worth range (%) | |||||||
| <$25 000 | 1831 (4.9) | 404 (6.1) | 2101 (5.7) | 561 (5.2) | 695 (5.2) | <0.001 | 5592 (5.3) |
| $25 000–149 000 | 5188 (14) | 1042 (16) | 5608 (15) | 1591 (15) | 2046 (15) | 15 475 (15) | |
| $150 000–249 000 | 4754 (13) | 837 (13) | 5182 (14) | 1499 (14) | 1862 (14) | 14 134 (14) | |
| $250 000–499 000 | 9694 (26) | 1779 (27) | 10 048 (28) | 3027 (28) | 3636 (27) | 28 184 (27) | |
| ≥$500 000 | 10 910 (29) | 2030 (30) | 10 860 (30) | 3208 (30) | 3852 (29) | 30 860 (30) | |
| Unknown/missing | 4840 (13) | 580 (8.7) | 2804 (7.7) | 886 (8.2) | 1229 (9.2) | 10 339 (9.9) | |
| 30-d complications (%) | |||||||
| Any | 5973 (16) | 1151 (17) | 6120 (17) | 1863 (17) | 2451 (18) | <0.001 | 17 558 (17) |
| Bleeding | 2547 (6.8) | 499 (7.5) | 2660 (7.3) | 860 (8.0) | 1083 (8.1) | <0.001 | 7649 (7.3) |
| Nonsepsis infection | 2978 (8.0) | 590 (8.8) | 2990 (8.2) | 875 (8.1) | 1236 (9.3) | <0.001 | 8669 (8.3) |
| Sepsis | 191 (0.5) | 50 (0.7) | 349 (1.0) | 105 (1.0) | 134 (1.0) | <0.001 | 829 (0.8) |
| Urinary retention | 980 (2.6) | 161 (2.4) | 1138 (3.1) | 341 (3.2) | 422 (3.2) | <0.001 | 3042 (2.9) |
| Hospitalization | 1019 (2.7) | 181 (2.7) | 1056 (2.9) | 320 (3.0) | 367 (2.8) | 0.6 | 2943 (2.8) |
AIDS = acquired immunodeficiency syndrome; AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; IQR = interquartile range; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer [screening trial]; PSA = prostate-specific antigen; USPTF = US Preventive Services Task Force.
Overall, 30-d complications occurred in 17% of biopsies. The most prevalent was nonsepsis genitourinary infections (8.3%), followed by bleeding complications (7.3%) and urinary retention (2.9%) (Table 1). The 30-d hospitalization rate was 2.8%.
We first examined unadjusted rates of prostate biopsy among men aged ≥40 yr (Fig. 1 and Table 2). From January 2005 to September 2014, biopsy rates fell 33% from 64.1 to 42.8 per 100 000 person-months. There was substantial seasonal variation in biopsy rates with reductions near the end of each calendar year. With the ITS approach, each intervention is potentially associated with an immediate change in event rates (ie, level shift or change in intercept) as well as a change in the rate of growth (ie, change in slope). At baseline, there was a statistically significant growth in prostate biopsy rates of 2.9 biopsies per 100 000 person-months per year (95% confidence interval [CI], 0.96–4.8; p < 0.001). Following the 2008 USPSTF recommendations, there was an immediate decrease (ie, level shift) in biopsy rates of −10.1 biopsies per 100 000 person-months (95% CI, −17.1 to −3.0; p < 0.001). Likewise, there were immediate decreases in biopsy rates following the 2012 USPSTF recommendations (−13.8 biopsies per 100 000 person-months; 95% CI, −21.0 to −6.7; p < 0.001) and 2013 AUA guidelines (−8.8 biopsies per 100 000 person-months; 95% CI, −16.7 to −0.92; p = 0.03). Analyses adjusted for case mix were similar, although the baseline growth rate was no longer significant, indicating this was related to population shifts over time (Table 2, model 2).
Fig. 1
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 2
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315)
| Prostate biopsy | Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Model 1* | ||||||||||||||
| Baseline rate | 57.8 (53.4–62.3) |
<0.001 | 8.6 (7.5–9.7) |
<0.001 | 4.0 (3.3–4.6) |
<0.001 | 0.17 (0.03–0.32) |
0.02 | 2.8 (2.3–3.3) |
<0.001 | 1.5 (1.3–1.6) |
<0.001 | 1.4 (1.1–1.7) |
<0.001 |
| Baseline trend in rate | 2.9 (0.96–4.8) |
<0.001 | 0.84 (0.48–1.3) |
<0.001 | 0.72 (0.48–1.1) |
<0.001 | 0.12 (0.12–0.12) |
<0.001 | 0.36 (0.12–0.48) |
<0.001 | 0.12 (0.00–0.12) |
<0.001 | 0.00 (−0.12 to 0.12) |
0.5 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.1 (−17.1 to −3.0) |
<0.001 | −0.61 (−3.1 to 1.9) |
0.6 | −1.4 (−2.8 to −0.03) |
0.053 | 0.04 (−0.35 to 0.42) |
0.8 | 0.56 (−0.70 to 1.8) |
0.4 | −0.06 (−0.69 to 0.57) |
0.9 | 0.16 (−0.58 to 0.90) |
0.7 |
| Trend change | 7.9 (−12.7 to 28.4) |
0.5 | −1.0 (−9.1 to 7.2) |
0.8 | −0.12 (−4.1 to 4.0) |
0.98 | −0.60 (−1.9 to 0.72) |
0.4 | −2.0 (−6.5 to 2.4) |
0.4 | −1.1 (−3.4 to 1.3) |
0.4 | −1.0 (−3.6 to 1.6) |
0.5 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.6 (−12 to 4.5) |
0.4 | −0.74 (−3.8 to 2.4) |
0.6 | −0.12 (−1.8 to 1.5) |
0.9 | 0.29 (−0.20 to 0.78) |
0.2 | −0.04 (−1.6 to 1.6) |
0.97 | 0.59 (−0.23 to 1.4) |
0.2 | 0.51 (−0.43 to 1.5) |
0.3 |
| Trend change | −11.2 (−32.0 to 9.6) |
0.3 | 0.0 (−8.2 to 8.0) |
0.99 | −0.96 (−5.0 to 3.1) |
0.7 | 0.36 (−0.96 to 1.8) |
0.5 | 1.7 (−2.8 to 6.1) |
0.5 | 1.0 (−1.4 to 3.4) |
0.4 | 0.84 (−1.7 to 3.5) |
0.5 |
| 2012 USPSTF | ||||||||||||||
| Level change | −13.8 (−21.0 to −6.7) |
<0.001 | −2.5 (−4.5 to −0.45) |
0.02 | −1.7 (−3.1 to −0.31) |
0.02 | −0.12 (−0.42 to 0.19) |
0.5 | −0.61 (−1.5 to 0.28) |
0.2 | −0.75 (−1.1 to −0.44) |
<0.001 | −0.25 (−0.77 to 0.27) |
0.4 |
| Trend change | 8.2 (−5.8 to 22.0) |
0.3 | 3.0 (−0.36 to 6.4) |
0.09 | 3.1 (0.72–5.5) |
0.01 | 0.12 (−0.36 to 0.60) |
0.7 | 0.84 (−0.72 to 2.3) |
0.3 | 0.48 (0.00 to 1.1) |
0.08 | 0.24 (−0.60 to 1.2) |
0.6 |
| 2013 AUA | ||||||||||||||
| Level change | −8.8 (−16.7 to −0.92) |
0.03 | −1.4 (−3.8 to 1.1) |
0.3 | −1.9 (−3.5 to −0.22) |
0.03 | −0.02 (−0.41 to 0.37) |
0.9 | −0.19 (−1.3 to 0.91) |
0.7 | −0.38 (−0.77 to 0.01) |
0.06 | −0.01 (−0.66 to 0.63) |
0.96 |
| Trend change | −10.4 (−30.0 to 9.2) |
0.3 | −4.1 (−7.9 to −0.24) |
0.04 | −2.9 (−6.0 to 0.12) |
0.07 | −0.24 (−0.72 to 0.36) |
0.5 | −1.8 (−3.5 to 0.0) |
0.046 | −0.60 (−1.2 to 0.12) |
0.09 | −0.60 (−1.7 to 0.36) |
0.2 |
| Model 2† | ||||||||||||||
| Baseline rate | 86.8 (82.1–91.4) |
<0.001 | 16.2 (14.9–17.6) |
<0.001 | 6.4 (4.9–7.9) |
<0.001 | 0.39 (−0.13 to 0.91) |
0.2 | 6.1 (4.6–7.7) |
<0.001 | 3.6 (2.8–4.4) |
<0.001 | 3.0 (2.2–3.8) |
<0.001 |
| Baseline trend in rate | 0.24 (−1.8 to 2.3) |
0.8 | −0.24 (−0.72 to 0.24) |
0.4 | 0.48 (−0.12 to 1.1) |
0.09 | 0.12 (−0.12 to 0.24) |
0.4 | −0.24 (−0.72 to 0.24) |
0.4 | −0.24 (−0.48 to 0.00) |
0.054 | −0.12 (−0.36 to 0.12) |
0.3 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.8 (−17.8 to −3.8) |
<0.001 | −5.4 (−9.6 to −1.2) |
0.01 | −2.5 (−5.6 to 0.71) |
0.1 | −0.61 (−1.9 to 0.71) |
0.4 | 0.14 (−3.8 to 4.1) |
0.9 | 1.1 (−0.98 to 3.1) |
0.3 | 0.14 (−1.9 to 2.1) |
0.9 |
| Trend change | 10.6 (−10.1 to 31.2) |
0.3 | 6.1 (−8.4 to 20.8) |
0.4 | −1.7 (−12.2 to 8.8) |
0.8 | 1.7 (−2.9 to 6.2) |
0.5 | 1.7 (−12.0 to 15.5) |
0.8 | −4.7 (−11.8 to 2.5) |
0.2 | −0.84 (−7.8 to 6.1) |
0.8 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.9 (−12 to 4.1) |
0.3 | −0.24 (−5.6 to 5.1) |
0.9 | 1.9 (−2.1 to 5.8) |
0.4 | −0.32 (−2.0 to 1.4) |
0.7 | −0.99 (−6.0 to 4.0) |
0.7 | 1.8 (−0.84 to 4.4) |
0.2 | 0.87 (−1.7 to 3.4) |
0.5 |
| Trend change | −16.2 (−37.1 to 4.7) |
0.1 | −6.8 (−21.4 to 7.8) |
0.4 | 0.36 (−10.2 to 10.9) |
0.95 | −1.9 (−6.5 to 2.8) |
0.4 | −1.6 (−15.2 to 12.2) |
0.8 | 4.7 (−2.4 to 11.9) |
0.2 | 0.48 (−6.4 to 7.4) |
0.9 |
| 2012 USPSTF | ||||||||||||||
| Level change | −11.0 (−18.1 to −4.0) |
0.001 | −2.8 (−6.0 to 0.38) |
0.09 | 0.53 (−1.9 to 3.0) |
0.7 | −0.02 (−0.95 to 0.91) |
0.97 | −1.8 (−4.6 to 0.96) |
0.2 | −1.5 (−3.1 to 0.17) |
0.08 | 0.16 (−1.3 to 1.6) |
0.8 |
| Trend change | 7.0 (−7.2 to 21.1) |
0.3 | 1.7 (−3.6 to 7.0) |
0.5 | 0.12 (−4.1 to 4.2) |
0.96 | 0.12 (−1.4 to 1.8) |
0.9 | −0.48 (−5.3 to 4.3) |
0.9 | 1.6 (−1.3 to 4.3) |
0.3 | 0.12 (−2.4 to 2.6) |
0.9 |
| 2013 AUA | ||||||||||||||
| Level change | −8.2 (−15.9 to −0.38) |
0.04 | −1.0 (−4.8 to 2.8) |
0.6 | −0.44 (−3.2 to 2.3) |
0.8 | −0.15 (−1.3 to 0.99) |
0.8 | 0.28 (−3.2 to 3.7) |
0.9 | −0.50 (−2.7 to 1.7) |
0.7 | 0.47 (−1.3 to 2.3) |
0.6 |
| Trend change | −4.2 (−24.5 to 16.1) |
0.7 | −0.7 (−7.6 to 6.2) |
0.9 | 1.1 (−4.2 to 6.2) |
0.7 | −0.12 (−1.8 to 1.7) |
0.9 | −0.60 (−5.9 to 4.8) |
0.8 | −1.8 (−4.9 to 1.2) |
0.3 | −0.48 (−3.1 to 2.2) |
0.7 |
* Model 1: Base model with constant, time trend, intervention indicator variables. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
† Model 2: Adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, and anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPTF = US Preventive Services Task Force.
Next, we examined 30-d postbiopsy complications among men aged ≥40 yr as a measure of the absolute morbidity of biopsy (Fig. 1, Supplementary Fig. 3, and Table 2). Across the study periods, overall complication rates decreased 10% from 8.7 to 7.8 per 100 000 person-months. In unadjusted ITS analyses, there was a baseline growth in the overall complication rate (0.84 per 100 000 person-months per year; 95% CI, 0.48–1.3; p < 0.001), driven by growth in the rate of nonsepsis infections (0.72 per 100 000 person-months per year; 95% CI, 0.48–1.1; p < 0.001). Only the 2012 USPSTF recommendations were associated with a significant change in complication rates, following which there was an immediate decrease in the overall complication rate of −2.5 per 100 000 person-months (95% CI, −4.5 to −0.45; p = 0.02), driven by reductions in nonsepsis infections (−1.7 per 100 000 person-months; 95% CI, −3.1 to −0.31; p = 0.02) and urinary retention (−0.75 per 100 000 person-months;95% CI, −1.1 to −0.44; p < 0.001). Following the 2013 AUA guidelines there was also a reduction in the overall complication baseline growth rate (−4.1 per 100 000 person-months per year; 95% CI, −7.9 to −0.24; p = 0.04). In analyses adjusted for case mix (Table 2, model 2), effects were overall attenuated, and only the 2008 USPSTF recommendations were associated with a reduction in overall complication rates (−5.4 per 100 000 person-months; 95% CI, −9.6 to −1.2; p = 0.01).
To examine whether there was a change in the relative morbidity of biopsy, we performed a secondary analysis examining complication rates among men undergoing biopsy (Fig. 2). The proportion of men undergoing biopsy who experienced a complication increased from 14% in January 2005 to 18% in September 2014, an increase of 36%. There was again a baseline growth in overall complication rates (0.72% per year; 95% CI, 0.36–1.1; p < 0.001), driven by an increase in nonsepsis infections and sepsis (Supplementary Table 2). None of the interventions affected this growth. Similar results were obtained after adjusting for case mix (Table 3).
Fig. 2
Interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 3
Adjusted* interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Baseline rate | 15.6 (14.3–16.8) |
<0.001 | 7.1 (6.5–7.7) |
<0.001 | 0.34 (0.11–0.57) |
<0.001 | 6.7 (5.9–7.5) |
<0.001 | 2.8 (2.5–3.2) |
<0.001 | 3.2 (2.8–3.6) |
<0.001 |
| Baseline trend in rate | 0.60 (0.12–1.1) |
0.02 | 0.72 (0.60–0.96) |
<0.001 | 0.12 (0.12–0.24) |
<0.001 | 0.12 (−0.12 to 0.36) |
0.3 | −0.12 (−0.24 to 0.00) |
0.3 | −0.12 (−0.24 to 0.00) |
0.0498 |
| 2008 USPSTF | ||||||||||||
| Level change | 1.1 (−2.0 to 4.1) |
0.5 | −1.3 (−3.2 to 0.58) |
0.2 | 0.13 (−0.45 to 0.72) |
0.7 | 1.9 (−0.11 to 3.9) |
0.07 | 0.46 (−0.50 to 1.4) |
0.4 | 0.27 (−0.69 to 1.2) |
0.6 |
| Trend change | −3.5 (−13.4 to 6.6) |
0.5 | −1.8 (−8.4 to 4.7) |
0.6 | −1.0 (−3.0 to 0.96) |
0.3 | −4.3 (−11.4 to 2.6) |
0.2 | −1.6 (−4.9 to 1.9) |
0.4 | −0.72 (−4.0 to 2.6) |
0.7 |
| ERSPC/PLCO | ||||||||||||
| Level change | −0.14 (−3.9 to 3.6) |
0.9 | 1.1 (−1.3 to 3.4) |
0.4 | 0.55 (−0.19 to 1.3) |
0.2 | 0.39 (−2.2 to 3.0) |
0.8 | 0.75 (−0.47 to 2.0) |
0.2 | 0.65 (−0.56 to 1.9) |
0.3 |
| Trend change | 2.9 (−7.1 to 13.0) |
0.6 | 0.72 (−5.8 to 7.2) |
0.8 | 0.84 (−1.2 to 2.9) |
0.4 | 4.2 (−2.8 to 11.3) |
0.2 | 1.7 (−1.8 to 5.0) |
0.4 | 0.72 (−2.6 to 4.0) |
0.7 |
| 2012 USPSTF | ||||||||||||
| Level change | −0.27 (−2.7 to 2.1) |
0.8 | −0.56 (−2.0 to 0.87) |
0.5 | 0.16 (−0.31 to 0.63) |
0.5 | 1.1 (−0.38 to 2.5) |
0.2 | −0.38 (−1.0 to 0.28) |
0.3 | 0.44 (−0.34 to 1.2) |
0.3 |
| Trend change | 1.3 (−2.6 to 5.3) |
0.5 | 2.9 (0.60–5.3) |
0.02 | −0.24 (−1.1 to 0.60) |
0.6 | −0.72 (−3.2 to 1.7) |
0.5 | 0.24 (−0.96 to 1.3) |
0.7 | −0.48 (−1.8 to 0.84) |
0.5 |
| 2013 AUA | ||||||||||||
| Level change | 0.25 (−2.6 to 3.1) |
0.9 | −0.64 (−2.3 to 1.1) |
0.5 | 0.22 (−0.40 to 0.83) |
0.5 | 1.1 (−0.68 to 2.8) |
0.2 | −0.29 (−1.1 to 0.48) |
0.5 | 0.47 (−0.55 to 1.5) |
0.4 |
| Trend change | −1.6 (−6.1 to 2.9) |
0.5 | −2.5 (−5.5 to 0.60) |
0.1 | 0.12 (−0.72 to 1.1) |
0.8 | −0.36 (−3.1 to 2.4) |
0.8 | 0.24 (−1.2 to 1.7) |
0.8 | 0.12 (−1.3 to 1.6) |
0.9 |
* Model adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 person-months. Baseline trend in rate and trend change per 100 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial]; USPTF = US Preventive Services Task Force.
Given the increase in the relative morbidity of biopsy, we performed logistic regression to identify predictors of each complication category (Supplementary Table 3 and Table 4). Due to the large sample size, we focused on those features associated with the greatest magnitude of effect. A prior diagnosis of cancer was associated with an increased risk of all complication categories including any complication (odds ratio [OR]: 1.30; 95% CI, 1.22–1.39; p < 0.001) and hospitalization (OR: 1.59; 95% CI, 1.39–1.82; p < 0.001). Similarly, prior fluoroquinolone use was associated with an increased risk of all complication categories including any complication (OR: 1.27; 95% CI, 1.22–1.32; p < 0.001) and sepsis (OR: 1.55; 95% CI, 1.33–1.81; p < 0.001). Age at biopsy ≥70 yr was associated with a nearly 4-fold increased risk of urinary retention. Although anticoagulant use was associated with increased risks of bleeding (OR: 1.28; 95% CI, 1.14–1.45; p < 0.001), sepsis (OR: 1.48; 95% CI, 1.06–2.07; p = 0.02), and hospitalization (OR: 1.39; 95% CI, 1.16–1.67; p < 0.001), the increase in the overall complication risk was modest (OR 1.14; 95% CI, 1.04–1.25; p = 0.004).
Table 4
Multivariable analysis* of 30-d postbiopsy complications among men aged ≥40 yr undergoing prostate biopsy (January 2005 to September 2014) (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Hospitalization | Urinary retention | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Age at biopsy, yr | ||||||||||||
| 40–49 | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| 50–59 | 1.01 (0.94–1.09) | 0.8 | 0.94 (0.85–1.04) | 0.2 | 1.16 (0.83–1.63) | 0.4 | 1.01 (0.90–1.14) | 0.8 | 1.12 (0.93–1.36) | 0.2 | 1.54 (1.20–1.97) | <0.001 |
| 60–69 | 1.05 (0.97–1.13) | 0.2 | 0.94 (0.85–1.03) | 0.2 | 0.98 (0.70–1.37) | 0.9 | 1.05 (0.94–1.18) | 0.4 | 1.16 (0.95–1.40) | 0.1 | 2.16 (1.68–2.77) | <0.001 |
| ≥70 | 1.25 (1.15–1.36) | <0.001 | 1.03 (0.93–1.15) | 0.6 | 1.05 (0.74–1.49) | 0.8 | 1.29 (1.15–1.46) | <0.001 | 1.31 (1.07–1.60) | 0.008 | 3.98 (3.08–5.15) | <0.001 |
| Race | ||||||||||||
| White | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| Asian | 1.27 (1.15–1.40) | <0.001 | 1.27 (1.11–1.44) | 0.0003 | 1.65 (1.19–2.29) | 0.003 | 1.36 (1.18–1.56) | <0.001 | 0.94 (0.76–1.17) | 0.6 | 1.15 (0.93–1.42) | 0.2 |
| African American | 1.09 (1.03–1.15) | 0.002 | 1.05 (0.97–1.13) | 0.2 | 1.34 (1.09–1.64) | 0.005 | 1.17 (1.09–1.27) | <0.001 | 1.13 (1.01–1.27) | 0.03 | 1.15 (1.03–1.28) | 0.02 |
| Hispanic | 1.32 (1.24–1.40) | <0.001 | 1.36 (1.26–1.47) | <0.001 | 1.52 (1.21–1.90) | <0.001 | 1.33 (1.22–1.45) | <0.001 | 1.12 (0.97–1.28) | 0.1 | 1.37 (1.21–1.56) | <0.001 |
| Unknown/missing | 0.99 (0.93–1.06) | 0.8 | 1.07 (0.99–1.17) | 0.1 | 1.23 (0.94–1.61) | 0.1 | 0.96 (0.88–1.05) | 0.4 | 0.94 (0.81–1.08) | 0.4 | 1.02 (0.88–1.18) | 0.8 |
| Diabetes mellitus | 1.03 (0.99–1.07) | 0.09 | 1.06 (1.00–1.11) | 0.045 | 1.08 (0.92–1.28) | 0.4 | 1.01 (0.96–1.07) | 0.7 | 1.20 (1.10–1.31) | <0.001 | 0.99 (0.91–1.08) | 0.8 |
| COPD | 1.14 (1.09–1.20) | <0.001 | 1.07 (1.00–1.14) | 0.04 | 1.23 (1.01–1.49) | 0.04 | 1.14 (1.06–1.22) | <0.001 | 1.36 (1.23–1.51) | <0.001 | 1.14 (1.03–1.25) | 0.01 |
| PVD | 1.24 (1.16–1.31) | <0.001 | 1.1 (1.02–1.2) | 0.02 | 1.19 (0.93–1.51) | 0.2 | 1.28 (1.18–1.40) | <0.001 | 1.31 (1.15–1.48) | <0.001 | 1.26 (1.12–1.42) | <0.001 |
| Cerebrovascular disease | 1.11 (1.04–1.19) | 0.001 | 1.12 (1.03–1.22) | 0.01 | 1.21 (0.93–1.57) | 0.2 | 1.03 (0.94–1.14) | 0.5 | 1.26 (1.10–1.44) | <0.001 | 1.30 (1.15–1.47) | <0.001 |
| Cancer | 1.30 (1.22–1.39) | <0.001 | 1.16 (1.06–1.27) | 0.001 | 1.37 (1.05–1.79) | 0.02 | 1.40 (1.27–1.53) | <0.001 | 1.59 (1.39–1.82) | <0.001 | 1.32 (1.16–1.52) | <0.001 |
| Anticoagulant use | 1.14 (1.04–1.25) | 0.004 | 1.01 (0.89–1.14) | 0.9 | 1.48 (1.06–2.07) | 0.02 | 1.28 (1.14–1.45) | <0.001 | 1.39 (1.16–1.67) | <0.001 | 1.11 (0.92–1.33) | 0.3 |
| Prior fluoroquinolones | 1.27 (1.22–1.32) | <0.001 | 1.37 (1.31–1.45) | <0.001 | 1.55 (1.33–1.81) | <0.001 | 1.21 (1.14–1.28) | <0.001 | 1.25 (1.14–1.36) | <0.001 | 1.27 (1.16–1.38) | <0.001 |
* Models adjusted for year of biopsy and the features listed in the table.
CI = confidence interval; COPD = chronic obstructive pulmonary disease; PVD = peripheral vascular disease.
We conducted sensitivity analyses where the postintervention run-in period was varied from 2 to 4 mo (Supplementary Table 4). Results overall were similar, although effects were attenuated for the 2013 AUA guidelines on biopsy rates and the 2012 USPSTF recommendations on overall complication rates, likely due to fewer time points for analysis. In a separate sensitivity analysis, we examined bleeding complications excluding microhematuria from 2009 to 2014, following introduction of a separate code for microhematuria. Bleeding rates paralleled those of the primary bleeding end point (Supplementary Fig. 4), and results of both ITS and logistic regression analyses were similar (Supplementary Table 5 and 6). As an exploratory analysis, we examined PCa diagnosis rates within 60 d following biopsy (Supplementary Fig. 5). These appeared to increase steadily across the study period.
To our knowledge, this is the largest study to examine the impact of landmark PSA screening publications on rates of prostate biopsy and the first to examine their impact on postbiopsy complications. Overall, biopsy rates have declined by 33%, which appears attributable to the publication of the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines. Interestingly, the magnitude of reduction in biopsy rates exceeded previously reported decreases in PSA screening rates [6], [7], [8], [9], [10], [11], [12], [22], and [23], which suggests that either the threshold for referral to a urologist or the threshold for performing a biopsy has changed. More importantly, there was a concurrent 22% decrease in 30-d complication rates that appears largely attributable to the 2012 USPSTF recommendations. Still, the reduction in complication rates was less than the reduction in biopsy rates due to an increase in the relative morbidity of biopsy, driven by growth in infectious complications. This may in part be related to a shift to older patients with greater comorbidity undergoing biopsy, although similar results were obtained in analyses adjusted for case mix, which suggests additional factors may exist.
To this end, we evaluated predictors of postbiopsy complications to identify actionable targets for further risk reduction. Some of these features represent modifiable risk factors (eg, anticoagulant use), and others represent opportunities for altering periprocedural management. For instance, prior fluoroquinolone use may identify patients who would benefit from prebiopsy rectal swabs to guide periprocedural antibiotics [24] and [25]. Nonmodifiable risk factors, including specific comorbidities such as cancer and peripheral vascular disease, may encourage reconsideration of the risk–benefit ratio before proceeding with biopsy.
It is also important to consider the concurrent impact of landmark PSA screening publications on rates of PCa diagnosis to contextualize the observed changes in biopsy and its morbidity. An exploratory analysis suggested that cancer diagnosis rates have steadily increased, although further studies are necessary to examine this issue comprehensively.
A number of studies have examined the impact of the 2008 USPSTF recommendations [7], [8], [9], and [22], ERSPC/PLCO trials [6] and [9], and 2012 USPSTF recommendations [6], [10], [11], [12], and [23] on PSA screening rates, generally noting reductions ranging from 0% to 8%. However, much less data are available on the impact of these publications on biopsy rates. One single-center study noted a 31% reduction in the number of biopsies performed following the 2012 USPSTF recommendations [13]; another noted no change in the proportion of referred men who underwent biopsy [26]. A report from a Canadian academic health network noted a reduction in biopsy rates from 58.0 to 35.5 biopsies per month following the 2012 USPSTF recommendations [27]. Our findings therefore confirm a reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines.
Limited data exist regarding secular trends in postbiopsy complications, and prior studies have focused exclusively on the relative morbidity among men undergoing biopsy rather than absolute complication rates. A Canadian population-based study observed that the 30-d hospitalization rate increased from 1.0% in 1996 to 4.1% in 2005, driven by an increase in infectious complications [28]. A Surveillance, Epidemiology and End Results–Medicare study also noted an increase in hospitalizations related to infectious complications over time [14]. Although we observed a similar increase in 30-d overall complications among men undergoing biopsy, likewise driven by infectious complications, there was an absolute 22% reduction in 30-d complications over the study period. Taken together, these findings support a reduction in the morbidity of biopsy following publication of revised PSA screening guidelines but highlight the continued need to reduce the relative morbidity of biopsy.
A number of studies have described complication rates following prostate biopsy [15], [16], [29], and [30]. Hematuria has been reported in 10–84% of biopsies, rectal bleeding in 1–45%, infectious complications in up to 6.3%, urinary retention in up to 1.7%, and hospitalization in up to 6.9% [14] and [29]. The wide range in complication rates is due to differences in definitions, ascertainment of end points, and variation in clinical practice. Given the heterogeneity in postbiopsy complication rates, our findings have two advantages. First, they represent a contemporary, geographically diverse population across the entire United States. Second, use of claims-based ascertainment limited complications to those cases in which patients sought medical care, and although this may underestimate the comprehensive morbidity of biopsy, it more accurately reflects the economic burden of serious complications.
This study has several limitations. Most importantly, a control population of men who were not exposed to the PSA screening publications was not available. This study is retrospective, and additional, unrecognized changes in policy or clinical practice may possibly have confounded the impact of the interventions examined. However, the ITS approach has been described as the strongest quasi-experimental design to evaluate policy changes when a randomized trial is not possible [19]. Because we relied on administrative claims for ascertainment of complications, minor complications that did not generate a claim were not captured, thus underestimating these end points. Although we adjusted for case mix, we were unable to adjust for PSA. However, population shifts in PSA may be confounded by changes in screening practices, and PSA adjustment may therefore under- or overestimate the impact of interventions. We were also unable to adjust for other factors that may be associated with biopsy or complication rates, such as prostate magnetic resonance imaging (MRI), ancillary biomarker testing, biopsy approach (eg, MRI-fusion biopsy, transperineal biopsy), indication for anticoagulation, use of periprocedural bridging, or operator biopsy volume/experience.
We observed an immediate reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines, with a smaller reduction in absolute 30-d complication rates. Still, the relative morbidity of biopsy has increased over time, driven by infectious complications. Accordingly, if health policy aims to reduce the morbidity of PSA screening related to biopsy, further efforts should focus on improving patient selection and periprocedural processes of care.
Author contributions: Boris Gershman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Gershman, van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Acquisition of data: Van Houten, Shah.
Analysis and interpretation of data: Gershman, Van Houten, Shah, Karnes.
Drafting of the manuscript: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Critical revision of the manuscript for important intellectual content: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Statistical analysis: Van Houten.
Obtaining funding: Karnes, Shah.
Administrative, technical, or material support: Karnes, Shah.
Supervision: Karnes, Shah.
Other (specify): None.
Financial disclosures: Boris Gershman certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: This publication was made possible by funding from the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery.
Funding/Support and role of the sponsor: None.
In 2009, two landmark randomized trials—the European Randomized Study of Screening for Prostate Cancer [ERSPC] [1] and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial [PLCO] [2]—reported smaller than anticipated reductions in prostate cancer (PCa) mortality and prompted reexamination of prostate-specific antigen (PSA) screening recommendations. The US Preventive Services Task Force (USPSTF), which in 2008 had recommended against PSA screening in men aged >75 yr [3], updated its position in 2012 to recommend against systematic PSA screening in the general population [4]. The American Urological Association (AUA) likewise updated its PSA screening guidelines in May 2013, recommending shared decision making for men aged 55–69 yr [5]. Both the 2012 USPSTF and 2013 AUA guidelines cited prostate biopsy and its potential complications as a significant harm associated with PSA screening.
Although several studies have noted a reduction in PSA screening rates following these publications [6], [7], [8], [9], [10], [11], and [12], there are little data on whether this has translated into the intended reductions in screening-related morbidity in the form of prostate biopsy and postbiopsy complications [13]. We evaluated whether publication of the 2008 USPSTF recommendations, 2009 ERSPC/PLCO trials, 2012 USPSTF recommendations, or 2013 AUA guidelines was associated with reductions in prostate biopsy or postbiopsy complications. Biopsy-related complications also remain understudied both with regard to incidence as well as identification of risk factors [14], [15], and [16]. Consequently, we also sought to characterize postbiopsy complications and evaluate predictors using a contemporary population-based cohort.
We conducted a retrospective analysis of administrative claims from the Optum Labs Data Warehouse that contains data on >100 million individuals enrolled in private health plans and Medicare Advantage Plans, from geographically diverse regions across the United States, as described in Supplementary Methods [17]. Because this study involved analysis of preexisting deidentified data, it was exempt from institutional review board approval.
We identified all prostate biopsies performed in men aged ≥40 yr between January 1, 2005, and September 30, 2014, using Current Procedural Terminology-4 code 55700 alone (11%) or in combination with 76872 (84%) or 76942 (0%), or International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 60.11 (5%). The first biopsy was selected as the index biopsy when more than one biopsy was performed. Patients with a prior diagnosis of PCa (ICD-9 code 185) were excluded. We restricted the cohort to patients with continuous medical and pharmacy enrollment for at least 12 mo prior to biopsy and 30 d after biopsy to ascertain comorbidities and 30-d complications (Supplementary Fig. 1).
We recorded sociodemographic characteristics, comorbid conditions, anticoagulant use within 30 d of biopsy, and fluoroquinolone use in the 12 mo prior to biopsy as described in Supplement 1. The Charlson Comorbidity Index was calculated as previously described [18]. Prebiopsy PSA values within 60 d of biopsy were available for a subset of patients.
The primary outcomes were crude and adjusted rates of prostate biopsy and 30-d complications. Complications were identified using inpatient and outpatient administrative claims as summarized in Supplementary Table 1.
Baseline demographic features were compared across study periods using the Kruskal-Wallis test for continuous variables and the chi-square test for discrete variables. To evaluate the impact of each publication on rates of prostate biopsy and 30-d complications, we used an interrupted time series (ITS) approach as described in Supplementary Methods [19] and [20]. Rates were estimated using all men aged ≥40 yr with at least 12 mo of continuous medical and pharmacy enrollment as the denominator. As secondary analyses, we estimated rates of 30-d complications among men who underwent biopsy. Two separate ITS analyses were performed: (1) unadjusted analyses, using crude rates of biopsy and complications aggregated for each month, and (2) analyses standardized for case mix, to adjust for population shifts over time as described in Supplementary Methods [20] and [21].
Logistic regression was used to evaluate predictors of 30-d complications. Multivariable models were adjusted for age at biopsy, year of biopsy, race, anticoagulant use, history of fluoroquinolone use, and the presence of the five most prevalent Charlson comorbidities.
Statistical analyses were performed using SAS v.9.3 (SAS Institute Inc., Cary, NC, USA). All tests were two sided with p < 0.05 considered statistically significant.
Between January 2005 and September 2014, a total of 104 584 men underwent prostate biopsy (Table 1). Median age at biopsy was 63.0 yr (interquartile range [IQR]: 57.0–70.0). Overall, 18% of men had used fluoroquinolones in the 12 mo preceding biopsy, and 2.8% were on anticoagulants. Median prebiopsy PSA was 5.1 ng/dl (IQR: 4.0–7.1) for the 18 749 men (18%) for whom this was available. Over the time periods, men undergoing biopsy were older, had increased prevalence of anticoagulant use and comorbid conditions, and had a higher PSA (Table 1 and Supplementary Fig. 2).
Table 1
Patient demographics for men who underwent prostate biopsy (January 2005 to September 2014) stratified by pre/postintervention time periods (n = 104 584)
| Baseline Jan 2005–Jul 2008 (n = 37 217) |
2008 USPSTF Aug 2008–Feb 2009 (n = 6672) |
ERSPC/PLCO Mar 2009–Apr 2012 (n = 36 603) |
2012 USPSTF May 2012–Apr 2013 (n = 10 772) |
2013 AUA May 2013–Sep 2014 (n = 13 320) |
p value | Total (n = 104 584) | |
|---|---|---|---|---|---|---|---|
| Age at biopsy, yr | |||||||
| Median (IQR) | 62.0 (56.0–68.0) | 62.0 (56.0–68.0) | 63.0 (57.0–70.0) | 65.0 (59.0–71.0) | 66.0 (59.0–71.0) | <0.001 | 63.0 (57.0–70.0) |
| Age group, yr (%) | |||||||
| 40–49 | 2269 (6.1) | 416 (6.2) | 1737 (4.7) | 420 (3.9) | 467 (3.5) | <0.001 | 5309 (5.1) |
| 50–59 | 12 366 (33) | 2201 (33) | 10 342 (28) | 2644 (25) | 3098 (23) | 30 651 (29) | |
| 60–69 | 14 671 (39) | 2697 (40) | 15 105 (41) | 4426 (41) | 5546 (42) | 42 445 (41) | |
| ≥70 | 7911 (21) | 1358 (20) | 9419 (26) | 3282 (31) | 4209 (32) | 26 179 (25) | |
| Race (%) | |||||||
| White | 25 508 (69) | 4643 (70) | 26 187 (72) | 7795 (72) | 9407 (71) | <0.001 | 73 540 (70) |
| African American | 3695 (9.9) | 780 (12) | 4423 (12) | 1283 (12) | 1589 (12) | 11 770 (11) | |
| Hispanic | 2549 (6.8) | 538 (8.1) | 2911 (8.0) | 837 (7.8) | 1186 (8.9) | 8021 (7.7) | |
| Asian | 1133 (3.0) | 236 (3.5) | 1142 (3.1) | 306 (2.8) | 387 (2.9) | 3204 (3.1) | |
| Unknown/Missing | 4332 (12) | 475 (7.1) | 1940 (5.3) | 551 (5.1) | 751 (5.6) | 8049 (7.7) | |
| Census region (%) | |||||||
| Northeast | 7094 (19) | 1196 (18) | 6120 (17) | 1886 (18) | 2424 (18) | <0.001 | 18 720 (18) |
| Midwest | 10 094 (27) | 1562 (23) | 8837 (24) | 2742 (26) | 3627 (27) | 26 862 (26) | |
| South | 15 330 (41) | 2889 (43) | 16 371 (45) | 4698 (44) | 5538 (42) | 44 826 (43) | |
| West | 4660 (13) | 993 (15) | 5147 (14) | 1442 (13) | 1728 (13) | 13 970 (13) | |
| Other/Missing | 39 (0.1) | 32 (0.5) | 128 (0.3) | 4 (0) | 3 (0) | 206 (0.2) | |
| PSA, ng/d | n = 5174 | n = 548 | n = 7345 | n = 2211 | n = 3471 | n = 18 749 | |
| Median (IQR) | 5.1 (3.9–7.1) | 4.9 (3.8–6.8) | 5.0 (4.0–6.8) | 5.3 (4.2–7.4) | 5.4 (4.2–7.8) | <0.001 | 5.1 (4.0–7.1) |
| Anticoagulant use (%) | 928 (2.5) | 184 (2.8) | 1103 (3.0) | 299 (2.8) | 415 (3.1) | <0.001 | 2929 (2.8) |
| History of fluoroquinolone use in prior 12 mo (%) | 6797 (18) | 1267 (19) | 6557 (18) | 2017 (19) | 2504 (19) | 0.054 | 19 142 (18) |
| Charlson Comorbidity Index (%) | |||||||
| 0 | 22 354 (60) | 3887 (58) | 20 269 (55) | 5758 (54) | 7039 (53) | <0.001 | 59 307 (57) |
| 1 | 8199 (22) | 1453 (22) | 8252 (23) | 2503 (23) | 2999 (23) | 23 406 (22) | |
| 2 | 3232 (8.7) | 664 (10) | 3520 (9.6) | 1108 (10) | 1359 (10) | 9883 (9.4) | |
| ≥3 | 3432 (9.2) | 668 (10) | 4562 (13) | 1403 (13) | 1923 (14) | 11 988 (12) | |
| Comorbidities (%) | |||||||
| Myocardial infarction | 763 (2.1) | 144 (2.2) | 812 (2.2) | 227 (2.1) | 273 (2.0) | 0.6 | 2219 (2.1) |
| Congestive heart failure | 1443 (3.9) | 223 (3.3) | 1548 (4.2) | 467 (4.3) | 586 (4.4) | <0.001 | 4267 (4.1) |
| Peripheral vascular disease | 2054 (5.5) | 399 (6.0) | 2660 (7.3) | 783 (7.3) | 1071 (8.0) | <0.001 | 6967 (6.7) |
| Cerebrovascular disease | 2007 (5.4) | 379 (5.7) | 2358 (6.4) | 681 (6.3) | 947 (7.1) | <0.001 | 6372 (6.1) |
| Dementia | 370 (1.0) | 63 (0.9) | 442 (1.2) | 141 (1.3) | 198 (1.5) | <0.001 | 1214 (1.2) |
| Chronic obstructive pulmonary disease | 4251 (11) | 758 (11) | 4293 (12) | 1254 (12) | 1686 (13) | 0.004 | 12 242 (12) |
| Ulcer | 362 (1.0) | 60 (0.9) | 297 (0.8) | 84 (0.8) | 102 (0.8) | 0.07 | 905 (0.9) |
| Mild liver disease | 1182 (3.2) | 223 (3.3) | 1332 (3.6) | 413 (3.8) | 515 (3.9) | <0.001 | 3665 (3.5) |
| Diabetes mellitus | 6475 (17) | 1290 (19) | 7823 (21) | 2442 (23) | 2973 (22) | <0.001 | 21 003 (20) |
| Diabetes with organ damage | 1105 (3.0) | 225 (3.4) | 1600 (4.4) | 492 (4.6) | 734 (5.5) | <0.001 | 4156 (4.0) |
| Hemiplegia | 128 (0.3) | 25 (0.4) | 150 (0.4) | 51 (0.5) | 56 (0.4) | 0.3 | 410 (0.4) |
| Moderate/severe renal disease | 1133 (3.0) | 254 (3.8) | 1950 (5.3) | 674 (6.3) | 908 (6.8) | <0.001 | 4919 (4.7) |
| Moderate/severe liver disease | 27 (0.1) | 7 (0.1) | 31 (0.1) | 9 (0.1) | 12 (0.1) | 0.9 | 86 (0.1) |
| Metastatic solid tumor | 175 (0.5) | 37 (0.6) | 199 (0.5) | 58 (0.5) | 80 (0.6) | 0.4 | 549 (0.5) |
| AIDS | 82 (0.2) | 16 (0.2) | 101 (0.3) | 27 (0.3) | 46 (0.3) | 0.2 | 272 (0.3) |
| Rheumatologic disease | 449 (1.2) | 71 (1.1) | 522 (1.4) | 158 (1.5) | 196 (1.5) | 0.007 | 1396 (1.3) |
| Cancer | 1736 (4.7) | 337 (5.1) | 1796 (4.9) | 571 (5.3) | 701 (5.3) | 0.02 | 5141 (4.9) |
| Net worth range (%) | |||||||
| <$25 000 | 1831 (4.9) | 404 (6.1) | 2101 (5.7) | 561 (5.2) | 695 (5.2) | <0.001 | 5592 (5.3) |
| $25 000–149 000 | 5188 (14) | 1042 (16) | 5608 (15) | 1591 (15) | 2046 (15) | 15 475 (15) | |
| $150 000–249 000 | 4754 (13) | 837 (13) | 5182 (14) | 1499 (14) | 1862 (14) | 14 134 (14) | |
| $250 000–499 000 | 9694 (26) | 1779 (27) | 10 048 (28) | 3027 (28) | 3636 (27) | 28 184 (27) | |
| ≥$500 000 | 10 910 (29) | 2030 (30) | 10 860 (30) | 3208 (30) | 3852 (29) | 30 860 (30) | |
| Unknown/missing | 4840 (13) | 580 (8.7) | 2804 (7.7) | 886 (8.2) | 1229 (9.2) | 10 339 (9.9) | |
| 30-d complications (%) | |||||||
| Any | 5973 (16) | 1151 (17) | 6120 (17) | 1863 (17) | 2451 (18) | <0.001 | 17 558 (17) |
| Bleeding | 2547 (6.8) | 499 (7.5) | 2660 (7.3) | 860 (8.0) | 1083 (8.1) | <0.001 | 7649 (7.3) |
| Nonsepsis infection | 2978 (8.0) | 590 (8.8) | 2990 (8.2) | 875 (8.1) | 1236 (9.3) | <0.001 | 8669 (8.3) |
| Sepsis | 191 (0.5) | 50 (0.7) | 349 (1.0) | 105 (1.0) | 134 (1.0) | <0.001 | 829 (0.8) |
| Urinary retention | 980 (2.6) | 161 (2.4) | 1138 (3.1) | 341 (3.2) | 422 (3.2) | <0.001 | 3042 (2.9) |
| Hospitalization | 1019 (2.7) | 181 (2.7) | 1056 (2.9) | 320 (3.0) | 367 (2.8) | 0.6 | 2943 (2.8) |
AIDS = acquired immunodeficiency syndrome; AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; IQR = interquartile range; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer [screening trial]; PSA = prostate-specific antigen; USPTF = US Preventive Services Task Force.
Overall, 30-d complications occurred in 17% of biopsies. The most prevalent was nonsepsis genitourinary infections (8.3%), followed by bleeding complications (7.3%) and urinary retention (2.9%) (Table 1). The 30-d hospitalization rate was 2.8%.
We first examined unadjusted rates of prostate biopsy among men aged ≥40 yr (Fig. 1 and Table 2). From January 2005 to September 2014, biopsy rates fell 33% from 64.1 to 42.8 per 100 000 person-months. There was substantial seasonal variation in biopsy rates with reductions near the end of each calendar year. With the ITS approach, each intervention is potentially associated with an immediate change in event rates (ie, level shift or change in intercept) as well as a change in the rate of growth (ie, change in slope). At baseline, there was a statistically significant growth in prostate biopsy rates of 2.9 biopsies per 100 000 person-months per year (95% confidence interval [CI], 0.96–4.8; p < 0.001). Following the 2008 USPSTF recommendations, there was an immediate decrease (ie, level shift) in biopsy rates of −10.1 biopsies per 100 000 person-months (95% CI, −17.1 to −3.0; p < 0.001). Likewise, there were immediate decreases in biopsy rates following the 2012 USPSTF recommendations (−13.8 biopsies per 100 000 person-months; 95% CI, −21.0 to −6.7; p < 0.001) and 2013 AUA guidelines (−8.8 biopsies per 100 000 person-months; 95% CI, −16.7 to −0.92; p = 0.03). Analyses adjusted for case mix were similar, although the baseline growth rate was no longer significant, indicating this was related to population shifts over time (Table 2, model 2).
Fig. 1
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 2
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315)
| Prostate biopsy | Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Model 1* | ||||||||||||||
| Baseline rate | 57.8 (53.4–62.3) |
<0.001 | 8.6 (7.5–9.7) |
<0.001 | 4.0 (3.3–4.6) |
<0.001 | 0.17 (0.03–0.32) |
0.02 | 2.8 (2.3–3.3) |
<0.001 | 1.5 (1.3–1.6) |
<0.001 | 1.4 (1.1–1.7) |
<0.001 |
| Baseline trend in rate | 2.9 (0.96–4.8) |
<0.001 | 0.84 (0.48–1.3) |
<0.001 | 0.72 (0.48–1.1) |
<0.001 | 0.12 (0.12–0.12) |
<0.001 | 0.36 (0.12–0.48) |
<0.001 | 0.12 (0.00–0.12) |
<0.001 | 0.00 (−0.12 to 0.12) |
0.5 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.1 (−17.1 to −3.0) |
<0.001 | −0.61 (−3.1 to 1.9) |
0.6 | −1.4 (−2.8 to −0.03) |
0.053 | 0.04 (−0.35 to 0.42) |
0.8 | 0.56 (−0.70 to 1.8) |
0.4 | −0.06 (−0.69 to 0.57) |
0.9 | 0.16 (−0.58 to 0.90) |
0.7 |
| Trend change | 7.9 (−12.7 to 28.4) |
0.5 | −1.0 (−9.1 to 7.2) |
0.8 | −0.12 (−4.1 to 4.0) |
0.98 | −0.60 (−1.9 to 0.72) |
0.4 | −2.0 (−6.5 to 2.4) |
0.4 | −1.1 (−3.4 to 1.3) |
0.4 | −1.0 (−3.6 to 1.6) |
0.5 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.6 (−12 to 4.5) |
0.4 | −0.74 (−3.8 to 2.4) |
0.6 | −0.12 (−1.8 to 1.5) |
0.9 | 0.29 (−0.20 to 0.78) |
0.2 | −0.04 (−1.6 to 1.6) |
0.97 | 0.59 (−0.23 to 1.4) |
0.2 | 0.51 (−0.43 to 1.5) |
0.3 |
| Trend change | −11.2 (−32.0 to 9.6) |
0.3 | 0.0 (−8.2 to 8.0) |
0.99 | −0.96 (−5.0 to 3.1) |
0.7 | 0.36 (−0.96 to 1.8) |
0.5 | 1.7 (−2.8 to 6.1) |
0.5 | 1.0 (−1.4 to 3.4) |
0.4 | 0.84 (−1.7 to 3.5) |
0.5 |
| 2012 USPSTF | ||||||||||||||
| Level change | −13.8 (−21.0 to −6.7) |
<0.001 | −2.5 (−4.5 to −0.45) |
0.02 | −1.7 (−3.1 to −0.31) |
0.02 | −0.12 (−0.42 to 0.19) |
0.5 | −0.61 (−1.5 to 0.28) |
0.2 | −0.75 (−1.1 to −0.44) |
<0.001 | −0.25 (−0.77 to 0.27) |
0.4 |
| Trend change | 8.2 (−5.8 to 22.0) |
0.3 | 3.0 (−0.36 to 6.4) |
0.09 | 3.1 (0.72–5.5) |
0.01 | 0.12 (−0.36 to 0.60) |
0.7 | 0.84 (−0.72 to 2.3) |
0.3 | 0.48 (0.00 to 1.1) |
0.08 | 0.24 (−0.60 to 1.2) |
0.6 |
| 2013 AUA | ||||||||||||||
| Level change | −8.8 (−16.7 to −0.92) |
0.03 | −1.4 (−3.8 to 1.1) |
0.3 | −1.9 (−3.5 to −0.22) |
0.03 | −0.02 (−0.41 to 0.37) |
0.9 | −0.19 (−1.3 to 0.91) |
0.7 | −0.38 (−0.77 to 0.01) |
0.06 | −0.01 (−0.66 to 0.63) |
0.96 |
| Trend change | −10.4 (−30.0 to 9.2) |
0.3 | −4.1 (−7.9 to −0.24) |
0.04 | −2.9 (−6.0 to 0.12) |
0.07 | −0.24 (−0.72 to 0.36) |
0.5 | −1.8 (−3.5 to 0.0) |
0.046 | −0.60 (−1.2 to 0.12) |
0.09 | −0.60 (−1.7 to 0.36) |
0.2 |
| Model 2† | ||||||||||||||
| Baseline rate | 86.8 (82.1–91.4) |
<0.001 | 16.2 (14.9–17.6) |
<0.001 | 6.4 (4.9–7.9) |
<0.001 | 0.39 (−0.13 to 0.91) |
0.2 | 6.1 (4.6–7.7) |
<0.001 | 3.6 (2.8–4.4) |
<0.001 | 3.0 (2.2–3.8) |
<0.001 |
| Baseline trend in rate | 0.24 (−1.8 to 2.3) |
0.8 | −0.24 (−0.72 to 0.24) |
0.4 | 0.48 (−0.12 to 1.1) |
0.09 | 0.12 (−0.12 to 0.24) |
0.4 | −0.24 (−0.72 to 0.24) |
0.4 | −0.24 (−0.48 to 0.00) |
0.054 | −0.12 (−0.36 to 0.12) |
0.3 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.8 (−17.8 to −3.8) |
<0.001 | −5.4 (−9.6 to −1.2) |
0.01 | −2.5 (−5.6 to 0.71) |
0.1 | −0.61 (−1.9 to 0.71) |
0.4 | 0.14 (−3.8 to 4.1) |
0.9 | 1.1 (−0.98 to 3.1) |
0.3 | 0.14 (−1.9 to 2.1) |
0.9 |
| Trend change | 10.6 (−10.1 to 31.2) |
0.3 | 6.1 (−8.4 to 20.8) |
0.4 | −1.7 (−12.2 to 8.8) |
0.8 | 1.7 (−2.9 to 6.2) |
0.5 | 1.7 (−12.0 to 15.5) |
0.8 | −4.7 (−11.8 to 2.5) |
0.2 | −0.84 (−7.8 to 6.1) |
0.8 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.9 (−12 to 4.1) |
0.3 | −0.24 (−5.6 to 5.1) |
0.9 | 1.9 (−2.1 to 5.8) |
0.4 | −0.32 (−2.0 to 1.4) |
0.7 | −0.99 (−6.0 to 4.0) |
0.7 | 1.8 (−0.84 to 4.4) |
0.2 | 0.87 (−1.7 to 3.4) |
0.5 |
| Trend change | −16.2 (−37.1 to 4.7) |
0.1 | −6.8 (−21.4 to 7.8) |
0.4 | 0.36 (−10.2 to 10.9) |
0.95 | −1.9 (−6.5 to 2.8) |
0.4 | −1.6 (−15.2 to 12.2) |
0.8 | 4.7 (−2.4 to 11.9) |
0.2 | 0.48 (−6.4 to 7.4) |
0.9 |
| 2012 USPSTF | ||||||||||||||
| Level change | −11.0 (−18.1 to −4.0) |
0.001 | −2.8 (−6.0 to 0.38) |
0.09 | 0.53 (−1.9 to 3.0) |
0.7 | −0.02 (−0.95 to 0.91) |
0.97 | −1.8 (−4.6 to 0.96) |
0.2 | −1.5 (−3.1 to 0.17) |
0.08 | 0.16 (−1.3 to 1.6) |
0.8 |
| Trend change | 7.0 (−7.2 to 21.1) |
0.3 | 1.7 (−3.6 to 7.0) |
0.5 | 0.12 (−4.1 to 4.2) |
0.96 | 0.12 (−1.4 to 1.8) |
0.9 | −0.48 (−5.3 to 4.3) |
0.9 | 1.6 (−1.3 to 4.3) |
0.3 | 0.12 (−2.4 to 2.6) |
0.9 |
| 2013 AUA | ||||||||||||||
| Level change | −8.2 (−15.9 to −0.38) |
0.04 | −1.0 (−4.8 to 2.8) |
0.6 | −0.44 (−3.2 to 2.3) |
0.8 | −0.15 (−1.3 to 0.99) |
0.8 | 0.28 (−3.2 to 3.7) |
0.9 | −0.50 (−2.7 to 1.7) |
0.7 | 0.47 (−1.3 to 2.3) |
0.6 |
| Trend change | −4.2 (−24.5 to 16.1) |
0.7 | −0.7 (−7.6 to 6.2) |
0.9 | 1.1 (−4.2 to 6.2) |
0.7 | −0.12 (−1.8 to 1.7) |
0.9 | −0.60 (−5.9 to 4.8) |
0.8 | −1.8 (−4.9 to 1.2) |
0.3 | −0.48 (−3.1 to 2.2) |
0.7 |
* Model 1: Base model with constant, time trend, intervention indicator variables. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
† Model 2: Adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, and anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPTF = US Preventive Services Task Force.
Next, we examined 30-d postbiopsy complications among men aged ≥40 yr as a measure of the absolute morbidity of biopsy (Fig. 1, Supplementary Fig. 3, and Table 2). Across the study periods, overall complication rates decreased 10% from 8.7 to 7.8 per 100 000 person-months. In unadjusted ITS analyses, there was a baseline growth in the overall complication rate (0.84 per 100 000 person-months per year; 95% CI, 0.48–1.3; p < 0.001), driven by growth in the rate of nonsepsis infections (0.72 per 100 000 person-months per year; 95% CI, 0.48–1.1; p < 0.001). Only the 2012 USPSTF recommendations were associated with a significant change in complication rates, following which there was an immediate decrease in the overall complication rate of −2.5 per 100 000 person-months (95% CI, −4.5 to −0.45; p = 0.02), driven by reductions in nonsepsis infections (−1.7 per 100 000 person-months; 95% CI, −3.1 to −0.31; p = 0.02) and urinary retention (−0.75 per 100 000 person-months;95% CI, −1.1 to −0.44; p < 0.001). Following the 2013 AUA guidelines there was also a reduction in the overall complication baseline growth rate (−4.1 per 100 000 person-months per year; 95% CI, −7.9 to −0.24; p = 0.04). In analyses adjusted for case mix (Table 2, model 2), effects were overall attenuated, and only the 2008 USPSTF recommendations were associated with a reduction in overall complication rates (−5.4 per 100 000 person-months; 95% CI, −9.6 to −1.2; p = 0.01).
To examine whether there was a change in the relative morbidity of biopsy, we performed a secondary analysis examining complication rates among men undergoing biopsy (Fig. 2). The proportion of men undergoing biopsy who experienced a complication increased from 14% in January 2005 to 18% in September 2014, an increase of 36%. There was again a baseline growth in overall complication rates (0.72% per year; 95% CI, 0.36–1.1; p < 0.001), driven by an increase in nonsepsis infections and sepsis (Supplementary Table 2). None of the interventions affected this growth. Similar results were obtained after adjusting for case mix (Table 3).
Fig. 2
Interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 3
Adjusted* interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Baseline rate | 15.6 (14.3–16.8) |
<0.001 | 7.1 (6.5–7.7) |
<0.001 | 0.34 (0.11–0.57) |
<0.001 | 6.7 (5.9–7.5) |
<0.001 | 2.8 (2.5–3.2) |
<0.001 | 3.2 (2.8–3.6) |
<0.001 |
| Baseline trend in rate | 0.60 (0.12–1.1) |
0.02 | 0.72 (0.60–0.96) |
<0.001 | 0.12 (0.12–0.24) |
<0.001 | 0.12 (−0.12 to 0.36) |
0.3 | −0.12 (−0.24 to 0.00) |
0.3 | −0.12 (−0.24 to 0.00) |
0.0498 |
| 2008 USPSTF | ||||||||||||
| Level change | 1.1 (−2.0 to 4.1) |
0.5 | −1.3 (−3.2 to 0.58) |
0.2 | 0.13 (−0.45 to 0.72) |
0.7 | 1.9 (−0.11 to 3.9) |
0.07 | 0.46 (−0.50 to 1.4) |
0.4 | 0.27 (−0.69 to 1.2) |
0.6 |
| Trend change | −3.5 (−13.4 to 6.6) |
0.5 | −1.8 (−8.4 to 4.7) |
0.6 | −1.0 (−3.0 to 0.96) |
0.3 | −4.3 (−11.4 to 2.6) |
0.2 | −1.6 (−4.9 to 1.9) |
0.4 | −0.72 (−4.0 to 2.6) |
0.7 |
| ERSPC/PLCO | ||||||||||||
| Level change | −0.14 (−3.9 to 3.6) |
0.9 | 1.1 (−1.3 to 3.4) |
0.4 | 0.55 (−0.19 to 1.3) |
0.2 | 0.39 (−2.2 to 3.0) |
0.8 | 0.75 (−0.47 to 2.0) |
0.2 | 0.65 (−0.56 to 1.9) |
0.3 |
| Trend change | 2.9 (−7.1 to 13.0) |
0.6 | 0.72 (−5.8 to 7.2) |
0.8 | 0.84 (−1.2 to 2.9) |
0.4 | 4.2 (−2.8 to 11.3) |
0.2 | 1.7 (−1.8 to 5.0) |
0.4 | 0.72 (−2.6 to 4.0) |
0.7 |
| 2012 USPSTF | ||||||||||||
| Level change | −0.27 (−2.7 to 2.1) |
0.8 | −0.56 (−2.0 to 0.87) |
0.5 | 0.16 (−0.31 to 0.63) |
0.5 | 1.1 (−0.38 to 2.5) |
0.2 | −0.38 (−1.0 to 0.28) |
0.3 | 0.44 (−0.34 to 1.2) |
0.3 |
| Trend change | 1.3 (−2.6 to 5.3) |
0.5 | 2.9 (0.60–5.3) |
0.02 | −0.24 (−1.1 to 0.60) |
0.6 | −0.72 (−3.2 to 1.7) |
0.5 | 0.24 (−0.96 to 1.3) |
0.7 | −0.48 (−1.8 to 0.84) |
0.5 |
| 2013 AUA | ||||||||||||
| Level change | 0.25 (−2.6 to 3.1) |
0.9 | −0.64 (−2.3 to 1.1) |
0.5 | 0.22 (−0.40 to 0.83) |
0.5 | 1.1 (−0.68 to 2.8) |
0.2 | −0.29 (−1.1 to 0.48) |
0.5 | 0.47 (−0.55 to 1.5) |
0.4 |
| Trend change | −1.6 (−6.1 to 2.9) |
0.5 | −2.5 (−5.5 to 0.60) |
0.1 | 0.12 (−0.72 to 1.1) |
0.8 | −0.36 (−3.1 to 2.4) |
0.8 | 0.24 (−1.2 to 1.7) |
0.8 | 0.12 (−1.3 to 1.6) |
0.9 |
* Model adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 person-months. Baseline trend in rate and trend change per 100 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial]; USPTF = US Preventive Services Task Force.
Given the increase in the relative morbidity of biopsy, we performed logistic regression to identify predictors of each complication category (Supplementary Table 3 and Table 4). Due to the large sample size, we focused on those features associated with the greatest magnitude of effect. A prior diagnosis of cancer was associated with an increased risk of all complication categories including any complication (odds ratio [OR]: 1.30; 95% CI, 1.22–1.39; p < 0.001) and hospitalization (OR: 1.59; 95% CI, 1.39–1.82; p < 0.001). Similarly, prior fluoroquinolone use was associated with an increased risk of all complication categories including any complication (OR: 1.27; 95% CI, 1.22–1.32; p < 0.001) and sepsis (OR: 1.55; 95% CI, 1.33–1.81; p < 0.001). Age at biopsy ≥70 yr was associated with a nearly 4-fold increased risk of urinary retention. Although anticoagulant use was associated with increased risks of bleeding (OR: 1.28; 95% CI, 1.14–1.45; p < 0.001), sepsis (OR: 1.48; 95% CI, 1.06–2.07; p = 0.02), and hospitalization (OR: 1.39; 95% CI, 1.16–1.67; p < 0.001), the increase in the overall complication risk was modest (OR 1.14; 95% CI, 1.04–1.25; p = 0.004).
Table 4
Multivariable analysis* of 30-d postbiopsy complications among men aged ≥40 yr undergoing prostate biopsy (January 2005 to September 2014) (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Hospitalization | Urinary retention | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Age at biopsy, yr | ||||||||||||
| 40–49 | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| 50–59 | 1.01 (0.94–1.09) | 0.8 | 0.94 (0.85–1.04) | 0.2 | 1.16 (0.83–1.63) | 0.4 | 1.01 (0.90–1.14) | 0.8 | 1.12 (0.93–1.36) | 0.2 | 1.54 (1.20–1.97) | <0.001 |
| 60–69 | 1.05 (0.97–1.13) | 0.2 | 0.94 (0.85–1.03) | 0.2 | 0.98 (0.70–1.37) | 0.9 | 1.05 (0.94–1.18) | 0.4 | 1.16 (0.95–1.40) | 0.1 | 2.16 (1.68–2.77) | <0.001 |
| ≥70 | 1.25 (1.15–1.36) | <0.001 | 1.03 (0.93–1.15) | 0.6 | 1.05 (0.74–1.49) | 0.8 | 1.29 (1.15–1.46) | <0.001 | 1.31 (1.07–1.60) | 0.008 | 3.98 (3.08–5.15) | <0.001 |
| Race | ||||||||||||
| White | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| Asian | 1.27 (1.15–1.40) | <0.001 | 1.27 (1.11–1.44) | 0.0003 | 1.65 (1.19–2.29) | 0.003 | 1.36 (1.18–1.56) | <0.001 | 0.94 (0.76–1.17) | 0.6 | 1.15 (0.93–1.42) | 0.2 |
| African American | 1.09 (1.03–1.15) | 0.002 | 1.05 (0.97–1.13) | 0.2 | 1.34 (1.09–1.64) | 0.005 | 1.17 (1.09–1.27) | <0.001 | 1.13 (1.01–1.27) | 0.03 | 1.15 (1.03–1.28) | 0.02 |
| Hispanic | 1.32 (1.24–1.40) | <0.001 | 1.36 (1.26–1.47) | <0.001 | 1.52 (1.21–1.90) | <0.001 | 1.33 (1.22–1.45) | <0.001 | 1.12 (0.97–1.28) | 0.1 | 1.37 (1.21–1.56) | <0.001 |
| Unknown/missing | 0.99 (0.93–1.06) | 0.8 | 1.07 (0.99–1.17) | 0.1 | 1.23 (0.94–1.61) | 0.1 | 0.96 (0.88–1.05) | 0.4 | 0.94 (0.81–1.08) | 0.4 | 1.02 (0.88–1.18) | 0.8 |
| Diabetes mellitus | 1.03 (0.99–1.07) | 0.09 | 1.06 (1.00–1.11) | 0.045 | 1.08 (0.92–1.28) | 0.4 | 1.01 (0.96–1.07) | 0.7 | 1.20 (1.10–1.31) | <0.001 | 0.99 (0.91–1.08) | 0.8 |
| COPD | 1.14 (1.09–1.20) | <0.001 | 1.07 (1.00–1.14) | 0.04 | 1.23 (1.01–1.49) | 0.04 | 1.14 (1.06–1.22) | <0.001 | 1.36 (1.23–1.51) | <0.001 | 1.14 (1.03–1.25) | 0.01 |
| PVD | 1.24 (1.16–1.31) | <0.001 | 1.1 (1.02–1.2) | 0.02 | 1.19 (0.93–1.51) | 0.2 | 1.28 (1.18–1.40) | <0.001 | 1.31 (1.15–1.48) | <0.001 | 1.26 (1.12–1.42) | <0.001 |
| Cerebrovascular disease | 1.11 (1.04–1.19) | 0.001 | 1.12 (1.03–1.22) | 0.01 | 1.21 (0.93–1.57) | 0.2 | 1.03 (0.94–1.14) | 0.5 | 1.26 (1.10–1.44) | <0.001 | 1.30 (1.15–1.47) | <0.001 |
| Cancer | 1.30 (1.22–1.39) | <0.001 | 1.16 (1.06–1.27) | 0.001 | 1.37 (1.05–1.79) | 0.02 | 1.40 (1.27–1.53) | <0.001 | 1.59 (1.39–1.82) | <0.001 | 1.32 (1.16–1.52) | <0.001 |
| Anticoagulant use | 1.14 (1.04–1.25) | 0.004 | 1.01 (0.89–1.14) | 0.9 | 1.48 (1.06–2.07) | 0.02 | 1.28 (1.14–1.45) | <0.001 | 1.39 (1.16–1.67) | <0.001 | 1.11 (0.92–1.33) | 0.3 |
| Prior fluoroquinolones | 1.27 (1.22–1.32) | <0.001 | 1.37 (1.31–1.45) | <0.001 | 1.55 (1.33–1.81) | <0.001 | 1.21 (1.14–1.28) | <0.001 | 1.25 (1.14–1.36) | <0.001 | 1.27 (1.16–1.38) | <0.001 |
* Models adjusted for year of biopsy and the features listed in the table.
CI = confidence interval; COPD = chronic obstructive pulmonary disease; PVD = peripheral vascular disease.
We conducted sensitivity analyses where the postintervention run-in period was varied from 2 to 4 mo (Supplementary Table 4). Results overall were similar, although effects were attenuated for the 2013 AUA guidelines on biopsy rates and the 2012 USPSTF recommendations on overall complication rates, likely due to fewer time points for analysis. In a separate sensitivity analysis, we examined bleeding complications excluding microhematuria from 2009 to 2014, following introduction of a separate code for microhematuria. Bleeding rates paralleled those of the primary bleeding end point (Supplementary Fig. 4), and results of both ITS and logistic regression analyses were similar (Supplementary Table 5 and 6). As an exploratory analysis, we examined PCa diagnosis rates within 60 d following biopsy (Supplementary Fig. 5). These appeared to increase steadily across the study period.
To our knowledge, this is the largest study to examine the impact of landmark PSA screening publications on rates of prostate biopsy and the first to examine their impact on postbiopsy complications. Overall, biopsy rates have declined by 33%, which appears attributable to the publication of the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines. Interestingly, the magnitude of reduction in biopsy rates exceeded previously reported decreases in PSA screening rates [6], [7], [8], [9], [10], [11], [12], [22], and [23], which suggests that either the threshold for referral to a urologist or the threshold for performing a biopsy has changed. More importantly, there was a concurrent 22% decrease in 30-d complication rates that appears largely attributable to the 2012 USPSTF recommendations. Still, the reduction in complication rates was less than the reduction in biopsy rates due to an increase in the relative morbidity of biopsy, driven by growth in infectious complications. This may in part be related to a shift to older patients with greater comorbidity undergoing biopsy, although similar results were obtained in analyses adjusted for case mix, which suggests additional factors may exist.
To this end, we evaluated predictors of postbiopsy complications to identify actionable targets for further risk reduction. Some of these features represent modifiable risk factors (eg, anticoagulant use), and others represent opportunities for altering periprocedural management. For instance, prior fluoroquinolone use may identify patients who would benefit from prebiopsy rectal swabs to guide periprocedural antibiotics [24] and [25]. Nonmodifiable risk factors, including specific comorbidities such as cancer and peripheral vascular disease, may encourage reconsideration of the risk–benefit ratio before proceeding with biopsy.
It is also important to consider the concurrent impact of landmark PSA screening publications on rates of PCa diagnosis to contextualize the observed changes in biopsy and its morbidity. An exploratory analysis suggested that cancer diagnosis rates have steadily increased, although further studies are necessary to examine this issue comprehensively.
A number of studies have examined the impact of the 2008 USPSTF recommendations [7], [8], [9], and [22], ERSPC/PLCO trials [6] and [9], and 2012 USPSTF recommendations [6], [10], [11], [12], and [23] on PSA screening rates, generally noting reductions ranging from 0% to 8%. However, much less data are available on the impact of these publications on biopsy rates. One single-center study noted a 31% reduction in the number of biopsies performed following the 2012 USPSTF recommendations [13]; another noted no change in the proportion of referred men who underwent biopsy [26]. A report from a Canadian academic health network noted a reduction in biopsy rates from 58.0 to 35.5 biopsies per month following the 2012 USPSTF recommendations [27]. Our findings therefore confirm a reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines.
Limited data exist regarding secular trends in postbiopsy complications, and prior studies have focused exclusively on the relative morbidity among men undergoing biopsy rather than absolute complication rates. A Canadian population-based study observed that the 30-d hospitalization rate increased from 1.0% in 1996 to 4.1% in 2005, driven by an increase in infectious complications [28]. A Surveillance, Epidemiology and End Results–Medicare study also noted an increase in hospitalizations related to infectious complications over time [14]. Although we observed a similar increase in 30-d overall complications among men undergoing biopsy, likewise driven by infectious complications, there was an absolute 22% reduction in 30-d complications over the study period. Taken together, these findings support a reduction in the morbidity of biopsy following publication of revised PSA screening guidelines but highlight the continued need to reduce the relative morbidity of biopsy.
A number of studies have described complication rates following prostate biopsy [15], [16], [29], and [30]. Hematuria has been reported in 10–84% of biopsies, rectal bleeding in 1–45%, infectious complications in up to 6.3%, urinary retention in up to 1.7%, and hospitalization in up to 6.9% [14] and [29]. The wide range in complication rates is due to differences in definitions, ascertainment of end points, and variation in clinical practice. Given the heterogeneity in postbiopsy complication rates, our findings have two advantages. First, they represent a contemporary, geographically diverse population across the entire United States. Second, use of claims-based ascertainment limited complications to those cases in which patients sought medical care, and although this may underestimate the comprehensive morbidity of biopsy, it more accurately reflects the economic burden of serious complications.
This study has several limitations. Most importantly, a control population of men who were not exposed to the PSA screening publications was not available. This study is retrospective, and additional, unrecognized changes in policy or clinical practice may possibly have confounded the impact of the interventions examined. However, the ITS approach has been described as the strongest quasi-experimental design to evaluate policy changes when a randomized trial is not possible [19]. Because we relied on administrative claims for ascertainment of complications, minor complications that did not generate a claim were not captured, thus underestimating these end points. Although we adjusted for case mix, we were unable to adjust for PSA. However, population shifts in PSA may be confounded by changes in screening practices, and PSA adjustment may therefore under- or overestimate the impact of interventions. We were also unable to adjust for other factors that may be associated with biopsy or complication rates, such as prostate magnetic resonance imaging (MRI), ancillary biomarker testing, biopsy approach (eg, MRI-fusion biopsy, transperineal biopsy), indication for anticoagulation, use of periprocedural bridging, or operator biopsy volume/experience.
We observed an immediate reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines, with a smaller reduction in absolute 30-d complication rates. Still, the relative morbidity of biopsy has increased over time, driven by infectious complications. Accordingly, if health policy aims to reduce the morbidity of PSA screening related to biopsy, further efforts should focus on improving patient selection and periprocedural processes of care.
Author contributions: Boris Gershman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Gershman, van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Acquisition of data: Van Houten, Shah.
Analysis and interpretation of data: Gershman, Van Houten, Shah, Karnes.
Drafting of the manuscript: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Critical revision of the manuscript for important intellectual content: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Statistical analysis: Van Houten.
Obtaining funding: Karnes, Shah.
Administrative, technical, or material support: Karnes, Shah.
Supervision: Karnes, Shah.
Other (specify): None.
Financial disclosures: Boris Gershman certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: This publication was made possible by funding from the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery.
Funding/Support and role of the sponsor: None.
In 2009, two landmark randomized trials—the European Randomized Study of Screening for Prostate Cancer [ERSPC] [1] and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial [PLCO] [2]—reported smaller than anticipated reductions in prostate cancer (PCa) mortality and prompted reexamination of prostate-specific antigen (PSA) screening recommendations. The US Preventive Services Task Force (USPSTF), which in 2008 had recommended against PSA screening in men aged >75 yr [3], updated its position in 2012 to recommend against systematic PSA screening in the general population [4]. The American Urological Association (AUA) likewise updated its PSA screening guidelines in May 2013, recommending shared decision making for men aged 55–69 yr [5]. Both the 2012 USPSTF and 2013 AUA guidelines cited prostate biopsy and its potential complications as a significant harm associated with PSA screening.
Although several studies have noted a reduction in PSA screening rates following these publications [6], [7], [8], [9], [10], [11], and [12], there are little data on whether this has translated into the intended reductions in screening-related morbidity in the form of prostate biopsy and postbiopsy complications [13]. We evaluated whether publication of the 2008 USPSTF recommendations, 2009 ERSPC/PLCO trials, 2012 USPSTF recommendations, or 2013 AUA guidelines was associated with reductions in prostate biopsy or postbiopsy complications. Biopsy-related complications also remain understudied both with regard to incidence as well as identification of risk factors [14], [15], and [16]. Consequently, we also sought to characterize postbiopsy complications and evaluate predictors using a contemporary population-based cohort.
We conducted a retrospective analysis of administrative claims from the Optum Labs Data Warehouse that contains data on >100 million individuals enrolled in private health plans and Medicare Advantage Plans, from geographically diverse regions across the United States, as described in Supplementary Methods [17]. Because this study involved analysis of preexisting deidentified data, it was exempt from institutional review board approval.
We identified all prostate biopsies performed in men aged ≥40 yr between January 1, 2005, and September 30, 2014, using Current Procedural Terminology-4 code 55700 alone (11%) or in combination with 76872 (84%) or 76942 (0%), or International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 60.11 (5%). The first biopsy was selected as the index biopsy when more than one biopsy was performed. Patients with a prior diagnosis of PCa (ICD-9 code 185) were excluded. We restricted the cohort to patients with continuous medical and pharmacy enrollment for at least 12 mo prior to biopsy and 30 d after biopsy to ascertain comorbidities and 30-d complications (Supplementary Fig. 1).
We recorded sociodemographic characteristics, comorbid conditions, anticoagulant use within 30 d of biopsy, and fluoroquinolone use in the 12 mo prior to biopsy as described in Supplement 1. The Charlson Comorbidity Index was calculated as previously described [18]. Prebiopsy PSA values within 60 d of biopsy were available for a subset of patients.
The primary outcomes were crude and adjusted rates of prostate biopsy and 30-d complications. Complications were identified using inpatient and outpatient administrative claims as summarized in Supplementary Table 1.
Baseline demographic features were compared across study periods using the Kruskal-Wallis test for continuous variables and the chi-square test for discrete variables. To evaluate the impact of each publication on rates of prostate biopsy and 30-d complications, we used an interrupted time series (ITS) approach as described in Supplementary Methods [19] and [20]. Rates were estimated using all men aged ≥40 yr with at least 12 mo of continuous medical and pharmacy enrollment as the denominator. As secondary analyses, we estimated rates of 30-d complications among men who underwent biopsy. Two separate ITS analyses were performed: (1) unadjusted analyses, using crude rates of biopsy and complications aggregated for each month, and (2) analyses standardized for case mix, to adjust for population shifts over time as described in Supplementary Methods [20] and [21].
Logistic regression was used to evaluate predictors of 30-d complications. Multivariable models were adjusted for age at biopsy, year of biopsy, race, anticoagulant use, history of fluoroquinolone use, and the presence of the five most prevalent Charlson comorbidities.
Statistical analyses were performed using SAS v.9.3 (SAS Institute Inc., Cary, NC, USA). All tests were two sided with p < 0.05 considered statistically significant.
Between January 2005 and September 2014, a total of 104 584 men underwent prostate biopsy (Table 1). Median age at biopsy was 63.0 yr (interquartile range [IQR]: 57.0–70.0). Overall, 18% of men had used fluoroquinolones in the 12 mo preceding biopsy, and 2.8% were on anticoagulants. Median prebiopsy PSA was 5.1 ng/dl (IQR: 4.0–7.1) for the 18 749 men (18%) for whom this was available. Over the time periods, men undergoing biopsy were older, had increased prevalence of anticoagulant use and comorbid conditions, and had a higher PSA (Table 1 and Supplementary Fig. 2).
Table 1
Patient demographics for men who underwent prostate biopsy (January 2005 to September 2014) stratified by pre/postintervention time periods (n = 104 584)
| Baseline Jan 2005–Jul 2008 (n = 37 217) |
2008 USPSTF Aug 2008–Feb 2009 (n = 6672) |
ERSPC/PLCO Mar 2009–Apr 2012 (n = 36 603) |
2012 USPSTF May 2012–Apr 2013 (n = 10 772) |
2013 AUA May 2013–Sep 2014 (n = 13 320) |
p value | Total (n = 104 584) | |
|---|---|---|---|---|---|---|---|
| Age at biopsy, yr | |||||||
| Median (IQR) | 62.0 (56.0–68.0) | 62.0 (56.0–68.0) | 63.0 (57.0–70.0) | 65.0 (59.0–71.0) | 66.0 (59.0–71.0) | <0.001 | 63.0 (57.0–70.0) |
| Age group, yr (%) | |||||||
| 40–49 | 2269 (6.1) | 416 (6.2) | 1737 (4.7) | 420 (3.9) | 467 (3.5) | <0.001 | 5309 (5.1) |
| 50–59 | 12 366 (33) | 2201 (33) | 10 342 (28) | 2644 (25) | 3098 (23) | 30 651 (29) | |
| 60–69 | 14 671 (39) | 2697 (40) | 15 105 (41) | 4426 (41) | 5546 (42) | 42 445 (41) | |
| ≥70 | 7911 (21) | 1358 (20) | 9419 (26) | 3282 (31) | 4209 (32) | 26 179 (25) | |
| Race (%) | |||||||
| White | 25 508 (69) | 4643 (70) | 26 187 (72) | 7795 (72) | 9407 (71) | <0.001 | 73 540 (70) |
| African American | 3695 (9.9) | 780 (12) | 4423 (12) | 1283 (12) | 1589 (12) | 11 770 (11) | |
| Hispanic | 2549 (6.8) | 538 (8.1) | 2911 (8.0) | 837 (7.8) | 1186 (8.9) | 8021 (7.7) | |
| Asian | 1133 (3.0) | 236 (3.5) | 1142 (3.1) | 306 (2.8) | 387 (2.9) | 3204 (3.1) | |
| Unknown/Missing | 4332 (12) | 475 (7.1) | 1940 (5.3) | 551 (5.1) | 751 (5.6) | 8049 (7.7) | |
| Census region (%) | |||||||
| Northeast | 7094 (19) | 1196 (18) | 6120 (17) | 1886 (18) | 2424 (18) | <0.001 | 18 720 (18) |
| Midwest | 10 094 (27) | 1562 (23) | 8837 (24) | 2742 (26) | 3627 (27) | 26 862 (26) | |
| South | 15 330 (41) | 2889 (43) | 16 371 (45) | 4698 (44) | 5538 (42) | 44 826 (43) | |
| West | 4660 (13) | 993 (15) | 5147 (14) | 1442 (13) | 1728 (13) | 13 970 (13) | |
| Other/Missing | 39 (0.1) | 32 (0.5) | 128 (0.3) | 4 (0) | 3 (0) | 206 (0.2) | |
| PSA, ng/d | n = 5174 | n = 548 | n = 7345 | n = 2211 | n = 3471 | n = 18 749 | |
| Median (IQR) | 5.1 (3.9–7.1) | 4.9 (3.8–6.8) | 5.0 (4.0–6.8) | 5.3 (4.2–7.4) | 5.4 (4.2–7.8) | <0.001 | 5.1 (4.0–7.1) |
| Anticoagulant use (%) | 928 (2.5) | 184 (2.8) | 1103 (3.0) | 299 (2.8) | 415 (3.1) | <0.001 | 2929 (2.8) |
| History of fluoroquinolone use in prior 12 mo (%) | 6797 (18) | 1267 (19) | 6557 (18) | 2017 (19) | 2504 (19) | 0.054 | 19 142 (18) |
| Charlson Comorbidity Index (%) | |||||||
| 0 | 22 354 (60) | 3887 (58) | 20 269 (55) | 5758 (54) | 7039 (53) | <0.001 | 59 307 (57) |
| 1 | 8199 (22) | 1453 (22) | 8252 (23) | 2503 (23) | 2999 (23) | 23 406 (22) | |
| 2 | 3232 (8.7) | 664 (10) | 3520 (9.6) | 1108 (10) | 1359 (10) | 9883 (9.4) | |
| ≥3 | 3432 (9.2) | 668 (10) | 4562 (13) | 1403 (13) | 1923 (14) | 11 988 (12) | |
| Comorbidities (%) | |||||||
| Myocardial infarction | 763 (2.1) | 144 (2.2) | 812 (2.2) | 227 (2.1) | 273 (2.0) | 0.6 | 2219 (2.1) |
| Congestive heart failure | 1443 (3.9) | 223 (3.3) | 1548 (4.2) | 467 (4.3) | 586 (4.4) | <0.001 | 4267 (4.1) |
| Peripheral vascular disease | 2054 (5.5) | 399 (6.0) | 2660 (7.3) | 783 (7.3) | 1071 (8.0) | <0.001 | 6967 (6.7) |
| Cerebrovascular disease | 2007 (5.4) | 379 (5.7) | 2358 (6.4) | 681 (6.3) | 947 (7.1) | <0.001 | 6372 (6.1) |
| Dementia | 370 (1.0) | 63 (0.9) | 442 (1.2) | 141 (1.3) | 198 (1.5) | <0.001 | 1214 (1.2) |
| Chronic obstructive pulmonary disease | 4251 (11) | 758 (11) | 4293 (12) | 1254 (12) | 1686 (13) | 0.004 | 12 242 (12) |
| Ulcer | 362 (1.0) | 60 (0.9) | 297 (0.8) | 84 (0.8) | 102 (0.8) | 0.07 | 905 (0.9) |
| Mild liver disease | 1182 (3.2) | 223 (3.3) | 1332 (3.6) | 413 (3.8) | 515 (3.9) | <0.001 | 3665 (3.5) |
| Diabetes mellitus | 6475 (17) | 1290 (19) | 7823 (21) | 2442 (23) | 2973 (22) | <0.001 | 21 003 (20) |
| Diabetes with organ damage | 1105 (3.0) | 225 (3.4) | 1600 (4.4) | 492 (4.6) | 734 (5.5) | <0.001 | 4156 (4.0) |
| Hemiplegia | 128 (0.3) | 25 (0.4) | 150 (0.4) | 51 (0.5) | 56 (0.4) | 0.3 | 410 (0.4) |
| Moderate/severe renal disease | 1133 (3.0) | 254 (3.8) | 1950 (5.3) | 674 (6.3) | 908 (6.8) | <0.001 | 4919 (4.7) |
| Moderate/severe liver disease | 27 (0.1) | 7 (0.1) | 31 (0.1) | 9 (0.1) | 12 (0.1) | 0.9 | 86 (0.1) |
| Metastatic solid tumor | 175 (0.5) | 37 (0.6) | 199 (0.5) | 58 (0.5) | 80 (0.6) | 0.4 | 549 (0.5) |
| AIDS | 82 (0.2) | 16 (0.2) | 101 (0.3) | 27 (0.3) | 46 (0.3) | 0.2 | 272 (0.3) |
| Rheumatologic disease | 449 (1.2) | 71 (1.1) | 522 (1.4) | 158 (1.5) | 196 (1.5) | 0.007 | 1396 (1.3) |
| Cancer | 1736 (4.7) | 337 (5.1) | 1796 (4.9) | 571 (5.3) | 701 (5.3) | 0.02 | 5141 (4.9) |
| Net worth range (%) | |||||||
| <$25 000 | 1831 (4.9) | 404 (6.1) | 2101 (5.7) | 561 (5.2) | 695 (5.2) | <0.001 | 5592 (5.3) |
| $25 000–149 000 | 5188 (14) | 1042 (16) | 5608 (15) | 1591 (15) | 2046 (15) | 15 475 (15) | |
| $150 000–249 000 | 4754 (13) | 837 (13) | 5182 (14) | 1499 (14) | 1862 (14) | 14 134 (14) | |
| $250 000–499 000 | 9694 (26) | 1779 (27) | 10 048 (28) | 3027 (28) | 3636 (27) | 28 184 (27) | |
| ≥$500 000 | 10 910 (29) | 2030 (30) | 10 860 (30) | 3208 (30) | 3852 (29) | 30 860 (30) | |
| Unknown/missing | 4840 (13) | 580 (8.7) | 2804 (7.7) | 886 (8.2) | 1229 (9.2) | 10 339 (9.9) | |
| 30-d complications (%) | |||||||
| Any | 5973 (16) | 1151 (17) | 6120 (17) | 1863 (17) | 2451 (18) | <0.001 | 17 558 (17) |
| Bleeding | 2547 (6.8) | 499 (7.5) | 2660 (7.3) | 860 (8.0) | 1083 (8.1) | <0.001 | 7649 (7.3) |
| Nonsepsis infection | 2978 (8.0) | 590 (8.8) | 2990 (8.2) | 875 (8.1) | 1236 (9.3) | <0.001 | 8669 (8.3) |
| Sepsis | 191 (0.5) | 50 (0.7) | 349 (1.0) | 105 (1.0) | 134 (1.0) | <0.001 | 829 (0.8) |
| Urinary retention | 980 (2.6) | 161 (2.4) | 1138 (3.1) | 341 (3.2) | 422 (3.2) | <0.001 | 3042 (2.9) |
| Hospitalization | 1019 (2.7) | 181 (2.7) | 1056 (2.9) | 320 (3.0) | 367 (2.8) | 0.6 | 2943 (2.8) |
AIDS = acquired immunodeficiency syndrome; AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; IQR = interquartile range; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer [screening trial]; PSA = prostate-specific antigen; USPTF = US Preventive Services Task Force.
Overall, 30-d complications occurred in 17% of biopsies. The most prevalent was nonsepsis genitourinary infections (8.3%), followed by bleeding complications (7.3%) and urinary retention (2.9%) (Table 1). The 30-d hospitalization rate was 2.8%.
We first examined unadjusted rates of prostate biopsy among men aged ≥40 yr (Fig. 1 and Table 2). From January 2005 to September 2014, biopsy rates fell 33% from 64.1 to 42.8 per 100 000 person-months. There was substantial seasonal variation in biopsy rates with reductions near the end of each calendar year. With the ITS approach, each intervention is potentially associated with an immediate change in event rates (ie, level shift or change in intercept) as well as a change in the rate of growth (ie, change in slope). At baseline, there was a statistically significant growth in prostate biopsy rates of 2.9 biopsies per 100 000 person-months per year (95% confidence interval [CI], 0.96–4.8; p < 0.001). Following the 2008 USPSTF recommendations, there was an immediate decrease (ie, level shift) in biopsy rates of −10.1 biopsies per 100 000 person-months (95% CI, −17.1 to −3.0; p < 0.001). Likewise, there were immediate decreases in biopsy rates following the 2012 USPSTF recommendations (−13.8 biopsies per 100 000 person-months; 95% CI, −21.0 to −6.7; p < 0.001) and 2013 AUA guidelines (−8.8 biopsies per 100 000 person-months; 95% CI, −16.7 to −0.92; p = 0.03). Analyses adjusted for case mix were similar, although the baseline growth rate was no longer significant, indicating this was related to population shifts over time (Table 2, model 2).
Fig. 1
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 2
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315)
| Prostate biopsy | Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Model 1* | ||||||||||||||
| Baseline rate | 57.8 (53.4–62.3) |
<0.001 | 8.6 (7.5–9.7) |
<0.001 | 4.0 (3.3–4.6) |
<0.001 | 0.17 (0.03–0.32) |
0.02 | 2.8 (2.3–3.3) |
<0.001 | 1.5 (1.3–1.6) |
<0.001 | 1.4 (1.1–1.7) |
<0.001 |
| Baseline trend in rate | 2.9 (0.96–4.8) |
<0.001 | 0.84 (0.48–1.3) |
<0.001 | 0.72 (0.48–1.1) |
<0.001 | 0.12 (0.12–0.12) |
<0.001 | 0.36 (0.12–0.48) |
<0.001 | 0.12 (0.00–0.12) |
<0.001 | 0.00 (−0.12 to 0.12) |
0.5 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.1 (−17.1 to −3.0) |
<0.001 | −0.61 (−3.1 to 1.9) |
0.6 | −1.4 (−2.8 to −0.03) |
0.053 | 0.04 (−0.35 to 0.42) |
0.8 | 0.56 (−0.70 to 1.8) |
0.4 | −0.06 (−0.69 to 0.57) |
0.9 | 0.16 (−0.58 to 0.90) |
0.7 |
| Trend change | 7.9 (−12.7 to 28.4) |
0.5 | −1.0 (−9.1 to 7.2) |
0.8 | −0.12 (−4.1 to 4.0) |
0.98 | −0.60 (−1.9 to 0.72) |
0.4 | −2.0 (−6.5 to 2.4) |
0.4 | −1.1 (−3.4 to 1.3) |
0.4 | −1.0 (−3.6 to 1.6) |
0.5 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.6 (−12 to 4.5) |
0.4 | −0.74 (−3.8 to 2.4) |
0.6 | −0.12 (−1.8 to 1.5) |
0.9 | 0.29 (−0.20 to 0.78) |
0.2 | −0.04 (−1.6 to 1.6) |
0.97 | 0.59 (−0.23 to 1.4) |
0.2 | 0.51 (−0.43 to 1.5) |
0.3 |
| Trend change | −11.2 (−32.0 to 9.6) |
0.3 | 0.0 (−8.2 to 8.0) |
0.99 | −0.96 (−5.0 to 3.1) |
0.7 | 0.36 (−0.96 to 1.8) |
0.5 | 1.7 (−2.8 to 6.1) |
0.5 | 1.0 (−1.4 to 3.4) |
0.4 | 0.84 (−1.7 to 3.5) |
0.5 |
| 2012 USPSTF | ||||||||||||||
| Level change | −13.8 (−21.0 to −6.7) |
<0.001 | −2.5 (−4.5 to −0.45) |
0.02 | −1.7 (−3.1 to −0.31) |
0.02 | −0.12 (−0.42 to 0.19) |
0.5 | −0.61 (−1.5 to 0.28) |
0.2 | −0.75 (−1.1 to −0.44) |
<0.001 | −0.25 (−0.77 to 0.27) |
0.4 |
| Trend change | 8.2 (−5.8 to 22.0) |
0.3 | 3.0 (−0.36 to 6.4) |
0.09 | 3.1 (0.72–5.5) |
0.01 | 0.12 (−0.36 to 0.60) |
0.7 | 0.84 (−0.72 to 2.3) |
0.3 | 0.48 (0.00 to 1.1) |
0.08 | 0.24 (−0.60 to 1.2) |
0.6 |
| 2013 AUA | ||||||||||||||
| Level change | −8.8 (−16.7 to −0.92) |
0.03 | −1.4 (−3.8 to 1.1) |
0.3 | −1.9 (−3.5 to −0.22) |
0.03 | −0.02 (−0.41 to 0.37) |
0.9 | −0.19 (−1.3 to 0.91) |
0.7 | −0.38 (−0.77 to 0.01) |
0.06 | −0.01 (−0.66 to 0.63) |
0.96 |
| Trend change | −10.4 (−30.0 to 9.2) |
0.3 | −4.1 (−7.9 to −0.24) |
0.04 | −2.9 (−6.0 to 0.12) |
0.07 | −0.24 (−0.72 to 0.36) |
0.5 | −1.8 (−3.5 to 0.0) |
0.046 | −0.60 (−1.2 to 0.12) |
0.09 | −0.60 (−1.7 to 0.36) |
0.2 |
| Model 2† | ||||||||||||||
| Baseline rate | 86.8 (82.1–91.4) |
<0.001 | 16.2 (14.9–17.6) |
<0.001 | 6.4 (4.9–7.9) |
<0.001 | 0.39 (−0.13 to 0.91) |
0.2 | 6.1 (4.6–7.7) |
<0.001 | 3.6 (2.8–4.4) |
<0.001 | 3.0 (2.2–3.8) |
<0.001 |
| Baseline trend in rate | 0.24 (−1.8 to 2.3) |
0.8 | −0.24 (−0.72 to 0.24) |
0.4 | 0.48 (−0.12 to 1.1) |
0.09 | 0.12 (−0.12 to 0.24) |
0.4 | −0.24 (−0.72 to 0.24) |
0.4 | −0.24 (−0.48 to 0.00) |
0.054 | −0.12 (−0.36 to 0.12) |
0.3 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.8 (−17.8 to −3.8) |
<0.001 | −5.4 (−9.6 to −1.2) |
0.01 | −2.5 (−5.6 to 0.71) |
0.1 | −0.61 (−1.9 to 0.71) |
0.4 | 0.14 (−3.8 to 4.1) |
0.9 | 1.1 (−0.98 to 3.1) |
0.3 | 0.14 (−1.9 to 2.1) |
0.9 |
| Trend change | 10.6 (−10.1 to 31.2) |
0.3 | 6.1 (−8.4 to 20.8) |
0.4 | −1.7 (−12.2 to 8.8) |
0.8 | 1.7 (−2.9 to 6.2) |
0.5 | 1.7 (−12.0 to 15.5) |
0.8 | −4.7 (−11.8 to 2.5) |
0.2 | −0.84 (−7.8 to 6.1) |
0.8 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.9 (−12 to 4.1) |
0.3 | −0.24 (−5.6 to 5.1) |
0.9 | 1.9 (−2.1 to 5.8) |
0.4 | −0.32 (−2.0 to 1.4) |
0.7 | −0.99 (−6.0 to 4.0) |
0.7 | 1.8 (−0.84 to 4.4) |
0.2 | 0.87 (−1.7 to 3.4) |
0.5 |
| Trend change | −16.2 (−37.1 to 4.7) |
0.1 | −6.8 (−21.4 to 7.8) |
0.4 | 0.36 (−10.2 to 10.9) |
0.95 | −1.9 (−6.5 to 2.8) |
0.4 | −1.6 (−15.2 to 12.2) |
0.8 | 4.7 (−2.4 to 11.9) |
0.2 | 0.48 (−6.4 to 7.4) |
0.9 |
| 2012 USPSTF | ||||||||||||||
| Level change | −11.0 (−18.1 to −4.0) |
0.001 | −2.8 (−6.0 to 0.38) |
0.09 | 0.53 (−1.9 to 3.0) |
0.7 | −0.02 (−0.95 to 0.91) |
0.97 | −1.8 (−4.6 to 0.96) |
0.2 | −1.5 (−3.1 to 0.17) |
0.08 | 0.16 (−1.3 to 1.6) |
0.8 |
| Trend change | 7.0 (−7.2 to 21.1) |
0.3 | 1.7 (−3.6 to 7.0) |
0.5 | 0.12 (−4.1 to 4.2) |
0.96 | 0.12 (−1.4 to 1.8) |
0.9 | −0.48 (−5.3 to 4.3) |
0.9 | 1.6 (−1.3 to 4.3) |
0.3 | 0.12 (−2.4 to 2.6) |
0.9 |
| 2013 AUA | ||||||||||||||
| Level change | −8.2 (−15.9 to −0.38) |
0.04 | −1.0 (−4.8 to 2.8) |
0.6 | −0.44 (−3.2 to 2.3) |
0.8 | −0.15 (−1.3 to 0.99) |
0.8 | 0.28 (−3.2 to 3.7) |
0.9 | −0.50 (−2.7 to 1.7) |
0.7 | 0.47 (−1.3 to 2.3) |
0.6 |
| Trend change | −4.2 (−24.5 to 16.1) |
0.7 | −0.7 (−7.6 to 6.2) |
0.9 | 1.1 (−4.2 to 6.2) |
0.7 | −0.12 (−1.8 to 1.7) |
0.9 | −0.60 (−5.9 to 4.8) |
0.8 | −1.8 (−4.9 to 1.2) |
0.3 | −0.48 (−3.1 to 2.2) |
0.7 |
* Model 1: Base model with constant, time trend, intervention indicator variables. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
† Model 2: Adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, and anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPTF = US Preventive Services Task Force.
Next, we examined 30-d postbiopsy complications among men aged ≥40 yr as a measure of the absolute morbidity of biopsy (Fig. 1, Supplementary Fig. 3, and Table 2). Across the study periods, overall complication rates decreased 10% from 8.7 to 7.8 per 100 000 person-months. In unadjusted ITS analyses, there was a baseline growth in the overall complication rate (0.84 per 100 000 person-months per year; 95% CI, 0.48–1.3; p < 0.001), driven by growth in the rate of nonsepsis infections (0.72 per 100 000 person-months per year; 95% CI, 0.48–1.1; p < 0.001). Only the 2012 USPSTF recommendations were associated with a significant change in complication rates, following which there was an immediate decrease in the overall complication rate of −2.5 per 100 000 person-months (95% CI, −4.5 to −0.45; p = 0.02), driven by reductions in nonsepsis infections (−1.7 per 100 000 person-months; 95% CI, −3.1 to −0.31; p = 0.02) and urinary retention (−0.75 per 100 000 person-months;95% CI, −1.1 to −0.44; p < 0.001). Following the 2013 AUA guidelines there was also a reduction in the overall complication baseline growth rate (−4.1 per 100 000 person-months per year; 95% CI, −7.9 to −0.24; p = 0.04). In analyses adjusted for case mix (Table 2, model 2), effects were overall attenuated, and only the 2008 USPSTF recommendations were associated with a reduction in overall complication rates (−5.4 per 100 000 person-months; 95% CI, −9.6 to −1.2; p = 0.01).
To examine whether there was a change in the relative morbidity of biopsy, we performed a secondary analysis examining complication rates among men undergoing biopsy (Fig. 2). The proportion of men undergoing biopsy who experienced a complication increased from 14% in January 2005 to 18% in September 2014, an increase of 36%. There was again a baseline growth in overall complication rates (0.72% per year; 95% CI, 0.36–1.1; p < 0.001), driven by an increase in nonsepsis infections and sepsis (Supplementary Table 2). None of the interventions affected this growth. Similar results were obtained after adjusting for case mix (Table 3).
Fig. 2
Interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 3
Adjusted* interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Baseline rate | 15.6 (14.3–16.8) |
<0.001 | 7.1 (6.5–7.7) |
<0.001 | 0.34 (0.11–0.57) |
<0.001 | 6.7 (5.9–7.5) |
<0.001 | 2.8 (2.5–3.2) |
<0.001 | 3.2 (2.8–3.6) |
<0.001 |
| Baseline trend in rate | 0.60 (0.12–1.1) |
0.02 | 0.72 (0.60–0.96) |
<0.001 | 0.12 (0.12–0.24) |
<0.001 | 0.12 (−0.12 to 0.36) |
0.3 | −0.12 (−0.24 to 0.00) |
0.3 | −0.12 (−0.24 to 0.00) |
0.0498 |
| 2008 USPSTF | ||||||||||||
| Level change | 1.1 (−2.0 to 4.1) |
0.5 | −1.3 (−3.2 to 0.58) |
0.2 | 0.13 (−0.45 to 0.72) |
0.7 | 1.9 (−0.11 to 3.9) |
0.07 | 0.46 (−0.50 to 1.4) |
0.4 | 0.27 (−0.69 to 1.2) |
0.6 |
| Trend change | −3.5 (−13.4 to 6.6) |
0.5 | −1.8 (−8.4 to 4.7) |
0.6 | −1.0 (−3.0 to 0.96) |
0.3 | −4.3 (−11.4 to 2.6) |
0.2 | −1.6 (−4.9 to 1.9) |
0.4 | −0.72 (−4.0 to 2.6) |
0.7 |
| ERSPC/PLCO | ||||||||||||
| Level change | −0.14 (−3.9 to 3.6) |
0.9 | 1.1 (−1.3 to 3.4) |
0.4 | 0.55 (−0.19 to 1.3) |
0.2 | 0.39 (−2.2 to 3.0) |
0.8 | 0.75 (−0.47 to 2.0) |
0.2 | 0.65 (−0.56 to 1.9) |
0.3 |
| Trend change | 2.9 (−7.1 to 13.0) |
0.6 | 0.72 (−5.8 to 7.2) |
0.8 | 0.84 (−1.2 to 2.9) |
0.4 | 4.2 (−2.8 to 11.3) |
0.2 | 1.7 (−1.8 to 5.0) |
0.4 | 0.72 (−2.6 to 4.0) |
0.7 |
| 2012 USPSTF | ||||||||||||
| Level change | −0.27 (−2.7 to 2.1) |
0.8 | −0.56 (−2.0 to 0.87) |
0.5 | 0.16 (−0.31 to 0.63) |
0.5 | 1.1 (−0.38 to 2.5) |
0.2 | −0.38 (−1.0 to 0.28) |
0.3 | 0.44 (−0.34 to 1.2) |
0.3 |
| Trend change | 1.3 (−2.6 to 5.3) |
0.5 | 2.9 (0.60–5.3) |
0.02 | −0.24 (−1.1 to 0.60) |
0.6 | −0.72 (−3.2 to 1.7) |
0.5 | 0.24 (−0.96 to 1.3) |
0.7 | −0.48 (−1.8 to 0.84) |
0.5 |
| 2013 AUA | ||||||||||||
| Level change | 0.25 (−2.6 to 3.1) |
0.9 | −0.64 (−2.3 to 1.1) |
0.5 | 0.22 (−0.40 to 0.83) |
0.5 | 1.1 (−0.68 to 2.8) |
0.2 | −0.29 (−1.1 to 0.48) |
0.5 | 0.47 (−0.55 to 1.5) |
0.4 |
| Trend change | −1.6 (−6.1 to 2.9) |
0.5 | −2.5 (−5.5 to 0.60) |
0.1 | 0.12 (−0.72 to 1.1) |
0.8 | −0.36 (−3.1 to 2.4) |
0.8 | 0.24 (−1.2 to 1.7) |
0.8 | 0.12 (−1.3 to 1.6) |
0.9 |
* Model adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 person-months. Baseline trend in rate and trend change per 100 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial]; USPTF = US Preventive Services Task Force.
Given the increase in the relative morbidity of biopsy, we performed logistic regression to identify predictors of each complication category (Supplementary Table 3 and Table 4). Due to the large sample size, we focused on those features associated with the greatest magnitude of effect. A prior diagnosis of cancer was associated with an increased risk of all complication categories including any complication (odds ratio [OR]: 1.30; 95% CI, 1.22–1.39; p < 0.001) and hospitalization (OR: 1.59; 95% CI, 1.39–1.82; p < 0.001). Similarly, prior fluoroquinolone use was associated with an increased risk of all complication categories including any complication (OR: 1.27; 95% CI, 1.22–1.32; p < 0.001) and sepsis (OR: 1.55; 95% CI, 1.33–1.81; p < 0.001). Age at biopsy ≥70 yr was associated with a nearly 4-fold increased risk of urinary retention. Although anticoagulant use was associated with increased risks of bleeding (OR: 1.28; 95% CI, 1.14–1.45; p < 0.001), sepsis (OR: 1.48; 95% CI, 1.06–2.07; p = 0.02), and hospitalization (OR: 1.39; 95% CI, 1.16–1.67; p < 0.001), the increase in the overall complication risk was modest (OR 1.14; 95% CI, 1.04–1.25; p = 0.004).
Table 4
Multivariable analysis* of 30-d postbiopsy complications among men aged ≥40 yr undergoing prostate biopsy (January 2005 to September 2014) (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Hospitalization | Urinary retention | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Age at biopsy, yr | ||||||||||||
| 40–49 | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| 50–59 | 1.01 (0.94–1.09) | 0.8 | 0.94 (0.85–1.04) | 0.2 | 1.16 (0.83–1.63) | 0.4 | 1.01 (0.90–1.14) | 0.8 | 1.12 (0.93–1.36) | 0.2 | 1.54 (1.20–1.97) | <0.001 |
| 60–69 | 1.05 (0.97–1.13) | 0.2 | 0.94 (0.85–1.03) | 0.2 | 0.98 (0.70–1.37) | 0.9 | 1.05 (0.94–1.18) | 0.4 | 1.16 (0.95–1.40) | 0.1 | 2.16 (1.68–2.77) | <0.001 |
| ≥70 | 1.25 (1.15–1.36) | <0.001 | 1.03 (0.93–1.15) | 0.6 | 1.05 (0.74–1.49) | 0.8 | 1.29 (1.15–1.46) | <0.001 | 1.31 (1.07–1.60) | 0.008 | 3.98 (3.08–5.15) | <0.001 |
| Race | ||||||||||||
| White | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| Asian | 1.27 (1.15–1.40) | <0.001 | 1.27 (1.11–1.44) | 0.0003 | 1.65 (1.19–2.29) | 0.003 | 1.36 (1.18–1.56) | <0.001 | 0.94 (0.76–1.17) | 0.6 | 1.15 (0.93–1.42) | 0.2 |
| African American | 1.09 (1.03–1.15) | 0.002 | 1.05 (0.97–1.13) | 0.2 | 1.34 (1.09–1.64) | 0.005 | 1.17 (1.09–1.27) | <0.001 | 1.13 (1.01–1.27) | 0.03 | 1.15 (1.03–1.28) | 0.02 |
| Hispanic | 1.32 (1.24–1.40) | <0.001 | 1.36 (1.26–1.47) | <0.001 | 1.52 (1.21–1.90) | <0.001 | 1.33 (1.22–1.45) | <0.001 | 1.12 (0.97–1.28) | 0.1 | 1.37 (1.21–1.56) | <0.001 |
| Unknown/missing | 0.99 (0.93–1.06) | 0.8 | 1.07 (0.99–1.17) | 0.1 | 1.23 (0.94–1.61) | 0.1 | 0.96 (0.88–1.05) | 0.4 | 0.94 (0.81–1.08) | 0.4 | 1.02 (0.88–1.18) | 0.8 |
| Diabetes mellitus | 1.03 (0.99–1.07) | 0.09 | 1.06 (1.00–1.11) | 0.045 | 1.08 (0.92–1.28) | 0.4 | 1.01 (0.96–1.07) | 0.7 | 1.20 (1.10–1.31) | <0.001 | 0.99 (0.91–1.08) | 0.8 |
| COPD | 1.14 (1.09–1.20) | <0.001 | 1.07 (1.00–1.14) | 0.04 | 1.23 (1.01–1.49) | 0.04 | 1.14 (1.06–1.22) | <0.001 | 1.36 (1.23–1.51) | <0.001 | 1.14 (1.03–1.25) | 0.01 |
| PVD | 1.24 (1.16–1.31) | <0.001 | 1.1 (1.02–1.2) | 0.02 | 1.19 (0.93–1.51) | 0.2 | 1.28 (1.18–1.40) | <0.001 | 1.31 (1.15–1.48) | <0.001 | 1.26 (1.12–1.42) | <0.001 |
| Cerebrovascular disease | 1.11 (1.04–1.19) | 0.001 | 1.12 (1.03–1.22) | 0.01 | 1.21 (0.93–1.57) | 0.2 | 1.03 (0.94–1.14) | 0.5 | 1.26 (1.10–1.44) | <0.001 | 1.30 (1.15–1.47) | <0.001 |
| Cancer | 1.30 (1.22–1.39) | <0.001 | 1.16 (1.06–1.27) | 0.001 | 1.37 (1.05–1.79) | 0.02 | 1.40 (1.27–1.53) | <0.001 | 1.59 (1.39–1.82) | <0.001 | 1.32 (1.16–1.52) | <0.001 |
| Anticoagulant use | 1.14 (1.04–1.25) | 0.004 | 1.01 (0.89–1.14) | 0.9 | 1.48 (1.06–2.07) | 0.02 | 1.28 (1.14–1.45) | <0.001 | 1.39 (1.16–1.67) | <0.001 | 1.11 (0.92–1.33) | 0.3 |
| Prior fluoroquinolones | 1.27 (1.22–1.32) | <0.001 | 1.37 (1.31–1.45) | <0.001 | 1.55 (1.33–1.81) | <0.001 | 1.21 (1.14–1.28) | <0.001 | 1.25 (1.14–1.36) | <0.001 | 1.27 (1.16–1.38) | <0.001 |
* Models adjusted for year of biopsy and the features listed in the table.
CI = confidence interval; COPD = chronic obstructive pulmonary disease; PVD = peripheral vascular disease.
We conducted sensitivity analyses where the postintervention run-in period was varied from 2 to 4 mo (Supplementary Table 4). Results overall were similar, although effects were attenuated for the 2013 AUA guidelines on biopsy rates and the 2012 USPSTF recommendations on overall complication rates, likely due to fewer time points for analysis. In a separate sensitivity analysis, we examined bleeding complications excluding microhematuria from 2009 to 2014, following introduction of a separate code for microhematuria. Bleeding rates paralleled those of the primary bleeding end point (Supplementary Fig. 4), and results of both ITS and logistic regression analyses were similar (Supplementary Table 5 and 6). As an exploratory analysis, we examined PCa diagnosis rates within 60 d following biopsy (Supplementary Fig. 5). These appeared to increase steadily across the study period.
To our knowledge, this is the largest study to examine the impact of landmark PSA screening publications on rates of prostate biopsy and the first to examine their impact on postbiopsy complications. Overall, biopsy rates have declined by 33%, which appears attributable to the publication of the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines. Interestingly, the magnitude of reduction in biopsy rates exceeded previously reported decreases in PSA screening rates [6], [7], [8], [9], [10], [11], [12], [22], and [23], which suggests that either the threshold for referral to a urologist or the threshold for performing a biopsy has changed. More importantly, there was a concurrent 22% decrease in 30-d complication rates that appears largely attributable to the 2012 USPSTF recommendations. Still, the reduction in complication rates was less than the reduction in biopsy rates due to an increase in the relative morbidity of biopsy, driven by growth in infectious complications. This may in part be related to a shift to older patients with greater comorbidity undergoing biopsy, although similar results were obtained in analyses adjusted for case mix, which suggests additional factors may exist.
To this end, we evaluated predictors of postbiopsy complications to identify actionable targets for further risk reduction. Some of these features represent modifiable risk factors (eg, anticoagulant use), and others represent opportunities for altering periprocedural management. For instance, prior fluoroquinolone use may identify patients who would benefit from prebiopsy rectal swabs to guide periprocedural antibiotics [24] and [25]. Nonmodifiable risk factors, including specific comorbidities such as cancer and peripheral vascular disease, may encourage reconsideration of the risk–benefit ratio before proceeding with biopsy.
It is also important to consider the concurrent impact of landmark PSA screening publications on rates of PCa diagnosis to contextualize the observed changes in biopsy and its morbidity. An exploratory analysis suggested that cancer diagnosis rates have steadily increased, although further studies are necessary to examine this issue comprehensively.
A number of studies have examined the impact of the 2008 USPSTF recommendations [7], [8], [9], and [22], ERSPC/PLCO trials [6] and [9], and 2012 USPSTF recommendations [6], [10], [11], [12], and [23] on PSA screening rates, generally noting reductions ranging from 0% to 8%. However, much less data are available on the impact of these publications on biopsy rates. One single-center study noted a 31% reduction in the number of biopsies performed following the 2012 USPSTF recommendations [13]; another noted no change in the proportion of referred men who underwent biopsy [26]. A report from a Canadian academic health network noted a reduction in biopsy rates from 58.0 to 35.5 biopsies per month following the 2012 USPSTF recommendations [27]. Our findings therefore confirm a reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines.
Limited data exist regarding secular trends in postbiopsy complications, and prior studies have focused exclusively on the relative morbidity among men undergoing biopsy rather than absolute complication rates. A Canadian population-based study observed that the 30-d hospitalization rate increased from 1.0% in 1996 to 4.1% in 2005, driven by an increase in infectious complications [28]. A Surveillance, Epidemiology and End Results–Medicare study also noted an increase in hospitalizations related to infectious complications over time [14]. Although we observed a similar increase in 30-d overall complications among men undergoing biopsy, likewise driven by infectious complications, there was an absolute 22% reduction in 30-d complications over the study period. Taken together, these findings support a reduction in the morbidity of biopsy following publication of revised PSA screening guidelines but highlight the continued need to reduce the relative morbidity of biopsy.
A number of studies have described complication rates following prostate biopsy [15], [16], [29], and [30]. Hematuria has been reported in 10–84% of biopsies, rectal bleeding in 1–45%, infectious complications in up to 6.3%, urinary retention in up to 1.7%, and hospitalization in up to 6.9% [14] and [29]. The wide range in complication rates is due to differences in definitions, ascertainment of end points, and variation in clinical practice. Given the heterogeneity in postbiopsy complication rates, our findings have two advantages. First, they represent a contemporary, geographically diverse population across the entire United States. Second, use of claims-based ascertainment limited complications to those cases in which patients sought medical care, and although this may underestimate the comprehensive morbidity of biopsy, it more accurately reflects the economic burden of serious complications.
This study has several limitations. Most importantly, a control population of men who were not exposed to the PSA screening publications was not available. This study is retrospective, and additional, unrecognized changes in policy or clinical practice may possibly have confounded the impact of the interventions examined. However, the ITS approach has been described as the strongest quasi-experimental design to evaluate policy changes when a randomized trial is not possible [19]. Because we relied on administrative claims for ascertainment of complications, minor complications that did not generate a claim were not captured, thus underestimating these end points. Although we adjusted for case mix, we were unable to adjust for PSA. However, population shifts in PSA may be confounded by changes in screening practices, and PSA adjustment may therefore under- or overestimate the impact of interventions. We were also unable to adjust for other factors that may be associated with biopsy or complication rates, such as prostate magnetic resonance imaging (MRI), ancillary biomarker testing, biopsy approach (eg, MRI-fusion biopsy, transperineal biopsy), indication for anticoagulation, use of periprocedural bridging, or operator biopsy volume/experience.
We observed an immediate reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines, with a smaller reduction in absolute 30-d complication rates. Still, the relative morbidity of biopsy has increased over time, driven by infectious complications. Accordingly, if health policy aims to reduce the morbidity of PSA screening related to biopsy, further efforts should focus on improving patient selection and periprocedural processes of care.
Author contributions: Boris Gershman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Gershman, van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Acquisition of data: Van Houten, Shah.
Analysis and interpretation of data: Gershman, Van Houten, Shah, Karnes.
Drafting of the manuscript: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Critical revision of the manuscript for important intellectual content: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Statistical analysis: Van Houten.
Obtaining funding: Karnes, Shah.
Administrative, technical, or material support: Karnes, Shah.
Supervision: Karnes, Shah.
Other (specify): None.
Financial disclosures: Boris Gershman certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: This publication was made possible by funding from the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery.
Funding/Support and role of the sponsor: None.
In 2009, two landmark randomized trials—the European Randomized Study of Screening for Prostate Cancer [ERSPC] [1] and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial [PLCO] [2]—reported smaller than anticipated reductions in prostate cancer (PCa) mortality and prompted reexamination of prostate-specific antigen (PSA) screening recommendations. The US Preventive Services Task Force (USPSTF), which in 2008 had recommended against PSA screening in men aged >75 yr [3], updated its position in 2012 to recommend against systematic PSA screening in the general population [4]. The American Urological Association (AUA) likewise updated its PSA screening guidelines in May 2013, recommending shared decision making for men aged 55–69 yr [5]. Both the 2012 USPSTF and 2013 AUA guidelines cited prostate biopsy and its potential complications as a significant harm associated with PSA screening.
Although several studies have noted a reduction in PSA screening rates following these publications [6], [7], [8], [9], [10], [11], and [12], there are little data on whether this has translated into the intended reductions in screening-related morbidity in the form of prostate biopsy and postbiopsy complications [13]. We evaluated whether publication of the 2008 USPSTF recommendations, 2009 ERSPC/PLCO trials, 2012 USPSTF recommendations, or 2013 AUA guidelines was associated with reductions in prostate biopsy or postbiopsy complications. Biopsy-related complications also remain understudied both with regard to incidence as well as identification of risk factors [14], [15], and [16]. Consequently, we also sought to characterize postbiopsy complications and evaluate predictors using a contemporary population-based cohort.
We conducted a retrospective analysis of administrative claims from the Optum Labs Data Warehouse that contains data on >100 million individuals enrolled in private health plans and Medicare Advantage Plans, from geographically diverse regions across the United States, as described in Supplementary Methods [17]. Because this study involved analysis of preexisting deidentified data, it was exempt from institutional review board approval.
We identified all prostate biopsies performed in men aged ≥40 yr between January 1, 2005, and September 30, 2014, using Current Procedural Terminology-4 code 55700 alone (11%) or in combination with 76872 (84%) or 76942 (0%), or International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 60.11 (5%). The first biopsy was selected as the index biopsy when more than one biopsy was performed. Patients with a prior diagnosis of PCa (ICD-9 code 185) were excluded. We restricted the cohort to patients with continuous medical and pharmacy enrollment for at least 12 mo prior to biopsy and 30 d after biopsy to ascertain comorbidities and 30-d complications (Supplementary Fig. 1).
We recorded sociodemographic characteristics, comorbid conditions, anticoagulant use within 30 d of biopsy, and fluoroquinolone use in the 12 mo prior to biopsy as described in Supplement 1. The Charlson Comorbidity Index was calculated as previously described [18]. Prebiopsy PSA values within 60 d of biopsy were available for a subset of patients.
The primary outcomes were crude and adjusted rates of prostate biopsy and 30-d complications. Complications were identified using inpatient and outpatient administrative claims as summarized in Supplementary Table 1.
Baseline demographic features were compared across study periods using the Kruskal-Wallis test for continuous variables and the chi-square test for discrete variables. To evaluate the impact of each publication on rates of prostate biopsy and 30-d complications, we used an interrupted time series (ITS) approach as described in Supplementary Methods [19] and [20]. Rates were estimated using all men aged ≥40 yr with at least 12 mo of continuous medical and pharmacy enrollment as the denominator. As secondary analyses, we estimated rates of 30-d complications among men who underwent biopsy. Two separate ITS analyses were performed: (1) unadjusted analyses, using crude rates of biopsy and complications aggregated for each month, and (2) analyses standardized for case mix, to adjust for population shifts over time as described in Supplementary Methods [20] and [21].
Logistic regression was used to evaluate predictors of 30-d complications. Multivariable models were adjusted for age at biopsy, year of biopsy, race, anticoagulant use, history of fluoroquinolone use, and the presence of the five most prevalent Charlson comorbidities.
Statistical analyses were performed using SAS v.9.3 (SAS Institute Inc., Cary, NC, USA). All tests were two sided with p < 0.05 considered statistically significant.
Between January 2005 and September 2014, a total of 104 584 men underwent prostate biopsy (Table 1). Median age at biopsy was 63.0 yr (interquartile range [IQR]: 57.0–70.0). Overall, 18% of men had used fluoroquinolones in the 12 mo preceding biopsy, and 2.8% were on anticoagulants. Median prebiopsy PSA was 5.1 ng/dl (IQR: 4.0–7.1) for the 18 749 men (18%) for whom this was available. Over the time periods, men undergoing biopsy were older, had increased prevalence of anticoagulant use and comorbid conditions, and had a higher PSA (Table 1 and Supplementary Fig. 2).
Table 1
Patient demographics for men who underwent prostate biopsy (January 2005 to September 2014) stratified by pre/postintervention time periods (n = 104 584)
| Baseline Jan 2005–Jul 2008 (n = 37 217) |
2008 USPSTF Aug 2008–Feb 2009 (n = 6672) |
ERSPC/PLCO Mar 2009–Apr 2012 (n = 36 603) |
2012 USPSTF May 2012–Apr 2013 (n = 10 772) |
2013 AUA May 2013–Sep 2014 (n = 13 320) |
p value | Total (n = 104 584) | |
|---|---|---|---|---|---|---|---|
| Age at biopsy, yr | |||||||
| Median (IQR) | 62.0 (56.0–68.0) | 62.0 (56.0–68.0) | 63.0 (57.0–70.0) | 65.0 (59.0–71.0) | 66.0 (59.0–71.0) | <0.001 | 63.0 (57.0–70.0) |
| Age group, yr (%) | |||||||
| 40–49 | 2269 (6.1) | 416 (6.2) | 1737 (4.7) | 420 (3.9) | 467 (3.5) | <0.001 | 5309 (5.1) |
| 50–59 | 12 366 (33) | 2201 (33) | 10 342 (28) | 2644 (25) | 3098 (23) | 30 651 (29) | |
| 60–69 | 14 671 (39) | 2697 (40) | 15 105 (41) | 4426 (41) | 5546 (42) | 42 445 (41) | |
| ≥70 | 7911 (21) | 1358 (20) | 9419 (26) | 3282 (31) | 4209 (32) | 26 179 (25) | |
| Race (%) | |||||||
| White | 25 508 (69) | 4643 (70) | 26 187 (72) | 7795 (72) | 9407 (71) | <0.001 | 73 540 (70) |
| African American | 3695 (9.9) | 780 (12) | 4423 (12) | 1283 (12) | 1589 (12) | 11 770 (11) | |
| Hispanic | 2549 (6.8) | 538 (8.1) | 2911 (8.0) | 837 (7.8) | 1186 (8.9) | 8021 (7.7) | |
| Asian | 1133 (3.0) | 236 (3.5) | 1142 (3.1) | 306 (2.8) | 387 (2.9) | 3204 (3.1) | |
| Unknown/Missing | 4332 (12) | 475 (7.1) | 1940 (5.3) | 551 (5.1) | 751 (5.6) | 8049 (7.7) | |
| Census region (%) | |||||||
| Northeast | 7094 (19) | 1196 (18) | 6120 (17) | 1886 (18) | 2424 (18) | <0.001 | 18 720 (18) |
| Midwest | 10 094 (27) | 1562 (23) | 8837 (24) | 2742 (26) | 3627 (27) | 26 862 (26) | |
| South | 15 330 (41) | 2889 (43) | 16 371 (45) | 4698 (44) | 5538 (42) | 44 826 (43) | |
| West | 4660 (13) | 993 (15) | 5147 (14) | 1442 (13) | 1728 (13) | 13 970 (13) | |
| Other/Missing | 39 (0.1) | 32 (0.5) | 128 (0.3) | 4 (0) | 3 (0) | 206 (0.2) | |
| PSA, ng/d | n = 5174 | n = 548 | n = 7345 | n = 2211 | n = 3471 | n = 18 749 | |
| Median (IQR) | 5.1 (3.9–7.1) | 4.9 (3.8–6.8) | 5.0 (4.0–6.8) | 5.3 (4.2–7.4) | 5.4 (4.2–7.8) | <0.001 | 5.1 (4.0–7.1) |
| Anticoagulant use (%) | 928 (2.5) | 184 (2.8) | 1103 (3.0) | 299 (2.8) | 415 (3.1) | <0.001 | 2929 (2.8) |
| History of fluoroquinolone use in prior 12 mo (%) | 6797 (18) | 1267 (19) | 6557 (18) | 2017 (19) | 2504 (19) | 0.054 | 19 142 (18) |
| Charlson Comorbidity Index (%) | |||||||
| 0 | 22 354 (60) | 3887 (58) | 20 269 (55) | 5758 (54) | 7039 (53) | <0.001 | 59 307 (57) |
| 1 | 8199 (22) | 1453 (22) | 8252 (23) | 2503 (23) | 2999 (23) | 23 406 (22) | |
| 2 | 3232 (8.7) | 664 (10) | 3520 (9.6) | 1108 (10) | 1359 (10) | 9883 (9.4) | |
| ≥3 | 3432 (9.2) | 668 (10) | 4562 (13) | 1403 (13) | 1923 (14) | 11 988 (12) | |
| Comorbidities (%) | |||||||
| Myocardial infarction | 763 (2.1) | 144 (2.2) | 812 (2.2) | 227 (2.1) | 273 (2.0) | 0.6 | 2219 (2.1) |
| Congestive heart failure | 1443 (3.9) | 223 (3.3) | 1548 (4.2) | 467 (4.3) | 586 (4.4) | <0.001 | 4267 (4.1) |
| Peripheral vascular disease | 2054 (5.5) | 399 (6.0) | 2660 (7.3) | 783 (7.3) | 1071 (8.0) | <0.001 | 6967 (6.7) |
| Cerebrovascular disease | 2007 (5.4) | 379 (5.7) | 2358 (6.4) | 681 (6.3) | 947 (7.1) | <0.001 | 6372 (6.1) |
| Dementia | 370 (1.0) | 63 (0.9) | 442 (1.2) | 141 (1.3) | 198 (1.5) | <0.001 | 1214 (1.2) |
| Chronic obstructive pulmonary disease | 4251 (11) | 758 (11) | 4293 (12) | 1254 (12) | 1686 (13) | 0.004 | 12 242 (12) |
| Ulcer | 362 (1.0) | 60 (0.9) | 297 (0.8) | 84 (0.8) | 102 (0.8) | 0.07 | 905 (0.9) |
| Mild liver disease | 1182 (3.2) | 223 (3.3) | 1332 (3.6) | 413 (3.8) | 515 (3.9) | <0.001 | 3665 (3.5) |
| Diabetes mellitus | 6475 (17) | 1290 (19) | 7823 (21) | 2442 (23) | 2973 (22) | <0.001 | 21 003 (20) |
| Diabetes with organ damage | 1105 (3.0) | 225 (3.4) | 1600 (4.4) | 492 (4.6) | 734 (5.5) | <0.001 | 4156 (4.0) |
| Hemiplegia | 128 (0.3) | 25 (0.4) | 150 (0.4) | 51 (0.5) | 56 (0.4) | 0.3 | 410 (0.4) |
| Moderate/severe renal disease | 1133 (3.0) | 254 (3.8) | 1950 (5.3) | 674 (6.3) | 908 (6.8) | <0.001 | 4919 (4.7) |
| Moderate/severe liver disease | 27 (0.1) | 7 (0.1) | 31 (0.1) | 9 (0.1) | 12 (0.1) | 0.9 | 86 (0.1) |
| Metastatic solid tumor | 175 (0.5) | 37 (0.6) | 199 (0.5) | 58 (0.5) | 80 (0.6) | 0.4 | 549 (0.5) |
| AIDS | 82 (0.2) | 16 (0.2) | 101 (0.3) | 27 (0.3) | 46 (0.3) | 0.2 | 272 (0.3) |
| Rheumatologic disease | 449 (1.2) | 71 (1.1) | 522 (1.4) | 158 (1.5) | 196 (1.5) | 0.007 | 1396 (1.3) |
| Cancer | 1736 (4.7) | 337 (5.1) | 1796 (4.9) | 571 (5.3) | 701 (5.3) | 0.02 | 5141 (4.9) |
| Net worth range (%) | |||||||
| <$25 000 | 1831 (4.9) | 404 (6.1) | 2101 (5.7) | 561 (5.2) | 695 (5.2) | <0.001 | 5592 (5.3) |
| $25 000–149 000 | 5188 (14) | 1042 (16) | 5608 (15) | 1591 (15) | 2046 (15) | 15 475 (15) | |
| $150 000–249 000 | 4754 (13) | 837 (13) | 5182 (14) | 1499 (14) | 1862 (14) | 14 134 (14) | |
| $250 000–499 000 | 9694 (26) | 1779 (27) | 10 048 (28) | 3027 (28) | 3636 (27) | 28 184 (27) | |
| ≥$500 000 | 10 910 (29) | 2030 (30) | 10 860 (30) | 3208 (30) | 3852 (29) | 30 860 (30) | |
| Unknown/missing | 4840 (13) | 580 (8.7) | 2804 (7.7) | 886 (8.2) | 1229 (9.2) | 10 339 (9.9) | |
| 30-d complications (%) | |||||||
| Any | 5973 (16) | 1151 (17) | 6120 (17) | 1863 (17) | 2451 (18) | <0.001 | 17 558 (17) |
| Bleeding | 2547 (6.8) | 499 (7.5) | 2660 (7.3) | 860 (8.0) | 1083 (8.1) | <0.001 | 7649 (7.3) |
| Nonsepsis infection | 2978 (8.0) | 590 (8.8) | 2990 (8.2) | 875 (8.1) | 1236 (9.3) | <0.001 | 8669 (8.3) |
| Sepsis | 191 (0.5) | 50 (0.7) | 349 (1.0) | 105 (1.0) | 134 (1.0) | <0.001 | 829 (0.8) |
| Urinary retention | 980 (2.6) | 161 (2.4) | 1138 (3.1) | 341 (3.2) | 422 (3.2) | <0.001 | 3042 (2.9) |
| Hospitalization | 1019 (2.7) | 181 (2.7) | 1056 (2.9) | 320 (3.0) | 367 (2.8) | 0.6 | 2943 (2.8) |
AIDS = acquired immunodeficiency syndrome; AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; IQR = interquartile range; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer [screening trial]; PSA = prostate-specific antigen; USPTF = US Preventive Services Task Force.
Overall, 30-d complications occurred in 17% of biopsies. The most prevalent was nonsepsis genitourinary infections (8.3%), followed by bleeding complications (7.3%) and urinary retention (2.9%) (Table 1). The 30-d hospitalization rate was 2.8%.
We first examined unadjusted rates of prostate biopsy among men aged ≥40 yr (Fig. 1 and Table 2). From January 2005 to September 2014, biopsy rates fell 33% from 64.1 to 42.8 per 100 000 person-months. There was substantial seasonal variation in biopsy rates with reductions near the end of each calendar year. With the ITS approach, each intervention is potentially associated with an immediate change in event rates (ie, level shift or change in intercept) as well as a change in the rate of growth (ie, change in slope). At baseline, there was a statistically significant growth in prostate biopsy rates of 2.9 biopsies per 100 000 person-months per year (95% confidence interval [CI], 0.96–4.8; p < 0.001). Following the 2008 USPSTF recommendations, there was an immediate decrease (ie, level shift) in biopsy rates of −10.1 biopsies per 100 000 person-months (95% CI, −17.1 to −3.0; p < 0.001). Likewise, there were immediate decreases in biopsy rates following the 2012 USPSTF recommendations (−13.8 biopsies per 100 000 person-months; 95% CI, −21.0 to −6.7; p < 0.001) and 2013 AUA guidelines (−8.8 biopsies per 100 000 person-months; 95% CI, −16.7 to −0.92; p = 0.03). Analyses adjusted for case mix were similar, although the baseline growth rate was no longer significant, indicating this was related to population shifts over time (Table 2, model 2).
Fig. 1
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 2
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315)
| Prostate biopsy | Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Model 1* | ||||||||||||||
| Baseline rate | 57.8 (53.4–62.3) |
<0.001 | 8.6 (7.5–9.7) |
<0.001 | 4.0 (3.3–4.6) |
<0.001 | 0.17 (0.03–0.32) |
0.02 | 2.8 (2.3–3.3) |
<0.001 | 1.5 (1.3–1.6) |
<0.001 | 1.4 (1.1–1.7) |
<0.001 |
| Baseline trend in rate | 2.9 (0.96–4.8) |
<0.001 | 0.84 (0.48–1.3) |
<0.001 | 0.72 (0.48–1.1) |
<0.001 | 0.12 (0.12–0.12) |
<0.001 | 0.36 (0.12–0.48) |
<0.001 | 0.12 (0.00–0.12) |
<0.001 | 0.00 (−0.12 to 0.12) |
0.5 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.1 (−17.1 to −3.0) |
<0.001 | −0.61 (−3.1 to 1.9) |
0.6 | −1.4 (−2.8 to −0.03) |
0.053 | 0.04 (−0.35 to 0.42) |
0.8 | 0.56 (−0.70 to 1.8) |
0.4 | −0.06 (−0.69 to 0.57) |
0.9 | 0.16 (−0.58 to 0.90) |
0.7 |
| Trend change | 7.9 (−12.7 to 28.4) |
0.5 | −1.0 (−9.1 to 7.2) |
0.8 | −0.12 (−4.1 to 4.0) |
0.98 | −0.60 (−1.9 to 0.72) |
0.4 | −2.0 (−6.5 to 2.4) |
0.4 | −1.1 (−3.4 to 1.3) |
0.4 | −1.0 (−3.6 to 1.6) |
0.5 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.6 (−12 to 4.5) |
0.4 | −0.74 (−3.8 to 2.4) |
0.6 | −0.12 (−1.8 to 1.5) |
0.9 | 0.29 (−0.20 to 0.78) |
0.2 | −0.04 (−1.6 to 1.6) |
0.97 | 0.59 (−0.23 to 1.4) |
0.2 | 0.51 (−0.43 to 1.5) |
0.3 |
| Trend change | −11.2 (−32.0 to 9.6) |
0.3 | 0.0 (−8.2 to 8.0) |
0.99 | −0.96 (−5.0 to 3.1) |
0.7 | 0.36 (−0.96 to 1.8) |
0.5 | 1.7 (−2.8 to 6.1) |
0.5 | 1.0 (−1.4 to 3.4) |
0.4 | 0.84 (−1.7 to 3.5) |
0.5 |
| 2012 USPSTF | ||||||||||||||
| Level change | −13.8 (−21.0 to −6.7) |
<0.001 | −2.5 (−4.5 to −0.45) |
0.02 | −1.7 (−3.1 to −0.31) |
0.02 | −0.12 (−0.42 to 0.19) |
0.5 | −0.61 (−1.5 to 0.28) |
0.2 | −0.75 (−1.1 to −0.44) |
<0.001 | −0.25 (−0.77 to 0.27) |
0.4 |
| Trend change | 8.2 (−5.8 to 22.0) |
0.3 | 3.0 (−0.36 to 6.4) |
0.09 | 3.1 (0.72–5.5) |
0.01 | 0.12 (−0.36 to 0.60) |
0.7 | 0.84 (−0.72 to 2.3) |
0.3 | 0.48 (0.00 to 1.1) |
0.08 | 0.24 (−0.60 to 1.2) |
0.6 |
| 2013 AUA | ||||||||||||||
| Level change | −8.8 (−16.7 to −0.92) |
0.03 | −1.4 (−3.8 to 1.1) |
0.3 | −1.9 (−3.5 to −0.22) |
0.03 | −0.02 (−0.41 to 0.37) |
0.9 | −0.19 (−1.3 to 0.91) |
0.7 | −0.38 (−0.77 to 0.01) |
0.06 | −0.01 (−0.66 to 0.63) |
0.96 |
| Trend change | −10.4 (−30.0 to 9.2) |
0.3 | −4.1 (−7.9 to −0.24) |
0.04 | −2.9 (−6.0 to 0.12) |
0.07 | −0.24 (−0.72 to 0.36) |
0.5 | −1.8 (−3.5 to 0.0) |
0.046 | −0.60 (−1.2 to 0.12) |
0.09 | −0.60 (−1.7 to 0.36) |
0.2 |
| Model 2† | ||||||||||||||
| Baseline rate | 86.8 (82.1–91.4) |
<0.001 | 16.2 (14.9–17.6) |
<0.001 | 6.4 (4.9–7.9) |
<0.001 | 0.39 (−0.13 to 0.91) |
0.2 | 6.1 (4.6–7.7) |
<0.001 | 3.6 (2.8–4.4) |
<0.001 | 3.0 (2.2–3.8) |
<0.001 |
| Baseline trend in rate | 0.24 (−1.8 to 2.3) |
0.8 | −0.24 (−0.72 to 0.24) |
0.4 | 0.48 (−0.12 to 1.1) |
0.09 | 0.12 (−0.12 to 0.24) |
0.4 | −0.24 (−0.72 to 0.24) |
0.4 | −0.24 (−0.48 to 0.00) |
0.054 | −0.12 (−0.36 to 0.12) |
0.3 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.8 (−17.8 to −3.8) |
<0.001 | −5.4 (−9.6 to −1.2) |
0.01 | −2.5 (−5.6 to 0.71) |
0.1 | −0.61 (−1.9 to 0.71) |
0.4 | 0.14 (−3.8 to 4.1) |
0.9 | 1.1 (−0.98 to 3.1) |
0.3 | 0.14 (−1.9 to 2.1) |
0.9 |
| Trend change | 10.6 (−10.1 to 31.2) |
0.3 | 6.1 (−8.4 to 20.8) |
0.4 | −1.7 (−12.2 to 8.8) |
0.8 | 1.7 (−2.9 to 6.2) |
0.5 | 1.7 (−12.0 to 15.5) |
0.8 | −4.7 (−11.8 to 2.5) |
0.2 | −0.84 (−7.8 to 6.1) |
0.8 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.9 (−12 to 4.1) |
0.3 | −0.24 (−5.6 to 5.1) |
0.9 | 1.9 (−2.1 to 5.8) |
0.4 | −0.32 (−2.0 to 1.4) |
0.7 | −0.99 (−6.0 to 4.0) |
0.7 | 1.8 (−0.84 to 4.4) |
0.2 | 0.87 (−1.7 to 3.4) |
0.5 |
| Trend change | −16.2 (−37.1 to 4.7) |
0.1 | −6.8 (−21.4 to 7.8) |
0.4 | 0.36 (−10.2 to 10.9) |
0.95 | −1.9 (−6.5 to 2.8) |
0.4 | −1.6 (−15.2 to 12.2) |
0.8 | 4.7 (−2.4 to 11.9) |
0.2 | 0.48 (−6.4 to 7.4) |
0.9 |
| 2012 USPSTF | ||||||||||||||
| Level change | −11.0 (−18.1 to −4.0) |
0.001 | −2.8 (−6.0 to 0.38) |
0.09 | 0.53 (−1.9 to 3.0) |
0.7 | −0.02 (−0.95 to 0.91) |
0.97 | −1.8 (−4.6 to 0.96) |
0.2 | −1.5 (−3.1 to 0.17) |
0.08 | 0.16 (−1.3 to 1.6) |
0.8 |
| Trend change | 7.0 (−7.2 to 21.1) |
0.3 | 1.7 (−3.6 to 7.0) |
0.5 | 0.12 (−4.1 to 4.2) |
0.96 | 0.12 (−1.4 to 1.8) |
0.9 | −0.48 (−5.3 to 4.3) |
0.9 | 1.6 (−1.3 to 4.3) |
0.3 | 0.12 (−2.4 to 2.6) |
0.9 |
| 2013 AUA | ||||||||||||||
| Level change | −8.2 (−15.9 to −0.38) |
0.04 | −1.0 (−4.8 to 2.8) |
0.6 | −0.44 (−3.2 to 2.3) |
0.8 | −0.15 (−1.3 to 0.99) |
0.8 | 0.28 (−3.2 to 3.7) |
0.9 | −0.50 (−2.7 to 1.7) |
0.7 | 0.47 (−1.3 to 2.3) |
0.6 |
| Trend change | −4.2 (−24.5 to 16.1) |
0.7 | −0.7 (−7.6 to 6.2) |
0.9 | 1.1 (−4.2 to 6.2) |
0.7 | −0.12 (−1.8 to 1.7) |
0.9 | −0.60 (−5.9 to 4.8) |
0.8 | −1.8 (−4.9 to 1.2) |
0.3 | −0.48 (−3.1 to 2.2) |
0.7 |
* Model 1: Base model with constant, time trend, intervention indicator variables. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
† Model 2: Adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, and anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPTF = US Preventive Services Task Force.
Next, we examined 30-d postbiopsy complications among men aged ≥40 yr as a measure of the absolute morbidity of biopsy (Fig. 1, Supplementary Fig. 3, and Table 2). Across the study periods, overall complication rates decreased 10% from 8.7 to 7.8 per 100 000 person-months. In unadjusted ITS analyses, there was a baseline growth in the overall complication rate (0.84 per 100 000 person-months per year; 95% CI, 0.48–1.3; p < 0.001), driven by growth in the rate of nonsepsis infections (0.72 per 100 000 person-months per year; 95% CI, 0.48–1.1; p < 0.001). Only the 2012 USPSTF recommendations were associated with a significant change in complication rates, following which there was an immediate decrease in the overall complication rate of −2.5 per 100 000 person-months (95% CI, −4.5 to −0.45; p = 0.02), driven by reductions in nonsepsis infections (−1.7 per 100 000 person-months; 95% CI, −3.1 to −0.31; p = 0.02) and urinary retention (−0.75 per 100 000 person-months;95% CI, −1.1 to −0.44; p < 0.001). Following the 2013 AUA guidelines there was also a reduction in the overall complication baseline growth rate (−4.1 per 100 000 person-months per year; 95% CI, −7.9 to −0.24; p = 0.04). In analyses adjusted for case mix (Table 2, model 2), effects were overall attenuated, and only the 2008 USPSTF recommendations were associated with a reduction in overall complication rates (−5.4 per 100 000 person-months; 95% CI, −9.6 to −1.2; p = 0.01).
To examine whether there was a change in the relative morbidity of biopsy, we performed a secondary analysis examining complication rates among men undergoing biopsy (Fig. 2). The proportion of men undergoing biopsy who experienced a complication increased from 14% in January 2005 to 18% in September 2014, an increase of 36%. There was again a baseline growth in overall complication rates (0.72% per year; 95% CI, 0.36–1.1; p < 0.001), driven by an increase in nonsepsis infections and sepsis (Supplementary Table 2). None of the interventions affected this growth. Similar results were obtained after adjusting for case mix (Table 3).
Fig. 2
Interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 3
Adjusted* interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Baseline rate | 15.6 (14.3–16.8) |
<0.001 | 7.1 (6.5–7.7) |
<0.001 | 0.34 (0.11–0.57) |
<0.001 | 6.7 (5.9–7.5) |
<0.001 | 2.8 (2.5–3.2) |
<0.001 | 3.2 (2.8–3.6) |
<0.001 |
| Baseline trend in rate | 0.60 (0.12–1.1) |
0.02 | 0.72 (0.60–0.96) |
<0.001 | 0.12 (0.12–0.24) |
<0.001 | 0.12 (−0.12 to 0.36) |
0.3 | −0.12 (−0.24 to 0.00) |
0.3 | −0.12 (−0.24 to 0.00) |
0.0498 |
| 2008 USPSTF | ||||||||||||
| Level change | 1.1 (−2.0 to 4.1) |
0.5 | −1.3 (−3.2 to 0.58) |
0.2 | 0.13 (−0.45 to 0.72) |
0.7 | 1.9 (−0.11 to 3.9) |
0.07 | 0.46 (−0.50 to 1.4) |
0.4 | 0.27 (−0.69 to 1.2) |
0.6 |
| Trend change | −3.5 (−13.4 to 6.6) |
0.5 | −1.8 (−8.4 to 4.7) |
0.6 | −1.0 (−3.0 to 0.96) |
0.3 | −4.3 (−11.4 to 2.6) |
0.2 | −1.6 (−4.9 to 1.9) |
0.4 | −0.72 (−4.0 to 2.6) |
0.7 |
| ERSPC/PLCO | ||||||||||||
| Level change | −0.14 (−3.9 to 3.6) |
0.9 | 1.1 (−1.3 to 3.4) |
0.4 | 0.55 (−0.19 to 1.3) |
0.2 | 0.39 (−2.2 to 3.0) |
0.8 | 0.75 (−0.47 to 2.0) |
0.2 | 0.65 (−0.56 to 1.9) |
0.3 |
| Trend change | 2.9 (−7.1 to 13.0) |
0.6 | 0.72 (−5.8 to 7.2) |
0.8 | 0.84 (−1.2 to 2.9) |
0.4 | 4.2 (−2.8 to 11.3) |
0.2 | 1.7 (−1.8 to 5.0) |
0.4 | 0.72 (−2.6 to 4.0) |
0.7 |
| 2012 USPSTF | ||||||||||||
| Level change | −0.27 (−2.7 to 2.1) |
0.8 | −0.56 (−2.0 to 0.87) |
0.5 | 0.16 (−0.31 to 0.63) |
0.5 | 1.1 (−0.38 to 2.5) |
0.2 | −0.38 (−1.0 to 0.28) |
0.3 | 0.44 (−0.34 to 1.2) |
0.3 |
| Trend change | 1.3 (−2.6 to 5.3) |
0.5 | 2.9 (0.60–5.3) |
0.02 | −0.24 (−1.1 to 0.60) |
0.6 | −0.72 (−3.2 to 1.7) |
0.5 | 0.24 (−0.96 to 1.3) |
0.7 | −0.48 (−1.8 to 0.84) |
0.5 |
| 2013 AUA | ||||||||||||
| Level change | 0.25 (−2.6 to 3.1) |
0.9 | −0.64 (−2.3 to 1.1) |
0.5 | 0.22 (−0.40 to 0.83) |
0.5 | 1.1 (−0.68 to 2.8) |
0.2 | −0.29 (−1.1 to 0.48) |
0.5 | 0.47 (−0.55 to 1.5) |
0.4 |
| Trend change | −1.6 (−6.1 to 2.9) |
0.5 | −2.5 (−5.5 to 0.60) |
0.1 | 0.12 (−0.72 to 1.1) |
0.8 | −0.36 (−3.1 to 2.4) |
0.8 | 0.24 (−1.2 to 1.7) |
0.8 | 0.12 (−1.3 to 1.6) |
0.9 |
* Model adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 person-months. Baseline trend in rate and trend change per 100 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial]; USPTF = US Preventive Services Task Force.
Given the increase in the relative morbidity of biopsy, we performed logistic regression to identify predictors of each complication category (Supplementary Table 3 and Table 4). Due to the large sample size, we focused on those features associated with the greatest magnitude of effect. A prior diagnosis of cancer was associated with an increased risk of all complication categories including any complication (odds ratio [OR]: 1.30; 95% CI, 1.22–1.39; p < 0.001) and hospitalization (OR: 1.59; 95% CI, 1.39–1.82; p < 0.001). Similarly, prior fluoroquinolone use was associated with an increased risk of all complication categories including any complication (OR: 1.27; 95% CI, 1.22–1.32; p < 0.001) and sepsis (OR: 1.55; 95% CI, 1.33–1.81; p < 0.001). Age at biopsy ≥70 yr was associated with a nearly 4-fold increased risk of urinary retention. Although anticoagulant use was associated with increased risks of bleeding (OR: 1.28; 95% CI, 1.14–1.45; p < 0.001), sepsis (OR: 1.48; 95% CI, 1.06–2.07; p = 0.02), and hospitalization (OR: 1.39; 95% CI, 1.16–1.67; p < 0.001), the increase in the overall complication risk was modest (OR 1.14; 95% CI, 1.04–1.25; p = 0.004).
Table 4
Multivariable analysis* of 30-d postbiopsy complications among men aged ≥40 yr undergoing prostate biopsy (January 2005 to September 2014) (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Hospitalization | Urinary retention | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Age at biopsy, yr | ||||||||||||
| 40–49 | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| 50–59 | 1.01 (0.94–1.09) | 0.8 | 0.94 (0.85–1.04) | 0.2 | 1.16 (0.83–1.63) | 0.4 | 1.01 (0.90–1.14) | 0.8 | 1.12 (0.93–1.36) | 0.2 | 1.54 (1.20–1.97) | <0.001 |
| 60–69 | 1.05 (0.97–1.13) | 0.2 | 0.94 (0.85–1.03) | 0.2 | 0.98 (0.70–1.37) | 0.9 | 1.05 (0.94–1.18) | 0.4 | 1.16 (0.95–1.40) | 0.1 | 2.16 (1.68–2.77) | <0.001 |
| ≥70 | 1.25 (1.15–1.36) | <0.001 | 1.03 (0.93–1.15) | 0.6 | 1.05 (0.74–1.49) | 0.8 | 1.29 (1.15–1.46) | <0.001 | 1.31 (1.07–1.60) | 0.008 | 3.98 (3.08–5.15) | <0.001 |
| Race | ||||||||||||
| White | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| Asian | 1.27 (1.15–1.40) | <0.001 | 1.27 (1.11–1.44) | 0.0003 | 1.65 (1.19–2.29) | 0.003 | 1.36 (1.18–1.56) | <0.001 | 0.94 (0.76–1.17) | 0.6 | 1.15 (0.93–1.42) | 0.2 |
| African American | 1.09 (1.03–1.15) | 0.002 | 1.05 (0.97–1.13) | 0.2 | 1.34 (1.09–1.64) | 0.005 | 1.17 (1.09–1.27) | <0.001 | 1.13 (1.01–1.27) | 0.03 | 1.15 (1.03–1.28) | 0.02 |
| Hispanic | 1.32 (1.24–1.40) | <0.001 | 1.36 (1.26–1.47) | <0.001 | 1.52 (1.21–1.90) | <0.001 | 1.33 (1.22–1.45) | <0.001 | 1.12 (0.97–1.28) | 0.1 | 1.37 (1.21–1.56) | <0.001 |
| Unknown/missing | 0.99 (0.93–1.06) | 0.8 | 1.07 (0.99–1.17) | 0.1 | 1.23 (0.94–1.61) | 0.1 | 0.96 (0.88–1.05) | 0.4 | 0.94 (0.81–1.08) | 0.4 | 1.02 (0.88–1.18) | 0.8 |
| Diabetes mellitus | 1.03 (0.99–1.07) | 0.09 | 1.06 (1.00–1.11) | 0.045 | 1.08 (0.92–1.28) | 0.4 | 1.01 (0.96–1.07) | 0.7 | 1.20 (1.10–1.31) | <0.001 | 0.99 (0.91–1.08) | 0.8 |
| COPD | 1.14 (1.09–1.20) | <0.001 | 1.07 (1.00–1.14) | 0.04 | 1.23 (1.01–1.49) | 0.04 | 1.14 (1.06–1.22) | <0.001 | 1.36 (1.23–1.51) | <0.001 | 1.14 (1.03–1.25) | 0.01 |
| PVD | 1.24 (1.16–1.31) | <0.001 | 1.1 (1.02–1.2) | 0.02 | 1.19 (0.93–1.51) | 0.2 | 1.28 (1.18–1.40) | <0.001 | 1.31 (1.15–1.48) | <0.001 | 1.26 (1.12–1.42) | <0.001 |
| Cerebrovascular disease | 1.11 (1.04–1.19) | 0.001 | 1.12 (1.03–1.22) | 0.01 | 1.21 (0.93–1.57) | 0.2 | 1.03 (0.94–1.14) | 0.5 | 1.26 (1.10–1.44) | <0.001 | 1.30 (1.15–1.47) | <0.001 |
| Cancer | 1.30 (1.22–1.39) | <0.001 | 1.16 (1.06–1.27) | 0.001 | 1.37 (1.05–1.79) | 0.02 | 1.40 (1.27–1.53) | <0.001 | 1.59 (1.39–1.82) | <0.001 | 1.32 (1.16–1.52) | <0.001 |
| Anticoagulant use | 1.14 (1.04–1.25) | 0.004 | 1.01 (0.89–1.14) | 0.9 | 1.48 (1.06–2.07) | 0.02 | 1.28 (1.14–1.45) | <0.001 | 1.39 (1.16–1.67) | <0.001 | 1.11 (0.92–1.33) | 0.3 |
| Prior fluoroquinolones | 1.27 (1.22–1.32) | <0.001 | 1.37 (1.31–1.45) | <0.001 | 1.55 (1.33–1.81) | <0.001 | 1.21 (1.14–1.28) | <0.001 | 1.25 (1.14–1.36) | <0.001 | 1.27 (1.16–1.38) | <0.001 |
* Models adjusted for year of biopsy and the features listed in the table.
CI = confidence interval; COPD = chronic obstructive pulmonary disease; PVD = peripheral vascular disease.
We conducted sensitivity analyses where the postintervention run-in period was varied from 2 to 4 mo (Supplementary Table 4). Results overall were similar, although effects were attenuated for the 2013 AUA guidelines on biopsy rates and the 2012 USPSTF recommendations on overall complication rates, likely due to fewer time points for analysis. In a separate sensitivity analysis, we examined bleeding complications excluding microhematuria from 2009 to 2014, following introduction of a separate code for microhematuria. Bleeding rates paralleled those of the primary bleeding end point (Supplementary Fig. 4), and results of both ITS and logistic regression analyses were similar (Supplementary Table 5 and 6). As an exploratory analysis, we examined PCa diagnosis rates within 60 d following biopsy (Supplementary Fig. 5). These appeared to increase steadily across the study period.
To our knowledge, this is the largest study to examine the impact of landmark PSA screening publications on rates of prostate biopsy and the first to examine their impact on postbiopsy complications. Overall, biopsy rates have declined by 33%, which appears attributable to the publication of the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines. Interestingly, the magnitude of reduction in biopsy rates exceeded previously reported decreases in PSA screening rates [6], [7], [8], [9], [10], [11], [12], [22], and [23], which suggests that either the threshold for referral to a urologist or the threshold for performing a biopsy has changed. More importantly, there was a concurrent 22% decrease in 30-d complication rates that appears largely attributable to the 2012 USPSTF recommendations. Still, the reduction in complication rates was less than the reduction in biopsy rates due to an increase in the relative morbidity of biopsy, driven by growth in infectious complications. This may in part be related to a shift to older patients with greater comorbidity undergoing biopsy, although similar results were obtained in analyses adjusted for case mix, which suggests additional factors may exist.
To this end, we evaluated predictors of postbiopsy complications to identify actionable targets for further risk reduction. Some of these features represent modifiable risk factors (eg, anticoagulant use), and others represent opportunities for altering periprocedural management. For instance, prior fluoroquinolone use may identify patients who would benefit from prebiopsy rectal swabs to guide periprocedural antibiotics [24] and [25]. Nonmodifiable risk factors, including specific comorbidities such as cancer and peripheral vascular disease, may encourage reconsideration of the risk–benefit ratio before proceeding with biopsy.
It is also important to consider the concurrent impact of landmark PSA screening publications on rates of PCa diagnosis to contextualize the observed changes in biopsy and its morbidity. An exploratory analysis suggested that cancer diagnosis rates have steadily increased, although further studies are necessary to examine this issue comprehensively.
A number of studies have examined the impact of the 2008 USPSTF recommendations [7], [8], [9], and [22], ERSPC/PLCO trials [6] and [9], and 2012 USPSTF recommendations [6], [10], [11], [12], and [23] on PSA screening rates, generally noting reductions ranging from 0% to 8%. However, much less data are available on the impact of these publications on biopsy rates. One single-center study noted a 31% reduction in the number of biopsies performed following the 2012 USPSTF recommendations [13]; another noted no change in the proportion of referred men who underwent biopsy [26]. A report from a Canadian academic health network noted a reduction in biopsy rates from 58.0 to 35.5 biopsies per month following the 2012 USPSTF recommendations [27]. Our findings therefore confirm a reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines.
Limited data exist regarding secular trends in postbiopsy complications, and prior studies have focused exclusively on the relative morbidity among men undergoing biopsy rather than absolute complication rates. A Canadian population-based study observed that the 30-d hospitalization rate increased from 1.0% in 1996 to 4.1% in 2005, driven by an increase in infectious complications [28]. A Surveillance, Epidemiology and End Results–Medicare study also noted an increase in hospitalizations related to infectious complications over time [14]. Although we observed a similar increase in 30-d overall complications among men undergoing biopsy, likewise driven by infectious complications, there was an absolute 22% reduction in 30-d complications over the study period. Taken together, these findings support a reduction in the morbidity of biopsy following publication of revised PSA screening guidelines but highlight the continued need to reduce the relative morbidity of biopsy.
A number of studies have described complication rates following prostate biopsy [15], [16], [29], and [30]. Hematuria has been reported in 10–84% of biopsies, rectal bleeding in 1–45%, infectious complications in up to 6.3%, urinary retention in up to 1.7%, and hospitalization in up to 6.9% [14] and [29]. The wide range in complication rates is due to differences in definitions, ascertainment of end points, and variation in clinical practice. Given the heterogeneity in postbiopsy complication rates, our findings have two advantages. First, they represent a contemporary, geographically diverse population across the entire United States. Second, use of claims-based ascertainment limited complications to those cases in which patients sought medical care, and although this may underestimate the comprehensive morbidity of biopsy, it more accurately reflects the economic burden of serious complications.
This study has several limitations. Most importantly, a control population of men who were not exposed to the PSA screening publications was not available. This study is retrospective, and additional, unrecognized changes in policy or clinical practice may possibly have confounded the impact of the interventions examined. However, the ITS approach has been described as the strongest quasi-experimental design to evaluate policy changes when a randomized trial is not possible [19]. Because we relied on administrative claims for ascertainment of complications, minor complications that did not generate a claim were not captured, thus underestimating these end points. Although we adjusted for case mix, we were unable to adjust for PSA. However, population shifts in PSA may be confounded by changes in screening practices, and PSA adjustment may therefore under- or overestimate the impact of interventions. We were also unable to adjust for other factors that may be associated with biopsy or complication rates, such as prostate magnetic resonance imaging (MRI), ancillary biomarker testing, biopsy approach (eg, MRI-fusion biopsy, transperineal biopsy), indication for anticoagulation, use of periprocedural bridging, or operator biopsy volume/experience.
We observed an immediate reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines, with a smaller reduction in absolute 30-d complication rates. Still, the relative morbidity of biopsy has increased over time, driven by infectious complications. Accordingly, if health policy aims to reduce the morbidity of PSA screening related to biopsy, further efforts should focus on improving patient selection and periprocedural processes of care.
Author contributions: Boris Gershman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Gershman, van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Acquisition of data: Van Houten, Shah.
Analysis and interpretation of data: Gershman, Van Houten, Shah, Karnes.
Drafting of the manuscript: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Critical revision of the manuscript for important intellectual content: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Statistical analysis: Van Houten.
Obtaining funding: Karnes, Shah.
Administrative, technical, or material support: Karnes, Shah.
Supervision: Karnes, Shah.
Other (specify): None.
Financial disclosures: Boris Gershman certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: This publication was made possible by funding from the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery.
Funding/Support and role of the sponsor: None.
In 2009, two landmark randomized trials—the European Randomized Study of Screening for Prostate Cancer [ERSPC] [1] and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial [PLCO] [2]—reported smaller than anticipated reductions in prostate cancer (PCa) mortality and prompted reexamination of prostate-specific antigen (PSA) screening recommendations. The US Preventive Services Task Force (USPSTF), which in 2008 had recommended against PSA screening in men aged >75 yr [3], updated its position in 2012 to recommend against systematic PSA screening in the general population [4]. The American Urological Association (AUA) likewise updated its PSA screening guidelines in May 2013, recommending shared decision making for men aged 55–69 yr [5]. Both the 2012 USPSTF and 2013 AUA guidelines cited prostate biopsy and its potential complications as a significant harm associated with PSA screening.
Although several studies have noted a reduction in PSA screening rates following these publications [6], [7], [8], [9], [10], [11], and [12], there are little data on whether this has translated into the intended reductions in screening-related morbidity in the form of prostate biopsy and postbiopsy complications [13]. We evaluated whether publication of the 2008 USPSTF recommendations, 2009 ERSPC/PLCO trials, 2012 USPSTF recommendations, or 2013 AUA guidelines was associated with reductions in prostate biopsy or postbiopsy complications. Biopsy-related complications also remain understudied both with regard to incidence as well as identification of risk factors [14], [15], and [16]. Consequently, we also sought to characterize postbiopsy complications and evaluate predictors using a contemporary population-based cohort.
We conducted a retrospective analysis of administrative claims from the Optum Labs Data Warehouse that contains data on >100 million individuals enrolled in private health plans and Medicare Advantage Plans, from geographically diverse regions across the United States, as described in Supplementary Methods [17]. Because this study involved analysis of preexisting deidentified data, it was exempt from institutional review board approval.
We identified all prostate biopsies performed in men aged ≥40 yr between January 1, 2005, and September 30, 2014, using Current Procedural Terminology-4 code 55700 alone (11%) or in combination with 76872 (84%) or 76942 (0%), or International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 60.11 (5%). The first biopsy was selected as the index biopsy when more than one biopsy was performed. Patients with a prior diagnosis of PCa (ICD-9 code 185) were excluded. We restricted the cohort to patients with continuous medical and pharmacy enrollment for at least 12 mo prior to biopsy and 30 d after biopsy to ascertain comorbidities and 30-d complications (Supplementary Fig. 1).
We recorded sociodemographic characteristics, comorbid conditions, anticoagulant use within 30 d of biopsy, and fluoroquinolone use in the 12 mo prior to biopsy as described in Supplement 1. The Charlson Comorbidity Index was calculated as previously described [18]. Prebiopsy PSA values within 60 d of biopsy were available for a subset of patients.
The primary outcomes were crude and adjusted rates of prostate biopsy and 30-d complications. Complications were identified using inpatient and outpatient administrative claims as summarized in Supplementary Table 1.
Baseline demographic features were compared across study periods using the Kruskal-Wallis test for continuous variables and the chi-square test for discrete variables. To evaluate the impact of each publication on rates of prostate biopsy and 30-d complications, we used an interrupted time series (ITS) approach as described in Supplementary Methods [19] and [20]. Rates were estimated using all men aged ≥40 yr with at least 12 mo of continuous medical and pharmacy enrollment as the denominator. As secondary analyses, we estimated rates of 30-d complications among men who underwent biopsy. Two separate ITS analyses were performed: (1) unadjusted analyses, using crude rates of biopsy and complications aggregated for each month, and (2) analyses standardized for case mix, to adjust for population shifts over time as described in Supplementary Methods [20] and [21].
Logistic regression was used to evaluate predictors of 30-d complications. Multivariable models were adjusted for age at biopsy, year of biopsy, race, anticoagulant use, history of fluoroquinolone use, and the presence of the five most prevalent Charlson comorbidities.
Statistical analyses were performed using SAS v.9.3 (SAS Institute Inc., Cary, NC, USA). All tests were two sided with p < 0.05 considered statistically significant.
Between January 2005 and September 2014, a total of 104 584 men underwent prostate biopsy (Table 1). Median age at biopsy was 63.0 yr (interquartile range [IQR]: 57.0–70.0). Overall, 18% of men had used fluoroquinolones in the 12 mo preceding biopsy, and 2.8% were on anticoagulants. Median prebiopsy PSA was 5.1 ng/dl (IQR: 4.0–7.1) for the 18 749 men (18%) for whom this was available. Over the time periods, men undergoing biopsy were older, had increased prevalence of anticoagulant use and comorbid conditions, and had a higher PSA (Table 1 and Supplementary Fig. 2).
Table 1
Patient demographics for men who underwent prostate biopsy (January 2005 to September 2014) stratified by pre/postintervention time periods (n = 104 584)
| Baseline Jan 2005–Jul 2008 (n = 37 217) |
2008 USPSTF Aug 2008–Feb 2009 (n = 6672) |
ERSPC/PLCO Mar 2009–Apr 2012 (n = 36 603) |
2012 USPSTF May 2012–Apr 2013 (n = 10 772) |
2013 AUA May 2013–Sep 2014 (n = 13 320) |
p value | Total (n = 104 584) | |
|---|---|---|---|---|---|---|---|
| Age at biopsy, yr | |||||||
| Median (IQR) | 62.0 (56.0–68.0) | 62.0 (56.0–68.0) | 63.0 (57.0–70.0) | 65.0 (59.0–71.0) | 66.0 (59.0–71.0) | <0.001 | 63.0 (57.0–70.0) |
| Age group, yr (%) | |||||||
| 40–49 | 2269 (6.1) | 416 (6.2) | 1737 (4.7) | 420 (3.9) | 467 (3.5) | <0.001 | 5309 (5.1) |
| 50–59 | 12 366 (33) | 2201 (33) | 10 342 (28) | 2644 (25) | 3098 (23) | 30 651 (29) | |
| 60–69 | 14 671 (39) | 2697 (40) | 15 105 (41) | 4426 (41) | 5546 (42) | 42 445 (41) | |
| ≥70 | 7911 (21) | 1358 (20) | 9419 (26) | 3282 (31) | 4209 (32) | 26 179 (25) | |
| Race (%) | |||||||
| White | 25 508 (69) | 4643 (70) | 26 187 (72) | 7795 (72) | 9407 (71) | <0.001 | 73 540 (70) |
| African American | 3695 (9.9) | 780 (12) | 4423 (12) | 1283 (12) | 1589 (12) | 11 770 (11) | |
| Hispanic | 2549 (6.8) | 538 (8.1) | 2911 (8.0) | 837 (7.8) | 1186 (8.9) | 8021 (7.7) | |
| Asian | 1133 (3.0) | 236 (3.5) | 1142 (3.1) | 306 (2.8) | 387 (2.9) | 3204 (3.1) | |
| Unknown/Missing | 4332 (12) | 475 (7.1) | 1940 (5.3) | 551 (5.1) | 751 (5.6) | 8049 (7.7) | |
| Census region (%) | |||||||
| Northeast | 7094 (19) | 1196 (18) | 6120 (17) | 1886 (18) | 2424 (18) | <0.001 | 18 720 (18) |
| Midwest | 10 094 (27) | 1562 (23) | 8837 (24) | 2742 (26) | 3627 (27) | 26 862 (26) | |
| South | 15 330 (41) | 2889 (43) | 16 371 (45) | 4698 (44) | 5538 (42) | 44 826 (43) | |
| West | 4660 (13) | 993 (15) | 5147 (14) | 1442 (13) | 1728 (13) | 13 970 (13) | |
| Other/Missing | 39 (0.1) | 32 (0.5) | 128 (0.3) | 4 (0) | 3 (0) | 206 (0.2) | |
| PSA, ng/d | n = 5174 | n = 548 | n = 7345 | n = 2211 | n = 3471 | n = 18 749 | |
| Median (IQR) | 5.1 (3.9–7.1) | 4.9 (3.8–6.8) | 5.0 (4.0–6.8) | 5.3 (4.2–7.4) | 5.4 (4.2–7.8) | <0.001 | 5.1 (4.0–7.1) |
| Anticoagulant use (%) | 928 (2.5) | 184 (2.8) | 1103 (3.0) | 299 (2.8) | 415 (3.1) | <0.001 | 2929 (2.8) |
| History of fluoroquinolone use in prior 12 mo (%) | 6797 (18) | 1267 (19) | 6557 (18) | 2017 (19) | 2504 (19) | 0.054 | 19 142 (18) |
| Charlson Comorbidity Index (%) | |||||||
| 0 | 22 354 (60) | 3887 (58) | 20 269 (55) | 5758 (54) | 7039 (53) | <0.001 | 59 307 (57) |
| 1 | 8199 (22) | 1453 (22) | 8252 (23) | 2503 (23) | 2999 (23) | 23 406 (22) | |
| 2 | 3232 (8.7) | 664 (10) | 3520 (9.6) | 1108 (10) | 1359 (10) | 9883 (9.4) | |
| ≥3 | 3432 (9.2) | 668 (10) | 4562 (13) | 1403 (13) | 1923 (14) | 11 988 (12) | |
| Comorbidities (%) | |||||||
| Myocardial infarction | 763 (2.1) | 144 (2.2) | 812 (2.2) | 227 (2.1) | 273 (2.0) | 0.6 | 2219 (2.1) |
| Congestive heart failure | 1443 (3.9) | 223 (3.3) | 1548 (4.2) | 467 (4.3) | 586 (4.4) | <0.001 | 4267 (4.1) |
| Peripheral vascular disease | 2054 (5.5) | 399 (6.0) | 2660 (7.3) | 783 (7.3) | 1071 (8.0) | <0.001 | 6967 (6.7) |
| Cerebrovascular disease | 2007 (5.4) | 379 (5.7) | 2358 (6.4) | 681 (6.3) | 947 (7.1) | <0.001 | 6372 (6.1) |
| Dementia | 370 (1.0) | 63 (0.9) | 442 (1.2) | 141 (1.3) | 198 (1.5) | <0.001 | 1214 (1.2) |
| Chronic obstructive pulmonary disease | 4251 (11) | 758 (11) | 4293 (12) | 1254 (12) | 1686 (13) | 0.004 | 12 242 (12) |
| Ulcer | 362 (1.0) | 60 (0.9) | 297 (0.8) | 84 (0.8) | 102 (0.8) | 0.07 | 905 (0.9) |
| Mild liver disease | 1182 (3.2) | 223 (3.3) | 1332 (3.6) | 413 (3.8) | 515 (3.9) | <0.001 | 3665 (3.5) |
| Diabetes mellitus | 6475 (17) | 1290 (19) | 7823 (21) | 2442 (23) | 2973 (22) | <0.001 | 21 003 (20) |
| Diabetes with organ damage | 1105 (3.0) | 225 (3.4) | 1600 (4.4) | 492 (4.6) | 734 (5.5) | <0.001 | 4156 (4.0) |
| Hemiplegia | 128 (0.3) | 25 (0.4) | 150 (0.4) | 51 (0.5) | 56 (0.4) | 0.3 | 410 (0.4) |
| Moderate/severe renal disease | 1133 (3.0) | 254 (3.8) | 1950 (5.3) | 674 (6.3) | 908 (6.8) | <0.001 | 4919 (4.7) |
| Moderate/severe liver disease | 27 (0.1) | 7 (0.1) | 31 (0.1) | 9 (0.1) | 12 (0.1) | 0.9 | 86 (0.1) |
| Metastatic solid tumor | 175 (0.5) | 37 (0.6) | 199 (0.5) | 58 (0.5) | 80 (0.6) | 0.4 | 549 (0.5) |
| AIDS | 82 (0.2) | 16 (0.2) | 101 (0.3) | 27 (0.3) | 46 (0.3) | 0.2 | 272 (0.3) |
| Rheumatologic disease | 449 (1.2) | 71 (1.1) | 522 (1.4) | 158 (1.5) | 196 (1.5) | 0.007 | 1396 (1.3) |
| Cancer | 1736 (4.7) | 337 (5.1) | 1796 (4.9) | 571 (5.3) | 701 (5.3) | 0.02 | 5141 (4.9) |
| Net worth range (%) | |||||||
| <$25 000 | 1831 (4.9) | 404 (6.1) | 2101 (5.7) | 561 (5.2) | 695 (5.2) | <0.001 | 5592 (5.3) |
| $25 000–149 000 | 5188 (14) | 1042 (16) | 5608 (15) | 1591 (15) | 2046 (15) | 15 475 (15) | |
| $150 000–249 000 | 4754 (13) | 837 (13) | 5182 (14) | 1499 (14) | 1862 (14) | 14 134 (14) | |
| $250 000–499 000 | 9694 (26) | 1779 (27) | 10 048 (28) | 3027 (28) | 3636 (27) | 28 184 (27) | |
| ≥$500 000 | 10 910 (29) | 2030 (30) | 10 860 (30) | 3208 (30) | 3852 (29) | 30 860 (30) | |
| Unknown/missing | 4840 (13) | 580 (8.7) | 2804 (7.7) | 886 (8.2) | 1229 (9.2) | 10 339 (9.9) | |
| 30-d complications (%) | |||||||
| Any | 5973 (16) | 1151 (17) | 6120 (17) | 1863 (17) | 2451 (18) | <0.001 | 17 558 (17) |
| Bleeding | 2547 (6.8) | 499 (7.5) | 2660 (7.3) | 860 (8.0) | 1083 (8.1) | <0.001 | 7649 (7.3) |
| Nonsepsis infection | 2978 (8.0) | 590 (8.8) | 2990 (8.2) | 875 (8.1) | 1236 (9.3) | <0.001 | 8669 (8.3) |
| Sepsis | 191 (0.5) | 50 (0.7) | 349 (1.0) | 105 (1.0) | 134 (1.0) | <0.001 | 829 (0.8) |
| Urinary retention | 980 (2.6) | 161 (2.4) | 1138 (3.1) | 341 (3.2) | 422 (3.2) | <0.001 | 3042 (2.9) |
| Hospitalization | 1019 (2.7) | 181 (2.7) | 1056 (2.9) | 320 (3.0) | 367 (2.8) | 0.6 | 2943 (2.8) |
AIDS = acquired immunodeficiency syndrome; AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; IQR = interquartile range; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer [screening trial]; PSA = prostate-specific antigen; USPTF = US Preventive Services Task Force.
Overall, 30-d complications occurred in 17% of biopsies. The most prevalent was nonsepsis genitourinary infections (8.3%), followed by bleeding complications (7.3%) and urinary retention (2.9%) (Table 1). The 30-d hospitalization rate was 2.8%.
We first examined unadjusted rates of prostate biopsy among men aged ≥40 yr (Fig. 1 and Table 2). From January 2005 to September 2014, biopsy rates fell 33% from 64.1 to 42.8 per 100 000 person-months. There was substantial seasonal variation in biopsy rates with reductions near the end of each calendar year. With the ITS approach, each intervention is potentially associated with an immediate change in event rates (ie, level shift or change in intercept) as well as a change in the rate of growth (ie, change in slope). At baseline, there was a statistically significant growth in prostate biopsy rates of 2.9 biopsies per 100 000 person-months per year (95% confidence interval [CI], 0.96–4.8; p < 0.001). Following the 2008 USPSTF recommendations, there was an immediate decrease (ie, level shift) in biopsy rates of −10.1 biopsies per 100 000 person-months (95% CI, −17.1 to −3.0; p < 0.001). Likewise, there were immediate decreases in biopsy rates following the 2012 USPSTF recommendations (−13.8 biopsies per 100 000 person-months; 95% CI, −21.0 to −6.7; p < 0.001) and 2013 AUA guidelines (−8.8 biopsies per 100 000 person-months; 95% CI, −16.7 to −0.92; p = 0.03). Analyses adjusted for case mix were similar, although the baseline growth rate was no longer significant, indicating this was related to population shifts over time (Table 2, model 2).
Fig. 1
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 2
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315)
| Prostate biopsy | Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Model 1* | ||||||||||||||
| Baseline rate | 57.8 (53.4–62.3) |
<0.001 | 8.6 (7.5–9.7) |
<0.001 | 4.0 (3.3–4.6) |
<0.001 | 0.17 (0.03–0.32) |
0.02 | 2.8 (2.3–3.3) |
<0.001 | 1.5 (1.3–1.6) |
<0.001 | 1.4 (1.1–1.7) |
<0.001 |
| Baseline trend in rate | 2.9 (0.96–4.8) |
<0.001 | 0.84 (0.48–1.3) |
<0.001 | 0.72 (0.48–1.1) |
<0.001 | 0.12 (0.12–0.12) |
<0.001 | 0.36 (0.12–0.48) |
<0.001 | 0.12 (0.00–0.12) |
<0.001 | 0.00 (−0.12 to 0.12) |
0.5 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.1 (−17.1 to −3.0) |
<0.001 | −0.61 (−3.1 to 1.9) |
0.6 | −1.4 (−2.8 to −0.03) |
0.053 | 0.04 (−0.35 to 0.42) |
0.8 | 0.56 (−0.70 to 1.8) |
0.4 | −0.06 (−0.69 to 0.57) |
0.9 | 0.16 (−0.58 to 0.90) |
0.7 |
| Trend change | 7.9 (−12.7 to 28.4) |
0.5 | −1.0 (−9.1 to 7.2) |
0.8 | −0.12 (−4.1 to 4.0) |
0.98 | −0.60 (−1.9 to 0.72) |
0.4 | −2.0 (−6.5 to 2.4) |
0.4 | −1.1 (−3.4 to 1.3) |
0.4 | −1.0 (−3.6 to 1.6) |
0.5 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.6 (−12 to 4.5) |
0.4 | −0.74 (−3.8 to 2.4) |
0.6 | −0.12 (−1.8 to 1.5) |
0.9 | 0.29 (−0.20 to 0.78) |
0.2 | −0.04 (−1.6 to 1.6) |
0.97 | 0.59 (−0.23 to 1.4) |
0.2 | 0.51 (−0.43 to 1.5) |
0.3 |
| Trend change | −11.2 (−32.0 to 9.6) |
0.3 | 0.0 (−8.2 to 8.0) |
0.99 | −0.96 (−5.0 to 3.1) |
0.7 | 0.36 (−0.96 to 1.8) |
0.5 | 1.7 (−2.8 to 6.1) |
0.5 | 1.0 (−1.4 to 3.4) |
0.4 | 0.84 (−1.7 to 3.5) |
0.5 |
| 2012 USPSTF | ||||||||||||||
| Level change | −13.8 (−21.0 to −6.7) |
<0.001 | −2.5 (−4.5 to −0.45) |
0.02 | −1.7 (−3.1 to −0.31) |
0.02 | −0.12 (−0.42 to 0.19) |
0.5 | −0.61 (−1.5 to 0.28) |
0.2 | −0.75 (−1.1 to −0.44) |
<0.001 | −0.25 (−0.77 to 0.27) |
0.4 |
| Trend change | 8.2 (−5.8 to 22.0) |
0.3 | 3.0 (−0.36 to 6.4) |
0.09 | 3.1 (0.72–5.5) |
0.01 | 0.12 (−0.36 to 0.60) |
0.7 | 0.84 (−0.72 to 2.3) |
0.3 | 0.48 (0.00 to 1.1) |
0.08 | 0.24 (−0.60 to 1.2) |
0.6 |
| 2013 AUA | ||||||||||||||
| Level change | −8.8 (−16.7 to −0.92) |
0.03 | −1.4 (−3.8 to 1.1) |
0.3 | −1.9 (−3.5 to −0.22) |
0.03 | −0.02 (−0.41 to 0.37) |
0.9 | −0.19 (−1.3 to 0.91) |
0.7 | −0.38 (−0.77 to 0.01) |
0.06 | −0.01 (−0.66 to 0.63) |
0.96 |
| Trend change | −10.4 (−30.0 to 9.2) |
0.3 | −4.1 (−7.9 to −0.24) |
0.04 | −2.9 (−6.0 to 0.12) |
0.07 | −0.24 (−0.72 to 0.36) |
0.5 | −1.8 (−3.5 to 0.0) |
0.046 | −0.60 (−1.2 to 0.12) |
0.09 | −0.60 (−1.7 to 0.36) |
0.2 |
| Model 2† | ||||||||||||||
| Baseline rate | 86.8 (82.1–91.4) |
<0.001 | 16.2 (14.9–17.6) |
<0.001 | 6.4 (4.9–7.9) |
<0.001 | 0.39 (−0.13 to 0.91) |
0.2 | 6.1 (4.6–7.7) |
<0.001 | 3.6 (2.8–4.4) |
<0.001 | 3.0 (2.2–3.8) |
<0.001 |
| Baseline trend in rate | 0.24 (−1.8 to 2.3) |
0.8 | −0.24 (−0.72 to 0.24) |
0.4 | 0.48 (−0.12 to 1.1) |
0.09 | 0.12 (−0.12 to 0.24) |
0.4 | −0.24 (−0.72 to 0.24) |
0.4 | −0.24 (−0.48 to 0.00) |
0.054 | −0.12 (−0.36 to 0.12) |
0.3 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.8 (−17.8 to −3.8) |
<0.001 | −5.4 (−9.6 to −1.2) |
0.01 | −2.5 (−5.6 to 0.71) |
0.1 | −0.61 (−1.9 to 0.71) |
0.4 | 0.14 (−3.8 to 4.1) |
0.9 | 1.1 (−0.98 to 3.1) |
0.3 | 0.14 (−1.9 to 2.1) |
0.9 |
| Trend change | 10.6 (−10.1 to 31.2) |
0.3 | 6.1 (−8.4 to 20.8) |
0.4 | −1.7 (−12.2 to 8.8) |
0.8 | 1.7 (−2.9 to 6.2) |
0.5 | 1.7 (−12.0 to 15.5) |
0.8 | −4.7 (−11.8 to 2.5) |
0.2 | −0.84 (−7.8 to 6.1) |
0.8 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.9 (−12 to 4.1) |
0.3 | −0.24 (−5.6 to 5.1) |
0.9 | 1.9 (−2.1 to 5.8) |
0.4 | −0.32 (−2.0 to 1.4) |
0.7 | −0.99 (−6.0 to 4.0) |
0.7 | 1.8 (−0.84 to 4.4) |
0.2 | 0.87 (−1.7 to 3.4) |
0.5 |
| Trend change | −16.2 (−37.1 to 4.7) |
0.1 | −6.8 (−21.4 to 7.8) |
0.4 | 0.36 (−10.2 to 10.9) |
0.95 | −1.9 (−6.5 to 2.8) |
0.4 | −1.6 (−15.2 to 12.2) |
0.8 | 4.7 (−2.4 to 11.9) |
0.2 | 0.48 (−6.4 to 7.4) |
0.9 |
| 2012 USPSTF | ||||||||||||||
| Level change | −11.0 (−18.1 to −4.0) |
0.001 | −2.8 (−6.0 to 0.38) |
0.09 | 0.53 (−1.9 to 3.0) |
0.7 | −0.02 (−0.95 to 0.91) |
0.97 | −1.8 (−4.6 to 0.96) |
0.2 | −1.5 (−3.1 to 0.17) |
0.08 | 0.16 (−1.3 to 1.6) |
0.8 |
| Trend change | 7.0 (−7.2 to 21.1) |
0.3 | 1.7 (−3.6 to 7.0) |
0.5 | 0.12 (−4.1 to 4.2) |
0.96 | 0.12 (−1.4 to 1.8) |
0.9 | −0.48 (−5.3 to 4.3) |
0.9 | 1.6 (−1.3 to 4.3) |
0.3 | 0.12 (−2.4 to 2.6) |
0.9 |
| 2013 AUA | ||||||||||||||
| Level change | −8.2 (−15.9 to −0.38) |
0.04 | −1.0 (−4.8 to 2.8) |
0.6 | −0.44 (−3.2 to 2.3) |
0.8 | −0.15 (−1.3 to 0.99) |
0.8 | 0.28 (−3.2 to 3.7) |
0.9 | −0.50 (−2.7 to 1.7) |
0.7 | 0.47 (−1.3 to 2.3) |
0.6 |
| Trend change | −4.2 (−24.5 to 16.1) |
0.7 | −0.7 (−7.6 to 6.2) |
0.9 | 1.1 (−4.2 to 6.2) |
0.7 | −0.12 (−1.8 to 1.7) |
0.9 | −0.60 (−5.9 to 4.8) |
0.8 | −1.8 (−4.9 to 1.2) |
0.3 | −0.48 (−3.1 to 2.2) |
0.7 |
* Model 1: Base model with constant, time trend, intervention indicator variables. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
† Model 2: Adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, and anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPTF = US Preventive Services Task Force.
Next, we examined 30-d postbiopsy complications among men aged ≥40 yr as a measure of the absolute morbidity of biopsy (Fig. 1, Supplementary Fig. 3, and Table 2). Across the study periods, overall complication rates decreased 10% from 8.7 to 7.8 per 100 000 person-months. In unadjusted ITS analyses, there was a baseline growth in the overall complication rate (0.84 per 100 000 person-months per year; 95% CI, 0.48–1.3; p < 0.001), driven by growth in the rate of nonsepsis infections (0.72 per 100 000 person-months per year; 95% CI, 0.48–1.1; p < 0.001). Only the 2012 USPSTF recommendations were associated with a significant change in complication rates, following which there was an immediate decrease in the overall complication rate of −2.5 per 100 000 person-months (95% CI, −4.5 to −0.45; p = 0.02), driven by reductions in nonsepsis infections (−1.7 per 100 000 person-months; 95% CI, −3.1 to −0.31; p = 0.02) and urinary retention (−0.75 per 100 000 person-months;95% CI, −1.1 to −0.44; p < 0.001). Following the 2013 AUA guidelines there was also a reduction in the overall complication baseline growth rate (−4.1 per 100 000 person-months per year; 95% CI, −7.9 to −0.24; p = 0.04). In analyses adjusted for case mix (Table 2, model 2), effects were overall attenuated, and only the 2008 USPSTF recommendations were associated with a reduction in overall complication rates (−5.4 per 100 000 person-months; 95% CI, −9.6 to −1.2; p = 0.01).
To examine whether there was a change in the relative morbidity of biopsy, we performed a secondary analysis examining complication rates among men undergoing biopsy (Fig. 2). The proportion of men undergoing biopsy who experienced a complication increased from 14% in January 2005 to 18% in September 2014, an increase of 36%. There was again a baseline growth in overall complication rates (0.72% per year; 95% CI, 0.36–1.1; p < 0.001), driven by an increase in nonsepsis infections and sepsis (Supplementary Table 2). None of the interventions affected this growth. Similar results were obtained after adjusting for case mix (Table 3).
Fig. 2
Interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 3
Adjusted* interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Baseline rate | 15.6 (14.3–16.8) |
<0.001 | 7.1 (6.5–7.7) |
<0.001 | 0.34 (0.11–0.57) |
<0.001 | 6.7 (5.9–7.5) |
<0.001 | 2.8 (2.5–3.2) |
<0.001 | 3.2 (2.8–3.6) |
<0.001 |
| Baseline trend in rate | 0.60 (0.12–1.1) |
0.02 | 0.72 (0.60–0.96) |
<0.001 | 0.12 (0.12–0.24) |
<0.001 | 0.12 (−0.12 to 0.36) |
0.3 | −0.12 (−0.24 to 0.00) |
0.3 | −0.12 (−0.24 to 0.00) |
0.0498 |
| 2008 USPSTF | ||||||||||||
| Level change | 1.1 (−2.0 to 4.1) |
0.5 | −1.3 (−3.2 to 0.58) |
0.2 | 0.13 (−0.45 to 0.72) |
0.7 | 1.9 (−0.11 to 3.9) |
0.07 | 0.46 (−0.50 to 1.4) |
0.4 | 0.27 (−0.69 to 1.2) |
0.6 |
| Trend change | −3.5 (−13.4 to 6.6) |
0.5 | −1.8 (−8.4 to 4.7) |
0.6 | −1.0 (−3.0 to 0.96) |
0.3 | −4.3 (−11.4 to 2.6) |
0.2 | −1.6 (−4.9 to 1.9) |
0.4 | −0.72 (−4.0 to 2.6) |
0.7 |
| ERSPC/PLCO | ||||||||||||
| Level change | −0.14 (−3.9 to 3.6) |
0.9 | 1.1 (−1.3 to 3.4) |
0.4 | 0.55 (−0.19 to 1.3) |
0.2 | 0.39 (−2.2 to 3.0) |
0.8 | 0.75 (−0.47 to 2.0) |
0.2 | 0.65 (−0.56 to 1.9) |
0.3 |
| Trend change | 2.9 (−7.1 to 13.0) |
0.6 | 0.72 (−5.8 to 7.2) |
0.8 | 0.84 (−1.2 to 2.9) |
0.4 | 4.2 (−2.8 to 11.3) |
0.2 | 1.7 (−1.8 to 5.0) |
0.4 | 0.72 (−2.6 to 4.0) |
0.7 |
| 2012 USPSTF | ||||||||||||
| Level change | −0.27 (−2.7 to 2.1) |
0.8 | −0.56 (−2.0 to 0.87) |
0.5 | 0.16 (−0.31 to 0.63) |
0.5 | 1.1 (−0.38 to 2.5) |
0.2 | −0.38 (−1.0 to 0.28) |
0.3 | 0.44 (−0.34 to 1.2) |
0.3 |
| Trend change | 1.3 (−2.6 to 5.3) |
0.5 | 2.9 (0.60–5.3) |
0.02 | −0.24 (−1.1 to 0.60) |
0.6 | −0.72 (−3.2 to 1.7) |
0.5 | 0.24 (−0.96 to 1.3) |
0.7 | −0.48 (−1.8 to 0.84) |
0.5 |
| 2013 AUA | ||||||||||||
| Level change | 0.25 (−2.6 to 3.1) |
0.9 | −0.64 (−2.3 to 1.1) |
0.5 | 0.22 (−0.40 to 0.83) |
0.5 | 1.1 (−0.68 to 2.8) |
0.2 | −0.29 (−1.1 to 0.48) |
0.5 | 0.47 (−0.55 to 1.5) |
0.4 |
| Trend change | −1.6 (−6.1 to 2.9) |
0.5 | −2.5 (−5.5 to 0.60) |
0.1 | 0.12 (−0.72 to 1.1) |
0.8 | −0.36 (−3.1 to 2.4) |
0.8 | 0.24 (−1.2 to 1.7) |
0.8 | 0.12 (−1.3 to 1.6) |
0.9 |
* Model adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 person-months. Baseline trend in rate and trend change per 100 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial]; USPTF = US Preventive Services Task Force.
Given the increase in the relative morbidity of biopsy, we performed logistic regression to identify predictors of each complication category (Supplementary Table 3 and Table 4). Due to the large sample size, we focused on those features associated with the greatest magnitude of effect. A prior diagnosis of cancer was associated with an increased risk of all complication categories including any complication (odds ratio [OR]: 1.30; 95% CI, 1.22–1.39; p < 0.001) and hospitalization (OR: 1.59; 95% CI, 1.39–1.82; p < 0.001). Similarly, prior fluoroquinolone use was associated with an increased risk of all complication categories including any complication (OR: 1.27; 95% CI, 1.22–1.32; p < 0.001) and sepsis (OR: 1.55; 95% CI, 1.33–1.81; p < 0.001). Age at biopsy ≥70 yr was associated with a nearly 4-fold increased risk of urinary retention. Although anticoagulant use was associated with increased risks of bleeding (OR: 1.28; 95% CI, 1.14–1.45; p < 0.001), sepsis (OR: 1.48; 95% CI, 1.06–2.07; p = 0.02), and hospitalization (OR: 1.39; 95% CI, 1.16–1.67; p < 0.001), the increase in the overall complication risk was modest (OR 1.14; 95% CI, 1.04–1.25; p = 0.004).
Table 4
Multivariable analysis* of 30-d postbiopsy complications among men aged ≥40 yr undergoing prostate biopsy (January 2005 to September 2014) (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Hospitalization | Urinary retention | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Age at biopsy, yr | ||||||||||||
| 40–49 | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| 50–59 | 1.01 (0.94–1.09) | 0.8 | 0.94 (0.85–1.04) | 0.2 | 1.16 (0.83–1.63) | 0.4 | 1.01 (0.90–1.14) | 0.8 | 1.12 (0.93–1.36) | 0.2 | 1.54 (1.20–1.97) | <0.001 |
| 60–69 | 1.05 (0.97–1.13) | 0.2 | 0.94 (0.85–1.03) | 0.2 | 0.98 (0.70–1.37) | 0.9 | 1.05 (0.94–1.18) | 0.4 | 1.16 (0.95–1.40) | 0.1 | 2.16 (1.68–2.77) | <0.001 |
| ≥70 | 1.25 (1.15–1.36) | <0.001 | 1.03 (0.93–1.15) | 0.6 | 1.05 (0.74–1.49) | 0.8 | 1.29 (1.15–1.46) | <0.001 | 1.31 (1.07–1.60) | 0.008 | 3.98 (3.08–5.15) | <0.001 |
| Race | ||||||||||||
| White | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| Asian | 1.27 (1.15–1.40) | <0.001 | 1.27 (1.11–1.44) | 0.0003 | 1.65 (1.19–2.29) | 0.003 | 1.36 (1.18–1.56) | <0.001 | 0.94 (0.76–1.17) | 0.6 | 1.15 (0.93–1.42) | 0.2 |
| African American | 1.09 (1.03–1.15) | 0.002 | 1.05 (0.97–1.13) | 0.2 | 1.34 (1.09–1.64) | 0.005 | 1.17 (1.09–1.27) | <0.001 | 1.13 (1.01–1.27) | 0.03 | 1.15 (1.03–1.28) | 0.02 |
| Hispanic | 1.32 (1.24–1.40) | <0.001 | 1.36 (1.26–1.47) | <0.001 | 1.52 (1.21–1.90) | <0.001 | 1.33 (1.22–1.45) | <0.001 | 1.12 (0.97–1.28) | 0.1 | 1.37 (1.21–1.56) | <0.001 |
| Unknown/missing | 0.99 (0.93–1.06) | 0.8 | 1.07 (0.99–1.17) | 0.1 | 1.23 (0.94–1.61) | 0.1 | 0.96 (0.88–1.05) | 0.4 | 0.94 (0.81–1.08) | 0.4 | 1.02 (0.88–1.18) | 0.8 |
| Diabetes mellitus | 1.03 (0.99–1.07) | 0.09 | 1.06 (1.00–1.11) | 0.045 | 1.08 (0.92–1.28) | 0.4 | 1.01 (0.96–1.07) | 0.7 | 1.20 (1.10–1.31) | <0.001 | 0.99 (0.91–1.08) | 0.8 |
| COPD | 1.14 (1.09–1.20) | <0.001 | 1.07 (1.00–1.14) | 0.04 | 1.23 (1.01–1.49) | 0.04 | 1.14 (1.06–1.22) | <0.001 | 1.36 (1.23–1.51) | <0.001 | 1.14 (1.03–1.25) | 0.01 |
| PVD | 1.24 (1.16–1.31) | <0.001 | 1.1 (1.02–1.2) | 0.02 | 1.19 (0.93–1.51) | 0.2 | 1.28 (1.18–1.40) | <0.001 | 1.31 (1.15–1.48) | <0.001 | 1.26 (1.12–1.42) | <0.001 |
| Cerebrovascular disease | 1.11 (1.04–1.19) | 0.001 | 1.12 (1.03–1.22) | 0.01 | 1.21 (0.93–1.57) | 0.2 | 1.03 (0.94–1.14) | 0.5 | 1.26 (1.10–1.44) | <0.001 | 1.30 (1.15–1.47) | <0.001 |
| Cancer | 1.30 (1.22–1.39) | <0.001 | 1.16 (1.06–1.27) | 0.001 | 1.37 (1.05–1.79) | 0.02 | 1.40 (1.27–1.53) | <0.001 | 1.59 (1.39–1.82) | <0.001 | 1.32 (1.16–1.52) | <0.001 |
| Anticoagulant use | 1.14 (1.04–1.25) | 0.004 | 1.01 (0.89–1.14) | 0.9 | 1.48 (1.06–2.07) | 0.02 | 1.28 (1.14–1.45) | <0.001 | 1.39 (1.16–1.67) | <0.001 | 1.11 (0.92–1.33) | 0.3 |
| Prior fluoroquinolones | 1.27 (1.22–1.32) | <0.001 | 1.37 (1.31–1.45) | <0.001 | 1.55 (1.33–1.81) | <0.001 | 1.21 (1.14–1.28) | <0.001 | 1.25 (1.14–1.36) | <0.001 | 1.27 (1.16–1.38) | <0.001 |
* Models adjusted for year of biopsy and the features listed in the table.
CI = confidence interval; COPD = chronic obstructive pulmonary disease; PVD = peripheral vascular disease.
We conducted sensitivity analyses where the postintervention run-in period was varied from 2 to 4 mo (Supplementary Table 4). Results overall were similar, although effects were attenuated for the 2013 AUA guidelines on biopsy rates and the 2012 USPSTF recommendations on overall complication rates, likely due to fewer time points for analysis. In a separate sensitivity analysis, we examined bleeding complications excluding microhematuria from 2009 to 2014, following introduction of a separate code for microhematuria. Bleeding rates paralleled those of the primary bleeding end point (Supplementary Fig. 4), and results of both ITS and logistic regression analyses were similar (Supplementary Table 5 and 6). As an exploratory analysis, we examined PCa diagnosis rates within 60 d following biopsy (Supplementary Fig. 5). These appeared to increase steadily across the study period.
To our knowledge, this is the largest study to examine the impact of landmark PSA screening publications on rates of prostate biopsy and the first to examine their impact on postbiopsy complications. Overall, biopsy rates have declined by 33%, which appears attributable to the publication of the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines. Interestingly, the magnitude of reduction in biopsy rates exceeded previously reported decreases in PSA screening rates [6], [7], [8], [9], [10], [11], [12], [22], and [23], which suggests that either the threshold for referral to a urologist or the threshold for performing a biopsy has changed. More importantly, there was a concurrent 22% decrease in 30-d complication rates that appears largely attributable to the 2012 USPSTF recommendations. Still, the reduction in complication rates was less than the reduction in biopsy rates due to an increase in the relative morbidity of biopsy, driven by growth in infectious complications. This may in part be related to a shift to older patients with greater comorbidity undergoing biopsy, although similar results were obtained in analyses adjusted for case mix, which suggests additional factors may exist.
To this end, we evaluated predictors of postbiopsy complications to identify actionable targets for further risk reduction. Some of these features represent modifiable risk factors (eg, anticoagulant use), and others represent opportunities for altering periprocedural management. For instance, prior fluoroquinolone use may identify patients who would benefit from prebiopsy rectal swabs to guide periprocedural antibiotics [24] and [25]. Nonmodifiable risk factors, including specific comorbidities such as cancer and peripheral vascular disease, may encourage reconsideration of the risk–benefit ratio before proceeding with biopsy.
It is also important to consider the concurrent impact of landmark PSA screening publications on rates of PCa diagnosis to contextualize the observed changes in biopsy and its morbidity. An exploratory analysis suggested that cancer diagnosis rates have steadily increased, although further studies are necessary to examine this issue comprehensively.
A number of studies have examined the impact of the 2008 USPSTF recommendations [7], [8], [9], and [22], ERSPC/PLCO trials [6] and [9], and 2012 USPSTF recommendations [6], [10], [11], [12], and [23] on PSA screening rates, generally noting reductions ranging from 0% to 8%. However, much less data are available on the impact of these publications on biopsy rates. One single-center study noted a 31% reduction in the number of biopsies performed following the 2012 USPSTF recommendations [13]; another noted no change in the proportion of referred men who underwent biopsy [26]. A report from a Canadian academic health network noted a reduction in biopsy rates from 58.0 to 35.5 biopsies per month following the 2012 USPSTF recommendations [27]. Our findings therefore confirm a reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines.
Limited data exist regarding secular trends in postbiopsy complications, and prior studies have focused exclusively on the relative morbidity among men undergoing biopsy rather than absolute complication rates. A Canadian population-based study observed that the 30-d hospitalization rate increased from 1.0% in 1996 to 4.1% in 2005, driven by an increase in infectious complications [28]. A Surveillance, Epidemiology and End Results–Medicare study also noted an increase in hospitalizations related to infectious complications over time [14]. Although we observed a similar increase in 30-d overall complications among men undergoing biopsy, likewise driven by infectious complications, there was an absolute 22% reduction in 30-d complications over the study period. Taken together, these findings support a reduction in the morbidity of biopsy following publication of revised PSA screening guidelines but highlight the continued need to reduce the relative morbidity of biopsy.
A number of studies have described complication rates following prostate biopsy [15], [16], [29], and [30]. Hematuria has been reported in 10–84% of biopsies, rectal bleeding in 1–45%, infectious complications in up to 6.3%, urinary retention in up to 1.7%, and hospitalization in up to 6.9% [14] and [29]. The wide range in complication rates is due to differences in definitions, ascertainment of end points, and variation in clinical practice. Given the heterogeneity in postbiopsy complication rates, our findings have two advantages. First, they represent a contemporary, geographically diverse population across the entire United States. Second, use of claims-based ascertainment limited complications to those cases in which patients sought medical care, and although this may underestimate the comprehensive morbidity of biopsy, it more accurately reflects the economic burden of serious complications.
This study has several limitations. Most importantly, a control population of men who were not exposed to the PSA screening publications was not available. This study is retrospective, and additional, unrecognized changes in policy or clinical practice may possibly have confounded the impact of the interventions examined. However, the ITS approach has been described as the strongest quasi-experimental design to evaluate policy changes when a randomized trial is not possible [19]. Because we relied on administrative claims for ascertainment of complications, minor complications that did not generate a claim were not captured, thus underestimating these end points. Although we adjusted for case mix, we were unable to adjust for PSA. However, population shifts in PSA may be confounded by changes in screening practices, and PSA adjustment may therefore under- or overestimate the impact of interventions. We were also unable to adjust for other factors that may be associated with biopsy or complication rates, such as prostate magnetic resonance imaging (MRI), ancillary biomarker testing, biopsy approach (eg, MRI-fusion biopsy, transperineal biopsy), indication for anticoagulation, use of periprocedural bridging, or operator biopsy volume/experience.
We observed an immediate reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines, with a smaller reduction in absolute 30-d complication rates. Still, the relative morbidity of biopsy has increased over time, driven by infectious complications. Accordingly, if health policy aims to reduce the morbidity of PSA screening related to biopsy, further efforts should focus on improving patient selection and periprocedural processes of care.
Author contributions: Boris Gershman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Gershman, van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Acquisition of data: Van Houten, Shah.
Analysis and interpretation of data: Gershman, Van Houten, Shah, Karnes.
Drafting of the manuscript: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Critical revision of the manuscript for important intellectual content: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Statistical analysis: Van Houten.
Obtaining funding: Karnes, Shah.
Administrative, technical, or material support: Karnes, Shah.
Supervision: Karnes, Shah.
Other (specify): None.
Financial disclosures: Boris Gershman certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: This publication was made possible by funding from the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery.
Funding/Support and role of the sponsor: None.
In 2009, two landmark randomized trials—the European Randomized Study of Screening for Prostate Cancer [ERSPC] [1] and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial [PLCO] [2]—reported smaller than anticipated reductions in prostate cancer (PCa) mortality and prompted reexamination of prostate-specific antigen (PSA) screening recommendations. The US Preventive Services Task Force (USPSTF), which in 2008 had recommended against PSA screening in men aged >75 yr [3], updated its position in 2012 to recommend against systematic PSA screening in the general population [4]. The American Urological Association (AUA) likewise updated its PSA screening guidelines in May 2013, recommending shared decision making for men aged 55–69 yr [5]. Both the 2012 USPSTF and 2013 AUA guidelines cited prostate biopsy and its potential complications as a significant harm associated with PSA screening.
Although several studies have noted a reduction in PSA screening rates following these publications [6], [7], [8], [9], [10], [11], and [12], there are little data on whether this has translated into the intended reductions in screening-related morbidity in the form of prostate biopsy and postbiopsy complications [13]. We evaluated whether publication of the 2008 USPSTF recommendations, 2009 ERSPC/PLCO trials, 2012 USPSTF recommendations, or 2013 AUA guidelines was associated with reductions in prostate biopsy or postbiopsy complications. Biopsy-related complications also remain understudied both with regard to incidence as well as identification of risk factors [14], [15], and [16]. Consequently, we also sought to characterize postbiopsy complications and evaluate predictors using a contemporary population-based cohort.
We conducted a retrospective analysis of administrative claims from the Optum Labs Data Warehouse that contains data on >100 million individuals enrolled in private health plans and Medicare Advantage Plans, from geographically diverse regions across the United States, as described in Supplementary Methods [17]. Because this study involved analysis of preexisting deidentified data, it was exempt from institutional review board approval.
We identified all prostate biopsies performed in men aged ≥40 yr between January 1, 2005, and September 30, 2014, using Current Procedural Terminology-4 code 55700 alone (11%) or in combination with 76872 (84%) or 76942 (0%), or International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 60.11 (5%). The first biopsy was selected as the index biopsy when more than one biopsy was performed. Patients with a prior diagnosis of PCa (ICD-9 code 185) were excluded. We restricted the cohort to patients with continuous medical and pharmacy enrollment for at least 12 mo prior to biopsy and 30 d after biopsy to ascertain comorbidities and 30-d complications (Supplementary Fig. 1).
We recorded sociodemographic characteristics, comorbid conditions, anticoagulant use within 30 d of biopsy, and fluoroquinolone use in the 12 mo prior to biopsy as described in Supplement 1. The Charlson Comorbidity Index was calculated as previously described [18]. Prebiopsy PSA values within 60 d of biopsy were available for a subset of patients.
The primary outcomes were crude and adjusted rates of prostate biopsy and 30-d complications. Complications were identified using inpatient and outpatient administrative claims as summarized in Supplementary Table 1.
Baseline demographic features were compared across study periods using the Kruskal-Wallis test for continuous variables and the chi-square test for discrete variables. To evaluate the impact of each publication on rates of prostate biopsy and 30-d complications, we used an interrupted time series (ITS) approach as described in Supplementary Methods [19] and [20]. Rates were estimated using all men aged ≥40 yr with at least 12 mo of continuous medical and pharmacy enrollment as the denominator. As secondary analyses, we estimated rates of 30-d complications among men who underwent biopsy. Two separate ITS analyses were performed: (1) unadjusted analyses, using crude rates of biopsy and complications aggregated for each month, and (2) analyses standardized for case mix, to adjust for population shifts over time as described in Supplementary Methods [20] and [21].
Logistic regression was used to evaluate predictors of 30-d complications. Multivariable models were adjusted for age at biopsy, year of biopsy, race, anticoagulant use, history of fluoroquinolone use, and the presence of the five most prevalent Charlson comorbidities.
Statistical analyses were performed using SAS v.9.3 (SAS Institute Inc., Cary, NC, USA). All tests were two sided with p < 0.05 considered statistically significant.
Between January 2005 and September 2014, a total of 104 584 men underwent prostate biopsy (Table 1). Median age at biopsy was 63.0 yr (interquartile range [IQR]: 57.0–70.0). Overall, 18% of men had used fluoroquinolones in the 12 mo preceding biopsy, and 2.8% were on anticoagulants. Median prebiopsy PSA was 5.1 ng/dl (IQR: 4.0–7.1) for the 18 749 men (18%) for whom this was available. Over the time periods, men undergoing biopsy were older, had increased prevalence of anticoagulant use and comorbid conditions, and had a higher PSA (Table 1 and Supplementary Fig. 2).
Table 1
Patient demographics for men who underwent prostate biopsy (January 2005 to September 2014) stratified by pre/postintervention time periods (n = 104 584)
| Baseline Jan 2005–Jul 2008 (n = 37 217) |
2008 USPSTF Aug 2008–Feb 2009 (n = 6672) |
ERSPC/PLCO Mar 2009–Apr 2012 (n = 36 603) |
2012 USPSTF May 2012–Apr 2013 (n = 10 772) |
2013 AUA May 2013–Sep 2014 (n = 13 320) |
p value | Total (n = 104 584) | |
|---|---|---|---|---|---|---|---|
| Age at biopsy, yr | |||||||
| Median (IQR) | 62.0 (56.0–68.0) | 62.0 (56.0–68.0) | 63.0 (57.0–70.0) | 65.0 (59.0–71.0) | 66.0 (59.0–71.0) | <0.001 | 63.0 (57.0–70.0) |
| Age group, yr (%) | |||||||
| 40–49 | 2269 (6.1) | 416 (6.2) | 1737 (4.7) | 420 (3.9) | 467 (3.5) | <0.001 | 5309 (5.1) |
| 50–59 | 12 366 (33) | 2201 (33) | 10 342 (28) | 2644 (25) | 3098 (23) | 30 651 (29) | |
| 60–69 | 14 671 (39) | 2697 (40) | 15 105 (41) | 4426 (41) | 5546 (42) | 42 445 (41) | |
| ≥70 | 7911 (21) | 1358 (20) | 9419 (26) | 3282 (31) | 4209 (32) | 26 179 (25) | |
| Race (%) | |||||||
| White | 25 508 (69) | 4643 (70) | 26 187 (72) | 7795 (72) | 9407 (71) | <0.001 | 73 540 (70) |
| African American | 3695 (9.9) | 780 (12) | 4423 (12) | 1283 (12) | 1589 (12) | 11 770 (11) | |
| Hispanic | 2549 (6.8) | 538 (8.1) | 2911 (8.0) | 837 (7.8) | 1186 (8.9) | 8021 (7.7) | |
| Asian | 1133 (3.0) | 236 (3.5) | 1142 (3.1) | 306 (2.8) | 387 (2.9) | 3204 (3.1) | |
| Unknown/Missing | 4332 (12) | 475 (7.1) | 1940 (5.3) | 551 (5.1) | 751 (5.6) | 8049 (7.7) | |
| Census region (%) | |||||||
| Northeast | 7094 (19) | 1196 (18) | 6120 (17) | 1886 (18) | 2424 (18) | <0.001 | 18 720 (18) |
| Midwest | 10 094 (27) | 1562 (23) | 8837 (24) | 2742 (26) | 3627 (27) | 26 862 (26) | |
| South | 15 330 (41) | 2889 (43) | 16 371 (45) | 4698 (44) | 5538 (42) | 44 826 (43) | |
| West | 4660 (13) | 993 (15) | 5147 (14) | 1442 (13) | 1728 (13) | 13 970 (13) | |
| Other/Missing | 39 (0.1) | 32 (0.5) | 128 (0.3) | 4 (0) | 3 (0) | 206 (0.2) | |
| PSA, ng/d | n = 5174 | n = 548 | n = 7345 | n = 2211 | n = 3471 | n = 18 749 | |
| Median (IQR) | 5.1 (3.9–7.1) | 4.9 (3.8–6.8) | 5.0 (4.0–6.8) | 5.3 (4.2–7.4) | 5.4 (4.2–7.8) | <0.001 | 5.1 (4.0–7.1) |
| Anticoagulant use (%) | 928 (2.5) | 184 (2.8) | 1103 (3.0) | 299 (2.8) | 415 (3.1) | <0.001 | 2929 (2.8) |
| History of fluoroquinolone use in prior 12 mo (%) | 6797 (18) | 1267 (19) | 6557 (18) | 2017 (19) | 2504 (19) | 0.054 | 19 142 (18) |
| Charlson Comorbidity Index (%) | |||||||
| 0 | 22 354 (60) | 3887 (58) | 20 269 (55) | 5758 (54) | 7039 (53) | <0.001 | 59 307 (57) |
| 1 | 8199 (22) | 1453 (22) | 8252 (23) | 2503 (23) | 2999 (23) | 23 406 (22) | |
| 2 | 3232 (8.7) | 664 (10) | 3520 (9.6) | 1108 (10) | 1359 (10) | 9883 (9.4) | |
| ≥3 | 3432 (9.2) | 668 (10) | 4562 (13) | 1403 (13) | 1923 (14) | 11 988 (12) | |
| Comorbidities (%) | |||||||
| Myocardial infarction | 763 (2.1) | 144 (2.2) | 812 (2.2) | 227 (2.1) | 273 (2.0) | 0.6 | 2219 (2.1) |
| Congestive heart failure | 1443 (3.9) | 223 (3.3) | 1548 (4.2) | 467 (4.3) | 586 (4.4) | <0.001 | 4267 (4.1) |
| Peripheral vascular disease | 2054 (5.5) | 399 (6.0) | 2660 (7.3) | 783 (7.3) | 1071 (8.0) | <0.001 | 6967 (6.7) |
| Cerebrovascular disease | 2007 (5.4) | 379 (5.7) | 2358 (6.4) | 681 (6.3) | 947 (7.1) | <0.001 | 6372 (6.1) |
| Dementia | 370 (1.0) | 63 (0.9) | 442 (1.2) | 141 (1.3) | 198 (1.5) | <0.001 | 1214 (1.2) |
| Chronic obstructive pulmonary disease | 4251 (11) | 758 (11) | 4293 (12) | 1254 (12) | 1686 (13) | 0.004 | 12 242 (12) |
| Ulcer | 362 (1.0) | 60 (0.9) | 297 (0.8) | 84 (0.8) | 102 (0.8) | 0.07 | 905 (0.9) |
| Mild liver disease | 1182 (3.2) | 223 (3.3) | 1332 (3.6) | 413 (3.8) | 515 (3.9) | <0.001 | 3665 (3.5) |
| Diabetes mellitus | 6475 (17) | 1290 (19) | 7823 (21) | 2442 (23) | 2973 (22) | <0.001 | 21 003 (20) |
| Diabetes with organ damage | 1105 (3.0) | 225 (3.4) | 1600 (4.4) | 492 (4.6) | 734 (5.5) | <0.001 | 4156 (4.0) |
| Hemiplegia | 128 (0.3) | 25 (0.4) | 150 (0.4) | 51 (0.5) | 56 (0.4) | 0.3 | 410 (0.4) |
| Moderate/severe renal disease | 1133 (3.0) | 254 (3.8) | 1950 (5.3) | 674 (6.3) | 908 (6.8) | <0.001 | 4919 (4.7) |
| Moderate/severe liver disease | 27 (0.1) | 7 (0.1) | 31 (0.1) | 9 (0.1) | 12 (0.1) | 0.9 | 86 (0.1) |
| Metastatic solid tumor | 175 (0.5) | 37 (0.6) | 199 (0.5) | 58 (0.5) | 80 (0.6) | 0.4 | 549 (0.5) |
| AIDS | 82 (0.2) | 16 (0.2) | 101 (0.3) | 27 (0.3) | 46 (0.3) | 0.2 | 272 (0.3) |
| Rheumatologic disease | 449 (1.2) | 71 (1.1) | 522 (1.4) | 158 (1.5) | 196 (1.5) | 0.007 | 1396 (1.3) |
| Cancer | 1736 (4.7) | 337 (5.1) | 1796 (4.9) | 571 (5.3) | 701 (5.3) | 0.02 | 5141 (4.9) |
| Net worth range (%) | |||||||
| <$25 000 | 1831 (4.9) | 404 (6.1) | 2101 (5.7) | 561 (5.2) | 695 (5.2) | <0.001 | 5592 (5.3) |
| $25 000–149 000 | 5188 (14) | 1042 (16) | 5608 (15) | 1591 (15) | 2046 (15) | 15 475 (15) | |
| $150 000–249 000 | 4754 (13) | 837 (13) | 5182 (14) | 1499 (14) | 1862 (14) | 14 134 (14) | |
| $250 000–499 000 | 9694 (26) | 1779 (27) | 10 048 (28) | 3027 (28) | 3636 (27) | 28 184 (27) | |
| ≥$500 000 | 10 910 (29) | 2030 (30) | 10 860 (30) | 3208 (30) | 3852 (29) | 30 860 (30) | |
| Unknown/missing | 4840 (13) | 580 (8.7) | 2804 (7.7) | 886 (8.2) | 1229 (9.2) | 10 339 (9.9) | |
| 30-d complications (%) | |||||||
| Any | 5973 (16) | 1151 (17) | 6120 (17) | 1863 (17) | 2451 (18) | <0.001 | 17 558 (17) |
| Bleeding | 2547 (6.8) | 499 (7.5) | 2660 (7.3) | 860 (8.0) | 1083 (8.1) | <0.001 | 7649 (7.3) |
| Nonsepsis infection | 2978 (8.0) | 590 (8.8) | 2990 (8.2) | 875 (8.1) | 1236 (9.3) | <0.001 | 8669 (8.3) |
| Sepsis | 191 (0.5) | 50 (0.7) | 349 (1.0) | 105 (1.0) | 134 (1.0) | <0.001 | 829 (0.8) |
| Urinary retention | 980 (2.6) | 161 (2.4) | 1138 (3.1) | 341 (3.2) | 422 (3.2) | <0.001 | 3042 (2.9) |
| Hospitalization | 1019 (2.7) | 181 (2.7) | 1056 (2.9) | 320 (3.0) | 367 (2.8) | 0.6 | 2943 (2.8) |
AIDS = acquired immunodeficiency syndrome; AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; IQR = interquartile range; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer [screening trial]; PSA = prostate-specific antigen; USPTF = US Preventive Services Task Force.
Overall, 30-d complications occurred in 17% of biopsies. The most prevalent was nonsepsis genitourinary infections (8.3%), followed by bleeding complications (7.3%) and urinary retention (2.9%) (Table 1). The 30-d hospitalization rate was 2.8%.
We first examined unadjusted rates of prostate biopsy among men aged ≥40 yr (Fig. 1 and Table 2). From January 2005 to September 2014, biopsy rates fell 33% from 64.1 to 42.8 per 100 000 person-months. There was substantial seasonal variation in biopsy rates with reductions near the end of each calendar year. With the ITS approach, each intervention is potentially associated with an immediate change in event rates (ie, level shift or change in intercept) as well as a change in the rate of growth (ie, change in slope). At baseline, there was a statistically significant growth in prostate biopsy rates of 2.9 biopsies per 100 000 person-months per year (95% confidence interval [CI], 0.96–4.8; p < 0.001). Following the 2008 USPSTF recommendations, there was an immediate decrease (ie, level shift) in biopsy rates of −10.1 biopsies per 100 000 person-months (95% CI, −17.1 to −3.0; p < 0.001). Likewise, there were immediate decreases in biopsy rates following the 2012 USPSTF recommendations (−13.8 biopsies per 100 000 person-months; 95% CI, −21.0 to −6.7; p < 0.001) and 2013 AUA guidelines (−8.8 biopsies per 100 000 person-months; 95% CI, −16.7 to −0.92; p = 0.03). Analyses adjusted for case mix were similar, although the baseline growth rate was no longer significant, indicating this was related to population shifts over time (Table 2, model 2).
Fig. 1
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 2
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315)
| Prostate biopsy | Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Model 1* | ||||||||||||||
| Baseline rate | 57.8 (53.4–62.3) |
<0.001 | 8.6 (7.5–9.7) |
<0.001 | 4.0 (3.3–4.6) |
<0.001 | 0.17 (0.03–0.32) |
0.02 | 2.8 (2.3–3.3) |
<0.001 | 1.5 (1.3–1.6) |
<0.001 | 1.4 (1.1–1.7) |
<0.001 |
| Baseline trend in rate | 2.9 (0.96–4.8) |
<0.001 | 0.84 (0.48–1.3) |
<0.001 | 0.72 (0.48–1.1) |
<0.001 | 0.12 (0.12–0.12) |
<0.001 | 0.36 (0.12–0.48) |
<0.001 | 0.12 (0.00–0.12) |
<0.001 | 0.00 (−0.12 to 0.12) |
0.5 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.1 (−17.1 to −3.0) |
<0.001 | −0.61 (−3.1 to 1.9) |
0.6 | −1.4 (−2.8 to −0.03) |
0.053 | 0.04 (−0.35 to 0.42) |
0.8 | 0.56 (−0.70 to 1.8) |
0.4 | −0.06 (−0.69 to 0.57) |
0.9 | 0.16 (−0.58 to 0.90) |
0.7 |
| Trend change | 7.9 (−12.7 to 28.4) |
0.5 | −1.0 (−9.1 to 7.2) |
0.8 | −0.12 (−4.1 to 4.0) |
0.98 | −0.60 (−1.9 to 0.72) |
0.4 | −2.0 (−6.5 to 2.4) |
0.4 | −1.1 (−3.4 to 1.3) |
0.4 | −1.0 (−3.6 to 1.6) |
0.5 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.6 (−12 to 4.5) |
0.4 | −0.74 (−3.8 to 2.4) |
0.6 | −0.12 (−1.8 to 1.5) |
0.9 | 0.29 (−0.20 to 0.78) |
0.2 | −0.04 (−1.6 to 1.6) |
0.97 | 0.59 (−0.23 to 1.4) |
0.2 | 0.51 (−0.43 to 1.5) |
0.3 |
| Trend change | −11.2 (−32.0 to 9.6) |
0.3 | 0.0 (−8.2 to 8.0) |
0.99 | −0.96 (−5.0 to 3.1) |
0.7 | 0.36 (−0.96 to 1.8) |
0.5 | 1.7 (−2.8 to 6.1) |
0.5 | 1.0 (−1.4 to 3.4) |
0.4 | 0.84 (−1.7 to 3.5) |
0.5 |
| 2012 USPSTF | ||||||||||||||
| Level change | −13.8 (−21.0 to −6.7) |
<0.001 | −2.5 (−4.5 to −0.45) |
0.02 | −1.7 (−3.1 to −0.31) |
0.02 | −0.12 (−0.42 to 0.19) |
0.5 | −0.61 (−1.5 to 0.28) |
0.2 | −0.75 (−1.1 to −0.44) |
<0.001 | −0.25 (−0.77 to 0.27) |
0.4 |
| Trend change | 8.2 (−5.8 to 22.0) |
0.3 | 3.0 (−0.36 to 6.4) |
0.09 | 3.1 (0.72–5.5) |
0.01 | 0.12 (−0.36 to 0.60) |
0.7 | 0.84 (−0.72 to 2.3) |
0.3 | 0.48 (0.00 to 1.1) |
0.08 | 0.24 (−0.60 to 1.2) |
0.6 |
| 2013 AUA | ||||||||||||||
| Level change | −8.8 (−16.7 to −0.92) |
0.03 | −1.4 (−3.8 to 1.1) |
0.3 | −1.9 (−3.5 to −0.22) |
0.03 | −0.02 (−0.41 to 0.37) |
0.9 | −0.19 (−1.3 to 0.91) |
0.7 | −0.38 (−0.77 to 0.01) |
0.06 | −0.01 (−0.66 to 0.63) |
0.96 |
| Trend change | −10.4 (−30.0 to 9.2) |
0.3 | −4.1 (−7.9 to −0.24) |
0.04 | −2.9 (−6.0 to 0.12) |
0.07 | −0.24 (−0.72 to 0.36) |
0.5 | −1.8 (−3.5 to 0.0) |
0.046 | −0.60 (−1.2 to 0.12) |
0.09 | −0.60 (−1.7 to 0.36) |
0.2 |
| Model 2† | ||||||||||||||
| Baseline rate | 86.8 (82.1–91.4) |
<0.001 | 16.2 (14.9–17.6) |
<0.001 | 6.4 (4.9–7.9) |
<0.001 | 0.39 (−0.13 to 0.91) |
0.2 | 6.1 (4.6–7.7) |
<0.001 | 3.6 (2.8–4.4) |
<0.001 | 3.0 (2.2–3.8) |
<0.001 |
| Baseline trend in rate | 0.24 (−1.8 to 2.3) |
0.8 | −0.24 (−0.72 to 0.24) |
0.4 | 0.48 (−0.12 to 1.1) |
0.09 | 0.12 (−0.12 to 0.24) |
0.4 | −0.24 (−0.72 to 0.24) |
0.4 | −0.24 (−0.48 to 0.00) |
0.054 | −0.12 (−0.36 to 0.12) |
0.3 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.8 (−17.8 to −3.8) |
<0.001 | −5.4 (−9.6 to −1.2) |
0.01 | −2.5 (−5.6 to 0.71) |
0.1 | −0.61 (−1.9 to 0.71) |
0.4 | 0.14 (−3.8 to 4.1) |
0.9 | 1.1 (−0.98 to 3.1) |
0.3 | 0.14 (−1.9 to 2.1) |
0.9 |
| Trend change | 10.6 (−10.1 to 31.2) |
0.3 | 6.1 (−8.4 to 20.8) |
0.4 | −1.7 (−12.2 to 8.8) |
0.8 | 1.7 (−2.9 to 6.2) |
0.5 | 1.7 (−12.0 to 15.5) |
0.8 | −4.7 (−11.8 to 2.5) |
0.2 | −0.84 (−7.8 to 6.1) |
0.8 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.9 (−12 to 4.1) |
0.3 | −0.24 (−5.6 to 5.1) |
0.9 | 1.9 (−2.1 to 5.8) |
0.4 | −0.32 (−2.0 to 1.4) |
0.7 | −0.99 (−6.0 to 4.0) |
0.7 | 1.8 (−0.84 to 4.4) |
0.2 | 0.87 (−1.7 to 3.4) |
0.5 |
| Trend change | −16.2 (−37.1 to 4.7) |
0.1 | −6.8 (−21.4 to 7.8) |
0.4 | 0.36 (−10.2 to 10.9) |
0.95 | −1.9 (−6.5 to 2.8) |
0.4 | −1.6 (−15.2 to 12.2) |
0.8 | 4.7 (−2.4 to 11.9) |
0.2 | 0.48 (−6.4 to 7.4) |
0.9 |
| 2012 USPSTF | ||||||||||||||
| Level change | −11.0 (−18.1 to −4.0) |
0.001 | −2.8 (−6.0 to 0.38) |
0.09 | 0.53 (−1.9 to 3.0) |
0.7 | −0.02 (−0.95 to 0.91) |
0.97 | −1.8 (−4.6 to 0.96) |
0.2 | −1.5 (−3.1 to 0.17) |
0.08 | 0.16 (−1.3 to 1.6) |
0.8 |
| Trend change | 7.0 (−7.2 to 21.1) |
0.3 | 1.7 (−3.6 to 7.0) |
0.5 | 0.12 (−4.1 to 4.2) |
0.96 | 0.12 (−1.4 to 1.8) |
0.9 | −0.48 (−5.3 to 4.3) |
0.9 | 1.6 (−1.3 to 4.3) |
0.3 | 0.12 (−2.4 to 2.6) |
0.9 |
| 2013 AUA | ||||||||||||||
| Level change | −8.2 (−15.9 to −0.38) |
0.04 | −1.0 (−4.8 to 2.8) |
0.6 | −0.44 (−3.2 to 2.3) |
0.8 | −0.15 (−1.3 to 0.99) |
0.8 | 0.28 (−3.2 to 3.7) |
0.9 | −0.50 (−2.7 to 1.7) |
0.7 | 0.47 (−1.3 to 2.3) |
0.6 |
| Trend change | −4.2 (−24.5 to 16.1) |
0.7 | −0.7 (−7.6 to 6.2) |
0.9 | 1.1 (−4.2 to 6.2) |
0.7 | −0.12 (−1.8 to 1.7) |
0.9 | −0.60 (−5.9 to 4.8) |
0.8 | −1.8 (−4.9 to 1.2) |
0.3 | −0.48 (−3.1 to 2.2) |
0.7 |
* Model 1: Base model with constant, time trend, intervention indicator variables. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
† Model 2: Adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, and anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPTF = US Preventive Services Task Force.
Next, we examined 30-d postbiopsy complications among men aged ≥40 yr as a measure of the absolute morbidity of biopsy (Fig. 1, Supplementary Fig. 3, and Table 2). Across the study periods, overall complication rates decreased 10% from 8.7 to 7.8 per 100 000 person-months. In unadjusted ITS analyses, there was a baseline growth in the overall complication rate (0.84 per 100 000 person-months per year; 95% CI, 0.48–1.3; p < 0.001), driven by growth in the rate of nonsepsis infections (0.72 per 100 000 person-months per year; 95% CI, 0.48–1.1; p < 0.001). Only the 2012 USPSTF recommendations were associated with a significant change in complication rates, following which there was an immediate decrease in the overall complication rate of −2.5 per 100 000 person-months (95% CI, −4.5 to −0.45; p = 0.02), driven by reductions in nonsepsis infections (−1.7 per 100 000 person-months; 95% CI, −3.1 to −0.31; p = 0.02) and urinary retention (−0.75 per 100 000 person-months;95% CI, −1.1 to −0.44; p < 0.001). Following the 2013 AUA guidelines there was also a reduction in the overall complication baseline growth rate (−4.1 per 100 000 person-months per year; 95% CI, −7.9 to −0.24; p = 0.04). In analyses adjusted for case mix (Table 2, model 2), effects were overall attenuated, and only the 2008 USPSTF recommendations were associated with a reduction in overall complication rates (−5.4 per 100 000 person-months; 95% CI, −9.6 to −1.2; p = 0.01).
To examine whether there was a change in the relative morbidity of biopsy, we performed a secondary analysis examining complication rates among men undergoing biopsy (Fig. 2). The proportion of men undergoing biopsy who experienced a complication increased from 14% in January 2005 to 18% in September 2014, an increase of 36%. There was again a baseline growth in overall complication rates (0.72% per year; 95% CI, 0.36–1.1; p < 0.001), driven by an increase in nonsepsis infections and sepsis (Supplementary Table 2). None of the interventions affected this growth. Similar results were obtained after adjusting for case mix (Table 3).
Fig. 2
Interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 3
Adjusted* interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Baseline rate | 15.6 (14.3–16.8) |
<0.001 | 7.1 (6.5–7.7) |
<0.001 | 0.34 (0.11–0.57) |
<0.001 | 6.7 (5.9–7.5) |
<0.001 | 2.8 (2.5–3.2) |
<0.001 | 3.2 (2.8–3.6) |
<0.001 |
| Baseline trend in rate | 0.60 (0.12–1.1) |
0.02 | 0.72 (0.60–0.96) |
<0.001 | 0.12 (0.12–0.24) |
<0.001 | 0.12 (−0.12 to 0.36) |
0.3 | −0.12 (−0.24 to 0.00) |
0.3 | −0.12 (−0.24 to 0.00) |
0.0498 |
| 2008 USPSTF | ||||||||||||
| Level change | 1.1 (−2.0 to 4.1) |
0.5 | −1.3 (−3.2 to 0.58) |
0.2 | 0.13 (−0.45 to 0.72) |
0.7 | 1.9 (−0.11 to 3.9) |
0.07 | 0.46 (−0.50 to 1.4) |
0.4 | 0.27 (−0.69 to 1.2) |
0.6 |
| Trend change | −3.5 (−13.4 to 6.6) |
0.5 | −1.8 (−8.4 to 4.7) |
0.6 | −1.0 (−3.0 to 0.96) |
0.3 | −4.3 (−11.4 to 2.6) |
0.2 | −1.6 (−4.9 to 1.9) |
0.4 | −0.72 (−4.0 to 2.6) |
0.7 |
| ERSPC/PLCO | ||||||||||||
| Level change | −0.14 (−3.9 to 3.6) |
0.9 | 1.1 (−1.3 to 3.4) |
0.4 | 0.55 (−0.19 to 1.3) |
0.2 | 0.39 (−2.2 to 3.0) |
0.8 | 0.75 (−0.47 to 2.0) |
0.2 | 0.65 (−0.56 to 1.9) |
0.3 |
| Trend change | 2.9 (−7.1 to 13.0) |
0.6 | 0.72 (−5.8 to 7.2) |
0.8 | 0.84 (−1.2 to 2.9) |
0.4 | 4.2 (−2.8 to 11.3) |
0.2 | 1.7 (−1.8 to 5.0) |
0.4 | 0.72 (−2.6 to 4.0) |
0.7 |
| 2012 USPSTF | ||||||||||||
| Level change | −0.27 (−2.7 to 2.1) |
0.8 | −0.56 (−2.0 to 0.87) |
0.5 | 0.16 (−0.31 to 0.63) |
0.5 | 1.1 (−0.38 to 2.5) |
0.2 | −0.38 (−1.0 to 0.28) |
0.3 | 0.44 (−0.34 to 1.2) |
0.3 |
| Trend change | 1.3 (−2.6 to 5.3) |
0.5 | 2.9 (0.60–5.3) |
0.02 | −0.24 (−1.1 to 0.60) |
0.6 | −0.72 (−3.2 to 1.7) |
0.5 | 0.24 (−0.96 to 1.3) |
0.7 | −0.48 (−1.8 to 0.84) |
0.5 |
| 2013 AUA | ||||||||||||
| Level change | 0.25 (−2.6 to 3.1) |
0.9 | −0.64 (−2.3 to 1.1) |
0.5 | 0.22 (−0.40 to 0.83) |
0.5 | 1.1 (−0.68 to 2.8) |
0.2 | −0.29 (−1.1 to 0.48) |
0.5 | 0.47 (−0.55 to 1.5) |
0.4 |
| Trend change | −1.6 (−6.1 to 2.9) |
0.5 | −2.5 (−5.5 to 0.60) |
0.1 | 0.12 (−0.72 to 1.1) |
0.8 | −0.36 (−3.1 to 2.4) |
0.8 | 0.24 (−1.2 to 1.7) |
0.8 | 0.12 (−1.3 to 1.6) |
0.9 |
* Model adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 person-months. Baseline trend in rate and trend change per 100 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial]; USPTF = US Preventive Services Task Force.
Given the increase in the relative morbidity of biopsy, we performed logistic regression to identify predictors of each complication category (Supplementary Table 3 and Table 4). Due to the large sample size, we focused on those features associated with the greatest magnitude of effect. A prior diagnosis of cancer was associated with an increased risk of all complication categories including any complication (odds ratio [OR]: 1.30; 95% CI, 1.22–1.39; p < 0.001) and hospitalization (OR: 1.59; 95% CI, 1.39–1.82; p < 0.001). Similarly, prior fluoroquinolone use was associated with an increased risk of all complication categories including any complication (OR: 1.27; 95% CI, 1.22–1.32; p < 0.001) and sepsis (OR: 1.55; 95% CI, 1.33–1.81; p < 0.001). Age at biopsy ≥70 yr was associated with a nearly 4-fold increased risk of urinary retention. Although anticoagulant use was associated with increased risks of bleeding (OR: 1.28; 95% CI, 1.14–1.45; p < 0.001), sepsis (OR: 1.48; 95% CI, 1.06–2.07; p = 0.02), and hospitalization (OR: 1.39; 95% CI, 1.16–1.67; p < 0.001), the increase in the overall complication risk was modest (OR 1.14; 95% CI, 1.04–1.25; p = 0.004).
Table 4
Multivariable analysis* of 30-d postbiopsy complications among men aged ≥40 yr undergoing prostate biopsy (January 2005 to September 2014) (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Hospitalization | Urinary retention | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Age at biopsy, yr | ||||||||||||
| 40–49 | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| 50–59 | 1.01 (0.94–1.09) | 0.8 | 0.94 (0.85–1.04) | 0.2 | 1.16 (0.83–1.63) | 0.4 | 1.01 (0.90–1.14) | 0.8 | 1.12 (0.93–1.36) | 0.2 | 1.54 (1.20–1.97) | <0.001 |
| 60–69 | 1.05 (0.97–1.13) | 0.2 | 0.94 (0.85–1.03) | 0.2 | 0.98 (0.70–1.37) | 0.9 | 1.05 (0.94–1.18) | 0.4 | 1.16 (0.95–1.40) | 0.1 | 2.16 (1.68–2.77) | <0.001 |
| ≥70 | 1.25 (1.15–1.36) | <0.001 | 1.03 (0.93–1.15) | 0.6 | 1.05 (0.74–1.49) | 0.8 | 1.29 (1.15–1.46) | <0.001 | 1.31 (1.07–1.60) | 0.008 | 3.98 (3.08–5.15) | <0.001 |
| Race | ||||||||||||
| White | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| Asian | 1.27 (1.15–1.40) | <0.001 | 1.27 (1.11–1.44) | 0.0003 | 1.65 (1.19–2.29) | 0.003 | 1.36 (1.18–1.56) | <0.001 | 0.94 (0.76–1.17) | 0.6 | 1.15 (0.93–1.42) | 0.2 |
| African American | 1.09 (1.03–1.15) | 0.002 | 1.05 (0.97–1.13) | 0.2 | 1.34 (1.09–1.64) | 0.005 | 1.17 (1.09–1.27) | <0.001 | 1.13 (1.01–1.27) | 0.03 | 1.15 (1.03–1.28) | 0.02 |
| Hispanic | 1.32 (1.24–1.40) | <0.001 | 1.36 (1.26–1.47) | <0.001 | 1.52 (1.21–1.90) | <0.001 | 1.33 (1.22–1.45) | <0.001 | 1.12 (0.97–1.28) | 0.1 | 1.37 (1.21–1.56) | <0.001 |
| Unknown/missing | 0.99 (0.93–1.06) | 0.8 | 1.07 (0.99–1.17) | 0.1 | 1.23 (0.94–1.61) | 0.1 | 0.96 (0.88–1.05) | 0.4 | 0.94 (0.81–1.08) | 0.4 | 1.02 (0.88–1.18) | 0.8 |
| Diabetes mellitus | 1.03 (0.99–1.07) | 0.09 | 1.06 (1.00–1.11) | 0.045 | 1.08 (0.92–1.28) | 0.4 | 1.01 (0.96–1.07) | 0.7 | 1.20 (1.10–1.31) | <0.001 | 0.99 (0.91–1.08) | 0.8 |
| COPD | 1.14 (1.09–1.20) | <0.001 | 1.07 (1.00–1.14) | 0.04 | 1.23 (1.01–1.49) | 0.04 | 1.14 (1.06–1.22) | <0.001 | 1.36 (1.23–1.51) | <0.001 | 1.14 (1.03–1.25) | 0.01 |
| PVD | 1.24 (1.16–1.31) | <0.001 | 1.1 (1.02–1.2) | 0.02 | 1.19 (0.93–1.51) | 0.2 | 1.28 (1.18–1.40) | <0.001 | 1.31 (1.15–1.48) | <0.001 | 1.26 (1.12–1.42) | <0.001 |
| Cerebrovascular disease | 1.11 (1.04–1.19) | 0.001 | 1.12 (1.03–1.22) | 0.01 | 1.21 (0.93–1.57) | 0.2 | 1.03 (0.94–1.14) | 0.5 | 1.26 (1.10–1.44) | <0.001 | 1.30 (1.15–1.47) | <0.001 |
| Cancer | 1.30 (1.22–1.39) | <0.001 | 1.16 (1.06–1.27) | 0.001 | 1.37 (1.05–1.79) | 0.02 | 1.40 (1.27–1.53) | <0.001 | 1.59 (1.39–1.82) | <0.001 | 1.32 (1.16–1.52) | <0.001 |
| Anticoagulant use | 1.14 (1.04–1.25) | 0.004 | 1.01 (0.89–1.14) | 0.9 | 1.48 (1.06–2.07) | 0.02 | 1.28 (1.14–1.45) | <0.001 | 1.39 (1.16–1.67) | <0.001 | 1.11 (0.92–1.33) | 0.3 |
| Prior fluoroquinolones | 1.27 (1.22–1.32) | <0.001 | 1.37 (1.31–1.45) | <0.001 | 1.55 (1.33–1.81) | <0.001 | 1.21 (1.14–1.28) | <0.001 | 1.25 (1.14–1.36) | <0.001 | 1.27 (1.16–1.38) | <0.001 |
* Models adjusted for year of biopsy and the features listed in the table.
CI = confidence interval; COPD = chronic obstructive pulmonary disease; PVD = peripheral vascular disease.
We conducted sensitivity analyses where the postintervention run-in period was varied from 2 to 4 mo (Supplementary Table 4). Results overall were similar, although effects were attenuated for the 2013 AUA guidelines on biopsy rates and the 2012 USPSTF recommendations on overall complication rates, likely due to fewer time points for analysis. In a separate sensitivity analysis, we examined bleeding complications excluding microhematuria from 2009 to 2014, following introduction of a separate code for microhematuria. Bleeding rates paralleled those of the primary bleeding end point (Supplementary Fig. 4), and results of both ITS and logistic regression analyses were similar (Supplementary Table 5 and 6). As an exploratory analysis, we examined PCa diagnosis rates within 60 d following biopsy (Supplementary Fig. 5). These appeared to increase steadily across the study period.
To our knowledge, this is the largest study to examine the impact of landmark PSA screening publications on rates of prostate biopsy and the first to examine their impact on postbiopsy complications. Overall, biopsy rates have declined by 33%, which appears attributable to the publication of the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines. Interestingly, the magnitude of reduction in biopsy rates exceeded previously reported decreases in PSA screening rates [6], [7], [8], [9], [10], [11], [12], [22], and [23], which suggests that either the threshold for referral to a urologist or the threshold for performing a biopsy has changed. More importantly, there was a concurrent 22% decrease in 30-d complication rates that appears largely attributable to the 2012 USPSTF recommendations. Still, the reduction in complication rates was less than the reduction in biopsy rates due to an increase in the relative morbidity of biopsy, driven by growth in infectious complications. This may in part be related to a shift to older patients with greater comorbidity undergoing biopsy, although similar results were obtained in analyses adjusted for case mix, which suggests additional factors may exist.
To this end, we evaluated predictors of postbiopsy complications to identify actionable targets for further risk reduction. Some of these features represent modifiable risk factors (eg, anticoagulant use), and others represent opportunities for altering periprocedural management. For instance, prior fluoroquinolone use may identify patients who would benefit from prebiopsy rectal swabs to guide periprocedural antibiotics [24] and [25]. Nonmodifiable risk factors, including specific comorbidities such as cancer and peripheral vascular disease, may encourage reconsideration of the risk–benefit ratio before proceeding with biopsy.
It is also important to consider the concurrent impact of landmark PSA screening publications on rates of PCa diagnosis to contextualize the observed changes in biopsy and its morbidity. An exploratory analysis suggested that cancer diagnosis rates have steadily increased, although further studies are necessary to examine this issue comprehensively.
A number of studies have examined the impact of the 2008 USPSTF recommendations [7], [8], [9], and [22], ERSPC/PLCO trials [6] and [9], and 2012 USPSTF recommendations [6], [10], [11], [12], and [23] on PSA screening rates, generally noting reductions ranging from 0% to 8%. However, much less data are available on the impact of these publications on biopsy rates. One single-center study noted a 31% reduction in the number of biopsies performed following the 2012 USPSTF recommendations [13]; another noted no change in the proportion of referred men who underwent biopsy [26]. A report from a Canadian academic health network noted a reduction in biopsy rates from 58.0 to 35.5 biopsies per month following the 2012 USPSTF recommendations [27]. Our findings therefore confirm a reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines.
Limited data exist regarding secular trends in postbiopsy complications, and prior studies have focused exclusively on the relative morbidity among men undergoing biopsy rather than absolute complication rates. A Canadian population-based study observed that the 30-d hospitalization rate increased from 1.0% in 1996 to 4.1% in 2005, driven by an increase in infectious complications [28]. A Surveillance, Epidemiology and End Results–Medicare study also noted an increase in hospitalizations related to infectious complications over time [14]. Although we observed a similar increase in 30-d overall complications among men undergoing biopsy, likewise driven by infectious complications, there was an absolute 22% reduction in 30-d complications over the study period. Taken together, these findings support a reduction in the morbidity of biopsy following publication of revised PSA screening guidelines but highlight the continued need to reduce the relative morbidity of biopsy.
A number of studies have described complication rates following prostate biopsy [15], [16], [29], and [30]. Hematuria has been reported in 10–84% of biopsies, rectal bleeding in 1–45%, infectious complications in up to 6.3%, urinary retention in up to 1.7%, and hospitalization in up to 6.9% [14] and [29]. The wide range in complication rates is due to differences in definitions, ascertainment of end points, and variation in clinical practice. Given the heterogeneity in postbiopsy complication rates, our findings have two advantages. First, they represent a contemporary, geographically diverse population across the entire United States. Second, use of claims-based ascertainment limited complications to those cases in which patients sought medical care, and although this may underestimate the comprehensive morbidity of biopsy, it more accurately reflects the economic burden of serious complications.
This study has several limitations. Most importantly, a control population of men who were not exposed to the PSA screening publications was not available. This study is retrospective, and additional, unrecognized changes in policy or clinical practice may possibly have confounded the impact of the interventions examined. However, the ITS approach has been described as the strongest quasi-experimental design to evaluate policy changes when a randomized trial is not possible [19]. Because we relied on administrative claims for ascertainment of complications, minor complications that did not generate a claim were not captured, thus underestimating these end points. Although we adjusted for case mix, we were unable to adjust for PSA. However, population shifts in PSA may be confounded by changes in screening practices, and PSA adjustment may therefore under- or overestimate the impact of interventions. We were also unable to adjust for other factors that may be associated with biopsy or complication rates, such as prostate magnetic resonance imaging (MRI), ancillary biomarker testing, biopsy approach (eg, MRI-fusion biopsy, transperineal biopsy), indication for anticoagulation, use of periprocedural bridging, or operator biopsy volume/experience.
We observed an immediate reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines, with a smaller reduction in absolute 30-d complication rates. Still, the relative morbidity of biopsy has increased over time, driven by infectious complications. Accordingly, if health policy aims to reduce the morbidity of PSA screening related to biopsy, further efforts should focus on improving patient selection and periprocedural processes of care.
Author contributions: Boris Gershman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Gershman, van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Acquisition of data: Van Houten, Shah.
Analysis and interpretation of data: Gershman, Van Houten, Shah, Karnes.
Drafting of the manuscript: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Critical revision of the manuscript for important intellectual content: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Statistical analysis: Van Houten.
Obtaining funding: Karnes, Shah.
Administrative, technical, or material support: Karnes, Shah.
Supervision: Karnes, Shah.
Other (specify): None.
Financial disclosures: Boris Gershman certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: This publication was made possible by funding from the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery.
Funding/Support and role of the sponsor: None.
In 2009, two landmark randomized trials—the European Randomized Study of Screening for Prostate Cancer [ERSPC] [1] and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial [PLCO] [2]—reported smaller than anticipated reductions in prostate cancer (PCa) mortality and prompted reexamination of prostate-specific antigen (PSA) screening recommendations. The US Preventive Services Task Force (USPSTF), which in 2008 had recommended against PSA screening in men aged >75 yr [3], updated its position in 2012 to recommend against systematic PSA screening in the general population [4]. The American Urological Association (AUA) likewise updated its PSA screening guidelines in May 2013, recommending shared decision making for men aged 55–69 yr [5]. Both the 2012 USPSTF and 2013 AUA guidelines cited prostate biopsy and its potential complications as a significant harm associated with PSA screening.
Although several studies have noted a reduction in PSA screening rates following these publications [6], [7], [8], [9], [10], [11], and [12], there are little data on whether this has translated into the intended reductions in screening-related morbidity in the form of prostate biopsy and postbiopsy complications [13]. We evaluated whether publication of the 2008 USPSTF recommendations, 2009 ERSPC/PLCO trials, 2012 USPSTF recommendations, or 2013 AUA guidelines was associated with reductions in prostate biopsy or postbiopsy complications. Biopsy-related complications also remain understudied both with regard to incidence as well as identification of risk factors [14], [15], and [16]. Consequently, we also sought to characterize postbiopsy complications and evaluate predictors using a contemporary population-based cohort.
We conducted a retrospective analysis of administrative claims from the Optum Labs Data Warehouse that contains data on >100 million individuals enrolled in private health plans and Medicare Advantage Plans, from geographically diverse regions across the United States, as described in Supplementary Methods [17]. Because this study involved analysis of preexisting deidentified data, it was exempt from institutional review board approval.
We identified all prostate biopsies performed in men aged ≥40 yr between January 1, 2005, and September 30, 2014, using Current Procedural Terminology-4 code 55700 alone (11%) or in combination with 76872 (84%) or 76942 (0%), or International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 60.11 (5%). The first biopsy was selected as the index biopsy when more than one biopsy was performed. Patients with a prior diagnosis of PCa (ICD-9 code 185) were excluded. We restricted the cohort to patients with continuous medical and pharmacy enrollment for at least 12 mo prior to biopsy and 30 d after biopsy to ascertain comorbidities and 30-d complications (Supplementary Fig. 1).
We recorded sociodemographic characteristics, comorbid conditions, anticoagulant use within 30 d of biopsy, and fluoroquinolone use in the 12 mo prior to biopsy as described in Supplement 1. The Charlson Comorbidity Index was calculated as previously described [18]. Prebiopsy PSA values within 60 d of biopsy were available for a subset of patients.
The primary outcomes were crude and adjusted rates of prostate biopsy and 30-d complications. Complications were identified using inpatient and outpatient administrative claims as summarized in Supplementary Table 1.
Baseline demographic features were compared across study periods using the Kruskal-Wallis test for continuous variables and the chi-square test for discrete variables. To evaluate the impact of each publication on rates of prostate biopsy and 30-d complications, we used an interrupted time series (ITS) approach as described in Supplementary Methods [19] and [20]. Rates were estimated using all men aged ≥40 yr with at least 12 mo of continuous medical and pharmacy enrollment as the denominator. As secondary analyses, we estimated rates of 30-d complications among men who underwent biopsy. Two separate ITS analyses were performed: (1) unadjusted analyses, using crude rates of biopsy and complications aggregated for each month, and (2) analyses standardized for case mix, to adjust for population shifts over time as described in Supplementary Methods [20] and [21].
Logistic regression was used to evaluate predictors of 30-d complications. Multivariable models were adjusted for age at biopsy, year of biopsy, race, anticoagulant use, history of fluoroquinolone use, and the presence of the five most prevalent Charlson comorbidities.
Statistical analyses were performed using SAS v.9.3 (SAS Institute Inc., Cary, NC, USA). All tests were two sided with p < 0.05 considered statistically significant.
Between January 2005 and September 2014, a total of 104 584 men underwent prostate biopsy (Table 1). Median age at biopsy was 63.0 yr (interquartile range [IQR]: 57.0–70.0). Overall, 18% of men had used fluoroquinolones in the 12 mo preceding biopsy, and 2.8% were on anticoagulants. Median prebiopsy PSA was 5.1 ng/dl (IQR: 4.0–7.1) for the 18 749 men (18%) for whom this was available. Over the time periods, men undergoing biopsy were older, had increased prevalence of anticoagulant use and comorbid conditions, and had a higher PSA (Table 1 and Supplementary Fig. 2).
Table 1
Patient demographics for men who underwent prostate biopsy (January 2005 to September 2014) stratified by pre/postintervention time periods (n = 104 584)
| Baseline Jan 2005–Jul 2008 (n = 37 217) |
2008 USPSTF Aug 2008–Feb 2009 (n = 6672) |
ERSPC/PLCO Mar 2009–Apr 2012 (n = 36 603) |
2012 USPSTF May 2012–Apr 2013 (n = 10 772) |
2013 AUA May 2013–Sep 2014 (n = 13 320) |
p value | Total (n = 104 584) | |
|---|---|---|---|---|---|---|---|
| Age at biopsy, yr | |||||||
| Median (IQR) | 62.0 (56.0–68.0) | 62.0 (56.0–68.0) | 63.0 (57.0–70.0) | 65.0 (59.0–71.0) | 66.0 (59.0–71.0) | <0.001 | 63.0 (57.0–70.0) |
| Age group, yr (%) | |||||||
| 40–49 | 2269 (6.1) | 416 (6.2) | 1737 (4.7) | 420 (3.9) | 467 (3.5) | <0.001 | 5309 (5.1) |
| 50–59 | 12 366 (33) | 2201 (33) | 10 342 (28) | 2644 (25) | 3098 (23) | 30 651 (29) | |
| 60–69 | 14 671 (39) | 2697 (40) | 15 105 (41) | 4426 (41) | 5546 (42) | 42 445 (41) | |
| ≥70 | 7911 (21) | 1358 (20) | 9419 (26) | 3282 (31) | 4209 (32) | 26 179 (25) | |
| Race (%) | |||||||
| White | 25 508 (69) | 4643 (70) | 26 187 (72) | 7795 (72) | 9407 (71) | <0.001 | 73 540 (70) |
| African American | 3695 (9.9) | 780 (12) | 4423 (12) | 1283 (12) | 1589 (12) | 11 770 (11) | |
| Hispanic | 2549 (6.8) | 538 (8.1) | 2911 (8.0) | 837 (7.8) | 1186 (8.9) | 8021 (7.7) | |
| Asian | 1133 (3.0) | 236 (3.5) | 1142 (3.1) | 306 (2.8) | 387 (2.9) | 3204 (3.1) | |
| Unknown/Missing | 4332 (12) | 475 (7.1) | 1940 (5.3) | 551 (5.1) | 751 (5.6) | 8049 (7.7) | |
| Census region (%) | |||||||
| Northeast | 7094 (19) | 1196 (18) | 6120 (17) | 1886 (18) | 2424 (18) | <0.001 | 18 720 (18) |
| Midwest | 10 094 (27) | 1562 (23) | 8837 (24) | 2742 (26) | 3627 (27) | 26 862 (26) | |
| South | 15 330 (41) | 2889 (43) | 16 371 (45) | 4698 (44) | 5538 (42) | 44 826 (43) | |
| West | 4660 (13) | 993 (15) | 5147 (14) | 1442 (13) | 1728 (13) | 13 970 (13) | |
| Other/Missing | 39 (0.1) | 32 (0.5) | 128 (0.3) | 4 (0) | 3 (0) | 206 (0.2) | |
| PSA, ng/d | n = 5174 | n = 548 | n = 7345 | n = 2211 | n = 3471 | n = 18 749 | |
| Median (IQR) | 5.1 (3.9–7.1) | 4.9 (3.8–6.8) | 5.0 (4.0–6.8) | 5.3 (4.2–7.4) | 5.4 (4.2–7.8) | <0.001 | 5.1 (4.0–7.1) |
| Anticoagulant use (%) | 928 (2.5) | 184 (2.8) | 1103 (3.0) | 299 (2.8) | 415 (3.1) | <0.001 | 2929 (2.8) |
| History of fluoroquinolone use in prior 12 mo (%) | 6797 (18) | 1267 (19) | 6557 (18) | 2017 (19) | 2504 (19) | 0.054 | 19 142 (18) |
| Charlson Comorbidity Index (%) | |||||||
| 0 | 22 354 (60) | 3887 (58) | 20 269 (55) | 5758 (54) | 7039 (53) | <0.001 | 59 307 (57) |
| 1 | 8199 (22) | 1453 (22) | 8252 (23) | 2503 (23) | 2999 (23) | 23 406 (22) | |
| 2 | 3232 (8.7) | 664 (10) | 3520 (9.6) | 1108 (10) | 1359 (10) | 9883 (9.4) | |
| ≥3 | 3432 (9.2) | 668 (10) | 4562 (13) | 1403 (13) | 1923 (14) | 11 988 (12) | |
| Comorbidities (%) | |||||||
| Myocardial infarction | 763 (2.1) | 144 (2.2) | 812 (2.2) | 227 (2.1) | 273 (2.0) | 0.6 | 2219 (2.1) |
| Congestive heart failure | 1443 (3.9) | 223 (3.3) | 1548 (4.2) | 467 (4.3) | 586 (4.4) | <0.001 | 4267 (4.1) |
| Peripheral vascular disease | 2054 (5.5) | 399 (6.0) | 2660 (7.3) | 783 (7.3) | 1071 (8.0) | <0.001 | 6967 (6.7) |
| Cerebrovascular disease | 2007 (5.4) | 379 (5.7) | 2358 (6.4) | 681 (6.3) | 947 (7.1) | <0.001 | 6372 (6.1) |
| Dementia | 370 (1.0) | 63 (0.9) | 442 (1.2) | 141 (1.3) | 198 (1.5) | <0.001 | 1214 (1.2) |
| Chronic obstructive pulmonary disease | 4251 (11) | 758 (11) | 4293 (12) | 1254 (12) | 1686 (13) | 0.004 | 12 242 (12) |
| Ulcer | 362 (1.0) | 60 (0.9) | 297 (0.8) | 84 (0.8) | 102 (0.8) | 0.07 | 905 (0.9) |
| Mild liver disease | 1182 (3.2) | 223 (3.3) | 1332 (3.6) | 413 (3.8) | 515 (3.9) | <0.001 | 3665 (3.5) |
| Diabetes mellitus | 6475 (17) | 1290 (19) | 7823 (21) | 2442 (23) | 2973 (22) | <0.001 | 21 003 (20) |
| Diabetes with organ damage | 1105 (3.0) | 225 (3.4) | 1600 (4.4) | 492 (4.6) | 734 (5.5) | <0.001 | 4156 (4.0) |
| Hemiplegia | 128 (0.3) | 25 (0.4) | 150 (0.4) | 51 (0.5) | 56 (0.4) | 0.3 | 410 (0.4) |
| Moderate/severe renal disease | 1133 (3.0) | 254 (3.8) | 1950 (5.3) | 674 (6.3) | 908 (6.8) | <0.001 | 4919 (4.7) |
| Moderate/severe liver disease | 27 (0.1) | 7 (0.1) | 31 (0.1) | 9 (0.1) | 12 (0.1) | 0.9 | 86 (0.1) |
| Metastatic solid tumor | 175 (0.5) | 37 (0.6) | 199 (0.5) | 58 (0.5) | 80 (0.6) | 0.4 | 549 (0.5) |
| AIDS | 82 (0.2) | 16 (0.2) | 101 (0.3) | 27 (0.3) | 46 (0.3) | 0.2 | 272 (0.3) |
| Rheumatologic disease | 449 (1.2) | 71 (1.1) | 522 (1.4) | 158 (1.5) | 196 (1.5) | 0.007 | 1396 (1.3) |
| Cancer | 1736 (4.7) | 337 (5.1) | 1796 (4.9) | 571 (5.3) | 701 (5.3) | 0.02 | 5141 (4.9) |
| Net worth range (%) | |||||||
| <$25 000 | 1831 (4.9) | 404 (6.1) | 2101 (5.7) | 561 (5.2) | 695 (5.2) | <0.001 | 5592 (5.3) |
| $25 000–149 000 | 5188 (14) | 1042 (16) | 5608 (15) | 1591 (15) | 2046 (15) | 15 475 (15) | |
| $150 000–249 000 | 4754 (13) | 837 (13) | 5182 (14) | 1499 (14) | 1862 (14) | 14 134 (14) | |
| $250 000–499 000 | 9694 (26) | 1779 (27) | 10 048 (28) | 3027 (28) | 3636 (27) | 28 184 (27) | |
| ≥$500 000 | 10 910 (29) | 2030 (30) | 10 860 (30) | 3208 (30) | 3852 (29) | 30 860 (30) | |
| Unknown/missing | 4840 (13) | 580 (8.7) | 2804 (7.7) | 886 (8.2) | 1229 (9.2) | 10 339 (9.9) | |
| 30-d complications (%) | |||||||
| Any | 5973 (16) | 1151 (17) | 6120 (17) | 1863 (17) | 2451 (18) | <0.001 | 17 558 (17) |
| Bleeding | 2547 (6.8) | 499 (7.5) | 2660 (7.3) | 860 (8.0) | 1083 (8.1) | <0.001 | 7649 (7.3) |
| Nonsepsis infection | 2978 (8.0) | 590 (8.8) | 2990 (8.2) | 875 (8.1) | 1236 (9.3) | <0.001 | 8669 (8.3) |
| Sepsis | 191 (0.5) | 50 (0.7) | 349 (1.0) | 105 (1.0) | 134 (1.0) | <0.001 | 829 (0.8) |
| Urinary retention | 980 (2.6) | 161 (2.4) | 1138 (3.1) | 341 (3.2) | 422 (3.2) | <0.001 | 3042 (2.9) |
| Hospitalization | 1019 (2.7) | 181 (2.7) | 1056 (2.9) | 320 (3.0) | 367 (2.8) | 0.6 | 2943 (2.8) |
AIDS = acquired immunodeficiency syndrome; AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; IQR = interquartile range; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer [screening trial]; PSA = prostate-specific antigen; USPTF = US Preventive Services Task Force.
Overall, 30-d complications occurred in 17% of biopsies. The most prevalent was nonsepsis genitourinary infections (8.3%), followed by bleeding complications (7.3%) and urinary retention (2.9%) (Table 1). The 30-d hospitalization rate was 2.8%.
We first examined unadjusted rates of prostate biopsy among men aged ≥40 yr (Fig. 1 and Table 2). From January 2005 to September 2014, biopsy rates fell 33% from 64.1 to 42.8 per 100 000 person-months. There was substantial seasonal variation in biopsy rates with reductions near the end of each calendar year. With the ITS approach, each intervention is potentially associated with an immediate change in event rates (ie, level shift or change in intercept) as well as a change in the rate of growth (ie, change in slope). At baseline, there was a statistically significant growth in prostate biopsy rates of 2.9 biopsies per 100 000 person-months per year (95% confidence interval [CI], 0.96–4.8; p < 0.001). Following the 2008 USPSTF recommendations, there was an immediate decrease (ie, level shift) in biopsy rates of −10.1 biopsies per 100 000 person-months (95% CI, −17.1 to −3.0; p < 0.001). Likewise, there were immediate decreases in biopsy rates following the 2012 USPSTF recommendations (−13.8 biopsies per 100 000 person-months; 95% CI, −21.0 to −6.7; p < 0.001) and 2013 AUA guidelines (−8.8 biopsies per 100 000 person-months; 95% CI, −16.7 to −0.92; p = 0.03). Analyses adjusted for case mix were similar, although the baseline growth rate was no longer significant, indicating this was related to population shifts over time (Table 2, model 2).
Fig. 1
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 2
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315)
| Prostate biopsy | Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Model 1* | ||||||||||||||
| Baseline rate | 57.8 (53.4–62.3) |
<0.001 | 8.6 (7.5–9.7) |
<0.001 | 4.0 (3.3–4.6) |
<0.001 | 0.17 (0.03–0.32) |
0.02 | 2.8 (2.3–3.3) |
<0.001 | 1.5 (1.3–1.6) |
<0.001 | 1.4 (1.1–1.7) |
<0.001 |
| Baseline trend in rate | 2.9 (0.96–4.8) |
<0.001 | 0.84 (0.48–1.3) |
<0.001 | 0.72 (0.48–1.1) |
<0.001 | 0.12 (0.12–0.12) |
<0.001 | 0.36 (0.12–0.48) |
<0.001 | 0.12 (0.00–0.12) |
<0.001 | 0.00 (−0.12 to 0.12) |
0.5 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.1 (−17.1 to −3.0) |
<0.001 | −0.61 (−3.1 to 1.9) |
0.6 | −1.4 (−2.8 to −0.03) |
0.053 | 0.04 (−0.35 to 0.42) |
0.8 | 0.56 (−0.70 to 1.8) |
0.4 | −0.06 (−0.69 to 0.57) |
0.9 | 0.16 (−0.58 to 0.90) |
0.7 |
| Trend change | 7.9 (−12.7 to 28.4) |
0.5 | −1.0 (−9.1 to 7.2) |
0.8 | −0.12 (−4.1 to 4.0) |
0.98 | −0.60 (−1.9 to 0.72) |
0.4 | −2.0 (−6.5 to 2.4) |
0.4 | −1.1 (−3.4 to 1.3) |
0.4 | −1.0 (−3.6 to 1.6) |
0.5 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.6 (−12 to 4.5) |
0.4 | −0.74 (−3.8 to 2.4) |
0.6 | −0.12 (−1.8 to 1.5) |
0.9 | 0.29 (−0.20 to 0.78) |
0.2 | −0.04 (−1.6 to 1.6) |
0.97 | 0.59 (−0.23 to 1.4) |
0.2 | 0.51 (−0.43 to 1.5) |
0.3 |
| Trend change | −11.2 (−32.0 to 9.6) |
0.3 | 0.0 (−8.2 to 8.0) |
0.99 | −0.96 (−5.0 to 3.1) |
0.7 | 0.36 (−0.96 to 1.8) |
0.5 | 1.7 (−2.8 to 6.1) |
0.5 | 1.0 (−1.4 to 3.4) |
0.4 | 0.84 (−1.7 to 3.5) |
0.5 |
| 2012 USPSTF | ||||||||||||||
| Level change | −13.8 (−21.0 to −6.7) |
<0.001 | −2.5 (−4.5 to −0.45) |
0.02 | −1.7 (−3.1 to −0.31) |
0.02 | −0.12 (−0.42 to 0.19) |
0.5 | −0.61 (−1.5 to 0.28) |
0.2 | −0.75 (−1.1 to −0.44) |
<0.001 | −0.25 (−0.77 to 0.27) |
0.4 |
| Trend change | 8.2 (−5.8 to 22.0) |
0.3 | 3.0 (−0.36 to 6.4) |
0.09 | 3.1 (0.72–5.5) |
0.01 | 0.12 (−0.36 to 0.60) |
0.7 | 0.84 (−0.72 to 2.3) |
0.3 | 0.48 (0.00 to 1.1) |
0.08 | 0.24 (−0.60 to 1.2) |
0.6 |
| 2013 AUA | ||||||||||||||
| Level change | −8.8 (−16.7 to −0.92) |
0.03 | −1.4 (−3.8 to 1.1) |
0.3 | −1.9 (−3.5 to −0.22) |
0.03 | −0.02 (−0.41 to 0.37) |
0.9 | −0.19 (−1.3 to 0.91) |
0.7 | −0.38 (−0.77 to 0.01) |
0.06 | −0.01 (−0.66 to 0.63) |
0.96 |
| Trend change | −10.4 (−30.0 to 9.2) |
0.3 | −4.1 (−7.9 to −0.24) |
0.04 | −2.9 (−6.0 to 0.12) |
0.07 | −0.24 (−0.72 to 0.36) |
0.5 | −1.8 (−3.5 to 0.0) |
0.046 | −0.60 (−1.2 to 0.12) |
0.09 | −0.60 (−1.7 to 0.36) |
0.2 |
| Model 2† | ||||||||||||||
| Baseline rate | 86.8 (82.1–91.4) |
<0.001 | 16.2 (14.9–17.6) |
<0.001 | 6.4 (4.9–7.9) |
<0.001 | 0.39 (−0.13 to 0.91) |
0.2 | 6.1 (4.6–7.7) |
<0.001 | 3.6 (2.8–4.4) |
<0.001 | 3.0 (2.2–3.8) |
<0.001 |
| Baseline trend in rate | 0.24 (−1.8 to 2.3) |
0.8 | −0.24 (−0.72 to 0.24) |
0.4 | 0.48 (−0.12 to 1.1) |
0.09 | 0.12 (−0.12 to 0.24) |
0.4 | −0.24 (−0.72 to 0.24) |
0.4 | −0.24 (−0.48 to 0.00) |
0.054 | −0.12 (−0.36 to 0.12) |
0.3 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.8 (−17.8 to −3.8) |
<0.001 | −5.4 (−9.6 to −1.2) |
0.01 | −2.5 (−5.6 to 0.71) |
0.1 | −0.61 (−1.9 to 0.71) |
0.4 | 0.14 (−3.8 to 4.1) |
0.9 | 1.1 (−0.98 to 3.1) |
0.3 | 0.14 (−1.9 to 2.1) |
0.9 |
| Trend change | 10.6 (−10.1 to 31.2) |
0.3 | 6.1 (−8.4 to 20.8) |
0.4 | −1.7 (−12.2 to 8.8) |
0.8 | 1.7 (−2.9 to 6.2) |
0.5 | 1.7 (−12.0 to 15.5) |
0.8 | −4.7 (−11.8 to 2.5) |
0.2 | −0.84 (−7.8 to 6.1) |
0.8 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.9 (−12 to 4.1) |
0.3 | −0.24 (−5.6 to 5.1) |
0.9 | 1.9 (−2.1 to 5.8) |
0.4 | −0.32 (−2.0 to 1.4) |
0.7 | −0.99 (−6.0 to 4.0) |
0.7 | 1.8 (−0.84 to 4.4) |
0.2 | 0.87 (−1.7 to 3.4) |
0.5 |
| Trend change | −16.2 (−37.1 to 4.7) |
0.1 | −6.8 (−21.4 to 7.8) |
0.4 | 0.36 (−10.2 to 10.9) |
0.95 | −1.9 (−6.5 to 2.8) |
0.4 | −1.6 (−15.2 to 12.2) |
0.8 | 4.7 (−2.4 to 11.9) |
0.2 | 0.48 (−6.4 to 7.4) |
0.9 |
| 2012 USPSTF | ||||||||||||||
| Level change | −11.0 (−18.1 to −4.0) |
0.001 | −2.8 (−6.0 to 0.38) |
0.09 | 0.53 (−1.9 to 3.0) |
0.7 | −0.02 (−0.95 to 0.91) |
0.97 | −1.8 (−4.6 to 0.96) |
0.2 | −1.5 (−3.1 to 0.17) |
0.08 | 0.16 (−1.3 to 1.6) |
0.8 |
| Trend change | 7.0 (−7.2 to 21.1) |
0.3 | 1.7 (−3.6 to 7.0) |
0.5 | 0.12 (−4.1 to 4.2) |
0.96 | 0.12 (−1.4 to 1.8) |
0.9 | −0.48 (−5.3 to 4.3) |
0.9 | 1.6 (−1.3 to 4.3) |
0.3 | 0.12 (−2.4 to 2.6) |
0.9 |
| 2013 AUA | ||||||||||||||
| Level change | −8.2 (−15.9 to −0.38) |
0.04 | −1.0 (−4.8 to 2.8) |
0.6 | −0.44 (−3.2 to 2.3) |
0.8 | −0.15 (−1.3 to 0.99) |
0.8 | 0.28 (−3.2 to 3.7) |
0.9 | −0.50 (−2.7 to 1.7) |
0.7 | 0.47 (−1.3 to 2.3) |
0.6 |
| Trend change | −4.2 (−24.5 to 16.1) |
0.7 | −0.7 (−7.6 to 6.2) |
0.9 | 1.1 (−4.2 to 6.2) |
0.7 | −0.12 (−1.8 to 1.7) |
0.9 | −0.60 (−5.9 to 4.8) |
0.8 | −1.8 (−4.9 to 1.2) |
0.3 | −0.48 (−3.1 to 2.2) |
0.7 |
* Model 1: Base model with constant, time trend, intervention indicator variables. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
† Model 2: Adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, and anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPTF = US Preventive Services Task Force.
Next, we examined 30-d postbiopsy complications among men aged ≥40 yr as a measure of the absolute morbidity of biopsy (Fig. 1, Supplementary Fig. 3, and Table 2). Across the study periods, overall complication rates decreased 10% from 8.7 to 7.8 per 100 000 person-months. In unadjusted ITS analyses, there was a baseline growth in the overall complication rate (0.84 per 100 000 person-months per year; 95% CI, 0.48–1.3; p < 0.001), driven by growth in the rate of nonsepsis infections (0.72 per 100 000 person-months per year; 95% CI, 0.48–1.1; p < 0.001). Only the 2012 USPSTF recommendations were associated with a significant change in complication rates, following which there was an immediate decrease in the overall complication rate of −2.5 per 100 000 person-months (95% CI, −4.5 to −0.45; p = 0.02), driven by reductions in nonsepsis infections (−1.7 per 100 000 person-months; 95% CI, −3.1 to −0.31; p = 0.02) and urinary retention (−0.75 per 100 000 person-months;95% CI, −1.1 to −0.44; p < 0.001). Following the 2013 AUA guidelines there was also a reduction in the overall complication baseline growth rate (−4.1 per 100 000 person-months per year; 95% CI, −7.9 to −0.24; p = 0.04). In analyses adjusted for case mix (Table 2, model 2), effects were overall attenuated, and only the 2008 USPSTF recommendations were associated with a reduction in overall complication rates (−5.4 per 100 000 person-months; 95% CI, −9.6 to −1.2; p = 0.01).
To examine whether there was a change in the relative morbidity of biopsy, we performed a secondary analysis examining complication rates among men undergoing biopsy (Fig. 2). The proportion of men undergoing biopsy who experienced a complication increased from 14% in January 2005 to 18% in September 2014, an increase of 36%. There was again a baseline growth in overall complication rates (0.72% per year; 95% CI, 0.36–1.1; p < 0.001), driven by an increase in nonsepsis infections and sepsis (Supplementary Table 2). None of the interventions affected this growth. Similar results were obtained after adjusting for case mix (Table 3).
Fig. 2
Interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 3
Adjusted* interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Baseline rate | 15.6 (14.3–16.8) |
<0.001 | 7.1 (6.5–7.7) |
<0.001 | 0.34 (0.11–0.57) |
<0.001 | 6.7 (5.9–7.5) |
<0.001 | 2.8 (2.5–3.2) |
<0.001 | 3.2 (2.8–3.6) |
<0.001 |
| Baseline trend in rate | 0.60 (0.12–1.1) |
0.02 | 0.72 (0.60–0.96) |
<0.001 | 0.12 (0.12–0.24) |
<0.001 | 0.12 (−0.12 to 0.36) |
0.3 | −0.12 (−0.24 to 0.00) |
0.3 | −0.12 (−0.24 to 0.00) |
0.0498 |
| 2008 USPSTF | ||||||||||||
| Level change | 1.1 (−2.0 to 4.1) |
0.5 | −1.3 (−3.2 to 0.58) |
0.2 | 0.13 (−0.45 to 0.72) |
0.7 | 1.9 (−0.11 to 3.9) |
0.07 | 0.46 (−0.50 to 1.4) |
0.4 | 0.27 (−0.69 to 1.2) |
0.6 |
| Trend change | −3.5 (−13.4 to 6.6) |
0.5 | −1.8 (−8.4 to 4.7) |
0.6 | −1.0 (−3.0 to 0.96) |
0.3 | −4.3 (−11.4 to 2.6) |
0.2 | −1.6 (−4.9 to 1.9) |
0.4 | −0.72 (−4.0 to 2.6) |
0.7 |
| ERSPC/PLCO | ||||||||||||
| Level change | −0.14 (−3.9 to 3.6) |
0.9 | 1.1 (−1.3 to 3.4) |
0.4 | 0.55 (−0.19 to 1.3) |
0.2 | 0.39 (−2.2 to 3.0) |
0.8 | 0.75 (−0.47 to 2.0) |
0.2 | 0.65 (−0.56 to 1.9) |
0.3 |
| Trend change | 2.9 (−7.1 to 13.0) |
0.6 | 0.72 (−5.8 to 7.2) |
0.8 | 0.84 (−1.2 to 2.9) |
0.4 | 4.2 (−2.8 to 11.3) |
0.2 | 1.7 (−1.8 to 5.0) |
0.4 | 0.72 (−2.6 to 4.0) |
0.7 |
| 2012 USPSTF | ||||||||||||
| Level change | −0.27 (−2.7 to 2.1) |
0.8 | −0.56 (−2.0 to 0.87) |
0.5 | 0.16 (−0.31 to 0.63) |
0.5 | 1.1 (−0.38 to 2.5) |
0.2 | −0.38 (−1.0 to 0.28) |
0.3 | 0.44 (−0.34 to 1.2) |
0.3 |
| Trend change | 1.3 (−2.6 to 5.3) |
0.5 | 2.9 (0.60–5.3) |
0.02 | −0.24 (−1.1 to 0.60) |
0.6 | −0.72 (−3.2 to 1.7) |
0.5 | 0.24 (−0.96 to 1.3) |
0.7 | −0.48 (−1.8 to 0.84) |
0.5 |
| 2013 AUA | ||||||||||||
| Level change | 0.25 (−2.6 to 3.1) |
0.9 | −0.64 (−2.3 to 1.1) |
0.5 | 0.22 (−0.40 to 0.83) |
0.5 | 1.1 (−0.68 to 2.8) |
0.2 | −0.29 (−1.1 to 0.48) |
0.5 | 0.47 (−0.55 to 1.5) |
0.4 |
| Trend change | −1.6 (−6.1 to 2.9) |
0.5 | −2.5 (−5.5 to 0.60) |
0.1 | 0.12 (−0.72 to 1.1) |
0.8 | −0.36 (−3.1 to 2.4) |
0.8 | 0.24 (−1.2 to 1.7) |
0.8 | 0.12 (−1.3 to 1.6) |
0.9 |
* Model adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 person-months. Baseline trend in rate and trend change per 100 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial]; USPTF = US Preventive Services Task Force.
Given the increase in the relative morbidity of biopsy, we performed logistic regression to identify predictors of each complication category (Supplementary Table 3 and Table 4). Due to the large sample size, we focused on those features associated with the greatest magnitude of effect. A prior diagnosis of cancer was associated with an increased risk of all complication categories including any complication (odds ratio [OR]: 1.30; 95% CI, 1.22–1.39; p < 0.001) and hospitalization (OR: 1.59; 95% CI, 1.39–1.82; p < 0.001). Similarly, prior fluoroquinolone use was associated with an increased risk of all complication categories including any complication (OR: 1.27; 95% CI, 1.22–1.32; p < 0.001) and sepsis (OR: 1.55; 95% CI, 1.33–1.81; p < 0.001). Age at biopsy ≥70 yr was associated with a nearly 4-fold increased risk of urinary retention. Although anticoagulant use was associated with increased risks of bleeding (OR: 1.28; 95% CI, 1.14–1.45; p < 0.001), sepsis (OR: 1.48; 95% CI, 1.06–2.07; p = 0.02), and hospitalization (OR: 1.39; 95% CI, 1.16–1.67; p < 0.001), the increase in the overall complication risk was modest (OR 1.14; 95% CI, 1.04–1.25; p = 0.004).
Table 4
Multivariable analysis* of 30-d postbiopsy complications among men aged ≥40 yr undergoing prostate biopsy (January 2005 to September 2014) (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Hospitalization | Urinary retention | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Age at biopsy, yr | ||||||||||||
| 40–49 | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| 50–59 | 1.01 (0.94–1.09) | 0.8 | 0.94 (0.85–1.04) | 0.2 | 1.16 (0.83–1.63) | 0.4 | 1.01 (0.90–1.14) | 0.8 | 1.12 (0.93–1.36) | 0.2 | 1.54 (1.20–1.97) | <0.001 |
| 60–69 | 1.05 (0.97–1.13) | 0.2 | 0.94 (0.85–1.03) | 0.2 | 0.98 (0.70–1.37) | 0.9 | 1.05 (0.94–1.18) | 0.4 | 1.16 (0.95–1.40) | 0.1 | 2.16 (1.68–2.77) | <0.001 |
| ≥70 | 1.25 (1.15–1.36) | <0.001 | 1.03 (0.93–1.15) | 0.6 | 1.05 (0.74–1.49) | 0.8 | 1.29 (1.15–1.46) | <0.001 | 1.31 (1.07–1.60) | 0.008 | 3.98 (3.08–5.15) | <0.001 |
| Race | ||||||||||||
| White | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| Asian | 1.27 (1.15–1.40) | <0.001 | 1.27 (1.11–1.44) | 0.0003 | 1.65 (1.19–2.29) | 0.003 | 1.36 (1.18–1.56) | <0.001 | 0.94 (0.76–1.17) | 0.6 | 1.15 (0.93–1.42) | 0.2 |
| African American | 1.09 (1.03–1.15) | 0.002 | 1.05 (0.97–1.13) | 0.2 | 1.34 (1.09–1.64) | 0.005 | 1.17 (1.09–1.27) | <0.001 | 1.13 (1.01–1.27) | 0.03 | 1.15 (1.03–1.28) | 0.02 |
| Hispanic | 1.32 (1.24–1.40) | <0.001 | 1.36 (1.26–1.47) | <0.001 | 1.52 (1.21–1.90) | <0.001 | 1.33 (1.22–1.45) | <0.001 | 1.12 (0.97–1.28) | 0.1 | 1.37 (1.21–1.56) | <0.001 |
| Unknown/missing | 0.99 (0.93–1.06) | 0.8 | 1.07 (0.99–1.17) | 0.1 | 1.23 (0.94–1.61) | 0.1 | 0.96 (0.88–1.05) | 0.4 | 0.94 (0.81–1.08) | 0.4 | 1.02 (0.88–1.18) | 0.8 |
| Diabetes mellitus | 1.03 (0.99–1.07) | 0.09 | 1.06 (1.00–1.11) | 0.045 | 1.08 (0.92–1.28) | 0.4 | 1.01 (0.96–1.07) | 0.7 | 1.20 (1.10–1.31) | <0.001 | 0.99 (0.91–1.08) | 0.8 |
| COPD | 1.14 (1.09–1.20) | <0.001 | 1.07 (1.00–1.14) | 0.04 | 1.23 (1.01–1.49) | 0.04 | 1.14 (1.06–1.22) | <0.001 | 1.36 (1.23–1.51) | <0.001 | 1.14 (1.03–1.25) | 0.01 |
| PVD | 1.24 (1.16–1.31) | <0.001 | 1.1 (1.02–1.2) | 0.02 | 1.19 (0.93–1.51) | 0.2 | 1.28 (1.18–1.40) | <0.001 | 1.31 (1.15–1.48) | <0.001 | 1.26 (1.12–1.42) | <0.001 |
| Cerebrovascular disease | 1.11 (1.04–1.19) | 0.001 | 1.12 (1.03–1.22) | 0.01 | 1.21 (0.93–1.57) | 0.2 | 1.03 (0.94–1.14) | 0.5 | 1.26 (1.10–1.44) | <0.001 | 1.30 (1.15–1.47) | <0.001 |
| Cancer | 1.30 (1.22–1.39) | <0.001 | 1.16 (1.06–1.27) | 0.001 | 1.37 (1.05–1.79) | 0.02 | 1.40 (1.27–1.53) | <0.001 | 1.59 (1.39–1.82) | <0.001 | 1.32 (1.16–1.52) | <0.001 |
| Anticoagulant use | 1.14 (1.04–1.25) | 0.004 | 1.01 (0.89–1.14) | 0.9 | 1.48 (1.06–2.07) | 0.02 | 1.28 (1.14–1.45) | <0.001 | 1.39 (1.16–1.67) | <0.001 | 1.11 (0.92–1.33) | 0.3 |
| Prior fluoroquinolones | 1.27 (1.22–1.32) | <0.001 | 1.37 (1.31–1.45) | <0.001 | 1.55 (1.33–1.81) | <0.001 | 1.21 (1.14–1.28) | <0.001 | 1.25 (1.14–1.36) | <0.001 | 1.27 (1.16–1.38) | <0.001 |
* Models adjusted for year of biopsy and the features listed in the table.
CI = confidence interval; COPD = chronic obstructive pulmonary disease; PVD = peripheral vascular disease.
We conducted sensitivity analyses where the postintervention run-in period was varied from 2 to 4 mo (Supplementary Table 4). Results overall were similar, although effects were attenuated for the 2013 AUA guidelines on biopsy rates and the 2012 USPSTF recommendations on overall complication rates, likely due to fewer time points for analysis. In a separate sensitivity analysis, we examined bleeding complications excluding microhematuria from 2009 to 2014, following introduction of a separate code for microhematuria. Bleeding rates paralleled those of the primary bleeding end point (Supplementary Fig. 4), and results of both ITS and logistic regression analyses were similar (Supplementary Table 5 and 6). As an exploratory analysis, we examined PCa diagnosis rates within 60 d following biopsy (Supplementary Fig. 5). These appeared to increase steadily across the study period.
To our knowledge, this is the largest study to examine the impact of landmark PSA screening publications on rates of prostate biopsy and the first to examine their impact on postbiopsy complications. Overall, biopsy rates have declined by 33%, which appears attributable to the publication of the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines. Interestingly, the magnitude of reduction in biopsy rates exceeded previously reported decreases in PSA screening rates [6], [7], [8], [9], [10], [11], [12], [22], and [23], which suggests that either the threshold for referral to a urologist or the threshold for performing a biopsy has changed. More importantly, there was a concurrent 22% decrease in 30-d complication rates that appears largely attributable to the 2012 USPSTF recommendations. Still, the reduction in complication rates was less than the reduction in biopsy rates due to an increase in the relative morbidity of biopsy, driven by growth in infectious complications. This may in part be related to a shift to older patients with greater comorbidity undergoing biopsy, although similar results were obtained in analyses adjusted for case mix, which suggests additional factors may exist.
To this end, we evaluated predictors of postbiopsy complications to identify actionable targets for further risk reduction. Some of these features represent modifiable risk factors (eg, anticoagulant use), and others represent opportunities for altering periprocedural management. For instance, prior fluoroquinolone use may identify patients who would benefit from prebiopsy rectal swabs to guide periprocedural antibiotics [24] and [25]. Nonmodifiable risk factors, including specific comorbidities such as cancer and peripheral vascular disease, may encourage reconsideration of the risk–benefit ratio before proceeding with biopsy.
It is also important to consider the concurrent impact of landmark PSA screening publications on rates of PCa diagnosis to contextualize the observed changes in biopsy and its morbidity. An exploratory analysis suggested that cancer diagnosis rates have steadily increased, although further studies are necessary to examine this issue comprehensively.
A number of studies have examined the impact of the 2008 USPSTF recommendations [7], [8], [9], and [22], ERSPC/PLCO trials [6] and [9], and 2012 USPSTF recommendations [6], [10], [11], [12], and [23] on PSA screening rates, generally noting reductions ranging from 0% to 8%. However, much less data are available on the impact of these publications on biopsy rates. One single-center study noted a 31% reduction in the number of biopsies performed following the 2012 USPSTF recommendations [13]; another noted no change in the proportion of referred men who underwent biopsy [26]. A report from a Canadian academic health network noted a reduction in biopsy rates from 58.0 to 35.5 biopsies per month following the 2012 USPSTF recommendations [27]. Our findings therefore confirm a reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines.
Limited data exist regarding secular trends in postbiopsy complications, and prior studies have focused exclusively on the relative morbidity among men undergoing biopsy rather than absolute complication rates. A Canadian population-based study observed that the 30-d hospitalization rate increased from 1.0% in 1996 to 4.1% in 2005, driven by an increase in infectious complications [28]. A Surveillance, Epidemiology and End Results–Medicare study also noted an increase in hospitalizations related to infectious complications over time [14]. Although we observed a similar increase in 30-d overall complications among men undergoing biopsy, likewise driven by infectious complications, there was an absolute 22% reduction in 30-d complications over the study period. Taken together, these findings support a reduction in the morbidity of biopsy following publication of revised PSA screening guidelines but highlight the continued need to reduce the relative morbidity of biopsy.
A number of studies have described complication rates following prostate biopsy [15], [16], [29], and [30]. Hematuria has been reported in 10–84% of biopsies, rectal bleeding in 1–45%, infectious complications in up to 6.3%, urinary retention in up to 1.7%, and hospitalization in up to 6.9% [14] and [29]. The wide range in complication rates is due to differences in definitions, ascertainment of end points, and variation in clinical practice. Given the heterogeneity in postbiopsy complication rates, our findings have two advantages. First, they represent a contemporary, geographically diverse population across the entire United States. Second, use of claims-based ascertainment limited complications to those cases in which patients sought medical care, and although this may underestimate the comprehensive morbidity of biopsy, it more accurately reflects the economic burden of serious complications.
This study has several limitations. Most importantly, a control population of men who were not exposed to the PSA screening publications was not available. This study is retrospective, and additional, unrecognized changes in policy or clinical practice may possibly have confounded the impact of the interventions examined. However, the ITS approach has been described as the strongest quasi-experimental design to evaluate policy changes when a randomized trial is not possible [19]. Because we relied on administrative claims for ascertainment of complications, minor complications that did not generate a claim were not captured, thus underestimating these end points. Although we adjusted for case mix, we were unable to adjust for PSA. However, population shifts in PSA may be confounded by changes in screening practices, and PSA adjustment may therefore under- or overestimate the impact of interventions. We were also unable to adjust for other factors that may be associated with biopsy or complication rates, such as prostate magnetic resonance imaging (MRI), ancillary biomarker testing, biopsy approach (eg, MRI-fusion biopsy, transperineal biopsy), indication for anticoagulation, use of periprocedural bridging, or operator biopsy volume/experience.
We observed an immediate reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines, with a smaller reduction in absolute 30-d complication rates. Still, the relative morbidity of biopsy has increased over time, driven by infectious complications. Accordingly, if health policy aims to reduce the morbidity of PSA screening related to biopsy, further efforts should focus on improving patient selection and periprocedural processes of care.
Author contributions: Boris Gershman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Gershman, van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Acquisition of data: Van Houten, Shah.
Analysis and interpretation of data: Gershman, Van Houten, Shah, Karnes.
Drafting of the manuscript: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Critical revision of the manuscript for important intellectual content: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Statistical analysis: Van Houten.
Obtaining funding: Karnes, Shah.
Administrative, technical, or material support: Karnes, Shah.
Supervision: Karnes, Shah.
Other (specify): None.
Financial disclosures: Boris Gershman certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: This publication was made possible by funding from the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery.
Funding/Support and role of the sponsor: None.
In 2009, two landmark randomized trials—the European Randomized Study of Screening for Prostate Cancer [ERSPC] [1] and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial [PLCO] [2]—reported smaller than anticipated reductions in prostate cancer (PCa) mortality and prompted reexamination of prostate-specific antigen (PSA) screening recommendations. The US Preventive Services Task Force (USPSTF), which in 2008 had recommended against PSA screening in men aged >75 yr [3], updated its position in 2012 to recommend against systematic PSA screening in the general population [4]. The American Urological Association (AUA) likewise updated its PSA screening guidelines in May 2013, recommending shared decision making for men aged 55–69 yr [5]. Both the 2012 USPSTF and 2013 AUA guidelines cited prostate biopsy and its potential complications as a significant harm associated with PSA screening.
Although several studies have noted a reduction in PSA screening rates following these publications [6], [7], [8], [9], [10], [11], and [12], there are little data on whether this has translated into the intended reductions in screening-related morbidity in the form of prostate biopsy and postbiopsy complications [13]. We evaluated whether publication of the 2008 USPSTF recommendations, 2009 ERSPC/PLCO trials, 2012 USPSTF recommendations, or 2013 AUA guidelines was associated with reductions in prostate biopsy or postbiopsy complications. Biopsy-related complications also remain understudied both with regard to incidence as well as identification of risk factors [14], [15], and [16]. Consequently, we also sought to characterize postbiopsy complications and evaluate predictors using a contemporary population-based cohort.
We conducted a retrospective analysis of administrative claims from the Optum Labs Data Warehouse that contains data on >100 million individuals enrolled in private health plans and Medicare Advantage Plans, from geographically diverse regions across the United States, as described in Supplementary Methods [17]. Because this study involved analysis of preexisting deidentified data, it was exempt from institutional review board approval.
We identified all prostate biopsies performed in men aged ≥40 yr between January 1, 2005, and September 30, 2014, using Current Procedural Terminology-4 code 55700 alone (11%) or in combination with 76872 (84%) or 76942 (0%), or International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 60.11 (5%). The first biopsy was selected as the index biopsy when more than one biopsy was performed. Patients with a prior diagnosis of PCa (ICD-9 code 185) were excluded. We restricted the cohort to patients with continuous medical and pharmacy enrollment for at least 12 mo prior to biopsy and 30 d after biopsy to ascertain comorbidities and 30-d complications (Supplementary Fig. 1).
We recorded sociodemographic characteristics, comorbid conditions, anticoagulant use within 30 d of biopsy, and fluoroquinolone use in the 12 mo prior to biopsy as described in Supplement 1. The Charlson Comorbidity Index was calculated as previously described [18]. Prebiopsy PSA values within 60 d of biopsy were available for a subset of patients.
The primary outcomes were crude and adjusted rates of prostate biopsy and 30-d complications. Complications were identified using inpatient and outpatient administrative claims as summarized in Supplementary Table 1.
Baseline demographic features were compared across study periods using the Kruskal-Wallis test for continuous variables and the chi-square test for discrete variables. To evaluate the impact of each publication on rates of prostate biopsy and 30-d complications, we used an interrupted time series (ITS) approach as described in Supplementary Methods [19] and [20]. Rates were estimated using all men aged ≥40 yr with at least 12 mo of continuous medical and pharmacy enrollment as the denominator. As secondary analyses, we estimated rates of 30-d complications among men who underwent biopsy. Two separate ITS analyses were performed: (1) unadjusted analyses, using crude rates of biopsy and complications aggregated for each month, and (2) analyses standardized for case mix, to adjust for population shifts over time as described in Supplementary Methods [20] and [21].
Logistic regression was used to evaluate predictors of 30-d complications. Multivariable models were adjusted for age at biopsy, year of biopsy, race, anticoagulant use, history of fluoroquinolone use, and the presence of the five most prevalent Charlson comorbidities.
Statistical analyses were performed using SAS v.9.3 (SAS Institute Inc., Cary, NC, USA). All tests were two sided with p < 0.05 considered statistically significant.
Between January 2005 and September 2014, a total of 104 584 men underwent prostate biopsy (Table 1). Median age at biopsy was 63.0 yr (interquartile range [IQR]: 57.0–70.0). Overall, 18% of men had used fluoroquinolones in the 12 mo preceding biopsy, and 2.8% were on anticoagulants. Median prebiopsy PSA was 5.1 ng/dl (IQR: 4.0–7.1) for the 18 749 men (18%) for whom this was available. Over the time periods, men undergoing biopsy were older, had increased prevalence of anticoagulant use and comorbid conditions, and had a higher PSA (Table 1 and Supplementary Fig. 2).
Table 1
Patient demographics for men who underwent prostate biopsy (January 2005 to September 2014) stratified by pre/postintervention time periods (n = 104 584)
| Baseline Jan 2005–Jul 2008 (n = 37 217) |
2008 USPSTF Aug 2008–Feb 2009 (n = 6672) |
ERSPC/PLCO Mar 2009–Apr 2012 (n = 36 603) |
2012 USPSTF May 2012–Apr 2013 (n = 10 772) |
2013 AUA May 2013–Sep 2014 (n = 13 320) |
p value | Total (n = 104 584) | |
|---|---|---|---|---|---|---|---|
| Age at biopsy, yr | |||||||
| Median (IQR) | 62.0 (56.0–68.0) | 62.0 (56.0–68.0) | 63.0 (57.0–70.0) | 65.0 (59.0–71.0) | 66.0 (59.0–71.0) | <0.001 | 63.0 (57.0–70.0) |
| Age group, yr (%) | |||||||
| 40–49 | 2269 (6.1) | 416 (6.2) | 1737 (4.7) | 420 (3.9) | 467 (3.5) | <0.001 | 5309 (5.1) |
| 50–59 | 12 366 (33) | 2201 (33) | 10 342 (28) | 2644 (25) | 3098 (23) | 30 651 (29) | |
| 60–69 | 14 671 (39) | 2697 (40) | 15 105 (41) | 4426 (41) | 5546 (42) | 42 445 (41) | |
| ≥70 | 7911 (21) | 1358 (20) | 9419 (26) | 3282 (31) | 4209 (32) | 26 179 (25) | |
| Race (%) | |||||||
| White | 25 508 (69) | 4643 (70) | 26 187 (72) | 7795 (72) | 9407 (71) | <0.001 | 73 540 (70) |
| African American | 3695 (9.9) | 780 (12) | 4423 (12) | 1283 (12) | 1589 (12) | 11 770 (11) | |
| Hispanic | 2549 (6.8) | 538 (8.1) | 2911 (8.0) | 837 (7.8) | 1186 (8.9) | 8021 (7.7) | |
| Asian | 1133 (3.0) | 236 (3.5) | 1142 (3.1) | 306 (2.8) | 387 (2.9) | 3204 (3.1) | |
| Unknown/Missing | 4332 (12) | 475 (7.1) | 1940 (5.3) | 551 (5.1) | 751 (5.6) | 8049 (7.7) | |
| Census region (%) | |||||||
| Northeast | 7094 (19) | 1196 (18) | 6120 (17) | 1886 (18) | 2424 (18) | <0.001 | 18 720 (18) |
| Midwest | 10 094 (27) | 1562 (23) | 8837 (24) | 2742 (26) | 3627 (27) | 26 862 (26) | |
| South | 15 330 (41) | 2889 (43) | 16 371 (45) | 4698 (44) | 5538 (42) | 44 826 (43) | |
| West | 4660 (13) | 993 (15) | 5147 (14) | 1442 (13) | 1728 (13) | 13 970 (13) | |
| Other/Missing | 39 (0.1) | 32 (0.5) | 128 (0.3) | 4 (0) | 3 (0) | 206 (0.2) | |
| PSA, ng/d | n = 5174 | n = 548 | n = 7345 | n = 2211 | n = 3471 | n = 18 749 | |
| Median (IQR) | 5.1 (3.9–7.1) | 4.9 (3.8–6.8) | 5.0 (4.0–6.8) | 5.3 (4.2–7.4) | 5.4 (4.2–7.8) | <0.001 | 5.1 (4.0–7.1) |
| Anticoagulant use (%) | 928 (2.5) | 184 (2.8) | 1103 (3.0) | 299 (2.8) | 415 (3.1) | <0.001 | 2929 (2.8) |
| History of fluoroquinolone use in prior 12 mo (%) | 6797 (18) | 1267 (19) | 6557 (18) | 2017 (19) | 2504 (19) | 0.054 | 19 142 (18) |
| Charlson Comorbidity Index (%) | |||||||
| 0 | 22 354 (60) | 3887 (58) | 20 269 (55) | 5758 (54) | 7039 (53) | <0.001 | 59 307 (57) |
| 1 | 8199 (22) | 1453 (22) | 8252 (23) | 2503 (23) | 2999 (23) | 23 406 (22) | |
| 2 | 3232 (8.7) | 664 (10) | 3520 (9.6) | 1108 (10) | 1359 (10) | 9883 (9.4) | |
| ≥3 | 3432 (9.2) | 668 (10) | 4562 (13) | 1403 (13) | 1923 (14) | 11 988 (12) | |
| Comorbidities (%) | |||||||
| Myocardial infarction | 763 (2.1) | 144 (2.2) | 812 (2.2) | 227 (2.1) | 273 (2.0) | 0.6 | 2219 (2.1) |
| Congestive heart failure | 1443 (3.9) | 223 (3.3) | 1548 (4.2) | 467 (4.3) | 586 (4.4) | <0.001 | 4267 (4.1) |
| Peripheral vascular disease | 2054 (5.5) | 399 (6.0) | 2660 (7.3) | 783 (7.3) | 1071 (8.0) | <0.001 | 6967 (6.7) |
| Cerebrovascular disease | 2007 (5.4) | 379 (5.7) | 2358 (6.4) | 681 (6.3) | 947 (7.1) | <0.001 | 6372 (6.1) |
| Dementia | 370 (1.0) | 63 (0.9) | 442 (1.2) | 141 (1.3) | 198 (1.5) | <0.001 | 1214 (1.2) |
| Chronic obstructive pulmonary disease | 4251 (11) | 758 (11) | 4293 (12) | 1254 (12) | 1686 (13) | 0.004 | 12 242 (12) |
| Ulcer | 362 (1.0) | 60 (0.9) | 297 (0.8) | 84 (0.8) | 102 (0.8) | 0.07 | 905 (0.9) |
| Mild liver disease | 1182 (3.2) | 223 (3.3) | 1332 (3.6) | 413 (3.8) | 515 (3.9) | <0.001 | 3665 (3.5) |
| Diabetes mellitus | 6475 (17) | 1290 (19) | 7823 (21) | 2442 (23) | 2973 (22) | <0.001 | 21 003 (20) |
| Diabetes with organ damage | 1105 (3.0) | 225 (3.4) | 1600 (4.4) | 492 (4.6) | 734 (5.5) | <0.001 | 4156 (4.0) |
| Hemiplegia | 128 (0.3) | 25 (0.4) | 150 (0.4) | 51 (0.5) | 56 (0.4) | 0.3 | 410 (0.4) |
| Moderate/severe renal disease | 1133 (3.0) | 254 (3.8) | 1950 (5.3) | 674 (6.3) | 908 (6.8) | <0.001 | 4919 (4.7) |
| Moderate/severe liver disease | 27 (0.1) | 7 (0.1) | 31 (0.1) | 9 (0.1) | 12 (0.1) | 0.9 | 86 (0.1) |
| Metastatic solid tumor | 175 (0.5) | 37 (0.6) | 199 (0.5) | 58 (0.5) | 80 (0.6) | 0.4 | 549 (0.5) |
| AIDS | 82 (0.2) | 16 (0.2) | 101 (0.3) | 27 (0.3) | 46 (0.3) | 0.2 | 272 (0.3) |
| Rheumatologic disease | 449 (1.2) | 71 (1.1) | 522 (1.4) | 158 (1.5) | 196 (1.5) | 0.007 | 1396 (1.3) |
| Cancer | 1736 (4.7) | 337 (5.1) | 1796 (4.9) | 571 (5.3) | 701 (5.3) | 0.02 | 5141 (4.9) |
| Net worth range (%) | |||||||
| <$25 000 | 1831 (4.9) | 404 (6.1) | 2101 (5.7) | 561 (5.2) | 695 (5.2) | <0.001 | 5592 (5.3) |
| $25 000–149 000 | 5188 (14) | 1042 (16) | 5608 (15) | 1591 (15) | 2046 (15) | 15 475 (15) | |
| $150 000–249 000 | 4754 (13) | 837 (13) | 5182 (14) | 1499 (14) | 1862 (14) | 14 134 (14) | |
| $250 000–499 000 | 9694 (26) | 1779 (27) | 10 048 (28) | 3027 (28) | 3636 (27) | 28 184 (27) | |
| ≥$500 000 | 10 910 (29) | 2030 (30) | 10 860 (30) | 3208 (30) | 3852 (29) | 30 860 (30) | |
| Unknown/missing | 4840 (13) | 580 (8.7) | 2804 (7.7) | 886 (8.2) | 1229 (9.2) | 10 339 (9.9) | |
| 30-d complications (%) | |||||||
| Any | 5973 (16) | 1151 (17) | 6120 (17) | 1863 (17) | 2451 (18) | <0.001 | 17 558 (17) |
| Bleeding | 2547 (6.8) | 499 (7.5) | 2660 (7.3) | 860 (8.0) | 1083 (8.1) | <0.001 | 7649 (7.3) |
| Nonsepsis infection | 2978 (8.0) | 590 (8.8) | 2990 (8.2) | 875 (8.1) | 1236 (9.3) | <0.001 | 8669 (8.3) |
| Sepsis | 191 (0.5) | 50 (0.7) | 349 (1.0) | 105 (1.0) | 134 (1.0) | <0.001 | 829 (0.8) |
| Urinary retention | 980 (2.6) | 161 (2.4) | 1138 (3.1) | 341 (3.2) | 422 (3.2) | <0.001 | 3042 (2.9) |
| Hospitalization | 1019 (2.7) | 181 (2.7) | 1056 (2.9) | 320 (3.0) | 367 (2.8) | 0.6 | 2943 (2.8) |
AIDS = acquired immunodeficiency syndrome; AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; IQR = interquartile range; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer [screening trial]; PSA = prostate-specific antigen; USPTF = US Preventive Services Task Force.
Overall, 30-d complications occurred in 17% of biopsies. The most prevalent was nonsepsis genitourinary infections (8.3%), followed by bleeding complications (7.3%) and urinary retention (2.9%) (Table 1). The 30-d hospitalization rate was 2.8%.
We first examined unadjusted rates of prostate biopsy among men aged ≥40 yr (Fig. 1 and Table 2). From January 2005 to September 2014, biopsy rates fell 33% from 64.1 to 42.8 per 100 000 person-months. There was substantial seasonal variation in biopsy rates with reductions near the end of each calendar year. With the ITS approach, each intervention is potentially associated with an immediate change in event rates (ie, level shift or change in intercept) as well as a change in the rate of growth (ie, change in slope). At baseline, there was a statistically significant growth in prostate biopsy rates of 2.9 biopsies per 100 000 person-months per year (95% confidence interval [CI], 0.96–4.8; p < 0.001). Following the 2008 USPSTF recommendations, there was an immediate decrease (ie, level shift) in biopsy rates of −10.1 biopsies per 100 000 person-months (95% CI, −17.1 to −3.0; p < 0.001). Likewise, there were immediate decreases in biopsy rates following the 2012 USPSTF recommendations (−13.8 biopsies per 100 000 person-months; 95% CI, −21.0 to −6.7; p < 0.001) and 2013 AUA guidelines (−8.8 biopsies per 100 000 person-months; 95% CI, −16.7 to −0.92; p = 0.03). Analyses adjusted for case mix were similar, although the baseline growth rate was no longer significant, indicating this was related to population shifts over time (Table 2, model 2).
Fig. 1
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 2
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315)
| Prostate biopsy | Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Model 1* | ||||||||||||||
| Baseline rate | 57.8 (53.4–62.3) |
<0.001 | 8.6 (7.5–9.7) |
<0.001 | 4.0 (3.3–4.6) |
<0.001 | 0.17 (0.03–0.32) |
0.02 | 2.8 (2.3–3.3) |
<0.001 | 1.5 (1.3–1.6) |
<0.001 | 1.4 (1.1–1.7) |
<0.001 |
| Baseline trend in rate | 2.9 (0.96–4.8) |
<0.001 | 0.84 (0.48–1.3) |
<0.001 | 0.72 (0.48–1.1) |
<0.001 | 0.12 (0.12–0.12) |
<0.001 | 0.36 (0.12–0.48) |
<0.001 | 0.12 (0.00–0.12) |
<0.001 | 0.00 (−0.12 to 0.12) |
0.5 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.1 (−17.1 to −3.0) |
<0.001 | −0.61 (−3.1 to 1.9) |
0.6 | −1.4 (−2.8 to −0.03) |
0.053 | 0.04 (−0.35 to 0.42) |
0.8 | 0.56 (−0.70 to 1.8) |
0.4 | −0.06 (−0.69 to 0.57) |
0.9 | 0.16 (−0.58 to 0.90) |
0.7 |
| Trend change | 7.9 (−12.7 to 28.4) |
0.5 | −1.0 (−9.1 to 7.2) |
0.8 | −0.12 (−4.1 to 4.0) |
0.98 | −0.60 (−1.9 to 0.72) |
0.4 | −2.0 (−6.5 to 2.4) |
0.4 | −1.1 (−3.4 to 1.3) |
0.4 | −1.0 (−3.6 to 1.6) |
0.5 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.6 (−12 to 4.5) |
0.4 | −0.74 (−3.8 to 2.4) |
0.6 | −0.12 (−1.8 to 1.5) |
0.9 | 0.29 (−0.20 to 0.78) |
0.2 | −0.04 (−1.6 to 1.6) |
0.97 | 0.59 (−0.23 to 1.4) |
0.2 | 0.51 (−0.43 to 1.5) |
0.3 |
| Trend change | −11.2 (−32.0 to 9.6) |
0.3 | 0.0 (−8.2 to 8.0) |
0.99 | −0.96 (−5.0 to 3.1) |
0.7 | 0.36 (−0.96 to 1.8) |
0.5 | 1.7 (−2.8 to 6.1) |
0.5 | 1.0 (−1.4 to 3.4) |
0.4 | 0.84 (−1.7 to 3.5) |
0.5 |
| 2012 USPSTF | ||||||||||||||
| Level change | −13.8 (−21.0 to −6.7) |
<0.001 | −2.5 (−4.5 to −0.45) |
0.02 | −1.7 (−3.1 to −0.31) |
0.02 | −0.12 (−0.42 to 0.19) |
0.5 | −0.61 (−1.5 to 0.28) |
0.2 | −0.75 (−1.1 to −0.44) |
<0.001 | −0.25 (−0.77 to 0.27) |
0.4 |
| Trend change | 8.2 (−5.8 to 22.0) |
0.3 | 3.0 (−0.36 to 6.4) |
0.09 | 3.1 (0.72–5.5) |
0.01 | 0.12 (−0.36 to 0.60) |
0.7 | 0.84 (−0.72 to 2.3) |
0.3 | 0.48 (0.00 to 1.1) |
0.08 | 0.24 (−0.60 to 1.2) |
0.6 |
| 2013 AUA | ||||||||||||||
| Level change | −8.8 (−16.7 to −0.92) |
0.03 | −1.4 (−3.8 to 1.1) |
0.3 | −1.9 (−3.5 to −0.22) |
0.03 | −0.02 (−0.41 to 0.37) |
0.9 | −0.19 (−1.3 to 0.91) |
0.7 | −0.38 (−0.77 to 0.01) |
0.06 | −0.01 (−0.66 to 0.63) |
0.96 |
| Trend change | −10.4 (−30.0 to 9.2) |
0.3 | −4.1 (−7.9 to −0.24) |
0.04 | −2.9 (−6.0 to 0.12) |
0.07 | −0.24 (−0.72 to 0.36) |
0.5 | −1.8 (−3.5 to 0.0) |
0.046 | −0.60 (−1.2 to 0.12) |
0.09 | −0.60 (−1.7 to 0.36) |
0.2 |
| Model 2† | ||||||||||||||
| Baseline rate | 86.8 (82.1–91.4) |
<0.001 | 16.2 (14.9–17.6) |
<0.001 | 6.4 (4.9–7.9) |
<0.001 | 0.39 (−0.13 to 0.91) |
0.2 | 6.1 (4.6–7.7) |
<0.001 | 3.6 (2.8–4.4) |
<0.001 | 3.0 (2.2–3.8) |
<0.001 |
| Baseline trend in rate | 0.24 (−1.8 to 2.3) |
0.8 | −0.24 (−0.72 to 0.24) |
0.4 | 0.48 (−0.12 to 1.1) |
0.09 | 0.12 (−0.12 to 0.24) |
0.4 | −0.24 (−0.72 to 0.24) |
0.4 | −0.24 (−0.48 to 0.00) |
0.054 | −0.12 (−0.36 to 0.12) |
0.3 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.8 (−17.8 to −3.8) |
<0.001 | −5.4 (−9.6 to −1.2) |
0.01 | −2.5 (−5.6 to 0.71) |
0.1 | −0.61 (−1.9 to 0.71) |
0.4 | 0.14 (−3.8 to 4.1) |
0.9 | 1.1 (−0.98 to 3.1) |
0.3 | 0.14 (−1.9 to 2.1) |
0.9 |
| Trend change | 10.6 (−10.1 to 31.2) |
0.3 | 6.1 (−8.4 to 20.8) |
0.4 | −1.7 (−12.2 to 8.8) |
0.8 | 1.7 (−2.9 to 6.2) |
0.5 | 1.7 (−12.0 to 15.5) |
0.8 | −4.7 (−11.8 to 2.5) |
0.2 | −0.84 (−7.8 to 6.1) |
0.8 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.9 (−12 to 4.1) |
0.3 | −0.24 (−5.6 to 5.1) |
0.9 | 1.9 (−2.1 to 5.8) |
0.4 | −0.32 (−2.0 to 1.4) |
0.7 | −0.99 (−6.0 to 4.0) |
0.7 | 1.8 (−0.84 to 4.4) |
0.2 | 0.87 (−1.7 to 3.4) |
0.5 |
| Trend change | −16.2 (−37.1 to 4.7) |
0.1 | −6.8 (−21.4 to 7.8) |
0.4 | 0.36 (−10.2 to 10.9) |
0.95 | −1.9 (−6.5 to 2.8) |
0.4 | −1.6 (−15.2 to 12.2) |
0.8 | 4.7 (−2.4 to 11.9) |
0.2 | 0.48 (−6.4 to 7.4) |
0.9 |
| 2012 USPSTF | ||||||||||||||
| Level change | −11.0 (−18.1 to −4.0) |
0.001 | −2.8 (−6.0 to 0.38) |
0.09 | 0.53 (−1.9 to 3.0) |
0.7 | −0.02 (−0.95 to 0.91) |
0.97 | −1.8 (−4.6 to 0.96) |
0.2 | −1.5 (−3.1 to 0.17) |
0.08 | 0.16 (−1.3 to 1.6) |
0.8 |
| Trend change | 7.0 (−7.2 to 21.1) |
0.3 | 1.7 (−3.6 to 7.0) |
0.5 | 0.12 (−4.1 to 4.2) |
0.96 | 0.12 (−1.4 to 1.8) |
0.9 | −0.48 (−5.3 to 4.3) |
0.9 | 1.6 (−1.3 to 4.3) |
0.3 | 0.12 (−2.4 to 2.6) |
0.9 |
| 2013 AUA | ||||||||||||||
| Level change | −8.2 (−15.9 to −0.38) |
0.04 | −1.0 (−4.8 to 2.8) |
0.6 | −0.44 (−3.2 to 2.3) |
0.8 | −0.15 (−1.3 to 0.99) |
0.8 | 0.28 (−3.2 to 3.7) |
0.9 | −0.50 (−2.7 to 1.7) |
0.7 | 0.47 (−1.3 to 2.3) |
0.6 |
| Trend change | −4.2 (−24.5 to 16.1) |
0.7 | −0.7 (−7.6 to 6.2) |
0.9 | 1.1 (−4.2 to 6.2) |
0.7 | −0.12 (−1.8 to 1.7) |
0.9 | −0.60 (−5.9 to 4.8) |
0.8 | −1.8 (−4.9 to 1.2) |
0.3 | −0.48 (−3.1 to 2.2) |
0.7 |
* Model 1: Base model with constant, time trend, intervention indicator variables. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
† Model 2: Adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, and anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPTF = US Preventive Services Task Force.
Next, we examined 30-d postbiopsy complications among men aged ≥40 yr as a measure of the absolute morbidity of biopsy (Fig. 1, Supplementary Fig. 3, and Table 2). Across the study periods, overall complication rates decreased 10% from 8.7 to 7.8 per 100 000 person-months. In unadjusted ITS analyses, there was a baseline growth in the overall complication rate (0.84 per 100 000 person-months per year; 95% CI, 0.48–1.3; p < 0.001), driven by growth in the rate of nonsepsis infections (0.72 per 100 000 person-months per year; 95% CI, 0.48–1.1; p < 0.001). Only the 2012 USPSTF recommendations were associated with a significant change in complication rates, following which there was an immediate decrease in the overall complication rate of −2.5 per 100 000 person-months (95% CI, −4.5 to −0.45; p = 0.02), driven by reductions in nonsepsis infections (−1.7 per 100 000 person-months; 95% CI, −3.1 to −0.31; p = 0.02) and urinary retention (−0.75 per 100 000 person-months;95% CI, −1.1 to −0.44; p < 0.001). Following the 2013 AUA guidelines there was also a reduction in the overall complication baseline growth rate (−4.1 per 100 000 person-months per year; 95% CI, −7.9 to −0.24; p = 0.04). In analyses adjusted for case mix (Table 2, model 2), effects were overall attenuated, and only the 2008 USPSTF recommendations were associated with a reduction in overall complication rates (−5.4 per 100 000 person-months; 95% CI, −9.6 to −1.2; p = 0.01).
To examine whether there was a change in the relative morbidity of biopsy, we performed a secondary analysis examining complication rates among men undergoing biopsy (Fig. 2). The proportion of men undergoing biopsy who experienced a complication increased from 14% in January 2005 to 18% in September 2014, an increase of 36%. There was again a baseline growth in overall complication rates (0.72% per year; 95% CI, 0.36–1.1; p < 0.001), driven by an increase in nonsepsis infections and sepsis (Supplementary Table 2). None of the interventions affected this growth. Similar results were obtained after adjusting for case mix (Table 3).
Fig. 2
Interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 3
Adjusted* interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Baseline rate | 15.6 (14.3–16.8) |
<0.001 | 7.1 (6.5–7.7) |
<0.001 | 0.34 (0.11–0.57) |
<0.001 | 6.7 (5.9–7.5) |
<0.001 | 2.8 (2.5–3.2) |
<0.001 | 3.2 (2.8–3.6) |
<0.001 |
| Baseline trend in rate | 0.60 (0.12–1.1) |
0.02 | 0.72 (0.60–0.96) |
<0.001 | 0.12 (0.12–0.24) |
<0.001 | 0.12 (−0.12 to 0.36) |
0.3 | −0.12 (−0.24 to 0.00) |
0.3 | −0.12 (−0.24 to 0.00) |
0.0498 |
| 2008 USPSTF | ||||||||||||
| Level change | 1.1 (−2.0 to 4.1) |
0.5 | −1.3 (−3.2 to 0.58) |
0.2 | 0.13 (−0.45 to 0.72) |
0.7 | 1.9 (−0.11 to 3.9) |
0.07 | 0.46 (−0.50 to 1.4) |
0.4 | 0.27 (−0.69 to 1.2) |
0.6 |
| Trend change | −3.5 (−13.4 to 6.6) |
0.5 | −1.8 (−8.4 to 4.7) |
0.6 | −1.0 (−3.0 to 0.96) |
0.3 | −4.3 (−11.4 to 2.6) |
0.2 | −1.6 (−4.9 to 1.9) |
0.4 | −0.72 (−4.0 to 2.6) |
0.7 |
| ERSPC/PLCO | ||||||||||||
| Level change | −0.14 (−3.9 to 3.6) |
0.9 | 1.1 (−1.3 to 3.4) |
0.4 | 0.55 (−0.19 to 1.3) |
0.2 | 0.39 (−2.2 to 3.0) |
0.8 | 0.75 (−0.47 to 2.0) |
0.2 | 0.65 (−0.56 to 1.9) |
0.3 |
| Trend change | 2.9 (−7.1 to 13.0) |
0.6 | 0.72 (−5.8 to 7.2) |
0.8 | 0.84 (−1.2 to 2.9) |
0.4 | 4.2 (−2.8 to 11.3) |
0.2 | 1.7 (−1.8 to 5.0) |
0.4 | 0.72 (−2.6 to 4.0) |
0.7 |
| 2012 USPSTF | ||||||||||||
| Level change | −0.27 (−2.7 to 2.1) |
0.8 | −0.56 (−2.0 to 0.87) |
0.5 | 0.16 (−0.31 to 0.63) |
0.5 | 1.1 (−0.38 to 2.5) |
0.2 | −0.38 (−1.0 to 0.28) |
0.3 | 0.44 (−0.34 to 1.2) |
0.3 |
| Trend change | 1.3 (−2.6 to 5.3) |
0.5 | 2.9 (0.60–5.3) |
0.02 | −0.24 (−1.1 to 0.60) |
0.6 | −0.72 (−3.2 to 1.7) |
0.5 | 0.24 (−0.96 to 1.3) |
0.7 | −0.48 (−1.8 to 0.84) |
0.5 |
| 2013 AUA | ||||||||||||
| Level change | 0.25 (−2.6 to 3.1) |
0.9 | −0.64 (−2.3 to 1.1) |
0.5 | 0.22 (−0.40 to 0.83) |
0.5 | 1.1 (−0.68 to 2.8) |
0.2 | −0.29 (−1.1 to 0.48) |
0.5 | 0.47 (−0.55 to 1.5) |
0.4 |
| Trend change | −1.6 (−6.1 to 2.9) |
0.5 | −2.5 (−5.5 to 0.60) |
0.1 | 0.12 (−0.72 to 1.1) |
0.8 | −0.36 (−3.1 to 2.4) |
0.8 | 0.24 (−1.2 to 1.7) |
0.8 | 0.12 (−1.3 to 1.6) |
0.9 |
* Model adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 person-months. Baseline trend in rate and trend change per 100 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial]; USPTF = US Preventive Services Task Force.
Given the increase in the relative morbidity of biopsy, we performed logistic regression to identify predictors of each complication category (Supplementary Table 3 and Table 4). Due to the large sample size, we focused on those features associated with the greatest magnitude of effect. A prior diagnosis of cancer was associated with an increased risk of all complication categories including any complication (odds ratio [OR]: 1.30; 95% CI, 1.22–1.39; p < 0.001) and hospitalization (OR: 1.59; 95% CI, 1.39–1.82; p < 0.001). Similarly, prior fluoroquinolone use was associated with an increased risk of all complication categories including any complication (OR: 1.27; 95% CI, 1.22–1.32; p < 0.001) and sepsis (OR: 1.55; 95% CI, 1.33–1.81; p < 0.001). Age at biopsy ≥70 yr was associated with a nearly 4-fold increased risk of urinary retention. Although anticoagulant use was associated with increased risks of bleeding (OR: 1.28; 95% CI, 1.14–1.45; p < 0.001), sepsis (OR: 1.48; 95% CI, 1.06–2.07; p = 0.02), and hospitalization (OR: 1.39; 95% CI, 1.16–1.67; p < 0.001), the increase in the overall complication risk was modest (OR 1.14; 95% CI, 1.04–1.25; p = 0.004).
Table 4
Multivariable analysis* of 30-d postbiopsy complications among men aged ≥40 yr undergoing prostate biopsy (January 2005 to September 2014) (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Hospitalization | Urinary retention | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Age at biopsy, yr | ||||||||||||
| 40–49 | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| 50–59 | 1.01 (0.94–1.09) | 0.8 | 0.94 (0.85–1.04) | 0.2 | 1.16 (0.83–1.63) | 0.4 | 1.01 (0.90–1.14) | 0.8 | 1.12 (0.93–1.36) | 0.2 | 1.54 (1.20–1.97) | <0.001 |
| 60–69 | 1.05 (0.97–1.13) | 0.2 | 0.94 (0.85–1.03) | 0.2 | 0.98 (0.70–1.37) | 0.9 | 1.05 (0.94–1.18) | 0.4 | 1.16 (0.95–1.40) | 0.1 | 2.16 (1.68–2.77) | <0.001 |
| ≥70 | 1.25 (1.15–1.36) | <0.001 | 1.03 (0.93–1.15) | 0.6 | 1.05 (0.74–1.49) | 0.8 | 1.29 (1.15–1.46) | <0.001 | 1.31 (1.07–1.60) | 0.008 | 3.98 (3.08–5.15) | <0.001 |
| Race | ||||||||||||
| White | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| Asian | 1.27 (1.15–1.40) | <0.001 | 1.27 (1.11–1.44) | 0.0003 | 1.65 (1.19–2.29) | 0.003 | 1.36 (1.18–1.56) | <0.001 | 0.94 (0.76–1.17) | 0.6 | 1.15 (0.93–1.42) | 0.2 |
| African American | 1.09 (1.03–1.15) | 0.002 | 1.05 (0.97–1.13) | 0.2 | 1.34 (1.09–1.64) | 0.005 | 1.17 (1.09–1.27) | <0.001 | 1.13 (1.01–1.27) | 0.03 | 1.15 (1.03–1.28) | 0.02 |
| Hispanic | 1.32 (1.24–1.40) | <0.001 | 1.36 (1.26–1.47) | <0.001 | 1.52 (1.21–1.90) | <0.001 | 1.33 (1.22–1.45) | <0.001 | 1.12 (0.97–1.28) | 0.1 | 1.37 (1.21–1.56) | <0.001 |
| Unknown/missing | 0.99 (0.93–1.06) | 0.8 | 1.07 (0.99–1.17) | 0.1 | 1.23 (0.94–1.61) | 0.1 | 0.96 (0.88–1.05) | 0.4 | 0.94 (0.81–1.08) | 0.4 | 1.02 (0.88–1.18) | 0.8 |
| Diabetes mellitus | 1.03 (0.99–1.07) | 0.09 | 1.06 (1.00–1.11) | 0.045 | 1.08 (0.92–1.28) | 0.4 | 1.01 (0.96–1.07) | 0.7 | 1.20 (1.10–1.31) | <0.001 | 0.99 (0.91–1.08) | 0.8 |
| COPD | 1.14 (1.09–1.20) | <0.001 | 1.07 (1.00–1.14) | 0.04 | 1.23 (1.01–1.49) | 0.04 | 1.14 (1.06–1.22) | <0.001 | 1.36 (1.23–1.51) | <0.001 | 1.14 (1.03–1.25) | 0.01 |
| PVD | 1.24 (1.16–1.31) | <0.001 | 1.1 (1.02–1.2) | 0.02 | 1.19 (0.93–1.51) | 0.2 | 1.28 (1.18–1.40) | <0.001 | 1.31 (1.15–1.48) | <0.001 | 1.26 (1.12–1.42) | <0.001 |
| Cerebrovascular disease | 1.11 (1.04–1.19) | 0.001 | 1.12 (1.03–1.22) | 0.01 | 1.21 (0.93–1.57) | 0.2 | 1.03 (0.94–1.14) | 0.5 | 1.26 (1.10–1.44) | <0.001 | 1.30 (1.15–1.47) | <0.001 |
| Cancer | 1.30 (1.22–1.39) | <0.001 | 1.16 (1.06–1.27) | 0.001 | 1.37 (1.05–1.79) | 0.02 | 1.40 (1.27–1.53) | <0.001 | 1.59 (1.39–1.82) | <0.001 | 1.32 (1.16–1.52) | <0.001 |
| Anticoagulant use | 1.14 (1.04–1.25) | 0.004 | 1.01 (0.89–1.14) | 0.9 | 1.48 (1.06–2.07) | 0.02 | 1.28 (1.14–1.45) | <0.001 | 1.39 (1.16–1.67) | <0.001 | 1.11 (0.92–1.33) | 0.3 |
| Prior fluoroquinolones | 1.27 (1.22–1.32) | <0.001 | 1.37 (1.31–1.45) | <0.001 | 1.55 (1.33–1.81) | <0.001 | 1.21 (1.14–1.28) | <0.001 | 1.25 (1.14–1.36) | <0.001 | 1.27 (1.16–1.38) | <0.001 |
* Models adjusted for year of biopsy and the features listed in the table.
CI = confidence interval; COPD = chronic obstructive pulmonary disease; PVD = peripheral vascular disease.
We conducted sensitivity analyses where the postintervention run-in period was varied from 2 to 4 mo (Supplementary Table 4). Results overall were similar, although effects were attenuated for the 2013 AUA guidelines on biopsy rates and the 2012 USPSTF recommendations on overall complication rates, likely due to fewer time points for analysis. In a separate sensitivity analysis, we examined bleeding complications excluding microhematuria from 2009 to 2014, following introduction of a separate code for microhematuria. Bleeding rates paralleled those of the primary bleeding end point (Supplementary Fig. 4), and results of both ITS and logistic regression analyses were similar (Supplementary Table 5 and 6). As an exploratory analysis, we examined PCa diagnosis rates within 60 d following biopsy (Supplementary Fig. 5). These appeared to increase steadily across the study period.
To our knowledge, this is the largest study to examine the impact of landmark PSA screening publications on rates of prostate biopsy and the first to examine their impact on postbiopsy complications. Overall, biopsy rates have declined by 33%, which appears attributable to the publication of the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines. Interestingly, the magnitude of reduction in biopsy rates exceeded previously reported decreases in PSA screening rates [6], [7], [8], [9], [10], [11], [12], [22], and [23], which suggests that either the threshold for referral to a urologist or the threshold for performing a biopsy has changed. More importantly, there was a concurrent 22% decrease in 30-d complication rates that appears largely attributable to the 2012 USPSTF recommendations. Still, the reduction in complication rates was less than the reduction in biopsy rates due to an increase in the relative morbidity of biopsy, driven by growth in infectious complications. This may in part be related to a shift to older patients with greater comorbidity undergoing biopsy, although similar results were obtained in analyses adjusted for case mix, which suggests additional factors may exist.
To this end, we evaluated predictors of postbiopsy complications to identify actionable targets for further risk reduction. Some of these features represent modifiable risk factors (eg, anticoagulant use), and others represent opportunities for altering periprocedural management. For instance, prior fluoroquinolone use may identify patients who would benefit from prebiopsy rectal swabs to guide periprocedural antibiotics [24] and [25]. Nonmodifiable risk factors, including specific comorbidities such as cancer and peripheral vascular disease, may encourage reconsideration of the risk–benefit ratio before proceeding with biopsy.
It is also important to consider the concurrent impact of landmark PSA screening publications on rates of PCa diagnosis to contextualize the observed changes in biopsy and its morbidity. An exploratory analysis suggested that cancer diagnosis rates have steadily increased, although further studies are necessary to examine this issue comprehensively.
A number of studies have examined the impact of the 2008 USPSTF recommendations [7], [8], [9], and [22], ERSPC/PLCO trials [6] and [9], and 2012 USPSTF recommendations [6], [10], [11], [12], and [23] on PSA screening rates, generally noting reductions ranging from 0% to 8%. However, much less data are available on the impact of these publications on biopsy rates. One single-center study noted a 31% reduction in the number of biopsies performed following the 2012 USPSTF recommendations [13]; another noted no change in the proportion of referred men who underwent biopsy [26]. A report from a Canadian academic health network noted a reduction in biopsy rates from 58.0 to 35.5 biopsies per month following the 2012 USPSTF recommendations [27]. Our findings therefore confirm a reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines.
Limited data exist regarding secular trends in postbiopsy complications, and prior studies have focused exclusively on the relative morbidity among men undergoing biopsy rather than absolute complication rates. A Canadian population-based study observed that the 30-d hospitalization rate increased from 1.0% in 1996 to 4.1% in 2005, driven by an increase in infectious complications [28]. A Surveillance, Epidemiology and End Results–Medicare study also noted an increase in hospitalizations related to infectious complications over time [14]. Although we observed a similar increase in 30-d overall complications among men undergoing biopsy, likewise driven by infectious complications, there was an absolute 22% reduction in 30-d complications over the study period. Taken together, these findings support a reduction in the morbidity of biopsy following publication of revised PSA screening guidelines but highlight the continued need to reduce the relative morbidity of biopsy.
A number of studies have described complication rates following prostate biopsy [15], [16], [29], and [30]. Hematuria has been reported in 10–84% of biopsies, rectal bleeding in 1–45%, infectious complications in up to 6.3%, urinary retention in up to 1.7%, and hospitalization in up to 6.9% [14] and [29]. The wide range in complication rates is due to differences in definitions, ascertainment of end points, and variation in clinical practice. Given the heterogeneity in postbiopsy complication rates, our findings have two advantages. First, they represent a contemporary, geographically diverse population across the entire United States. Second, use of claims-based ascertainment limited complications to those cases in which patients sought medical care, and although this may underestimate the comprehensive morbidity of biopsy, it more accurately reflects the economic burden of serious complications.
This study has several limitations. Most importantly, a control population of men who were not exposed to the PSA screening publications was not available. This study is retrospective, and additional, unrecognized changes in policy or clinical practice may possibly have confounded the impact of the interventions examined. However, the ITS approach has been described as the strongest quasi-experimental design to evaluate policy changes when a randomized trial is not possible [19]. Because we relied on administrative claims for ascertainment of complications, minor complications that did not generate a claim were not captured, thus underestimating these end points. Although we adjusted for case mix, we were unable to adjust for PSA. However, population shifts in PSA may be confounded by changes in screening practices, and PSA adjustment may therefore under- or overestimate the impact of interventions. We were also unable to adjust for other factors that may be associated with biopsy or complication rates, such as prostate magnetic resonance imaging (MRI), ancillary biomarker testing, biopsy approach (eg, MRI-fusion biopsy, transperineal biopsy), indication for anticoagulation, use of periprocedural bridging, or operator biopsy volume/experience.
We observed an immediate reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines, with a smaller reduction in absolute 30-d complication rates. Still, the relative morbidity of biopsy has increased over time, driven by infectious complications. Accordingly, if health policy aims to reduce the morbidity of PSA screening related to biopsy, further efforts should focus on improving patient selection and periprocedural processes of care.
Author contributions: Boris Gershman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Gershman, van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Acquisition of data: Van Houten, Shah.
Analysis and interpretation of data: Gershman, Van Houten, Shah, Karnes.
Drafting of the manuscript: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Critical revision of the manuscript for important intellectual content: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Statistical analysis: Van Houten.
Obtaining funding: Karnes, Shah.
Administrative, technical, or material support: Karnes, Shah.
Supervision: Karnes, Shah.
Other (specify): None.
Financial disclosures: Boris Gershman certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: This publication was made possible by funding from the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery.
Funding/Support and role of the sponsor: None.
In 2009, two landmark randomized trials—the European Randomized Study of Screening for Prostate Cancer [ERSPC] [1] and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial [PLCO] [2]—reported smaller than anticipated reductions in prostate cancer (PCa) mortality and prompted reexamination of prostate-specific antigen (PSA) screening recommendations. The US Preventive Services Task Force (USPSTF), which in 2008 had recommended against PSA screening in men aged >75 yr [3], updated its position in 2012 to recommend against systematic PSA screening in the general population [4]. The American Urological Association (AUA) likewise updated its PSA screening guidelines in May 2013, recommending shared decision making for men aged 55–69 yr [5]. Both the 2012 USPSTF and 2013 AUA guidelines cited prostate biopsy and its potential complications as a significant harm associated with PSA screening.
Although several studies have noted a reduction in PSA screening rates following these publications [6], [7], [8], [9], [10], [11], and [12], there are little data on whether this has translated into the intended reductions in screening-related morbidity in the form of prostate biopsy and postbiopsy complications [13]. We evaluated whether publication of the 2008 USPSTF recommendations, 2009 ERSPC/PLCO trials, 2012 USPSTF recommendations, or 2013 AUA guidelines was associated with reductions in prostate biopsy or postbiopsy complications. Biopsy-related complications also remain understudied both with regard to incidence as well as identification of risk factors [14], [15], and [16]. Consequently, we also sought to characterize postbiopsy complications and evaluate predictors using a contemporary population-based cohort.
We conducted a retrospective analysis of administrative claims from the Optum Labs Data Warehouse that contains data on >100 million individuals enrolled in private health plans and Medicare Advantage Plans, from geographically diverse regions across the United States, as described in Supplementary Methods [17]. Because this study involved analysis of preexisting deidentified data, it was exempt from institutional review board approval.
We identified all prostate biopsies performed in men aged ≥40 yr between January 1, 2005, and September 30, 2014, using Current Procedural Terminology-4 code 55700 alone (11%) or in combination with 76872 (84%) or 76942 (0%), or International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 60.11 (5%). The first biopsy was selected as the index biopsy when more than one biopsy was performed. Patients with a prior diagnosis of PCa (ICD-9 code 185) were excluded. We restricted the cohort to patients with continuous medical and pharmacy enrollment for at least 12 mo prior to biopsy and 30 d after biopsy to ascertain comorbidities and 30-d complications (Supplementary Fig. 1).
We recorded sociodemographic characteristics, comorbid conditions, anticoagulant use within 30 d of biopsy, and fluoroquinolone use in the 12 mo prior to biopsy as described in Supplement 1. The Charlson Comorbidity Index was calculated as previously described [18]. Prebiopsy PSA values within 60 d of biopsy were available for a subset of patients.
The primary outcomes were crude and adjusted rates of prostate biopsy and 30-d complications. Complications were identified using inpatient and outpatient administrative claims as summarized in Supplementary Table 1.
Baseline demographic features were compared across study periods using the Kruskal-Wallis test for continuous variables and the chi-square test for discrete variables. To evaluate the impact of each publication on rates of prostate biopsy and 30-d complications, we used an interrupted time series (ITS) approach as described in Supplementary Methods [19] and [20]. Rates were estimated using all men aged ≥40 yr with at least 12 mo of continuous medical and pharmacy enrollment as the denominator. As secondary analyses, we estimated rates of 30-d complications among men who underwent biopsy. Two separate ITS analyses were performed: (1) unadjusted analyses, using crude rates of biopsy and complications aggregated for each month, and (2) analyses standardized for case mix, to adjust for population shifts over time as described in Supplementary Methods [20] and [21].
Logistic regression was used to evaluate predictors of 30-d complications. Multivariable models were adjusted for age at biopsy, year of biopsy, race, anticoagulant use, history of fluoroquinolone use, and the presence of the five most prevalent Charlson comorbidities.
Statistical analyses were performed using SAS v.9.3 (SAS Institute Inc., Cary, NC, USA). All tests were two sided with p < 0.05 considered statistically significant.
Between January 2005 and September 2014, a total of 104 584 men underwent prostate biopsy (Table 1). Median age at biopsy was 63.0 yr (interquartile range [IQR]: 57.0–70.0). Overall, 18% of men had used fluoroquinolones in the 12 mo preceding biopsy, and 2.8% were on anticoagulants. Median prebiopsy PSA was 5.1 ng/dl (IQR: 4.0–7.1) for the 18 749 men (18%) for whom this was available. Over the time periods, men undergoing biopsy were older, had increased prevalence of anticoagulant use and comorbid conditions, and had a higher PSA (Table 1 and Supplementary Fig. 2).
Table 1
Patient demographics for men who underwent prostate biopsy (January 2005 to September 2014) stratified by pre/postintervention time periods (n = 104 584)
| Baseline Jan 2005–Jul 2008 (n = 37 217) |
2008 USPSTF Aug 2008–Feb 2009 (n = 6672) |
ERSPC/PLCO Mar 2009–Apr 2012 (n = 36 603) |
2012 USPSTF May 2012–Apr 2013 (n = 10 772) |
2013 AUA May 2013–Sep 2014 (n = 13 320) |
p value | Total (n = 104 584) | |
|---|---|---|---|---|---|---|---|
| Age at biopsy, yr | |||||||
| Median (IQR) | 62.0 (56.0–68.0) | 62.0 (56.0–68.0) | 63.0 (57.0–70.0) | 65.0 (59.0–71.0) | 66.0 (59.0–71.0) | <0.001 | 63.0 (57.0–70.0) |
| Age group, yr (%) | |||||||
| 40–49 | 2269 (6.1) | 416 (6.2) | 1737 (4.7) | 420 (3.9) | 467 (3.5) | <0.001 | 5309 (5.1) |
| 50–59 | 12 366 (33) | 2201 (33) | 10 342 (28) | 2644 (25) | 3098 (23) | 30 651 (29) | |
| 60–69 | 14 671 (39) | 2697 (40) | 15 105 (41) | 4426 (41) | 5546 (42) | 42 445 (41) | |
| ≥70 | 7911 (21) | 1358 (20) | 9419 (26) | 3282 (31) | 4209 (32) | 26 179 (25) | |
| Race (%) | |||||||
| White | 25 508 (69) | 4643 (70) | 26 187 (72) | 7795 (72) | 9407 (71) | <0.001 | 73 540 (70) |
| African American | 3695 (9.9) | 780 (12) | 4423 (12) | 1283 (12) | 1589 (12) | 11 770 (11) | |
| Hispanic | 2549 (6.8) | 538 (8.1) | 2911 (8.0) | 837 (7.8) | 1186 (8.9) | 8021 (7.7) | |
| Asian | 1133 (3.0) | 236 (3.5) | 1142 (3.1) | 306 (2.8) | 387 (2.9) | 3204 (3.1) | |
| Unknown/Missing | 4332 (12) | 475 (7.1) | 1940 (5.3) | 551 (5.1) | 751 (5.6) | 8049 (7.7) | |
| Census region (%) | |||||||
| Northeast | 7094 (19) | 1196 (18) | 6120 (17) | 1886 (18) | 2424 (18) | <0.001 | 18 720 (18) |
| Midwest | 10 094 (27) | 1562 (23) | 8837 (24) | 2742 (26) | 3627 (27) | 26 862 (26) | |
| South | 15 330 (41) | 2889 (43) | 16 371 (45) | 4698 (44) | 5538 (42) | 44 826 (43) | |
| West | 4660 (13) | 993 (15) | 5147 (14) | 1442 (13) | 1728 (13) | 13 970 (13) | |
| Other/Missing | 39 (0.1) | 32 (0.5) | 128 (0.3) | 4 (0) | 3 (0) | 206 (0.2) | |
| PSA, ng/d | n = 5174 | n = 548 | n = 7345 | n = 2211 | n = 3471 | n = 18 749 | |
| Median (IQR) | 5.1 (3.9–7.1) | 4.9 (3.8–6.8) | 5.0 (4.0–6.8) | 5.3 (4.2–7.4) | 5.4 (4.2–7.8) | <0.001 | 5.1 (4.0–7.1) |
| Anticoagulant use (%) | 928 (2.5) | 184 (2.8) | 1103 (3.0) | 299 (2.8) | 415 (3.1) | <0.001 | 2929 (2.8) |
| History of fluoroquinolone use in prior 12 mo (%) | 6797 (18) | 1267 (19) | 6557 (18) | 2017 (19) | 2504 (19) | 0.054 | 19 142 (18) |
| Charlson Comorbidity Index (%) | |||||||
| 0 | 22 354 (60) | 3887 (58) | 20 269 (55) | 5758 (54) | 7039 (53) | <0.001 | 59 307 (57) |
| 1 | 8199 (22) | 1453 (22) | 8252 (23) | 2503 (23) | 2999 (23) | 23 406 (22) | |
| 2 | 3232 (8.7) | 664 (10) | 3520 (9.6) | 1108 (10) | 1359 (10) | 9883 (9.4) | |
| ≥3 | 3432 (9.2) | 668 (10) | 4562 (13) | 1403 (13) | 1923 (14) | 11 988 (12) | |
| Comorbidities (%) | |||||||
| Myocardial infarction | 763 (2.1) | 144 (2.2) | 812 (2.2) | 227 (2.1) | 273 (2.0) | 0.6 | 2219 (2.1) |
| Congestive heart failure | 1443 (3.9) | 223 (3.3) | 1548 (4.2) | 467 (4.3) | 586 (4.4) | <0.001 | 4267 (4.1) |
| Peripheral vascular disease | 2054 (5.5) | 399 (6.0) | 2660 (7.3) | 783 (7.3) | 1071 (8.0) | <0.001 | 6967 (6.7) |
| Cerebrovascular disease | 2007 (5.4) | 379 (5.7) | 2358 (6.4) | 681 (6.3) | 947 (7.1) | <0.001 | 6372 (6.1) |
| Dementia | 370 (1.0) | 63 (0.9) | 442 (1.2) | 141 (1.3) | 198 (1.5) | <0.001 | 1214 (1.2) |
| Chronic obstructive pulmonary disease | 4251 (11) | 758 (11) | 4293 (12) | 1254 (12) | 1686 (13) | 0.004 | 12 242 (12) |
| Ulcer | 362 (1.0) | 60 (0.9) | 297 (0.8) | 84 (0.8) | 102 (0.8) | 0.07 | 905 (0.9) |
| Mild liver disease | 1182 (3.2) | 223 (3.3) | 1332 (3.6) | 413 (3.8) | 515 (3.9) | <0.001 | 3665 (3.5) |
| Diabetes mellitus | 6475 (17) | 1290 (19) | 7823 (21) | 2442 (23) | 2973 (22) | <0.001 | 21 003 (20) |
| Diabetes with organ damage | 1105 (3.0) | 225 (3.4) | 1600 (4.4) | 492 (4.6) | 734 (5.5) | <0.001 | 4156 (4.0) |
| Hemiplegia | 128 (0.3) | 25 (0.4) | 150 (0.4) | 51 (0.5) | 56 (0.4) | 0.3 | 410 (0.4) |
| Moderate/severe renal disease | 1133 (3.0) | 254 (3.8) | 1950 (5.3) | 674 (6.3) | 908 (6.8) | <0.001 | 4919 (4.7) |
| Moderate/severe liver disease | 27 (0.1) | 7 (0.1) | 31 (0.1) | 9 (0.1) | 12 (0.1) | 0.9 | 86 (0.1) |
| Metastatic solid tumor | 175 (0.5) | 37 (0.6) | 199 (0.5) | 58 (0.5) | 80 (0.6) | 0.4 | 549 (0.5) |
| AIDS | 82 (0.2) | 16 (0.2) | 101 (0.3) | 27 (0.3) | 46 (0.3) | 0.2 | 272 (0.3) |
| Rheumatologic disease | 449 (1.2) | 71 (1.1) | 522 (1.4) | 158 (1.5) | 196 (1.5) | 0.007 | 1396 (1.3) |
| Cancer | 1736 (4.7) | 337 (5.1) | 1796 (4.9) | 571 (5.3) | 701 (5.3) | 0.02 | 5141 (4.9) |
| Net worth range (%) | |||||||
| <$25 000 | 1831 (4.9) | 404 (6.1) | 2101 (5.7) | 561 (5.2) | 695 (5.2) | <0.001 | 5592 (5.3) |
| $25 000–149 000 | 5188 (14) | 1042 (16) | 5608 (15) | 1591 (15) | 2046 (15) | 15 475 (15) | |
| $150 000–249 000 | 4754 (13) | 837 (13) | 5182 (14) | 1499 (14) | 1862 (14) | 14 134 (14) | |
| $250 000–499 000 | 9694 (26) | 1779 (27) | 10 048 (28) | 3027 (28) | 3636 (27) | 28 184 (27) | |
| ≥$500 000 | 10 910 (29) | 2030 (30) | 10 860 (30) | 3208 (30) | 3852 (29) | 30 860 (30) | |
| Unknown/missing | 4840 (13) | 580 (8.7) | 2804 (7.7) | 886 (8.2) | 1229 (9.2) | 10 339 (9.9) | |
| 30-d complications (%) | |||||||
| Any | 5973 (16) | 1151 (17) | 6120 (17) | 1863 (17) | 2451 (18) | <0.001 | 17 558 (17) |
| Bleeding | 2547 (6.8) | 499 (7.5) | 2660 (7.3) | 860 (8.0) | 1083 (8.1) | <0.001 | 7649 (7.3) |
| Nonsepsis infection | 2978 (8.0) | 590 (8.8) | 2990 (8.2) | 875 (8.1) | 1236 (9.3) | <0.001 | 8669 (8.3) |
| Sepsis | 191 (0.5) | 50 (0.7) | 349 (1.0) | 105 (1.0) | 134 (1.0) | <0.001 | 829 (0.8) |
| Urinary retention | 980 (2.6) | 161 (2.4) | 1138 (3.1) | 341 (3.2) | 422 (3.2) | <0.001 | 3042 (2.9) |
| Hospitalization | 1019 (2.7) | 181 (2.7) | 1056 (2.9) | 320 (3.0) | 367 (2.8) | 0.6 | 2943 (2.8) |
AIDS = acquired immunodeficiency syndrome; AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; IQR = interquartile range; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer [screening trial]; PSA = prostate-specific antigen; USPTF = US Preventive Services Task Force.
Overall, 30-d complications occurred in 17% of biopsies. The most prevalent was nonsepsis genitourinary infections (8.3%), followed by bleeding complications (7.3%) and urinary retention (2.9%) (Table 1). The 30-d hospitalization rate was 2.8%.
We first examined unadjusted rates of prostate biopsy among men aged ≥40 yr (Fig. 1 and Table 2). From January 2005 to September 2014, biopsy rates fell 33% from 64.1 to 42.8 per 100 000 person-months. There was substantial seasonal variation in biopsy rates with reductions near the end of each calendar year. With the ITS approach, each intervention is potentially associated with an immediate change in event rates (ie, level shift or change in intercept) as well as a change in the rate of growth (ie, change in slope). At baseline, there was a statistically significant growth in prostate biopsy rates of 2.9 biopsies per 100 000 person-months per year (95% confidence interval [CI], 0.96–4.8; p < 0.001). Following the 2008 USPSTF recommendations, there was an immediate decrease (ie, level shift) in biopsy rates of −10.1 biopsies per 100 000 person-months (95% CI, −17.1 to −3.0; p < 0.001). Likewise, there were immediate decreases in biopsy rates following the 2012 USPSTF recommendations (−13.8 biopsies per 100 000 person-months; 95% CI, −21.0 to −6.7; p < 0.001) and 2013 AUA guidelines (−8.8 biopsies per 100 000 person-months; 95% CI, −16.7 to −0.92; p = 0.03). Analyses adjusted for case mix were similar, although the baseline growth rate was no longer significant, indicating this was related to population shifts over time (Table 2, model 2).
Fig. 1
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 2
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315)
| Prostate biopsy | Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Model 1* | ||||||||||||||
| Baseline rate | 57.8 (53.4–62.3) |
<0.001 | 8.6 (7.5–9.7) |
<0.001 | 4.0 (3.3–4.6) |
<0.001 | 0.17 (0.03–0.32) |
0.02 | 2.8 (2.3–3.3) |
<0.001 | 1.5 (1.3–1.6) |
<0.001 | 1.4 (1.1–1.7) |
<0.001 |
| Baseline trend in rate | 2.9 (0.96–4.8) |
<0.001 | 0.84 (0.48–1.3) |
<0.001 | 0.72 (0.48–1.1) |
<0.001 | 0.12 (0.12–0.12) |
<0.001 | 0.36 (0.12–0.48) |
<0.001 | 0.12 (0.00–0.12) |
<0.001 | 0.00 (−0.12 to 0.12) |
0.5 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.1 (−17.1 to −3.0) |
<0.001 | −0.61 (−3.1 to 1.9) |
0.6 | −1.4 (−2.8 to −0.03) |
0.053 | 0.04 (−0.35 to 0.42) |
0.8 | 0.56 (−0.70 to 1.8) |
0.4 | −0.06 (−0.69 to 0.57) |
0.9 | 0.16 (−0.58 to 0.90) |
0.7 |
| Trend change | 7.9 (−12.7 to 28.4) |
0.5 | −1.0 (−9.1 to 7.2) |
0.8 | −0.12 (−4.1 to 4.0) |
0.98 | −0.60 (−1.9 to 0.72) |
0.4 | −2.0 (−6.5 to 2.4) |
0.4 | −1.1 (−3.4 to 1.3) |
0.4 | −1.0 (−3.6 to 1.6) |
0.5 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.6 (−12 to 4.5) |
0.4 | −0.74 (−3.8 to 2.4) |
0.6 | −0.12 (−1.8 to 1.5) |
0.9 | 0.29 (−0.20 to 0.78) |
0.2 | −0.04 (−1.6 to 1.6) |
0.97 | 0.59 (−0.23 to 1.4) |
0.2 | 0.51 (−0.43 to 1.5) |
0.3 |
| Trend change | −11.2 (−32.0 to 9.6) |
0.3 | 0.0 (−8.2 to 8.0) |
0.99 | −0.96 (−5.0 to 3.1) |
0.7 | 0.36 (−0.96 to 1.8) |
0.5 | 1.7 (−2.8 to 6.1) |
0.5 | 1.0 (−1.4 to 3.4) |
0.4 | 0.84 (−1.7 to 3.5) |
0.5 |
| 2012 USPSTF | ||||||||||||||
| Level change | −13.8 (−21.0 to −6.7) |
<0.001 | −2.5 (−4.5 to −0.45) |
0.02 | −1.7 (−3.1 to −0.31) |
0.02 | −0.12 (−0.42 to 0.19) |
0.5 | −0.61 (−1.5 to 0.28) |
0.2 | −0.75 (−1.1 to −0.44) |
<0.001 | −0.25 (−0.77 to 0.27) |
0.4 |
| Trend change | 8.2 (−5.8 to 22.0) |
0.3 | 3.0 (−0.36 to 6.4) |
0.09 | 3.1 (0.72–5.5) |
0.01 | 0.12 (−0.36 to 0.60) |
0.7 | 0.84 (−0.72 to 2.3) |
0.3 | 0.48 (0.00 to 1.1) |
0.08 | 0.24 (−0.60 to 1.2) |
0.6 |
| 2013 AUA | ||||||||||||||
| Level change | −8.8 (−16.7 to −0.92) |
0.03 | −1.4 (−3.8 to 1.1) |
0.3 | −1.9 (−3.5 to −0.22) |
0.03 | −0.02 (−0.41 to 0.37) |
0.9 | −0.19 (−1.3 to 0.91) |
0.7 | −0.38 (−0.77 to 0.01) |
0.06 | −0.01 (−0.66 to 0.63) |
0.96 |
| Trend change | −10.4 (−30.0 to 9.2) |
0.3 | −4.1 (−7.9 to −0.24) |
0.04 | −2.9 (−6.0 to 0.12) |
0.07 | −0.24 (−0.72 to 0.36) |
0.5 | −1.8 (−3.5 to 0.0) |
0.046 | −0.60 (−1.2 to 0.12) |
0.09 | −0.60 (−1.7 to 0.36) |
0.2 |
| Model 2† | ||||||||||||||
| Baseline rate | 86.8 (82.1–91.4) |
<0.001 | 16.2 (14.9–17.6) |
<0.001 | 6.4 (4.9–7.9) |
<0.001 | 0.39 (−0.13 to 0.91) |
0.2 | 6.1 (4.6–7.7) |
<0.001 | 3.6 (2.8–4.4) |
<0.001 | 3.0 (2.2–3.8) |
<0.001 |
| Baseline trend in rate | 0.24 (−1.8 to 2.3) |
0.8 | −0.24 (−0.72 to 0.24) |
0.4 | 0.48 (−0.12 to 1.1) |
0.09 | 0.12 (−0.12 to 0.24) |
0.4 | −0.24 (−0.72 to 0.24) |
0.4 | −0.24 (−0.48 to 0.00) |
0.054 | −0.12 (−0.36 to 0.12) |
0.3 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.8 (−17.8 to −3.8) |
<0.001 | −5.4 (−9.6 to −1.2) |
0.01 | −2.5 (−5.6 to 0.71) |
0.1 | −0.61 (−1.9 to 0.71) |
0.4 | 0.14 (−3.8 to 4.1) |
0.9 | 1.1 (−0.98 to 3.1) |
0.3 | 0.14 (−1.9 to 2.1) |
0.9 |
| Trend change | 10.6 (−10.1 to 31.2) |
0.3 | 6.1 (−8.4 to 20.8) |
0.4 | −1.7 (−12.2 to 8.8) |
0.8 | 1.7 (−2.9 to 6.2) |
0.5 | 1.7 (−12.0 to 15.5) |
0.8 | −4.7 (−11.8 to 2.5) |
0.2 | −0.84 (−7.8 to 6.1) |
0.8 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.9 (−12 to 4.1) |
0.3 | −0.24 (−5.6 to 5.1) |
0.9 | 1.9 (−2.1 to 5.8) |
0.4 | −0.32 (−2.0 to 1.4) |
0.7 | −0.99 (−6.0 to 4.0) |
0.7 | 1.8 (−0.84 to 4.4) |
0.2 | 0.87 (−1.7 to 3.4) |
0.5 |
| Trend change | −16.2 (−37.1 to 4.7) |
0.1 | −6.8 (−21.4 to 7.8) |
0.4 | 0.36 (−10.2 to 10.9) |
0.95 | −1.9 (−6.5 to 2.8) |
0.4 | −1.6 (−15.2 to 12.2) |
0.8 | 4.7 (−2.4 to 11.9) |
0.2 | 0.48 (−6.4 to 7.4) |
0.9 |
| 2012 USPSTF | ||||||||||||||
| Level change | −11.0 (−18.1 to −4.0) |
0.001 | −2.8 (−6.0 to 0.38) |
0.09 | 0.53 (−1.9 to 3.0) |
0.7 | −0.02 (−0.95 to 0.91) |
0.97 | −1.8 (−4.6 to 0.96) |
0.2 | −1.5 (−3.1 to 0.17) |
0.08 | 0.16 (−1.3 to 1.6) |
0.8 |
| Trend change | 7.0 (−7.2 to 21.1) |
0.3 | 1.7 (−3.6 to 7.0) |
0.5 | 0.12 (−4.1 to 4.2) |
0.96 | 0.12 (−1.4 to 1.8) |
0.9 | −0.48 (−5.3 to 4.3) |
0.9 | 1.6 (−1.3 to 4.3) |
0.3 | 0.12 (−2.4 to 2.6) |
0.9 |
| 2013 AUA | ||||||||||||||
| Level change | −8.2 (−15.9 to −0.38) |
0.04 | −1.0 (−4.8 to 2.8) |
0.6 | −0.44 (−3.2 to 2.3) |
0.8 | −0.15 (−1.3 to 0.99) |
0.8 | 0.28 (−3.2 to 3.7) |
0.9 | −0.50 (−2.7 to 1.7) |
0.7 | 0.47 (−1.3 to 2.3) |
0.6 |
| Trend change | −4.2 (−24.5 to 16.1) |
0.7 | −0.7 (−7.6 to 6.2) |
0.9 | 1.1 (−4.2 to 6.2) |
0.7 | −0.12 (−1.8 to 1.7) |
0.9 | −0.60 (−5.9 to 4.8) |
0.8 | −1.8 (−4.9 to 1.2) |
0.3 | −0.48 (−3.1 to 2.2) |
0.7 |
* Model 1: Base model with constant, time trend, intervention indicator variables. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
† Model 2: Adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, and anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPTF = US Preventive Services Task Force.
Next, we examined 30-d postbiopsy complications among men aged ≥40 yr as a measure of the absolute morbidity of biopsy (Fig. 1, Supplementary Fig. 3, and Table 2). Across the study periods, overall complication rates decreased 10% from 8.7 to 7.8 per 100 000 person-months. In unadjusted ITS analyses, there was a baseline growth in the overall complication rate (0.84 per 100 000 person-months per year; 95% CI, 0.48–1.3; p < 0.001), driven by growth in the rate of nonsepsis infections (0.72 per 100 000 person-months per year; 95% CI, 0.48–1.1; p < 0.001). Only the 2012 USPSTF recommendations were associated with a significant change in complication rates, following which there was an immediate decrease in the overall complication rate of −2.5 per 100 000 person-months (95% CI, −4.5 to −0.45; p = 0.02), driven by reductions in nonsepsis infections (−1.7 per 100 000 person-months; 95% CI, −3.1 to −0.31; p = 0.02) and urinary retention (−0.75 per 100 000 person-months;95% CI, −1.1 to −0.44; p < 0.001). Following the 2013 AUA guidelines there was also a reduction in the overall complication baseline growth rate (−4.1 per 100 000 person-months per year; 95% CI, −7.9 to −0.24; p = 0.04). In analyses adjusted for case mix (Table 2, model 2), effects were overall attenuated, and only the 2008 USPSTF recommendations were associated with a reduction in overall complication rates (−5.4 per 100 000 person-months; 95% CI, −9.6 to −1.2; p = 0.01).
To examine whether there was a change in the relative morbidity of biopsy, we performed a secondary analysis examining complication rates among men undergoing biopsy (Fig. 2). The proportion of men undergoing biopsy who experienced a complication increased from 14% in January 2005 to 18% in September 2014, an increase of 36%. There was again a baseline growth in overall complication rates (0.72% per year; 95% CI, 0.36–1.1; p < 0.001), driven by an increase in nonsepsis infections and sepsis (Supplementary Table 2). None of the interventions affected this growth. Similar results were obtained after adjusting for case mix (Table 3).
Fig. 2
Interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 3
Adjusted* interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Baseline rate | 15.6 (14.3–16.8) |
<0.001 | 7.1 (6.5–7.7) |
<0.001 | 0.34 (0.11–0.57) |
<0.001 | 6.7 (5.9–7.5) |
<0.001 | 2.8 (2.5–3.2) |
<0.001 | 3.2 (2.8–3.6) |
<0.001 |
| Baseline trend in rate | 0.60 (0.12–1.1) |
0.02 | 0.72 (0.60–0.96) |
<0.001 | 0.12 (0.12–0.24) |
<0.001 | 0.12 (−0.12 to 0.36) |
0.3 | −0.12 (−0.24 to 0.00) |
0.3 | −0.12 (−0.24 to 0.00) |
0.0498 |
| 2008 USPSTF | ||||||||||||
| Level change | 1.1 (−2.0 to 4.1) |
0.5 | −1.3 (−3.2 to 0.58) |
0.2 | 0.13 (−0.45 to 0.72) |
0.7 | 1.9 (−0.11 to 3.9) |
0.07 | 0.46 (−0.50 to 1.4) |
0.4 | 0.27 (−0.69 to 1.2) |
0.6 |
| Trend change | −3.5 (−13.4 to 6.6) |
0.5 | −1.8 (−8.4 to 4.7) |
0.6 | −1.0 (−3.0 to 0.96) |
0.3 | −4.3 (−11.4 to 2.6) |
0.2 | −1.6 (−4.9 to 1.9) |
0.4 | −0.72 (−4.0 to 2.6) |
0.7 |
| ERSPC/PLCO | ||||||||||||
| Level change | −0.14 (−3.9 to 3.6) |
0.9 | 1.1 (−1.3 to 3.4) |
0.4 | 0.55 (−0.19 to 1.3) |
0.2 | 0.39 (−2.2 to 3.0) |
0.8 | 0.75 (−0.47 to 2.0) |
0.2 | 0.65 (−0.56 to 1.9) |
0.3 |
| Trend change | 2.9 (−7.1 to 13.0) |
0.6 | 0.72 (−5.8 to 7.2) |
0.8 | 0.84 (−1.2 to 2.9) |
0.4 | 4.2 (−2.8 to 11.3) |
0.2 | 1.7 (−1.8 to 5.0) |
0.4 | 0.72 (−2.6 to 4.0) |
0.7 |
| 2012 USPSTF | ||||||||||||
| Level change | −0.27 (−2.7 to 2.1) |
0.8 | −0.56 (−2.0 to 0.87) |
0.5 | 0.16 (−0.31 to 0.63) |
0.5 | 1.1 (−0.38 to 2.5) |
0.2 | −0.38 (−1.0 to 0.28) |
0.3 | 0.44 (−0.34 to 1.2) |
0.3 |
| Trend change | 1.3 (−2.6 to 5.3) |
0.5 | 2.9 (0.60–5.3) |
0.02 | −0.24 (−1.1 to 0.60) |
0.6 | −0.72 (−3.2 to 1.7) |
0.5 | 0.24 (−0.96 to 1.3) |
0.7 | −0.48 (−1.8 to 0.84) |
0.5 |
| 2013 AUA | ||||||||||||
| Level change | 0.25 (−2.6 to 3.1) |
0.9 | −0.64 (−2.3 to 1.1) |
0.5 | 0.22 (−0.40 to 0.83) |
0.5 | 1.1 (−0.68 to 2.8) |
0.2 | −0.29 (−1.1 to 0.48) |
0.5 | 0.47 (−0.55 to 1.5) |
0.4 |
| Trend change | −1.6 (−6.1 to 2.9) |
0.5 | −2.5 (−5.5 to 0.60) |
0.1 | 0.12 (−0.72 to 1.1) |
0.8 | −0.36 (−3.1 to 2.4) |
0.8 | 0.24 (−1.2 to 1.7) |
0.8 | 0.12 (−1.3 to 1.6) |
0.9 |
* Model adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 person-months. Baseline trend in rate and trend change per 100 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial]; USPTF = US Preventive Services Task Force.
Given the increase in the relative morbidity of biopsy, we performed logistic regression to identify predictors of each complication category (Supplementary Table 3 and Table 4). Due to the large sample size, we focused on those features associated with the greatest magnitude of effect. A prior diagnosis of cancer was associated with an increased risk of all complication categories including any complication (odds ratio [OR]: 1.30; 95% CI, 1.22–1.39; p < 0.001) and hospitalization (OR: 1.59; 95% CI, 1.39–1.82; p < 0.001). Similarly, prior fluoroquinolone use was associated with an increased risk of all complication categories including any complication (OR: 1.27; 95% CI, 1.22–1.32; p < 0.001) and sepsis (OR: 1.55; 95% CI, 1.33–1.81; p < 0.001). Age at biopsy ≥70 yr was associated with a nearly 4-fold increased risk of urinary retention. Although anticoagulant use was associated with increased risks of bleeding (OR: 1.28; 95% CI, 1.14–1.45; p < 0.001), sepsis (OR: 1.48; 95% CI, 1.06–2.07; p = 0.02), and hospitalization (OR: 1.39; 95% CI, 1.16–1.67; p < 0.001), the increase in the overall complication risk was modest (OR 1.14; 95% CI, 1.04–1.25; p = 0.004).
Table 4
Multivariable analysis* of 30-d postbiopsy complications among men aged ≥40 yr undergoing prostate biopsy (January 2005 to September 2014) (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Hospitalization | Urinary retention | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Age at biopsy, yr | ||||||||||||
| 40–49 | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| 50–59 | 1.01 (0.94–1.09) | 0.8 | 0.94 (0.85–1.04) | 0.2 | 1.16 (0.83–1.63) | 0.4 | 1.01 (0.90–1.14) | 0.8 | 1.12 (0.93–1.36) | 0.2 | 1.54 (1.20–1.97) | <0.001 |
| 60–69 | 1.05 (0.97–1.13) | 0.2 | 0.94 (0.85–1.03) | 0.2 | 0.98 (0.70–1.37) | 0.9 | 1.05 (0.94–1.18) | 0.4 | 1.16 (0.95–1.40) | 0.1 | 2.16 (1.68–2.77) | <0.001 |
| ≥70 | 1.25 (1.15–1.36) | <0.001 | 1.03 (0.93–1.15) | 0.6 | 1.05 (0.74–1.49) | 0.8 | 1.29 (1.15–1.46) | <0.001 | 1.31 (1.07–1.60) | 0.008 | 3.98 (3.08–5.15) | <0.001 |
| Race | ||||||||||||
| White | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| Asian | 1.27 (1.15–1.40) | <0.001 | 1.27 (1.11–1.44) | 0.0003 | 1.65 (1.19–2.29) | 0.003 | 1.36 (1.18–1.56) | <0.001 | 0.94 (0.76–1.17) | 0.6 | 1.15 (0.93–1.42) | 0.2 |
| African American | 1.09 (1.03–1.15) | 0.002 | 1.05 (0.97–1.13) | 0.2 | 1.34 (1.09–1.64) | 0.005 | 1.17 (1.09–1.27) | <0.001 | 1.13 (1.01–1.27) | 0.03 | 1.15 (1.03–1.28) | 0.02 |
| Hispanic | 1.32 (1.24–1.40) | <0.001 | 1.36 (1.26–1.47) | <0.001 | 1.52 (1.21–1.90) | <0.001 | 1.33 (1.22–1.45) | <0.001 | 1.12 (0.97–1.28) | 0.1 | 1.37 (1.21–1.56) | <0.001 |
| Unknown/missing | 0.99 (0.93–1.06) | 0.8 | 1.07 (0.99–1.17) | 0.1 | 1.23 (0.94–1.61) | 0.1 | 0.96 (0.88–1.05) | 0.4 | 0.94 (0.81–1.08) | 0.4 | 1.02 (0.88–1.18) | 0.8 |
| Diabetes mellitus | 1.03 (0.99–1.07) | 0.09 | 1.06 (1.00–1.11) | 0.045 | 1.08 (0.92–1.28) | 0.4 | 1.01 (0.96–1.07) | 0.7 | 1.20 (1.10–1.31) | <0.001 | 0.99 (0.91–1.08) | 0.8 |
| COPD | 1.14 (1.09–1.20) | <0.001 | 1.07 (1.00–1.14) | 0.04 | 1.23 (1.01–1.49) | 0.04 | 1.14 (1.06–1.22) | <0.001 | 1.36 (1.23–1.51) | <0.001 | 1.14 (1.03–1.25) | 0.01 |
| PVD | 1.24 (1.16–1.31) | <0.001 | 1.1 (1.02–1.2) | 0.02 | 1.19 (0.93–1.51) | 0.2 | 1.28 (1.18–1.40) | <0.001 | 1.31 (1.15–1.48) | <0.001 | 1.26 (1.12–1.42) | <0.001 |
| Cerebrovascular disease | 1.11 (1.04–1.19) | 0.001 | 1.12 (1.03–1.22) | 0.01 | 1.21 (0.93–1.57) | 0.2 | 1.03 (0.94–1.14) | 0.5 | 1.26 (1.10–1.44) | <0.001 | 1.30 (1.15–1.47) | <0.001 |
| Cancer | 1.30 (1.22–1.39) | <0.001 | 1.16 (1.06–1.27) | 0.001 | 1.37 (1.05–1.79) | 0.02 | 1.40 (1.27–1.53) | <0.001 | 1.59 (1.39–1.82) | <0.001 | 1.32 (1.16–1.52) | <0.001 |
| Anticoagulant use | 1.14 (1.04–1.25) | 0.004 | 1.01 (0.89–1.14) | 0.9 | 1.48 (1.06–2.07) | 0.02 | 1.28 (1.14–1.45) | <0.001 | 1.39 (1.16–1.67) | <0.001 | 1.11 (0.92–1.33) | 0.3 |
| Prior fluoroquinolones | 1.27 (1.22–1.32) | <0.001 | 1.37 (1.31–1.45) | <0.001 | 1.55 (1.33–1.81) | <0.001 | 1.21 (1.14–1.28) | <0.001 | 1.25 (1.14–1.36) | <0.001 | 1.27 (1.16–1.38) | <0.001 |
* Models adjusted for year of biopsy and the features listed in the table.
CI = confidence interval; COPD = chronic obstructive pulmonary disease; PVD = peripheral vascular disease.
We conducted sensitivity analyses where the postintervention run-in period was varied from 2 to 4 mo (Supplementary Table 4). Results overall were similar, although effects were attenuated for the 2013 AUA guidelines on biopsy rates and the 2012 USPSTF recommendations on overall complication rates, likely due to fewer time points for analysis. In a separate sensitivity analysis, we examined bleeding complications excluding microhematuria from 2009 to 2014, following introduction of a separate code for microhematuria. Bleeding rates paralleled those of the primary bleeding end point (Supplementary Fig. 4), and results of both ITS and logistic regression analyses were similar (Supplementary Table 5 and 6). As an exploratory analysis, we examined PCa diagnosis rates within 60 d following biopsy (Supplementary Fig. 5). These appeared to increase steadily across the study period.
To our knowledge, this is the largest study to examine the impact of landmark PSA screening publications on rates of prostate biopsy and the first to examine their impact on postbiopsy complications. Overall, biopsy rates have declined by 33%, which appears attributable to the publication of the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines. Interestingly, the magnitude of reduction in biopsy rates exceeded previously reported decreases in PSA screening rates [6], [7], [8], [9], [10], [11], [12], [22], and [23], which suggests that either the threshold for referral to a urologist or the threshold for performing a biopsy has changed. More importantly, there was a concurrent 22% decrease in 30-d complication rates that appears largely attributable to the 2012 USPSTF recommendations. Still, the reduction in complication rates was less than the reduction in biopsy rates due to an increase in the relative morbidity of biopsy, driven by growth in infectious complications. This may in part be related to a shift to older patients with greater comorbidity undergoing biopsy, although similar results were obtained in analyses adjusted for case mix, which suggests additional factors may exist.
To this end, we evaluated predictors of postbiopsy complications to identify actionable targets for further risk reduction. Some of these features represent modifiable risk factors (eg, anticoagulant use), and others represent opportunities for altering periprocedural management. For instance, prior fluoroquinolone use may identify patients who would benefit from prebiopsy rectal swabs to guide periprocedural antibiotics [24] and [25]. Nonmodifiable risk factors, including specific comorbidities such as cancer and peripheral vascular disease, may encourage reconsideration of the risk–benefit ratio before proceeding with biopsy.
It is also important to consider the concurrent impact of landmark PSA screening publications on rates of PCa diagnosis to contextualize the observed changes in biopsy and its morbidity. An exploratory analysis suggested that cancer diagnosis rates have steadily increased, although further studies are necessary to examine this issue comprehensively.
A number of studies have examined the impact of the 2008 USPSTF recommendations [7], [8], [9], and [22], ERSPC/PLCO trials [6] and [9], and 2012 USPSTF recommendations [6], [10], [11], [12], and [23] on PSA screening rates, generally noting reductions ranging from 0% to 8%. However, much less data are available on the impact of these publications on biopsy rates. One single-center study noted a 31% reduction in the number of biopsies performed following the 2012 USPSTF recommendations [13]; another noted no change in the proportion of referred men who underwent biopsy [26]. A report from a Canadian academic health network noted a reduction in biopsy rates from 58.0 to 35.5 biopsies per month following the 2012 USPSTF recommendations [27]. Our findings therefore confirm a reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines.
Limited data exist regarding secular trends in postbiopsy complications, and prior studies have focused exclusively on the relative morbidity among men undergoing biopsy rather than absolute complication rates. A Canadian population-based study observed that the 30-d hospitalization rate increased from 1.0% in 1996 to 4.1% in 2005, driven by an increase in infectious complications [28]. A Surveillance, Epidemiology and End Results–Medicare study also noted an increase in hospitalizations related to infectious complications over time [14]. Although we observed a similar increase in 30-d overall complications among men undergoing biopsy, likewise driven by infectious complications, there was an absolute 22% reduction in 30-d complications over the study period. Taken together, these findings support a reduction in the morbidity of biopsy following publication of revised PSA screening guidelines but highlight the continued need to reduce the relative morbidity of biopsy.
A number of studies have described complication rates following prostate biopsy [15], [16], [29], and [30]. Hematuria has been reported in 10–84% of biopsies, rectal bleeding in 1–45%, infectious complications in up to 6.3%, urinary retention in up to 1.7%, and hospitalization in up to 6.9% [14] and [29]. The wide range in complication rates is due to differences in definitions, ascertainment of end points, and variation in clinical practice. Given the heterogeneity in postbiopsy complication rates, our findings have two advantages. First, they represent a contemporary, geographically diverse population across the entire United States. Second, use of claims-based ascertainment limited complications to those cases in which patients sought medical care, and although this may underestimate the comprehensive morbidity of biopsy, it more accurately reflects the economic burden of serious complications.
This study has several limitations. Most importantly, a control population of men who were not exposed to the PSA screening publications was not available. This study is retrospective, and additional, unrecognized changes in policy or clinical practice may possibly have confounded the impact of the interventions examined. However, the ITS approach has been described as the strongest quasi-experimental design to evaluate policy changes when a randomized trial is not possible [19]. Because we relied on administrative claims for ascertainment of complications, minor complications that did not generate a claim were not captured, thus underestimating these end points. Although we adjusted for case mix, we were unable to adjust for PSA. However, population shifts in PSA may be confounded by changes in screening practices, and PSA adjustment may therefore under- or overestimate the impact of interventions. We were also unable to adjust for other factors that may be associated with biopsy or complication rates, such as prostate magnetic resonance imaging (MRI), ancillary biomarker testing, biopsy approach (eg, MRI-fusion biopsy, transperineal biopsy), indication for anticoagulation, use of periprocedural bridging, or operator biopsy volume/experience.
We observed an immediate reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines, with a smaller reduction in absolute 30-d complication rates. Still, the relative morbidity of biopsy has increased over time, driven by infectious complications. Accordingly, if health policy aims to reduce the morbidity of PSA screening related to biopsy, further efforts should focus on improving patient selection and periprocedural processes of care.
Author contributions: Boris Gershman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Gershman, van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Acquisition of data: Van Houten, Shah.
Analysis and interpretation of data: Gershman, Van Houten, Shah, Karnes.
Drafting of the manuscript: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Critical revision of the manuscript for important intellectual content: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Statistical analysis: Van Houten.
Obtaining funding: Karnes, Shah.
Administrative, technical, or material support: Karnes, Shah.
Supervision: Karnes, Shah.
Other (specify): None.
Financial disclosures: Boris Gershman certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: This publication was made possible by funding from the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery.
Funding/Support and role of the sponsor: None.
In 2009, two landmark randomized trials—the European Randomized Study of Screening for Prostate Cancer [ERSPC] [1] and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial [PLCO] [2]—reported smaller than anticipated reductions in prostate cancer (PCa) mortality and prompted reexamination of prostate-specific antigen (PSA) screening recommendations. The US Preventive Services Task Force (USPSTF), which in 2008 had recommended against PSA screening in men aged >75 yr [3], updated its position in 2012 to recommend against systematic PSA screening in the general population [4]. The American Urological Association (AUA) likewise updated its PSA screening guidelines in May 2013, recommending shared decision making for men aged 55–69 yr [5]. Both the 2012 USPSTF and 2013 AUA guidelines cited prostate biopsy and its potential complications as a significant harm associated with PSA screening.
Although several studies have noted a reduction in PSA screening rates following these publications [6], [7], [8], [9], [10], [11], and [12], there are little data on whether this has translated into the intended reductions in screening-related morbidity in the form of prostate biopsy and postbiopsy complications [13]. We evaluated whether publication of the 2008 USPSTF recommendations, 2009 ERSPC/PLCO trials, 2012 USPSTF recommendations, or 2013 AUA guidelines was associated with reductions in prostate biopsy or postbiopsy complications. Biopsy-related complications also remain understudied both with regard to incidence as well as identification of risk factors [14], [15], and [16]. Consequently, we also sought to characterize postbiopsy complications and evaluate predictors using a contemporary population-based cohort.
We conducted a retrospective analysis of administrative claims from the Optum Labs Data Warehouse that contains data on >100 million individuals enrolled in private health plans and Medicare Advantage Plans, from geographically diverse regions across the United States, as described in Supplementary Methods [17]. Because this study involved analysis of preexisting deidentified data, it was exempt from institutional review board approval.
We identified all prostate biopsies performed in men aged ≥40 yr between January 1, 2005, and September 30, 2014, using Current Procedural Terminology-4 code 55700 alone (11%) or in combination with 76872 (84%) or 76942 (0%), or International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 60.11 (5%). The first biopsy was selected as the index biopsy when more than one biopsy was performed. Patients with a prior diagnosis of PCa (ICD-9 code 185) were excluded. We restricted the cohort to patients with continuous medical and pharmacy enrollment for at least 12 mo prior to biopsy and 30 d after biopsy to ascertain comorbidities and 30-d complications (Supplementary Fig. 1).
We recorded sociodemographic characteristics, comorbid conditions, anticoagulant use within 30 d of biopsy, and fluoroquinolone use in the 12 mo prior to biopsy as described in Supplement 1. The Charlson Comorbidity Index was calculated as previously described [18]. Prebiopsy PSA values within 60 d of biopsy were available for a subset of patients.
The primary outcomes were crude and adjusted rates of prostate biopsy and 30-d complications. Complications were identified using inpatient and outpatient administrative claims as summarized in Supplementary Table 1.
Baseline demographic features were compared across study periods using the Kruskal-Wallis test for continuous variables and the chi-square test for discrete variables. To evaluate the impact of each publication on rates of prostate biopsy and 30-d complications, we used an interrupted time series (ITS) approach as described in Supplementary Methods [19] and [20]. Rates were estimated using all men aged ≥40 yr with at least 12 mo of continuous medical and pharmacy enrollment as the denominator. As secondary analyses, we estimated rates of 30-d complications among men who underwent biopsy. Two separate ITS analyses were performed: (1) unadjusted analyses, using crude rates of biopsy and complications aggregated for each month, and (2) analyses standardized for case mix, to adjust for population shifts over time as described in Supplementary Methods [20] and [21].
Logistic regression was used to evaluate predictors of 30-d complications. Multivariable models were adjusted for age at biopsy, year of biopsy, race, anticoagulant use, history of fluoroquinolone use, and the presence of the five most prevalent Charlson comorbidities.
Statistical analyses were performed using SAS v.9.3 (SAS Institute Inc., Cary, NC, USA). All tests were two sided with p < 0.05 considered statistically significant.
Between January 2005 and September 2014, a total of 104 584 men underwent prostate biopsy (Table 1). Median age at biopsy was 63.0 yr (interquartile range [IQR]: 57.0–70.0). Overall, 18% of men had used fluoroquinolones in the 12 mo preceding biopsy, and 2.8% were on anticoagulants. Median prebiopsy PSA was 5.1 ng/dl (IQR: 4.0–7.1) for the 18 749 men (18%) for whom this was available. Over the time periods, men undergoing biopsy were older, had increased prevalence of anticoagulant use and comorbid conditions, and had a higher PSA (Table 1 and Supplementary Fig. 2).
Table 1
Patient demographics for men who underwent prostate biopsy (January 2005 to September 2014) stratified by pre/postintervention time periods (n = 104 584)
| Baseline Jan 2005–Jul 2008 (n = 37 217) |
2008 USPSTF Aug 2008–Feb 2009 (n = 6672) |
ERSPC/PLCO Mar 2009–Apr 2012 (n = 36 603) |
2012 USPSTF May 2012–Apr 2013 (n = 10 772) |
2013 AUA May 2013–Sep 2014 (n = 13 320) |
p value | Total (n = 104 584) | |
|---|---|---|---|---|---|---|---|
| Age at biopsy, yr | |||||||
| Median (IQR) | 62.0 (56.0–68.0) | 62.0 (56.0–68.0) | 63.0 (57.0–70.0) | 65.0 (59.0–71.0) | 66.0 (59.0–71.0) | <0.001 | 63.0 (57.0–70.0) |
| Age group, yr (%) | |||||||
| 40–49 | 2269 (6.1) | 416 (6.2) | 1737 (4.7) | 420 (3.9) | 467 (3.5) | <0.001 | 5309 (5.1) |
| 50–59 | 12 366 (33) | 2201 (33) | 10 342 (28) | 2644 (25) | 3098 (23) | 30 651 (29) | |
| 60–69 | 14 671 (39) | 2697 (40) | 15 105 (41) | 4426 (41) | 5546 (42) | 42 445 (41) | |
| ≥70 | 7911 (21) | 1358 (20) | 9419 (26) | 3282 (31) | 4209 (32) | 26 179 (25) | |
| Race (%) | |||||||
| White | 25 508 (69) | 4643 (70) | 26 187 (72) | 7795 (72) | 9407 (71) | <0.001 | 73 540 (70) |
| African American | 3695 (9.9) | 780 (12) | 4423 (12) | 1283 (12) | 1589 (12) | 11 770 (11) | |
| Hispanic | 2549 (6.8) | 538 (8.1) | 2911 (8.0) | 837 (7.8) | 1186 (8.9) | 8021 (7.7) | |
| Asian | 1133 (3.0) | 236 (3.5) | 1142 (3.1) | 306 (2.8) | 387 (2.9) | 3204 (3.1) | |
| Unknown/Missing | 4332 (12) | 475 (7.1) | 1940 (5.3) | 551 (5.1) | 751 (5.6) | 8049 (7.7) | |
| Census region (%) | |||||||
| Northeast | 7094 (19) | 1196 (18) | 6120 (17) | 1886 (18) | 2424 (18) | <0.001 | 18 720 (18) |
| Midwest | 10 094 (27) | 1562 (23) | 8837 (24) | 2742 (26) | 3627 (27) | 26 862 (26) | |
| South | 15 330 (41) | 2889 (43) | 16 371 (45) | 4698 (44) | 5538 (42) | 44 826 (43) | |
| West | 4660 (13) | 993 (15) | 5147 (14) | 1442 (13) | 1728 (13) | 13 970 (13) | |
| Other/Missing | 39 (0.1) | 32 (0.5) | 128 (0.3) | 4 (0) | 3 (0) | 206 (0.2) | |
| PSA, ng/d | n = 5174 | n = 548 | n = 7345 | n = 2211 | n = 3471 | n = 18 749 | |
| Median (IQR) | 5.1 (3.9–7.1) | 4.9 (3.8–6.8) | 5.0 (4.0–6.8) | 5.3 (4.2–7.4) | 5.4 (4.2–7.8) | <0.001 | 5.1 (4.0–7.1) |
| Anticoagulant use (%) | 928 (2.5) | 184 (2.8) | 1103 (3.0) | 299 (2.8) | 415 (3.1) | <0.001 | 2929 (2.8) |
| History of fluoroquinolone use in prior 12 mo (%) | 6797 (18) | 1267 (19) | 6557 (18) | 2017 (19) | 2504 (19) | 0.054 | 19 142 (18) |
| Charlson Comorbidity Index (%) | |||||||
| 0 | 22 354 (60) | 3887 (58) | 20 269 (55) | 5758 (54) | 7039 (53) | <0.001 | 59 307 (57) |
| 1 | 8199 (22) | 1453 (22) | 8252 (23) | 2503 (23) | 2999 (23) | 23 406 (22) | |
| 2 | 3232 (8.7) | 664 (10) | 3520 (9.6) | 1108 (10) | 1359 (10) | 9883 (9.4) | |
| ≥3 | 3432 (9.2) | 668 (10) | 4562 (13) | 1403 (13) | 1923 (14) | 11 988 (12) | |
| Comorbidities (%) | |||||||
| Myocardial infarction | 763 (2.1) | 144 (2.2) | 812 (2.2) | 227 (2.1) | 273 (2.0) | 0.6 | 2219 (2.1) |
| Congestive heart failure | 1443 (3.9) | 223 (3.3) | 1548 (4.2) | 467 (4.3) | 586 (4.4) | <0.001 | 4267 (4.1) |
| Peripheral vascular disease | 2054 (5.5) | 399 (6.0) | 2660 (7.3) | 783 (7.3) | 1071 (8.0) | <0.001 | 6967 (6.7) |
| Cerebrovascular disease | 2007 (5.4) | 379 (5.7) | 2358 (6.4) | 681 (6.3) | 947 (7.1) | <0.001 | 6372 (6.1) |
| Dementia | 370 (1.0) | 63 (0.9) | 442 (1.2) | 141 (1.3) | 198 (1.5) | <0.001 | 1214 (1.2) |
| Chronic obstructive pulmonary disease | 4251 (11) | 758 (11) | 4293 (12) | 1254 (12) | 1686 (13) | 0.004 | 12 242 (12) |
| Ulcer | 362 (1.0) | 60 (0.9) | 297 (0.8) | 84 (0.8) | 102 (0.8) | 0.07 | 905 (0.9) |
| Mild liver disease | 1182 (3.2) | 223 (3.3) | 1332 (3.6) | 413 (3.8) | 515 (3.9) | <0.001 | 3665 (3.5) |
| Diabetes mellitus | 6475 (17) | 1290 (19) | 7823 (21) | 2442 (23) | 2973 (22) | <0.001 | 21 003 (20) |
| Diabetes with organ damage | 1105 (3.0) | 225 (3.4) | 1600 (4.4) | 492 (4.6) | 734 (5.5) | <0.001 | 4156 (4.0) |
| Hemiplegia | 128 (0.3) | 25 (0.4) | 150 (0.4) | 51 (0.5) | 56 (0.4) | 0.3 | 410 (0.4) |
| Moderate/severe renal disease | 1133 (3.0) | 254 (3.8) | 1950 (5.3) | 674 (6.3) | 908 (6.8) | <0.001 | 4919 (4.7) |
| Moderate/severe liver disease | 27 (0.1) | 7 (0.1) | 31 (0.1) | 9 (0.1) | 12 (0.1) | 0.9 | 86 (0.1) |
| Metastatic solid tumor | 175 (0.5) | 37 (0.6) | 199 (0.5) | 58 (0.5) | 80 (0.6) | 0.4 | 549 (0.5) |
| AIDS | 82 (0.2) | 16 (0.2) | 101 (0.3) | 27 (0.3) | 46 (0.3) | 0.2 | 272 (0.3) |
| Rheumatologic disease | 449 (1.2) | 71 (1.1) | 522 (1.4) | 158 (1.5) | 196 (1.5) | 0.007 | 1396 (1.3) |
| Cancer | 1736 (4.7) | 337 (5.1) | 1796 (4.9) | 571 (5.3) | 701 (5.3) | 0.02 | 5141 (4.9) |
| Net worth range (%) | |||||||
| <$25 000 | 1831 (4.9) | 404 (6.1) | 2101 (5.7) | 561 (5.2) | 695 (5.2) | <0.001 | 5592 (5.3) |
| $25 000–149 000 | 5188 (14) | 1042 (16) | 5608 (15) | 1591 (15) | 2046 (15) | 15 475 (15) | |
| $150 000–249 000 | 4754 (13) | 837 (13) | 5182 (14) | 1499 (14) | 1862 (14) | 14 134 (14) | |
| $250 000–499 000 | 9694 (26) | 1779 (27) | 10 048 (28) | 3027 (28) | 3636 (27) | 28 184 (27) | |
| ≥$500 000 | 10 910 (29) | 2030 (30) | 10 860 (30) | 3208 (30) | 3852 (29) | 30 860 (30) | |
| Unknown/missing | 4840 (13) | 580 (8.7) | 2804 (7.7) | 886 (8.2) | 1229 (9.2) | 10 339 (9.9) | |
| 30-d complications (%) | |||||||
| Any | 5973 (16) | 1151 (17) | 6120 (17) | 1863 (17) | 2451 (18) | <0.001 | 17 558 (17) |
| Bleeding | 2547 (6.8) | 499 (7.5) | 2660 (7.3) | 860 (8.0) | 1083 (8.1) | <0.001 | 7649 (7.3) |
| Nonsepsis infection | 2978 (8.0) | 590 (8.8) | 2990 (8.2) | 875 (8.1) | 1236 (9.3) | <0.001 | 8669 (8.3) |
| Sepsis | 191 (0.5) | 50 (0.7) | 349 (1.0) | 105 (1.0) | 134 (1.0) | <0.001 | 829 (0.8) |
| Urinary retention | 980 (2.6) | 161 (2.4) | 1138 (3.1) | 341 (3.2) | 422 (3.2) | <0.001 | 3042 (2.9) |
| Hospitalization | 1019 (2.7) | 181 (2.7) | 1056 (2.9) | 320 (3.0) | 367 (2.8) | 0.6 | 2943 (2.8) |
AIDS = acquired immunodeficiency syndrome; AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; IQR = interquartile range; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer [screening trial]; PSA = prostate-specific antigen; USPTF = US Preventive Services Task Force.
Overall, 30-d complications occurred in 17% of biopsies. The most prevalent was nonsepsis genitourinary infections (8.3%), followed by bleeding complications (7.3%) and urinary retention (2.9%) (Table 1). The 30-d hospitalization rate was 2.8%.
We first examined unadjusted rates of prostate biopsy among men aged ≥40 yr (Fig. 1 and Table 2). From January 2005 to September 2014, biopsy rates fell 33% from 64.1 to 42.8 per 100 000 person-months. There was substantial seasonal variation in biopsy rates with reductions near the end of each calendar year. With the ITS approach, each intervention is potentially associated with an immediate change in event rates (ie, level shift or change in intercept) as well as a change in the rate of growth (ie, change in slope). At baseline, there was a statistically significant growth in prostate biopsy rates of 2.9 biopsies per 100 000 person-months per year (95% confidence interval [CI], 0.96–4.8; p < 0.001). Following the 2008 USPSTF recommendations, there was an immediate decrease (ie, level shift) in biopsy rates of −10.1 biopsies per 100 000 person-months (95% CI, −17.1 to −3.0; p < 0.001). Likewise, there were immediate decreases in biopsy rates following the 2012 USPSTF recommendations (−13.8 biopsies per 100 000 person-months; 95% CI, −21.0 to −6.7; p < 0.001) and 2013 AUA guidelines (−8.8 biopsies per 100 000 person-months; 95% CI, −16.7 to −0.92; p = 0.03). Analyses adjusted for case mix were similar, although the baseline growth rate was no longer significant, indicating this was related to population shifts over time (Table 2, model 2).
Fig. 1
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 2
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315)
| Prostate biopsy | Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Model 1* | ||||||||||||||
| Baseline rate | 57.8 (53.4–62.3) |
<0.001 | 8.6 (7.5–9.7) |
<0.001 | 4.0 (3.3–4.6) |
<0.001 | 0.17 (0.03–0.32) |
0.02 | 2.8 (2.3–3.3) |
<0.001 | 1.5 (1.3–1.6) |
<0.001 | 1.4 (1.1–1.7) |
<0.001 |
| Baseline trend in rate | 2.9 (0.96–4.8) |
<0.001 | 0.84 (0.48–1.3) |
<0.001 | 0.72 (0.48–1.1) |
<0.001 | 0.12 (0.12–0.12) |
<0.001 | 0.36 (0.12–0.48) |
<0.001 | 0.12 (0.00–0.12) |
<0.001 | 0.00 (−0.12 to 0.12) |
0.5 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.1 (−17.1 to −3.0) |
<0.001 | −0.61 (−3.1 to 1.9) |
0.6 | −1.4 (−2.8 to −0.03) |
0.053 | 0.04 (−0.35 to 0.42) |
0.8 | 0.56 (−0.70 to 1.8) |
0.4 | −0.06 (−0.69 to 0.57) |
0.9 | 0.16 (−0.58 to 0.90) |
0.7 |
| Trend change | 7.9 (−12.7 to 28.4) |
0.5 | −1.0 (−9.1 to 7.2) |
0.8 | −0.12 (−4.1 to 4.0) |
0.98 | −0.60 (−1.9 to 0.72) |
0.4 | −2.0 (−6.5 to 2.4) |
0.4 | −1.1 (−3.4 to 1.3) |
0.4 | −1.0 (−3.6 to 1.6) |
0.5 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.6 (−12 to 4.5) |
0.4 | −0.74 (−3.8 to 2.4) |
0.6 | −0.12 (−1.8 to 1.5) |
0.9 | 0.29 (−0.20 to 0.78) |
0.2 | −0.04 (−1.6 to 1.6) |
0.97 | 0.59 (−0.23 to 1.4) |
0.2 | 0.51 (−0.43 to 1.5) |
0.3 |
| Trend change | −11.2 (−32.0 to 9.6) |
0.3 | 0.0 (−8.2 to 8.0) |
0.99 | −0.96 (−5.0 to 3.1) |
0.7 | 0.36 (−0.96 to 1.8) |
0.5 | 1.7 (−2.8 to 6.1) |
0.5 | 1.0 (−1.4 to 3.4) |
0.4 | 0.84 (−1.7 to 3.5) |
0.5 |
| 2012 USPSTF | ||||||||||||||
| Level change | −13.8 (−21.0 to −6.7) |
<0.001 | −2.5 (−4.5 to −0.45) |
0.02 | −1.7 (−3.1 to −0.31) |
0.02 | −0.12 (−0.42 to 0.19) |
0.5 | −0.61 (−1.5 to 0.28) |
0.2 | −0.75 (−1.1 to −0.44) |
<0.001 | −0.25 (−0.77 to 0.27) |
0.4 |
| Trend change | 8.2 (−5.8 to 22.0) |
0.3 | 3.0 (−0.36 to 6.4) |
0.09 | 3.1 (0.72–5.5) |
0.01 | 0.12 (−0.36 to 0.60) |
0.7 | 0.84 (−0.72 to 2.3) |
0.3 | 0.48 (0.00 to 1.1) |
0.08 | 0.24 (−0.60 to 1.2) |
0.6 |
| 2013 AUA | ||||||||||||||
| Level change | −8.8 (−16.7 to −0.92) |
0.03 | −1.4 (−3.8 to 1.1) |
0.3 | −1.9 (−3.5 to −0.22) |
0.03 | −0.02 (−0.41 to 0.37) |
0.9 | −0.19 (−1.3 to 0.91) |
0.7 | −0.38 (−0.77 to 0.01) |
0.06 | −0.01 (−0.66 to 0.63) |
0.96 |
| Trend change | −10.4 (−30.0 to 9.2) |
0.3 | −4.1 (−7.9 to −0.24) |
0.04 | −2.9 (−6.0 to 0.12) |
0.07 | −0.24 (−0.72 to 0.36) |
0.5 | −1.8 (−3.5 to 0.0) |
0.046 | −0.60 (−1.2 to 0.12) |
0.09 | −0.60 (−1.7 to 0.36) |
0.2 |
| Model 2† | ||||||||||||||
| Baseline rate | 86.8 (82.1–91.4) |
<0.001 | 16.2 (14.9–17.6) |
<0.001 | 6.4 (4.9–7.9) |
<0.001 | 0.39 (−0.13 to 0.91) |
0.2 | 6.1 (4.6–7.7) |
<0.001 | 3.6 (2.8–4.4) |
<0.001 | 3.0 (2.2–3.8) |
<0.001 |
| Baseline trend in rate | 0.24 (−1.8 to 2.3) |
0.8 | −0.24 (−0.72 to 0.24) |
0.4 | 0.48 (−0.12 to 1.1) |
0.09 | 0.12 (−0.12 to 0.24) |
0.4 | −0.24 (−0.72 to 0.24) |
0.4 | −0.24 (−0.48 to 0.00) |
0.054 | −0.12 (−0.36 to 0.12) |
0.3 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.8 (−17.8 to −3.8) |
<0.001 | −5.4 (−9.6 to −1.2) |
0.01 | −2.5 (−5.6 to 0.71) |
0.1 | −0.61 (−1.9 to 0.71) |
0.4 | 0.14 (−3.8 to 4.1) |
0.9 | 1.1 (−0.98 to 3.1) |
0.3 | 0.14 (−1.9 to 2.1) |
0.9 |
| Trend change | 10.6 (−10.1 to 31.2) |
0.3 | 6.1 (−8.4 to 20.8) |
0.4 | −1.7 (−12.2 to 8.8) |
0.8 | 1.7 (−2.9 to 6.2) |
0.5 | 1.7 (−12.0 to 15.5) |
0.8 | −4.7 (−11.8 to 2.5) |
0.2 | −0.84 (−7.8 to 6.1) |
0.8 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.9 (−12 to 4.1) |
0.3 | −0.24 (−5.6 to 5.1) |
0.9 | 1.9 (−2.1 to 5.8) |
0.4 | −0.32 (−2.0 to 1.4) |
0.7 | −0.99 (−6.0 to 4.0) |
0.7 | 1.8 (−0.84 to 4.4) |
0.2 | 0.87 (−1.7 to 3.4) |
0.5 |
| Trend change | −16.2 (−37.1 to 4.7) |
0.1 | −6.8 (−21.4 to 7.8) |
0.4 | 0.36 (−10.2 to 10.9) |
0.95 | −1.9 (−6.5 to 2.8) |
0.4 | −1.6 (−15.2 to 12.2) |
0.8 | 4.7 (−2.4 to 11.9) |
0.2 | 0.48 (−6.4 to 7.4) |
0.9 |
| 2012 USPSTF | ||||||||||||||
| Level change | −11.0 (−18.1 to −4.0) |
0.001 | −2.8 (−6.0 to 0.38) |
0.09 | 0.53 (−1.9 to 3.0) |
0.7 | −0.02 (−0.95 to 0.91) |
0.97 | −1.8 (−4.6 to 0.96) |
0.2 | −1.5 (−3.1 to 0.17) |
0.08 | 0.16 (−1.3 to 1.6) |
0.8 |
| Trend change | 7.0 (−7.2 to 21.1) |
0.3 | 1.7 (−3.6 to 7.0) |
0.5 | 0.12 (−4.1 to 4.2) |
0.96 | 0.12 (−1.4 to 1.8) |
0.9 | −0.48 (−5.3 to 4.3) |
0.9 | 1.6 (−1.3 to 4.3) |
0.3 | 0.12 (−2.4 to 2.6) |
0.9 |
| 2013 AUA | ||||||||||||||
| Level change | −8.2 (−15.9 to −0.38) |
0.04 | −1.0 (−4.8 to 2.8) |
0.6 | −0.44 (−3.2 to 2.3) |
0.8 | −0.15 (−1.3 to 0.99) |
0.8 | 0.28 (−3.2 to 3.7) |
0.9 | −0.50 (−2.7 to 1.7) |
0.7 | 0.47 (−1.3 to 2.3) |
0.6 |
| Trend change | −4.2 (−24.5 to 16.1) |
0.7 | −0.7 (−7.6 to 6.2) |
0.9 | 1.1 (−4.2 to 6.2) |
0.7 | −0.12 (−1.8 to 1.7) |
0.9 | −0.60 (−5.9 to 4.8) |
0.8 | −1.8 (−4.9 to 1.2) |
0.3 | −0.48 (−3.1 to 2.2) |
0.7 |
* Model 1: Base model with constant, time trend, intervention indicator variables. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
† Model 2: Adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, and anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPTF = US Preventive Services Task Force.
Next, we examined 30-d postbiopsy complications among men aged ≥40 yr as a measure of the absolute morbidity of biopsy (Fig. 1, Supplementary Fig. 3, and Table 2). Across the study periods, overall complication rates decreased 10% from 8.7 to 7.8 per 100 000 person-months. In unadjusted ITS analyses, there was a baseline growth in the overall complication rate (0.84 per 100 000 person-months per year; 95% CI, 0.48–1.3; p < 0.001), driven by growth in the rate of nonsepsis infections (0.72 per 100 000 person-months per year; 95% CI, 0.48–1.1; p < 0.001). Only the 2012 USPSTF recommendations were associated with a significant change in complication rates, following which there was an immediate decrease in the overall complication rate of −2.5 per 100 000 person-months (95% CI, −4.5 to −0.45; p = 0.02), driven by reductions in nonsepsis infections (−1.7 per 100 000 person-months; 95% CI, −3.1 to −0.31; p = 0.02) and urinary retention (−0.75 per 100 000 person-months;95% CI, −1.1 to −0.44; p < 0.001). Following the 2013 AUA guidelines there was also a reduction in the overall complication baseline growth rate (−4.1 per 100 000 person-months per year; 95% CI, −7.9 to −0.24; p = 0.04). In analyses adjusted for case mix (Table 2, model 2), effects were overall attenuated, and only the 2008 USPSTF recommendations were associated with a reduction in overall complication rates (−5.4 per 100 000 person-months; 95% CI, −9.6 to −1.2; p = 0.01).
To examine whether there was a change in the relative morbidity of biopsy, we performed a secondary analysis examining complication rates among men undergoing biopsy (Fig. 2). The proportion of men undergoing biopsy who experienced a complication increased from 14% in January 2005 to 18% in September 2014, an increase of 36%. There was again a baseline growth in overall complication rates (0.72% per year; 95% CI, 0.36–1.1; p < 0.001), driven by an increase in nonsepsis infections and sepsis (Supplementary Table 2). None of the interventions affected this growth. Similar results were obtained after adjusting for case mix (Table 3).
Fig. 2
Interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 3
Adjusted* interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Baseline rate | 15.6 (14.3–16.8) |
<0.001 | 7.1 (6.5–7.7) |
<0.001 | 0.34 (0.11–0.57) |
<0.001 | 6.7 (5.9–7.5) |
<0.001 | 2.8 (2.5–3.2) |
<0.001 | 3.2 (2.8–3.6) |
<0.001 |
| Baseline trend in rate | 0.60 (0.12–1.1) |
0.02 | 0.72 (0.60–0.96) |
<0.001 | 0.12 (0.12–0.24) |
<0.001 | 0.12 (−0.12 to 0.36) |
0.3 | −0.12 (−0.24 to 0.00) |
0.3 | −0.12 (−0.24 to 0.00) |
0.0498 |
| 2008 USPSTF | ||||||||||||
| Level change | 1.1 (−2.0 to 4.1) |
0.5 | −1.3 (−3.2 to 0.58) |
0.2 | 0.13 (−0.45 to 0.72) |
0.7 | 1.9 (−0.11 to 3.9) |
0.07 | 0.46 (−0.50 to 1.4) |
0.4 | 0.27 (−0.69 to 1.2) |
0.6 |
| Trend change | −3.5 (−13.4 to 6.6) |
0.5 | −1.8 (−8.4 to 4.7) |
0.6 | −1.0 (−3.0 to 0.96) |
0.3 | −4.3 (−11.4 to 2.6) |
0.2 | −1.6 (−4.9 to 1.9) |
0.4 | −0.72 (−4.0 to 2.6) |
0.7 |
| ERSPC/PLCO | ||||||||||||
| Level change | −0.14 (−3.9 to 3.6) |
0.9 | 1.1 (−1.3 to 3.4) |
0.4 | 0.55 (−0.19 to 1.3) |
0.2 | 0.39 (−2.2 to 3.0) |
0.8 | 0.75 (−0.47 to 2.0) |
0.2 | 0.65 (−0.56 to 1.9) |
0.3 |
| Trend change | 2.9 (−7.1 to 13.0) |
0.6 | 0.72 (−5.8 to 7.2) |
0.8 | 0.84 (−1.2 to 2.9) |
0.4 | 4.2 (−2.8 to 11.3) |
0.2 | 1.7 (−1.8 to 5.0) |
0.4 | 0.72 (−2.6 to 4.0) |
0.7 |
| 2012 USPSTF | ||||||||||||
| Level change | −0.27 (−2.7 to 2.1) |
0.8 | −0.56 (−2.0 to 0.87) |
0.5 | 0.16 (−0.31 to 0.63) |
0.5 | 1.1 (−0.38 to 2.5) |
0.2 | −0.38 (−1.0 to 0.28) |
0.3 | 0.44 (−0.34 to 1.2) |
0.3 |
| Trend change | 1.3 (−2.6 to 5.3) |
0.5 | 2.9 (0.60–5.3) |
0.02 | −0.24 (−1.1 to 0.60) |
0.6 | −0.72 (−3.2 to 1.7) |
0.5 | 0.24 (−0.96 to 1.3) |
0.7 | −0.48 (−1.8 to 0.84) |
0.5 |
| 2013 AUA | ||||||||||||
| Level change | 0.25 (−2.6 to 3.1) |
0.9 | −0.64 (−2.3 to 1.1) |
0.5 | 0.22 (−0.40 to 0.83) |
0.5 | 1.1 (−0.68 to 2.8) |
0.2 | −0.29 (−1.1 to 0.48) |
0.5 | 0.47 (−0.55 to 1.5) |
0.4 |
| Trend change | −1.6 (−6.1 to 2.9) |
0.5 | −2.5 (−5.5 to 0.60) |
0.1 | 0.12 (−0.72 to 1.1) |
0.8 | −0.36 (−3.1 to 2.4) |
0.8 | 0.24 (−1.2 to 1.7) |
0.8 | 0.12 (−1.3 to 1.6) |
0.9 |
* Model adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 person-months. Baseline trend in rate and trend change per 100 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial]; USPTF = US Preventive Services Task Force.
Given the increase in the relative morbidity of biopsy, we performed logistic regression to identify predictors of each complication category (Supplementary Table 3 and Table 4). Due to the large sample size, we focused on those features associated with the greatest magnitude of effect. A prior diagnosis of cancer was associated with an increased risk of all complication categories including any complication (odds ratio [OR]: 1.30; 95% CI, 1.22–1.39; p < 0.001) and hospitalization (OR: 1.59; 95% CI, 1.39–1.82; p < 0.001). Similarly, prior fluoroquinolone use was associated with an increased risk of all complication categories including any complication (OR: 1.27; 95% CI, 1.22–1.32; p < 0.001) and sepsis (OR: 1.55; 95% CI, 1.33–1.81; p < 0.001). Age at biopsy ≥70 yr was associated with a nearly 4-fold increased risk of urinary retention. Although anticoagulant use was associated with increased risks of bleeding (OR: 1.28; 95% CI, 1.14–1.45; p < 0.001), sepsis (OR: 1.48; 95% CI, 1.06–2.07; p = 0.02), and hospitalization (OR: 1.39; 95% CI, 1.16–1.67; p < 0.001), the increase in the overall complication risk was modest (OR 1.14; 95% CI, 1.04–1.25; p = 0.004).
Table 4
Multivariable analysis* of 30-d postbiopsy complications among men aged ≥40 yr undergoing prostate biopsy (January 2005 to September 2014) (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Hospitalization | Urinary retention | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Age at biopsy, yr | ||||||||||||
| 40–49 | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| 50–59 | 1.01 (0.94–1.09) | 0.8 | 0.94 (0.85–1.04) | 0.2 | 1.16 (0.83–1.63) | 0.4 | 1.01 (0.90–1.14) | 0.8 | 1.12 (0.93–1.36) | 0.2 | 1.54 (1.20–1.97) | <0.001 |
| 60–69 | 1.05 (0.97–1.13) | 0.2 | 0.94 (0.85–1.03) | 0.2 | 0.98 (0.70–1.37) | 0.9 | 1.05 (0.94–1.18) | 0.4 | 1.16 (0.95–1.40) | 0.1 | 2.16 (1.68–2.77) | <0.001 |
| ≥70 | 1.25 (1.15–1.36) | <0.001 | 1.03 (0.93–1.15) | 0.6 | 1.05 (0.74–1.49) | 0.8 | 1.29 (1.15–1.46) | <0.001 | 1.31 (1.07–1.60) | 0.008 | 3.98 (3.08–5.15) | <0.001 |
| Race | ||||||||||||
| White | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| Asian | 1.27 (1.15–1.40) | <0.001 | 1.27 (1.11–1.44) | 0.0003 | 1.65 (1.19–2.29) | 0.003 | 1.36 (1.18–1.56) | <0.001 | 0.94 (0.76–1.17) | 0.6 | 1.15 (0.93–1.42) | 0.2 |
| African American | 1.09 (1.03–1.15) | 0.002 | 1.05 (0.97–1.13) | 0.2 | 1.34 (1.09–1.64) | 0.005 | 1.17 (1.09–1.27) | <0.001 | 1.13 (1.01–1.27) | 0.03 | 1.15 (1.03–1.28) | 0.02 |
| Hispanic | 1.32 (1.24–1.40) | <0.001 | 1.36 (1.26–1.47) | <0.001 | 1.52 (1.21–1.90) | <0.001 | 1.33 (1.22–1.45) | <0.001 | 1.12 (0.97–1.28) | 0.1 | 1.37 (1.21–1.56) | <0.001 |
| Unknown/missing | 0.99 (0.93–1.06) | 0.8 | 1.07 (0.99–1.17) | 0.1 | 1.23 (0.94–1.61) | 0.1 | 0.96 (0.88–1.05) | 0.4 | 0.94 (0.81–1.08) | 0.4 | 1.02 (0.88–1.18) | 0.8 |
| Diabetes mellitus | 1.03 (0.99–1.07) | 0.09 | 1.06 (1.00–1.11) | 0.045 | 1.08 (0.92–1.28) | 0.4 | 1.01 (0.96–1.07) | 0.7 | 1.20 (1.10–1.31) | <0.001 | 0.99 (0.91–1.08) | 0.8 |
| COPD | 1.14 (1.09–1.20) | <0.001 | 1.07 (1.00–1.14) | 0.04 | 1.23 (1.01–1.49) | 0.04 | 1.14 (1.06–1.22) | <0.001 | 1.36 (1.23–1.51) | <0.001 | 1.14 (1.03–1.25) | 0.01 |
| PVD | 1.24 (1.16–1.31) | <0.001 | 1.1 (1.02–1.2) | 0.02 | 1.19 (0.93–1.51) | 0.2 | 1.28 (1.18–1.40) | <0.001 | 1.31 (1.15–1.48) | <0.001 | 1.26 (1.12–1.42) | <0.001 |
| Cerebrovascular disease | 1.11 (1.04–1.19) | 0.001 | 1.12 (1.03–1.22) | 0.01 | 1.21 (0.93–1.57) | 0.2 | 1.03 (0.94–1.14) | 0.5 | 1.26 (1.10–1.44) | <0.001 | 1.30 (1.15–1.47) | <0.001 |
| Cancer | 1.30 (1.22–1.39) | <0.001 | 1.16 (1.06–1.27) | 0.001 | 1.37 (1.05–1.79) | 0.02 | 1.40 (1.27–1.53) | <0.001 | 1.59 (1.39–1.82) | <0.001 | 1.32 (1.16–1.52) | <0.001 |
| Anticoagulant use | 1.14 (1.04–1.25) | 0.004 | 1.01 (0.89–1.14) | 0.9 | 1.48 (1.06–2.07) | 0.02 | 1.28 (1.14–1.45) | <0.001 | 1.39 (1.16–1.67) | <0.001 | 1.11 (0.92–1.33) | 0.3 |
| Prior fluoroquinolones | 1.27 (1.22–1.32) | <0.001 | 1.37 (1.31–1.45) | <0.001 | 1.55 (1.33–1.81) | <0.001 | 1.21 (1.14–1.28) | <0.001 | 1.25 (1.14–1.36) | <0.001 | 1.27 (1.16–1.38) | <0.001 |
* Models adjusted for year of biopsy and the features listed in the table.
CI = confidence interval; COPD = chronic obstructive pulmonary disease; PVD = peripheral vascular disease.
We conducted sensitivity analyses where the postintervention run-in period was varied from 2 to 4 mo (Supplementary Table 4). Results overall were similar, although effects were attenuated for the 2013 AUA guidelines on biopsy rates and the 2012 USPSTF recommendations on overall complication rates, likely due to fewer time points for analysis. In a separate sensitivity analysis, we examined bleeding complications excluding microhematuria from 2009 to 2014, following introduction of a separate code for microhematuria. Bleeding rates paralleled those of the primary bleeding end point (Supplementary Fig. 4), and results of both ITS and logistic regression analyses were similar (Supplementary Table 5 and 6). As an exploratory analysis, we examined PCa diagnosis rates within 60 d following biopsy (Supplementary Fig. 5). These appeared to increase steadily across the study period.
To our knowledge, this is the largest study to examine the impact of landmark PSA screening publications on rates of prostate biopsy and the first to examine their impact on postbiopsy complications. Overall, biopsy rates have declined by 33%, which appears attributable to the publication of the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines. Interestingly, the magnitude of reduction in biopsy rates exceeded previously reported decreases in PSA screening rates [6], [7], [8], [9], [10], [11], [12], [22], and [23], which suggests that either the threshold for referral to a urologist or the threshold for performing a biopsy has changed. More importantly, there was a concurrent 22% decrease in 30-d complication rates that appears largely attributable to the 2012 USPSTF recommendations. Still, the reduction in complication rates was less than the reduction in biopsy rates due to an increase in the relative morbidity of biopsy, driven by growth in infectious complications. This may in part be related to a shift to older patients with greater comorbidity undergoing biopsy, although similar results were obtained in analyses adjusted for case mix, which suggests additional factors may exist.
To this end, we evaluated predictors of postbiopsy complications to identify actionable targets for further risk reduction. Some of these features represent modifiable risk factors (eg, anticoagulant use), and others represent opportunities for altering periprocedural management. For instance, prior fluoroquinolone use may identify patients who would benefit from prebiopsy rectal swabs to guide periprocedural antibiotics [24] and [25]. Nonmodifiable risk factors, including specific comorbidities such as cancer and peripheral vascular disease, may encourage reconsideration of the risk–benefit ratio before proceeding with biopsy.
It is also important to consider the concurrent impact of landmark PSA screening publications on rates of PCa diagnosis to contextualize the observed changes in biopsy and its morbidity. An exploratory analysis suggested that cancer diagnosis rates have steadily increased, although further studies are necessary to examine this issue comprehensively.
A number of studies have examined the impact of the 2008 USPSTF recommendations [7], [8], [9], and [22], ERSPC/PLCO trials [6] and [9], and 2012 USPSTF recommendations [6], [10], [11], [12], and [23] on PSA screening rates, generally noting reductions ranging from 0% to 8%. However, much less data are available on the impact of these publications on biopsy rates. One single-center study noted a 31% reduction in the number of biopsies performed following the 2012 USPSTF recommendations [13]; another noted no change in the proportion of referred men who underwent biopsy [26]. A report from a Canadian academic health network noted a reduction in biopsy rates from 58.0 to 35.5 biopsies per month following the 2012 USPSTF recommendations [27]. Our findings therefore confirm a reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines.
Limited data exist regarding secular trends in postbiopsy complications, and prior studies have focused exclusively on the relative morbidity among men undergoing biopsy rather than absolute complication rates. A Canadian population-based study observed that the 30-d hospitalization rate increased from 1.0% in 1996 to 4.1% in 2005, driven by an increase in infectious complications [28]. A Surveillance, Epidemiology and End Results–Medicare study also noted an increase in hospitalizations related to infectious complications over time [14]. Although we observed a similar increase in 30-d overall complications among men undergoing biopsy, likewise driven by infectious complications, there was an absolute 22% reduction in 30-d complications over the study period. Taken together, these findings support a reduction in the morbidity of biopsy following publication of revised PSA screening guidelines but highlight the continued need to reduce the relative morbidity of biopsy.
A number of studies have described complication rates following prostate biopsy [15], [16], [29], and [30]. Hematuria has been reported in 10–84% of biopsies, rectal bleeding in 1–45%, infectious complications in up to 6.3%, urinary retention in up to 1.7%, and hospitalization in up to 6.9% [14] and [29]. The wide range in complication rates is due to differences in definitions, ascertainment of end points, and variation in clinical practice. Given the heterogeneity in postbiopsy complication rates, our findings have two advantages. First, they represent a contemporary, geographically diverse population across the entire United States. Second, use of claims-based ascertainment limited complications to those cases in which patients sought medical care, and although this may underestimate the comprehensive morbidity of biopsy, it more accurately reflects the economic burden of serious complications.
This study has several limitations. Most importantly, a control population of men who were not exposed to the PSA screening publications was not available. This study is retrospective, and additional, unrecognized changes in policy or clinical practice may possibly have confounded the impact of the interventions examined. However, the ITS approach has been described as the strongest quasi-experimental design to evaluate policy changes when a randomized trial is not possible [19]. Because we relied on administrative claims for ascertainment of complications, minor complications that did not generate a claim were not captured, thus underestimating these end points. Although we adjusted for case mix, we were unable to adjust for PSA. However, population shifts in PSA may be confounded by changes in screening practices, and PSA adjustment may therefore under- or overestimate the impact of interventions. We were also unable to adjust for other factors that may be associated with biopsy or complication rates, such as prostate magnetic resonance imaging (MRI), ancillary biomarker testing, biopsy approach (eg, MRI-fusion biopsy, transperineal biopsy), indication for anticoagulation, use of periprocedural bridging, or operator biopsy volume/experience.
We observed an immediate reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines, with a smaller reduction in absolute 30-d complication rates. Still, the relative morbidity of biopsy has increased over time, driven by infectious complications. Accordingly, if health policy aims to reduce the morbidity of PSA screening related to biopsy, further efforts should focus on improving patient selection and periprocedural processes of care.
Author contributions: Boris Gershman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Gershman, van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Acquisition of data: Van Houten, Shah.
Analysis and interpretation of data: Gershman, Van Houten, Shah, Karnes.
Drafting of the manuscript: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Critical revision of the manuscript for important intellectual content: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Statistical analysis: Van Houten.
Obtaining funding: Karnes, Shah.
Administrative, technical, or material support: Karnes, Shah.
Supervision: Karnes, Shah.
Other (specify): None.
Financial disclosures: Boris Gershman certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: This publication was made possible by funding from the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery.
Funding/Support and role of the sponsor: None.
In 2009, two landmark randomized trials—the European Randomized Study of Screening for Prostate Cancer [ERSPC] [1] and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial [PLCO] [2]—reported smaller than anticipated reductions in prostate cancer (PCa) mortality and prompted reexamination of prostate-specific antigen (PSA) screening recommendations. The US Preventive Services Task Force (USPSTF), which in 2008 had recommended against PSA screening in men aged >75 yr [3], updated its position in 2012 to recommend against systematic PSA screening in the general population [4]. The American Urological Association (AUA) likewise updated its PSA screening guidelines in May 2013, recommending shared decision making for men aged 55–69 yr [5]. Both the 2012 USPSTF and 2013 AUA guidelines cited prostate biopsy and its potential complications as a significant harm associated with PSA screening.
Although several studies have noted a reduction in PSA screening rates following these publications [6], [7], [8], [9], [10], [11], and [12], there are little data on whether this has translated into the intended reductions in screening-related morbidity in the form of prostate biopsy and postbiopsy complications [13]. We evaluated whether publication of the 2008 USPSTF recommendations, 2009 ERSPC/PLCO trials, 2012 USPSTF recommendations, or 2013 AUA guidelines was associated with reductions in prostate biopsy or postbiopsy complications. Biopsy-related complications also remain understudied both with regard to incidence as well as identification of risk factors [14], [15], and [16]. Consequently, we also sought to characterize postbiopsy complications and evaluate predictors using a contemporary population-based cohort.
We conducted a retrospective analysis of administrative claims from the Optum Labs Data Warehouse that contains data on >100 million individuals enrolled in private health plans and Medicare Advantage Plans, from geographically diverse regions across the United States, as described in Supplementary Methods [17]. Because this study involved analysis of preexisting deidentified data, it was exempt from institutional review board approval.
We identified all prostate biopsies performed in men aged ≥40 yr between January 1, 2005, and September 30, 2014, using Current Procedural Terminology-4 code 55700 alone (11%) or in combination with 76872 (84%) or 76942 (0%), or International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 60.11 (5%). The first biopsy was selected as the index biopsy when more than one biopsy was performed. Patients with a prior diagnosis of PCa (ICD-9 code 185) were excluded. We restricted the cohort to patients with continuous medical and pharmacy enrollment for at least 12 mo prior to biopsy and 30 d after biopsy to ascertain comorbidities and 30-d complications (Supplementary Fig. 1).
We recorded sociodemographic characteristics, comorbid conditions, anticoagulant use within 30 d of biopsy, and fluoroquinolone use in the 12 mo prior to biopsy as described in Supplement 1. The Charlson Comorbidity Index was calculated as previously described [18]. Prebiopsy PSA values within 60 d of biopsy were available for a subset of patients.
The primary outcomes were crude and adjusted rates of prostate biopsy and 30-d complications. Complications were identified using inpatient and outpatient administrative claims as summarized in Supplementary Table 1.
Baseline demographic features were compared across study periods using the Kruskal-Wallis test for continuous variables and the chi-square test for discrete variables. To evaluate the impact of each publication on rates of prostate biopsy and 30-d complications, we used an interrupted time series (ITS) approach as described in Supplementary Methods [19] and [20]. Rates were estimated using all men aged ≥40 yr with at least 12 mo of continuous medical and pharmacy enrollment as the denominator. As secondary analyses, we estimated rates of 30-d complications among men who underwent biopsy. Two separate ITS analyses were performed: (1) unadjusted analyses, using crude rates of biopsy and complications aggregated for each month, and (2) analyses standardized for case mix, to adjust for population shifts over time as described in Supplementary Methods [20] and [21].
Logistic regression was used to evaluate predictors of 30-d complications. Multivariable models were adjusted for age at biopsy, year of biopsy, race, anticoagulant use, history of fluoroquinolone use, and the presence of the five most prevalent Charlson comorbidities.
Statistical analyses were performed using SAS v.9.3 (SAS Institute Inc., Cary, NC, USA). All tests were two sided with p < 0.05 considered statistically significant.
Between January 2005 and September 2014, a total of 104 584 men underwent prostate biopsy (Table 1). Median age at biopsy was 63.0 yr (interquartile range [IQR]: 57.0–70.0). Overall, 18% of men had used fluoroquinolones in the 12 mo preceding biopsy, and 2.8% were on anticoagulants. Median prebiopsy PSA was 5.1 ng/dl (IQR: 4.0–7.1) for the 18 749 men (18%) for whom this was available. Over the time periods, men undergoing biopsy were older, had increased prevalence of anticoagulant use and comorbid conditions, and had a higher PSA (Table 1 and Supplementary Fig. 2).
Table 1
Patient demographics for men who underwent prostate biopsy (January 2005 to September 2014) stratified by pre/postintervention time periods (n = 104 584)
| Baseline Jan 2005–Jul 2008 (n = 37 217) |
2008 USPSTF Aug 2008–Feb 2009 (n = 6672) |
ERSPC/PLCO Mar 2009–Apr 2012 (n = 36 603) |
2012 USPSTF May 2012–Apr 2013 (n = 10 772) |
2013 AUA May 2013–Sep 2014 (n = 13 320) |
p value | Total (n = 104 584) | |
|---|---|---|---|---|---|---|---|
| Age at biopsy, yr | |||||||
| Median (IQR) | 62.0 (56.0–68.0) | 62.0 (56.0–68.0) | 63.0 (57.0–70.0) | 65.0 (59.0–71.0) | 66.0 (59.0–71.0) | <0.001 | 63.0 (57.0–70.0) |
| Age group, yr (%) | |||||||
| 40–49 | 2269 (6.1) | 416 (6.2) | 1737 (4.7) | 420 (3.9) | 467 (3.5) | <0.001 | 5309 (5.1) |
| 50–59 | 12 366 (33) | 2201 (33) | 10 342 (28) | 2644 (25) | 3098 (23) | 30 651 (29) | |
| 60–69 | 14 671 (39) | 2697 (40) | 15 105 (41) | 4426 (41) | 5546 (42) | 42 445 (41) | |
| ≥70 | 7911 (21) | 1358 (20) | 9419 (26) | 3282 (31) | 4209 (32) | 26 179 (25) | |
| Race (%) | |||||||
| White | 25 508 (69) | 4643 (70) | 26 187 (72) | 7795 (72) | 9407 (71) | <0.001 | 73 540 (70) |
| African American | 3695 (9.9) | 780 (12) | 4423 (12) | 1283 (12) | 1589 (12) | 11 770 (11) | |
| Hispanic | 2549 (6.8) | 538 (8.1) | 2911 (8.0) | 837 (7.8) | 1186 (8.9) | 8021 (7.7) | |
| Asian | 1133 (3.0) | 236 (3.5) | 1142 (3.1) | 306 (2.8) | 387 (2.9) | 3204 (3.1) | |
| Unknown/Missing | 4332 (12) | 475 (7.1) | 1940 (5.3) | 551 (5.1) | 751 (5.6) | 8049 (7.7) | |
| Census region (%) | |||||||
| Northeast | 7094 (19) | 1196 (18) | 6120 (17) | 1886 (18) | 2424 (18) | <0.001 | 18 720 (18) |
| Midwest | 10 094 (27) | 1562 (23) | 8837 (24) | 2742 (26) | 3627 (27) | 26 862 (26) | |
| South | 15 330 (41) | 2889 (43) | 16 371 (45) | 4698 (44) | 5538 (42) | 44 826 (43) | |
| West | 4660 (13) | 993 (15) | 5147 (14) | 1442 (13) | 1728 (13) | 13 970 (13) | |
| Other/Missing | 39 (0.1) | 32 (0.5) | 128 (0.3) | 4 (0) | 3 (0) | 206 (0.2) | |
| PSA, ng/d | n = 5174 | n = 548 | n = 7345 | n = 2211 | n = 3471 | n = 18 749 | |
| Median (IQR) | 5.1 (3.9–7.1) | 4.9 (3.8–6.8) | 5.0 (4.0–6.8) | 5.3 (4.2–7.4) | 5.4 (4.2–7.8) | <0.001 | 5.1 (4.0–7.1) |
| Anticoagulant use (%) | 928 (2.5) | 184 (2.8) | 1103 (3.0) | 299 (2.8) | 415 (3.1) | <0.001 | 2929 (2.8) |
| History of fluoroquinolone use in prior 12 mo (%) | 6797 (18) | 1267 (19) | 6557 (18) | 2017 (19) | 2504 (19) | 0.054 | 19 142 (18) |
| Charlson Comorbidity Index (%) | |||||||
| 0 | 22 354 (60) | 3887 (58) | 20 269 (55) | 5758 (54) | 7039 (53) | <0.001 | 59 307 (57) |
| 1 | 8199 (22) | 1453 (22) | 8252 (23) | 2503 (23) | 2999 (23) | 23 406 (22) | |
| 2 | 3232 (8.7) | 664 (10) | 3520 (9.6) | 1108 (10) | 1359 (10) | 9883 (9.4) | |
| ≥3 | 3432 (9.2) | 668 (10) | 4562 (13) | 1403 (13) | 1923 (14) | 11 988 (12) | |
| Comorbidities (%) | |||||||
| Myocardial infarction | 763 (2.1) | 144 (2.2) | 812 (2.2) | 227 (2.1) | 273 (2.0) | 0.6 | 2219 (2.1) |
| Congestive heart failure | 1443 (3.9) | 223 (3.3) | 1548 (4.2) | 467 (4.3) | 586 (4.4) | <0.001 | 4267 (4.1) |
| Peripheral vascular disease | 2054 (5.5) | 399 (6.0) | 2660 (7.3) | 783 (7.3) | 1071 (8.0) | <0.001 | 6967 (6.7) |
| Cerebrovascular disease | 2007 (5.4) | 379 (5.7) | 2358 (6.4) | 681 (6.3) | 947 (7.1) | <0.001 | 6372 (6.1) |
| Dementia | 370 (1.0) | 63 (0.9) | 442 (1.2) | 141 (1.3) | 198 (1.5) | <0.001 | 1214 (1.2) |
| Chronic obstructive pulmonary disease | 4251 (11) | 758 (11) | 4293 (12) | 1254 (12) | 1686 (13) | 0.004 | 12 242 (12) |
| Ulcer | 362 (1.0) | 60 (0.9) | 297 (0.8) | 84 (0.8) | 102 (0.8) | 0.07 | 905 (0.9) |
| Mild liver disease | 1182 (3.2) | 223 (3.3) | 1332 (3.6) | 413 (3.8) | 515 (3.9) | <0.001 | 3665 (3.5) |
| Diabetes mellitus | 6475 (17) | 1290 (19) | 7823 (21) | 2442 (23) | 2973 (22) | <0.001 | 21 003 (20) |
| Diabetes with organ damage | 1105 (3.0) | 225 (3.4) | 1600 (4.4) | 492 (4.6) | 734 (5.5) | <0.001 | 4156 (4.0) |
| Hemiplegia | 128 (0.3) | 25 (0.4) | 150 (0.4) | 51 (0.5) | 56 (0.4) | 0.3 | 410 (0.4) |
| Moderate/severe renal disease | 1133 (3.0) | 254 (3.8) | 1950 (5.3) | 674 (6.3) | 908 (6.8) | <0.001 | 4919 (4.7) |
| Moderate/severe liver disease | 27 (0.1) | 7 (0.1) | 31 (0.1) | 9 (0.1) | 12 (0.1) | 0.9 | 86 (0.1) |
| Metastatic solid tumor | 175 (0.5) | 37 (0.6) | 199 (0.5) | 58 (0.5) | 80 (0.6) | 0.4 | 549 (0.5) |
| AIDS | 82 (0.2) | 16 (0.2) | 101 (0.3) | 27 (0.3) | 46 (0.3) | 0.2 | 272 (0.3) |
| Rheumatologic disease | 449 (1.2) | 71 (1.1) | 522 (1.4) | 158 (1.5) | 196 (1.5) | 0.007 | 1396 (1.3) |
| Cancer | 1736 (4.7) | 337 (5.1) | 1796 (4.9) | 571 (5.3) | 701 (5.3) | 0.02 | 5141 (4.9) |
| Net worth range (%) | |||||||
| <$25 000 | 1831 (4.9) | 404 (6.1) | 2101 (5.7) | 561 (5.2) | 695 (5.2) | <0.001 | 5592 (5.3) |
| $25 000–149 000 | 5188 (14) | 1042 (16) | 5608 (15) | 1591 (15) | 2046 (15) | 15 475 (15) | |
| $150 000–249 000 | 4754 (13) | 837 (13) | 5182 (14) | 1499 (14) | 1862 (14) | 14 134 (14) | |
| $250 000–499 000 | 9694 (26) | 1779 (27) | 10 048 (28) | 3027 (28) | 3636 (27) | 28 184 (27) | |
| ≥$500 000 | 10 910 (29) | 2030 (30) | 10 860 (30) | 3208 (30) | 3852 (29) | 30 860 (30) | |
| Unknown/missing | 4840 (13) | 580 (8.7) | 2804 (7.7) | 886 (8.2) | 1229 (9.2) | 10 339 (9.9) | |
| 30-d complications (%) | |||||||
| Any | 5973 (16) | 1151 (17) | 6120 (17) | 1863 (17) | 2451 (18) | <0.001 | 17 558 (17) |
| Bleeding | 2547 (6.8) | 499 (7.5) | 2660 (7.3) | 860 (8.0) | 1083 (8.1) | <0.001 | 7649 (7.3) |
| Nonsepsis infection | 2978 (8.0) | 590 (8.8) | 2990 (8.2) | 875 (8.1) | 1236 (9.3) | <0.001 | 8669 (8.3) |
| Sepsis | 191 (0.5) | 50 (0.7) | 349 (1.0) | 105 (1.0) | 134 (1.0) | <0.001 | 829 (0.8) |
| Urinary retention | 980 (2.6) | 161 (2.4) | 1138 (3.1) | 341 (3.2) | 422 (3.2) | <0.001 | 3042 (2.9) |
| Hospitalization | 1019 (2.7) | 181 (2.7) | 1056 (2.9) | 320 (3.0) | 367 (2.8) | 0.6 | 2943 (2.8) |
AIDS = acquired immunodeficiency syndrome; AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; IQR = interquartile range; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer [screening trial]; PSA = prostate-specific antigen; USPTF = US Preventive Services Task Force.
Overall, 30-d complications occurred in 17% of biopsies. The most prevalent was nonsepsis genitourinary infections (8.3%), followed by bleeding complications (7.3%) and urinary retention (2.9%) (Table 1). The 30-d hospitalization rate was 2.8%.
We first examined unadjusted rates of prostate biopsy among men aged ≥40 yr (Fig. 1 and Table 2). From January 2005 to September 2014, biopsy rates fell 33% from 64.1 to 42.8 per 100 000 person-months. There was substantial seasonal variation in biopsy rates with reductions near the end of each calendar year. With the ITS approach, each intervention is potentially associated with an immediate change in event rates (ie, level shift or change in intercept) as well as a change in the rate of growth (ie, change in slope). At baseline, there was a statistically significant growth in prostate biopsy rates of 2.9 biopsies per 100 000 person-months per year (95% confidence interval [CI], 0.96–4.8; p < 0.001). Following the 2008 USPSTF recommendations, there was an immediate decrease (ie, level shift) in biopsy rates of −10.1 biopsies per 100 000 person-months (95% CI, −17.1 to −3.0; p < 0.001). Likewise, there were immediate decreases in biopsy rates following the 2012 USPSTF recommendations (−13.8 biopsies per 100 000 person-months; 95% CI, −21.0 to −6.7; p < 0.001) and 2013 AUA guidelines (−8.8 biopsies per 100 000 person-months; 95% CI, −16.7 to −0.92; p = 0.03). Analyses adjusted for case mix were similar, although the baseline growth rate was no longer significant, indicating this was related to population shifts over time (Table 2, model 2).
Fig. 1
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 2
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315)
| Prostate biopsy | Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Model 1* | ||||||||||||||
| Baseline rate | 57.8 (53.4–62.3) |
<0.001 | 8.6 (7.5–9.7) |
<0.001 | 4.0 (3.3–4.6) |
<0.001 | 0.17 (0.03–0.32) |
0.02 | 2.8 (2.3–3.3) |
<0.001 | 1.5 (1.3–1.6) |
<0.001 | 1.4 (1.1–1.7) |
<0.001 |
| Baseline trend in rate | 2.9 (0.96–4.8) |
<0.001 | 0.84 (0.48–1.3) |
<0.001 | 0.72 (0.48–1.1) |
<0.001 | 0.12 (0.12–0.12) |
<0.001 | 0.36 (0.12–0.48) |
<0.001 | 0.12 (0.00–0.12) |
<0.001 | 0.00 (−0.12 to 0.12) |
0.5 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.1 (−17.1 to −3.0) |
<0.001 | −0.61 (−3.1 to 1.9) |
0.6 | −1.4 (−2.8 to −0.03) |
0.053 | 0.04 (−0.35 to 0.42) |
0.8 | 0.56 (−0.70 to 1.8) |
0.4 | −0.06 (−0.69 to 0.57) |
0.9 | 0.16 (−0.58 to 0.90) |
0.7 |
| Trend change | 7.9 (−12.7 to 28.4) |
0.5 | −1.0 (−9.1 to 7.2) |
0.8 | −0.12 (−4.1 to 4.0) |
0.98 | −0.60 (−1.9 to 0.72) |
0.4 | −2.0 (−6.5 to 2.4) |
0.4 | −1.1 (−3.4 to 1.3) |
0.4 | −1.0 (−3.6 to 1.6) |
0.5 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.6 (−12 to 4.5) |
0.4 | −0.74 (−3.8 to 2.4) |
0.6 | −0.12 (−1.8 to 1.5) |
0.9 | 0.29 (−0.20 to 0.78) |
0.2 | −0.04 (−1.6 to 1.6) |
0.97 | 0.59 (−0.23 to 1.4) |
0.2 | 0.51 (−0.43 to 1.5) |
0.3 |
| Trend change | −11.2 (−32.0 to 9.6) |
0.3 | 0.0 (−8.2 to 8.0) |
0.99 | −0.96 (−5.0 to 3.1) |
0.7 | 0.36 (−0.96 to 1.8) |
0.5 | 1.7 (−2.8 to 6.1) |
0.5 | 1.0 (−1.4 to 3.4) |
0.4 | 0.84 (−1.7 to 3.5) |
0.5 |
| 2012 USPSTF | ||||||||||||||
| Level change | −13.8 (−21.0 to −6.7) |
<0.001 | −2.5 (−4.5 to −0.45) |
0.02 | −1.7 (−3.1 to −0.31) |
0.02 | −0.12 (−0.42 to 0.19) |
0.5 | −0.61 (−1.5 to 0.28) |
0.2 | −0.75 (−1.1 to −0.44) |
<0.001 | −0.25 (−0.77 to 0.27) |
0.4 |
| Trend change | 8.2 (−5.8 to 22.0) |
0.3 | 3.0 (−0.36 to 6.4) |
0.09 | 3.1 (0.72–5.5) |
0.01 | 0.12 (−0.36 to 0.60) |
0.7 | 0.84 (−0.72 to 2.3) |
0.3 | 0.48 (0.00 to 1.1) |
0.08 | 0.24 (−0.60 to 1.2) |
0.6 |
| 2013 AUA | ||||||||||||||
| Level change | −8.8 (−16.7 to −0.92) |
0.03 | −1.4 (−3.8 to 1.1) |
0.3 | −1.9 (−3.5 to −0.22) |
0.03 | −0.02 (−0.41 to 0.37) |
0.9 | −0.19 (−1.3 to 0.91) |
0.7 | −0.38 (−0.77 to 0.01) |
0.06 | −0.01 (−0.66 to 0.63) |
0.96 |
| Trend change | −10.4 (−30.0 to 9.2) |
0.3 | −4.1 (−7.9 to −0.24) |
0.04 | −2.9 (−6.0 to 0.12) |
0.07 | −0.24 (−0.72 to 0.36) |
0.5 | −1.8 (−3.5 to 0.0) |
0.046 | −0.60 (−1.2 to 0.12) |
0.09 | −0.60 (−1.7 to 0.36) |
0.2 |
| Model 2† | ||||||||||||||
| Baseline rate | 86.8 (82.1–91.4) |
<0.001 | 16.2 (14.9–17.6) |
<0.001 | 6.4 (4.9–7.9) |
<0.001 | 0.39 (−0.13 to 0.91) |
0.2 | 6.1 (4.6–7.7) |
<0.001 | 3.6 (2.8–4.4) |
<0.001 | 3.0 (2.2–3.8) |
<0.001 |
| Baseline trend in rate | 0.24 (−1.8 to 2.3) |
0.8 | −0.24 (−0.72 to 0.24) |
0.4 | 0.48 (−0.12 to 1.1) |
0.09 | 0.12 (−0.12 to 0.24) |
0.4 | −0.24 (−0.72 to 0.24) |
0.4 | −0.24 (−0.48 to 0.00) |
0.054 | −0.12 (−0.36 to 0.12) |
0.3 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.8 (−17.8 to −3.8) |
<0.001 | −5.4 (−9.6 to −1.2) |
0.01 | −2.5 (−5.6 to 0.71) |
0.1 | −0.61 (−1.9 to 0.71) |
0.4 | 0.14 (−3.8 to 4.1) |
0.9 | 1.1 (−0.98 to 3.1) |
0.3 | 0.14 (−1.9 to 2.1) |
0.9 |
| Trend change | 10.6 (−10.1 to 31.2) |
0.3 | 6.1 (−8.4 to 20.8) |
0.4 | −1.7 (−12.2 to 8.8) |
0.8 | 1.7 (−2.9 to 6.2) |
0.5 | 1.7 (−12.0 to 15.5) |
0.8 | −4.7 (−11.8 to 2.5) |
0.2 | −0.84 (−7.8 to 6.1) |
0.8 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.9 (−12 to 4.1) |
0.3 | −0.24 (−5.6 to 5.1) |
0.9 | 1.9 (−2.1 to 5.8) |
0.4 | −0.32 (−2.0 to 1.4) |
0.7 | −0.99 (−6.0 to 4.0) |
0.7 | 1.8 (−0.84 to 4.4) |
0.2 | 0.87 (−1.7 to 3.4) |
0.5 |
| Trend change | −16.2 (−37.1 to 4.7) |
0.1 | −6.8 (−21.4 to 7.8) |
0.4 | 0.36 (−10.2 to 10.9) |
0.95 | −1.9 (−6.5 to 2.8) |
0.4 | −1.6 (−15.2 to 12.2) |
0.8 | 4.7 (−2.4 to 11.9) |
0.2 | 0.48 (−6.4 to 7.4) |
0.9 |
| 2012 USPSTF | ||||||||||||||
| Level change | −11.0 (−18.1 to −4.0) |
0.001 | −2.8 (−6.0 to 0.38) |
0.09 | 0.53 (−1.9 to 3.0) |
0.7 | −0.02 (−0.95 to 0.91) |
0.97 | −1.8 (−4.6 to 0.96) |
0.2 | −1.5 (−3.1 to 0.17) |
0.08 | 0.16 (−1.3 to 1.6) |
0.8 |
| Trend change | 7.0 (−7.2 to 21.1) |
0.3 | 1.7 (−3.6 to 7.0) |
0.5 | 0.12 (−4.1 to 4.2) |
0.96 | 0.12 (−1.4 to 1.8) |
0.9 | −0.48 (−5.3 to 4.3) |
0.9 | 1.6 (−1.3 to 4.3) |
0.3 | 0.12 (−2.4 to 2.6) |
0.9 |
| 2013 AUA | ||||||||||||||
| Level change | −8.2 (−15.9 to −0.38) |
0.04 | −1.0 (−4.8 to 2.8) |
0.6 | −0.44 (−3.2 to 2.3) |
0.8 | −0.15 (−1.3 to 0.99) |
0.8 | 0.28 (−3.2 to 3.7) |
0.9 | −0.50 (−2.7 to 1.7) |
0.7 | 0.47 (−1.3 to 2.3) |
0.6 |
| Trend change | −4.2 (−24.5 to 16.1) |
0.7 | −0.7 (−7.6 to 6.2) |
0.9 | 1.1 (−4.2 to 6.2) |
0.7 | −0.12 (−1.8 to 1.7) |
0.9 | −0.60 (−5.9 to 4.8) |
0.8 | −1.8 (−4.9 to 1.2) |
0.3 | −0.48 (−3.1 to 2.2) |
0.7 |
* Model 1: Base model with constant, time trend, intervention indicator variables. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
† Model 2: Adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, and anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPTF = US Preventive Services Task Force.
Next, we examined 30-d postbiopsy complications among men aged ≥40 yr as a measure of the absolute morbidity of biopsy (Fig. 1, Supplementary Fig. 3, and Table 2). Across the study periods, overall complication rates decreased 10% from 8.7 to 7.8 per 100 000 person-months. In unadjusted ITS analyses, there was a baseline growth in the overall complication rate (0.84 per 100 000 person-months per year; 95% CI, 0.48–1.3; p < 0.001), driven by growth in the rate of nonsepsis infections (0.72 per 100 000 person-months per year; 95% CI, 0.48–1.1; p < 0.001). Only the 2012 USPSTF recommendations were associated with a significant change in complication rates, following which there was an immediate decrease in the overall complication rate of −2.5 per 100 000 person-months (95% CI, −4.5 to −0.45; p = 0.02), driven by reductions in nonsepsis infections (−1.7 per 100 000 person-months; 95% CI, −3.1 to −0.31; p = 0.02) and urinary retention (−0.75 per 100 000 person-months;95% CI, −1.1 to −0.44; p < 0.001). Following the 2013 AUA guidelines there was also a reduction in the overall complication baseline growth rate (−4.1 per 100 000 person-months per year; 95% CI, −7.9 to −0.24; p = 0.04). In analyses adjusted for case mix (Table 2, model 2), effects were overall attenuated, and only the 2008 USPSTF recommendations were associated with a reduction in overall complication rates (−5.4 per 100 000 person-months; 95% CI, −9.6 to −1.2; p = 0.01).
To examine whether there was a change in the relative morbidity of biopsy, we performed a secondary analysis examining complication rates among men undergoing biopsy (Fig. 2). The proportion of men undergoing biopsy who experienced a complication increased from 14% in January 2005 to 18% in September 2014, an increase of 36%. There was again a baseline growth in overall complication rates (0.72% per year; 95% CI, 0.36–1.1; p < 0.001), driven by an increase in nonsepsis infections and sepsis (Supplementary Table 2). None of the interventions affected this growth. Similar results were obtained after adjusting for case mix (Table 3).
Fig. 2
Interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 3
Adjusted* interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Baseline rate | 15.6 (14.3–16.8) |
<0.001 | 7.1 (6.5–7.7) |
<0.001 | 0.34 (0.11–0.57) |
<0.001 | 6.7 (5.9–7.5) |
<0.001 | 2.8 (2.5–3.2) |
<0.001 | 3.2 (2.8–3.6) |
<0.001 |
| Baseline trend in rate | 0.60 (0.12–1.1) |
0.02 | 0.72 (0.60–0.96) |
<0.001 | 0.12 (0.12–0.24) |
<0.001 | 0.12 (−0.12 to 0.36) |
0.3 | −0.12 (−0.24 to 0.00) |
0.3 | −0.12 (−0.24 to 0.00) |
0.0498 |
| 2008 USPSTF | ||||||||||||
| Level change | 1.1 (−2.0 to 4.1) |
0.5 | −1.3 (−3.2 to 0.58) |
0.2 | 0.13 (−0.45 to 0.72) |
0.7 | 1.9 (−0.11 to 3.9) |
0.07 | 0.46 (−0.50 to 1.4) |
0.4 | 0.27 (−0.69 to 1.2) |
0.6 |
| Trend change | −3.5 (−13.4 to 6.6) |
0.5 | −1.8 (−8.4 to 4.7) |
0.6 | −1.0 (−3.0 to 0.96) |
0.3 | −4.3 (−11.4 to 2.6) |
0.2 | −1.6 (−4.9 to 1.9) |
0.4 | −0.72 (−4.0 to 2.6) |
0.7 |
| ERSPC/PLCO | ||||||||||||
| Level change | −0.14 (−3.9 to 3.6) |
0.9 | 1.1 (−1.3 to 3.4) |
0.4 | 0.55 (−0.19 to 1.3) |
0.2 | 0.39 (−2.2 to 3.0) |
0.8 | 0.75 (−0.47 to 2.0) |
0.2 | 0.65 (−0.56 to 1.9) |
0.3 |
| Trend change | 2.9 (−7.1 to 13.0) |
0.6 | 0.72 (−5.8 to 7.2) |
0.8 | 0.84 (−1.2 to 2.9) |
0.4 | 4.2 (−2.8 to 11.3) |
0.2 | 1.7 (−1.8 to 5.0) |
0.4 | 0.72 (−2.6 to 4.0) |
0.7 |
| 2012 USPSTF | ||||||||||||
| Level change | −0.27 (−2.7 to 2.1) |
0.8 | −0.56 (−2.0 to 0.87) |
0.5 | 0.16 (−0.31 to 0.63) |
0.5 | 1.1 (−0.38 to 2.5) |
0.2 | −0.38 (−1.0 to 0.28) |
0.3 | 0.44 (−0.34 to 1.2) |
0.3 |
| Trend change | 1.3 (−2.6 to 5.3) |
0.5 | 2.9 (0.60–5.3) |
0.02 | −0.24 (−1.1 to 0.60) |
0.6 | −0.72 (−3.2 to 1.7) |
0.5 | 0.24 (−0.96 to 1.3) |
0.7 | −0.48 (−1.8 to 0.84) |
0.5 |
| 2013 AUA | ||||||||||||
| Level change | 0.25 (−2.6 to 3.1) |
0.9 | −0.64 (−2.3 to 1.1) |
0.5 | 0.22 (−0.40 to 0.83) |
0.5 | 1.1 (−0.68 to 2.8) |
0.2 | −0.29 (−1.1 to 0.48) |
0.5 | 0.47 (−0.55 to 1.5) |
0.4 |
| Trend change | −1.6 (−6.1 to 2.9) |
0.5 | −2.5 (−5.5 to 0.60) |
0.1 | 0.12 (−0.72 to 1.1) |
0.8 | −0.36 (−3.1 to 2.4) |
0.8 | 0.24 (−1.2 to 1.7) |
0.8 | 0.12 (−1.3 to 1.6) |
0.9 |
* Model adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 person-months. Baseline trend in rate and trend change per 100 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial]; USPTF = US Preventive Services Task Force.
Given the increase in the relative morbidity of biopsy, we performed logistic regression to identify predictors of each complication category (Supplementary Table 3 and Table 4). Due to the large sample size, we focused on those features associated with the greatest magnitude of effect. A prior diagnosis of cancer was associated with an increased risk of all complication categories including any complication (odds ratio [OR]: 1.30; 95% CI, 1.22–1.39; p < 0.001) and hospitalization (OR: 1.59; 95% CI, 1.39–1.82; p < 0.001). Similarly, prior fluoroquinolone use was associated with an increased risk of all complication categories including any complication (OR: 1.27; 95% CI, 1.22–1.32; p < 0.001) and sepsis (OR: 1.55; 95% CI, 1.33–1.81; p < 0.001). Age at biopsy ≥70 yr was associated with a nearly 4-fold increased risk of urinary retention. Although anticoagulant use was associated with increased risks of bleeding (OR: 1.28; 95% CI, 1.14–1.45; p < 0.001), sepsis (OR: 1.48; 95% CI, 1.06–2.07; p = 0.02), and hospitalization (OR: 1.39; 95% CI, 1.16–1.67; p < 0.001), the increase in the overall complication risk was modest (OR 1.14; 95% CI, 1.04–1.25; p = 0.004).
Table 4
Multivariable analysis* of 30-d postbiopsy complications among men aged ≥40 yr undergoing prostate biopsy (January 2005 to September 2014) (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Hospitalization | Urinary retention | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Age at biopsy, yr | ||||||||||||
| 40–49 | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| 50–59 | 1.01 (0.94–1.09) | 0.8 | 0.94 (0.85–1.04) | 0.2 | 1.16 (0.83–1.63) | 0.4 | 1.01 (0.90–1.14) | 0.8 | 1.12 (0.93–1.36) | 0.2 | 1.54 (1.20–1.97) | <0.001 |
| 60–69 | 1.05 (0.97–1.13) | 0.2 | 0.94 (0.85–1.03) | 0.2 | 0.98 (0.70–1.37) | 0.9 | 1.05 (0.94–1.18) | 0.4 | 1.16 (0.95–1.40) | 0.1 | 2.16 (1.68–2.77) | <0.001 |
| ≥70 | 1.25 (1.15–1.36) | <0.001 | 1.03 (0.93–1.15) | 0.6 | 1.05 (0.74–1.49) | 0.8 | 1.29 (1.15–1.46) | <0.001 | 1.31 (1.07–1.60) | 0.008 | 3.98 (3.08–5.15) | <0.001 |
| Race | ||||||||||||
| White | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| Asian | 1.27 (1.15–1.40) | <0.001 | 1.27 (1.11–1.44) | 0.0003 | 1.65 (1.19–2.29) | 0.003 | 1.36 (1.18–1.56) | <0.001 | 0.94 (0.76–1.17) | 0.6 | 1.15 (0.93–1.42) | 0.2 |
| African American | 1.09 (1.03–1.15) | 0.002 | 1.05 (0.97–1.13) | 0.2 | 1.34 (1.09–1.64) | 0.005 | 1.17 (1.09–1.27) | <0.001 | 1.13 (1.01–1.27) | 0.03 | 1.15 (1.03–1.28) | 0.02 |
| Hispanic | 1.32 (1.24–1.40) | <0.001 | 1.36 (1.26–1.47) | <0.001 | 1.52 (1.21–1.90) | <0.001 | 1.33 (1.22–1.45) | <0.001 | 1.12 (0.97–1.28) | 0.1 | 1.37 (1.21–1.56) | <0.001 |
| Unknown/missing | 0.99 (0.93–1.06) | 0.8 | 1.07 (0.99–1.17) | 0.1 | 1.23 (0.94–1.61) | 0.1 | 0.96 (0.88–1.05) | 0.4 | 0.94 (0.81–1.08) | 0.4 | 1.02 (0.88–1.18) | 0.8 |
| Diabetes mellitus | 1.03 (0.99–1.07) | 0.09 | 1.06 (1.00–1.11) | 0.045 | 1.08 (0.92–1.28) | 0.4 | 1.01 (0.96–1.07) | 0.7 | 1.20 (1.10–1.31) | <0.001 | 0.99 (0.91–1.08) | 0.8 |
| COPD | 1.14 (1.09–1.20) | <0.001 | 1.07 (1.00–1.14) | 0.04 | 1.23 (1.01–1.49) | 0.04 | 1.14 (1.06–1.22) | <0.001 | 1.36 (1.23–1.51) | <0.001 | 1.14 (1.03–1.25) | 0.01 |
| PVD | 1.24 (1.16–1.31) | <0.001 | 1.1 (1.02–1.2) | 0.02 | 1.19 (0.93–1.51) | 0.2 | 1.28 (1.18–1.40) | <0.001 | 1.31 (1.15–1.48) | <0.001 | 1.26 (1.12–1.42) | <0.001 |
| Cerebrovascular disease | 1.11 (1.04–1.19) | 0.001 | 1.12 (1.03–1.22) | 0.01 | 1.21 (0.93–1.57) | 0.2 | 1.03 (0.94–1.14) | 0.5 | 1.26 (1.10–1.44) | <0.001 | 1.30 (1.15–1.47) | <0.001 |
| Cancer | 1.30 (1.22–1.39) | <0.001 | 1.16 (1.06–1.27) | 0.001 | 1.37 (1.05–1.79) | 0.02 | 1.40 (1.27–1.53) | <0.001 | 1.59 (1.39–1.82) | <0.001 | 1.32 (1.16–1.52) | <0.001 |
| Anticoagulant use | 1.14 (1.04–1.25) | 0.004 | 1.01 (0.89–1.14) | 0.9 | 1.48 (1.06–2.07) | 0.02 | 1.28 (1.14–1.45) | <0.001 | 1.39 (1.16–1.67) | <0.001 | 1.11 (0.92–1.33) | 0.3 |
| Prior fluoroquinolones | 1.27 (1.22–1.32) | <0.001 | 1.37 (1.31–1.45) | <0.001 | 1.55 (1.33–1.81) | <0.001 | 1.21 (1.14–1.28) | <0.001 | 1.25 (1.14–1.36) | <0.001 | 1.27 (1.16–1.38) | <0.001 |
* Models adjusted for year of biopsy and the features listed in the table.
CI = confidence interval; COPD = chronic obstructive pulmonary disease; PVD = peripheral vascular disease.
We conducted sensitivity analyses where the postintervention run-in period was varied from 2 to 4 mo (Supplementary Table 4). Results overall were similar, although effects were attenuated for the 2013 AUA guidelines on biopsy rates and the 2012 USPSTF recommendations on overall complication rates, likely due to fewer time points for analysis. In a separate sensitivity analysis, we examined bleeding complications excluding microhematuria from 2009 to 2014, following introduction of a separate code for microhematuria. Bleeding rates paralleled those of the primary bleeding end point (Supplementary Fig. 4), and results of both ITS and logistic regression analyses were similar (Supplementary Table 5 and 6). As an exploratory analysis, we examined PCa diagnosis rates within 60 d following biopsy (Supplementary Fig. 5). These appeared to increase steadily across the study period.
To our knowledge, this is the largest study to examine the impact of landmark PSA screening publications on rates of prostate biopsy and the first to examine their impact on postbiopsy complications. Overall, biopsy rates have declined by 33%, which appears attributable to the publication of the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines. Interestingly, the magnitude of reduction in biopsy rates exceeded previously reported decreases in PSA screening rates [6], [7], [8], [9], [10], [11], [12], [22], and [23], which suggests that either the threshold for referral to a urologist or the threshold for performing a biopsy has changed. More importantly, there was a concurrent 22% decrease in 30-d complication rates that appears largely attributable to the 2012 USPSTF recommendations. Still, the reduction in complication rates was less than the reduction in biopsy rates due to an increase in the relative morbidity of biopsy, driven by growth in infectious complications. This may in part be related to a shift to older patients with greater comorbidity undergoing biopsy, although similar results were obtained in analyses adjusted for case mix, which suggests additional factors may exist.
To this end, we evaluated predictors of postbiopsy complications to identify actionable targets for further risk reduction. Some of these features represent modifiable risk factors (eg, anticoagulant use), and others represent opportunities for altering periprocedural management. For instance, prior fluoroquinolone use may identify patients who would benefit from prebiopsy rectal swabs to guide periprocedural antibiotics [24] and [25]. Nonmodifiable risk factors, including specific comorbidities such as cancer and peripheral vascular disease, may encourage reconsideration of the risk–benefit ratio before proceeding with biopsy.
It is also important to consider the concurrent impact of landmark PSA screening publications on rates of PCa diagnosis to contextualize the observed changes in biopsy and its morbidity. An exploratory analysis suggested that cancer diagnosis rates have steadily increased, although further studies are necessary to examine this issue comprehensively.
A number of studies have examined the impact of the 2008 USPSTF recommendations [7], [8], [9], and [22], ERSPC/PLCO trials [6] and [9], and 2012 USPSTF recommendations [6], [10], [11], [12], and [23] on PSA screening rates, generally noting reductions ranging from 0% to 8%. However, much less data are available on the impact of these publications on biopsy rates. One single-center study noted a 31% reduction in the number of biopsies performed following the 2012 USPSTF recommendations [13]; another noted no change in the proportion of referred men who underwent biopsy [26]. A report from a Canadian academic health network noted a reduction in biopsy rates from 58.0 to 35.5 biopsies per month following the 2012 USPSTF recommendations [27]. Our findings therefore confirm a reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines.
Limited data exist regarding secular trends in postbiopsy complications, and prior studies have focused exclusively on the relative morbidity among men undergoing biopsy rather than absolute complication rates. A Canadian population-based study observed that the 30-d hospitalization rate increased from 1.0% in 1996 to 4.1% in 2005, driven by an increase in infectious complications [28]. A Surveillance, Epidemiology and End Results–Medicare study also noted an increase in hospitalizations related to infectious complications over time [14]. Although we observed a similar increase in 30-d overall complications among men undergoing biopsy, likewise driven by infectious complications, there was an absolute 22% reduction in 30-d complications over the study period. Taken together, these findings support a reduction in the morbidity of biopsy following publication of revised PSA screening guidelines but highlight the continued need to reduce the relative morbidity of biopsy.
A number of studies have described complication rates following prostate biopsy [15], [16], [29], and [30]. Hematuria has been reported in 10–84% of biopsies, rectal bleeding in 1–45%, infectious complications in up to 6.3%, urinary retention in up to 1.7%, and hospitalization in up to 6.9% [14] and [29]. The wide range in complication rates is due to differences in definitions, ascertainment of end points, and variation in clinical practice. Given the heterogeneity in postbiopsy complication rates, our findings have two advantages. First, they represent a contemporary, geographically diverse population across the entire United States. Second, use of claims-based ascertainment limited complications to those cases in which patients sought medical care, and although this may underestimate the comprehensive morbidity of biopsy, it more accurately reflects the economic burden of serious complications.
This study has several limitations. Most importantly, a control population of men who were not exposed to the PSA screening publications was not available. This study is retrospective, and additional, unrecognized changes in policy or clinical practice may possibly have confounded the impact of the interventions examined. However, the ITS approach has been described as the strongest quasi-experimental design to evaluate policy changes when a randomized trial is not possible [19]. Because we relied on administrative claims for ascertainment of complications, minor complications that did not generate a claim were not captured, thus underestimating these end points. Although we adjusted for case mix, we were unable to adjust for PSA. However, population shifts in PSA may be confounded by changes in screening practices, and PSA adjustment may therefore under- or overestimate the impact of interventions. We were also unable to adjust for other factors that may be associated with biopsy or complication rates, such as prostate magnetic resonance imaging (MRI), ancillary biomarker testing, biopsy approach (eg, MRI-fusion biopsy, transperineal biopsy), indication for anticoagulation, use of periprocedural bridging, or operator biopsy volume/experience.
We observed an immediate reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines, with a smaller reduction in absolute 30-d complication rates. Still, the relative morbidity of biopsy has increased over time, driven by infectious complications. Accordingly, if health policy aims to reduce the morbidity of PSA screening related to biopsy, further efforts should focus on improving patient selection and periprocedural processes of care.
Author contributions: Boris Gershman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Gershman, van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Acquisition of data: Van Houten, Shah.
Analysis and interpretation of data: Gershman, Van Houten, Shah, Karnes.
Drafting of the manuscript: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Critical revision of the manuscript for important intellectual content: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Statistical analysis: Van Houten.
Obtaining funding: Karnes, Shah.
Administrative, technical, or material support: Karnes, Shah.
Supervision: Karnes, Shah.
Other (specify): None.
Financial disclosures: Boris Gershman certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: This publication was made possible by funding from the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery.
Funding/Support and role of the sponsor: None.
In 2009, two landmark randomized trials—the European Randomized Study of Screening for Prostate Cancer [ERSPC] [1] and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial [PLCO] [2]—reported smaller than anticipated reductions in prostate cancer (PCa) mortality and prompted reexamination of prostate-specific antigen (PSA) screening recommendations. The US Preventive Services Task Force (USPSTF), which in 2008 had recommended against PSA screening in men aged >75 yr [3], updated its position in 2012 to recommend against systematic PSA screening in the general population [4]. The American Urological Association (AUA) likewise updated its PSA screening guidelines in May 2013, recommending shared decision making for men aged 55–69 yr [5]. Both the 2012 USPSTF and 2013 AUA guidelines cited prostate biopsy and its potential complications as a significant harm associated with PSA screening.
Although several studies have noted a reduction in PSA screening rates following these publications [6], [7], [8], [9], [10], [11], and [12], there are little data on whether this has translated into the intended reductions in screening-related morbidity in the form of prostate biopsy and postbiopsy complications [13]. We evaluated whether publication of the 2008 USPSTF recommendations, 2009 ERSPC/PLCO trials, 2012 USPSTF recommendations, or 2013 AUA guidelines was associated with reductions in prostate biopsy or postbiopsy complications. Biopsy-related complications also remain understudied both with regard to incidence as well as identification of risk factors [14], [15], and [16]. Consequently, we also sought to characterize postbiopsy complications and evaluate predictors using a contemporary population-based cohort.
We conducted a retrospective analysis of administrative claims from the Optum Labs Data Warehouse that contains data on >100 million individuals enrolled in private health plans and Medicare Advantage Plans, from geographically diverse regions across the United States, as described in Supplementary Methods [17]. Because this study involved analysis of preexisting deidentified data, it was exempt from institutional review board approval.
We identified all prostate biopsies performed in men aged ≥40 yr between January 1, 2005, and September 30, 2014, using Current Procedural Terminology-4 code 55700 alone (11%) or in combination with 76872 (84%) or 76942 (0%), or International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 60.11 (5%). The first biopsy was selected as the index biopsy when more than one biopsy was performed. Patients with a prior diagnosis of PCa (ICD-9 code 185) were excluded. We restricted the cohort to patients with continuous medical and pharmacy enrollment for at least 12 mo prior to biopsy and 30 d after biopsy to ascertain comorbidities and 30-d complications (Supplementary Fig. 1).
We recorded sociodemographic characteristics, comorbid conditions, anticoagulant use within 30 d of biopsy, and fluoroquinolone use in the 12 mo prior to biopsy as described in Supplement 1. The Charlson Comorbidity Index was calculated as previously described [18]. Prebiopsy PSA values within 60 d of biopsy were available for a subset of patients.
The primary outcomes were crude and adjusted rates of prostate biopsy and 30-d complications. Complications were identified using inpatient and outpatient administrative claims as summarized in Supplementary Table 1.
Baseline demographic features were compared across study periods using the Kruskal-Wallis test for continuous variables and the chi-square test for discrete variables. To evaluate the impact of each publication on rates of prostate biopsy and 30-d complications, we used an interrupted time series (ITS) approach as described in Supplementary Methods [19] and [20]. Rates were estimated using all men aged ≥40 yr with at least 12 mo of continuous medical and pharmacy enrollment as the denominator. As secondary analyses, we estimated rates of 30-d complications among men who underwent biopsy. Two separate ITS analyses were performed: (1) unadjusted analyses, using crude rates of biopsy and complications aggregated for each month, and (2) analyses standardized for case mix, to adjust for population shifts over time as described in Supplementary Methods [20] and [21].
Logistic regression was used to evaluate predictors of 30-d complications. Multivariable models were adjusted for age at biopsy, year of biopsy, race, anticoagulant use, history of fluoroquinolone use, and the presence of the five most prevalent Charlson comorbidities.
Statistical analyses were performed using SAS v.9.3 (SAS Institute Inc., Cary, NC, USA). All tests were two sided with p < 0.05 considered statistically significant.
Between January 2005 and September 2014, a total of 104 584 men underwent prostate biopsy (Table 1). Median age at biopsy was 63.0 yr (interquartile range [IQR]: 57.0–70.0). Overall, 18% of men had used fluoroquinolones in the 12 mo preceding biopsy, and 2.8% were on anticoagulants. Median prebiopsy PSA was 5.1 ng/dl (IQR: 4.0–7.1) for the 18 749 men (18%) for whom this was available. Over the time periods, men undergoing biopsy were older, had increased prevalence of anticoagulant use and comorbid conditions, and had a higher PSA (Table 1 and Supplementary Fig. 2).
Table 1
Patient demographics for men who underwent prostate biopsy (January 2005 to September 2014) stratified by pre/postintervention time periods (n = 104 584)
| Baseline Jan 2005–Jul 2008 (n = 37 217) |
2008 USPSTF Aug 2008–Feb 2009 (n = 6672) |
ERSPC/PLCO Mar 2009–Apr 2012 (n = 36 603) |
2012 USPSTF May 2012–Apr 2013 (n = 10 772) |
2013 AUA May 2013–Sep 2014 (n = 13 320) |
p value | Total (n = 104 584) | |
|---|---|---|---|---|---|---|---|
| Age at biopsy, yr | |||||||
| Median (IQR) | 62.0 (56.0–68.0) | 62.0 (56.0–68.0) | 63.0 (57.0–70.0) | 65.0 (59.0–71.0) | 66.0 (59.0–71.0) | <0.001 | 63.0 (57.0–70.0) |
| Age group, yr (%) | |||||||
| 40–49 | 2269 (6.1) | 416 (6.2) | 1737 (4.7) | 420 (3.9) | 467 (3.5) | <0.001 | 5309 (5.1) |
| 50–59 | 12 366 (33) | 2201 (33) | 10 342 (28) | 2644 (25) | 3098 (23) | 30 651 (29) | |
| 60–69 | 14 671 (39) | 2697 (40) | 15 105 (41) | 4426 (41) | 5546 (42) | 42 445 (41) | |
| ≥70 | 7911 (21) | 1358 (20) | 9419 (26) | 3282 (31) | 4209 (32) | 26 179 (25) | |
| Race (%) | |||||||
| White | 25 508 (69) | 4643 (70) | 26 187 (72) | 7795 (72) | 9407 (71) | <0.001 | 73 540 (70) |
| African American | 3695 (9.9) | 780 (12) | 4423 (12) | 1283 (12) | 1589 (12) | 11 770 (11) | |
| Hispanic | 2549 (6.8) | 538 (8.1) | 2911 (8.0) | 837 (7.8) | 1186 (8.9) | 8021 (7.7) | |
| Asian | 1133 (3.0) | 236 (3.5) | 1142 (3.1) | 306 (2.8) | 387 (2.9) | 3204 (3.1) | |
| Unknown/Missing | 4332 (12) | 475 (7.1) | 1940 (5.3) | 551 (5.1) | 751 (5.6) | 8049 (7.7) | |
| Census region (%) | |||||||
| Northeast | 7094 (19) | 1196 (18) | 6120 (17) | 1886 (18) | 2424 (18) | <0.001 | 18 720 (18) |
| Midwest | 10 094 (27) | 1562 (23) | 8837 (24) | 2742 (26) | 3627 (27) | 26 862 (26) | |
| South | 15 330 (41) | 2889 (43) | 16 371 (45) | 4698 (44) | 5538 (42) | 44 826 (43) | |
| West | 4660 (13) | 993 (15) | 5147 (14) | 1442 (13) | 1728 (13) | 13 970 (13) | |
| Other/Missing | 39 (0.1) | 32 (0.5) | 128 (0.3) | 4 (0) | 3 (0) | 206 (0.2) | |
| PSA, ng/d | n = 5174 | n = 548 | n = 7345 | n = 2211 | n = 3471 | n = 18 749 | |
| Median (IQR) | 5.1 (3.9–7.1) | 4.9 (3.8–6.8) | 5.0 (4.0–6.8) | 5.3 (4.2–7.4) | 5.4 (4.2–7.8) | <0.001 | 5.1 (4.0–7.1) |
| Anticoagulant use (%) | 928 (2.5) | 184 (2.8) | 1103 (3.0) | 299 (2.8) | 415 (3.1) | <0.001 | 2929 (2.8) |
| History of fluoroquinolone use in prior 12 mo (%) | 6797 (18) | 1267 (19) | 6557 (18) | 2017 (19) | 2504 (19) | 0.054 | 19 142 (18) |
| Charlson Comorbidity Index (%) | |||||||
| 0 | 22 354 (60) | 3887 (58) | 20 269 (55) | 5758 (54) | 7039 (53) | <0.001 | 59 307 (57) |
| 1 | 8199 (22) | 1453 (22) | 8252 (23) | 2503 (23) | 2999 (23) | 23 406 (22) | |
| 2 | 3232 (8.7) | 664 (10) | 3520 (9.6) | 1108 (10) | 1359 (10) | 9883 (9.4) | |
| ≥3 | 3432 (9.2) | 668 (10) | 4562 (13) | 1403 (13) | 1923 (14) | 11 988 (12) | |
| Comorbidities (%) | |||||||
| Myocardial infarction | 763 (2.1) | 144 (2.2) | 812 (2.2) | 227 (2.1) | 273 (2.0) | 0.6 | 2219 (2.1) |
| Congestive heart failure | 1443 (3.9) | 223 (3.3) | 1548 (4.2) | 467 (4.3) | 586 (4.4) | <0.001 | 4267 (4.1) |
| Peripheral vascular disease | 2054 (5.5) | 399 (6.0) | 2660 (7.3) | 783 (7.3) | 1071 (8.0) | <0.001 | 6967 (6.7) |
| Cerebrovascular disease | 2007 (5.4) | 379 (5.7) | 2358 (6.4) | 681 (6.3) | 947 (7.1) | <0.001 | 6372 (6.1) |
| Dementia | 370 (1.0) | 63 (0.9) | 442 (1.2) | 141 (1.3) | 198 (1.5) | <0.001 | 1214 (1.2) |
| Chronic obstructive pulmonary disease | 4251 (11) | 758 (11) | 4293 (12) | 1254 (12) | 1686 (13) | 0.004 | 12 242 (12) |
| Ulcer | 362 (1.0) | 60 (0.9) | 297 (0.8) | 84 (0.8) | 102 (0.8) | 0.07 | 905 (0.9) |
| Mild liver disease | 1182 (3.2) | 223 (3.3) | 1332 (3.6) | 413 (3.8) | 515 (3.9) | <0.001 | 3665 (3.5) |
| Diabetes mellitus | 6475 (17) | 1290 (19) | 7823 (21) | 2442 (23) | 2973 (22) | <0.001 | 21 003 (20) |
| Diabetes with organ damage | 1105 (3.0) | 225 (3.4) | 1600 (4.4) | 492 (4.6) | 734 (5.5) | <0.001 | 4156 (4.0) |
| Hemiplegia | 128 (0.3) | 25 (0.4) | 150 (0.4) | 51 (0.5) | 56 (0.4) | 0.3 | 410 (0.4) |
| Moderate/severe renal disease | 1133 (3.0) | 254 (3.8) | 1950 (5.3) | 674 (6.3) | 908 (6.8) | <0.001 | 4919 (4.7) |
| Moderate/severe liver disease | 27 (0.1) | 7 (0.1) | 31 (0.1) | 9 (0.1) | 12 (0.1) | 0.9 | 86 (0.1) |
| Metastatic solid tumor | 175 (0.5) | 37 (0.6) | 199 (0.5) | 58 (0.5) | 80 (0.6) | 0.4 | 549 (0.5) |
| AIDS | 82 (0.2) | 16 (0.2) | 101 (0.3) | 27 (0.3) | 46 (0.3) | 0.2 | 272 (0.3) |
| Rheumatologic disease | 449 (1.2) | 71 (1.1) | 522 (1.4) | 158 (1.5) | 196 (1.5) | 0.007 | 1396 (1.3) |
| Cancer | 1736 (4.7) | 337 (5.1) | 1796 (4.9) | 571 (5.3) | 701 (5.3) | 0.02 | 5141 (4.9) |
| Net worth range (%) | |||||||
| <$25 000 | 1831 (4.9) | 404 (6.1) | 2101 (5.7) | 561 (5.2) | 695 (5.2) | <0.001 | 5592 (5.3) |
| $25 000–149 000 | 5188 (14) | 1042 (16) | 5608 (15) | 1591 (15) | 2046 (15) | 15 475 (15) | |
| $150 000–249 000 | 4754 (13) | 837 (13) | 5182 (14) | 1499 (14) | 1862 (14) | 14 134 (14) | |
| $250 000–499 000 | 9694 (26) | 1779 (27) | 10 048 (28) | 3027 (28) | 3636 (27) | 28 184 (27) | |
| ≥$500 000 | 10 910 (29) | 2030 (30) | 10 860 (30) | 3208 (30) | 3852 (29) | 30 860 (30) | |
| Unknown/missing | 4840 (13) | 580 (8.7) | 2804 (7.7) | 886 (8.2) | 1229 (9.2) | 10 339 (9.9) | |
| 30-d complications (%) | |||||||
| Any | 5973 (16) | 1151 (17) | 6120 (17) | 1863 (17) | 2451 (18) | <0.001 | 17 558 (17) |
| Bleeding | 2547 (6.8) | 499 (7.5) | 2660 (7.3) | 860 (8.0) | 1083 (8.1) | <0.001 | 7649 (7.3) |
| Nonsepsis infection | 2978 (8.0) | 590 (8.8) | 2990 (8.2) | 875 (8.1) | 1236 (9.3) | <0.001 | 8669 (8.3) |
| Sepsis | 191 (0.5) | 50 (0.7) | 349 (1.0) | 105 (1.0) | 134 (1.0) | <0.001 | 829 (0.8) |
| Urinary retention | 980 (2.6) | 161 (2.4) | 1138 (3.1) | 341 (3.2) | 422 (3.2) | <0.001 | 3042 (2.9) |
| Hospitalization | 1019 (2.7) | 181 (2.7) | 1056 (2.9) | 320 (3.0) | 367 (2.8) | 0.6 | 2943 (2.8) |
AIDS = acquired immunodeficiency syndrome; AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; IQR = interquartile range; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer [screening trial]; PSA = prostate-specific antigen; USPTF = US Preventive Services Task Force.
Overall, 30-d complications occurred in 17% of biopsies. The most prevalent was nonsepsis genitourinary infections (8.3%), followed by bleeding complications (7.3%) and urinary retention (2.9%) (Table 1). The 30-d hospitalization rate was 2.8%.
We first examined unadjusted rates of prostate biopsy among men aged ≥40 yr (Fig. 1 and Table 2). From January 2005 to September 2014, biopsy rates fell 33% from 64.1 to 42.8 per 100 000 person-months. There was substantial seasonal variation in biopsy rates with reductions near the end of each calendar year. With the ITS approach, each intervention is potentially associated with an immediate change in event rates (ie, level shift or change in intercept) as well as a change in the rate of growth (ie, change in slope). At baseline, there was a statistically significant growth in prostate biopsy rates of 2.9 biopsies per 100 000 person-months per year (95% confidence interval [CI], 0.96–4.8; p < 0.001). Following the 2008 USPSTF recommendations, there was an immediate decrease (ie, level shift) in biopsy rates of −10.1 biopsies per 100 000 person-months (95% CI, −17.1 to −3.0; p < 0.001). Likewise, there were immediate decreases in biopsy rates following the 2012 USPSTF recommendations (−13.8 biopsies per 100 000 person-months; 95% CI, −21.0 to −6.7; p < 0.001) and 2013 AUA guidelines (−8.8 biopsies per 100 000 person-months; 95% CI, −16.7 to −0.92; p = 0.03). Analyses adjusted for case mix were similar, although the baseline growth rate was no longer significant, indicating this was related to population shifts over time (Table 2, model 2).
Fig. 1
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 2
Interrupted time series analysis of monthly rates of prostate biopsy and 30-d complications (January 2005 to September 2014) among men aged ≥40 yr (n = 5 279 315)
| Prostate biopsy | Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Model 1* | ||||||||||||||
| Baseline rate | 57.8 (53.4–62.3) |
<0.001 | 8.6 (7.5–9.7) |
<0.001 | 4.0 (3.3–4.6) |
<0.001 | 0.17 (0.03–0.32) |
0.02 | 2.8 (2.3–3.3) |
<0.001 | 1.5 (1.3–1.6) |
<0.001 | 1.4 (1.1–1.7) |
<0.001 |
| Baseline trend in rate | 2.9 (0.96–4.8) |
<0.001 | 0.84 (0.48–1.3) |
<0.001 | 0.72 (0.48–1.1) |
<0.001 | 0.12 (0.12–0.12) |
<0.001 | 0.36 (0.12–0.48) |
<0.001 | 0.12 (0.00–0.12) |
<0.001 | 0.00 (−0.12 to 0.12) |
0.5 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.1 (−17.1 to −3.0) |
<0.001 | −0.61 (−3.1 to 1.9) |
0.6 | −1.4 (−2.8 to −0.03) |
0.053 | 0.04 (−0.35 to 0.42) |
0.8 | 0.56 (−0.70 to 1.8) |
0.4 | −0.06 (−0.69 to 0.57) |
0.9 | 0.16 (−0.58 to 0.90) |
0.7 |
| Trend change | 7.9 (−12.7 to 28.4) |
0.5 | −1.0 (−9.1 to 7.2) |
0.8 | −0.12 (−4.1 to 4.0) |
0.98 | −0.60 (−1.9 to 0.72) |
0.4 | −2.0 (−6.5 to 2.4) |
0.4 | −1.1 (−3.4 to 1.3) |
0.4 | −1.0 (−3.6 to 1.6) |
0.5 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.6 (−12 to 4.5) |
0.4 | −0.74 (−3.8 to 2.4) |
0.6 | −0.12 (−1.8 to 1.5) |
0.9 | 0.29 (−0.20 to 0.78) |
0.2 | −0.04 (−1.6 to 1.6) |
0.97 | 0.59 (−0.23 to 1.4) |
0.2 | 0.51 (−0.43 to 1.5) |
0.3 |
| Trend change | −11.2 (−32.0 to 9.6) |
0.3 | 0.0 (−8.2 to 8.0) |
0.99 | −0.96 (−5.0 to 3.1) |
0.7 | 0.36 (−0.96 to 1.8) |
0.5 | 1.7 (−2.8 to 6.1) |
0.5 | 1.0 (−1.4 to 3.4) |
0.4 | 0.84 (−1.7 to 3.5) |
0.5 |
| 2012 USPSTF | ||||||||||||||
| Level change | −13.8 (−21.0 to −6.7) |
<0.001 | −2.5 (−4.5 to −0.45) |
0.02 | −1.7 (−3.1 to −0.31) |
0.02 | −0.12 (−0.42 to 0.19) |
0.5 | −0.61 (−1.5 to 0.28) |
0.2 | −0.75 (−1.1 to −0.44) |
<0.001 | −0.25 (−0.77 to 0.27) |
0.4 |
| Trend change | 8.2 (−5.8 to 22.0) |
0.3 | 3.0 (−0.36 to 6.4) |
0.09 | 3.1 (0.72–5.5) |
0.01 | 0.12 (−0.36 to 0.60) |
0.7 | 0.84 (−0.72 to 2.3) |
0.3 | 0.48 (0.00 to 1.1) |
0.08 | 0.24 (−0.60 to 1.2) |
0.6 |
| 2013 AUA | ||||||||||||||
| Level change | −8.8 (−16.7 to −0.92) |
0.03 | −1.4 (−3.8 to 1.1) |
0.3 | −1.9 (−3.5 to −0.22) |
0.03 | −0.02 (−0.41 to 0.37) |
0.9 | −0.19 (−1.3 to 0.91) |
0.7 | −0.38 (−0.77 to 0.01) |
0.06 | −0.01 (−0.66 to 0.63) |
0.96 |
| Trend change | −10.4 (−30.0 to 9.2) |
0.3 | −4.1 (−7.9 to −0.24) |
0.04 | −2.9 (−6.0 to 0.12) |
0.07 | −0.24 (−0.72 to 0.36) |
0.5 | −1.8 (−3.5 to 0.0) |
0.046 | −0.60 (−1.2 to 0.12) |
0.09 | −0.60 (−1.7 to 0.36) |
0.2 |
| Model 2† | ||||||||||||||
| Baseline rate | 86.8 (82.1–91.4) |
<0.001 | 16.2 (14.9–17.6) |
<0.001 | 6.4 (4.9–7.9) |
<0.001 | 0.39 (−0.13 to 0.91) |
0.2 | 6.1 (4.6–7.7) |
<0.001 | 3.6 (2.8–4.4) |
<0.001 | 3.0 (2.2–3.8) |
<0.001 |
| Baseline trend in rate | 0.24 (−1.8 to 2.3) |
0.8 | −0.24 (−0.72 to 0.24) |
0.4 | 0.48 (−0.12 to 1.1) |
0.09 | 0.12 (−0.12 to 0.24) |
0.4 | −0.24 (−0.72 to 0.24) |
0.4 | −0.24 (−0.48 to 0.00) |
0.054 | −0.12 (−0.36 to 0.12) |
0.3 |
| 2008 USPSTF | ||||||||||||||
| Level change | −10.8 (−17.8 to −3.8) |
<0.001 | −5.4 (−9.6 to −1.2) |
0.01 | −2.5 (−5.6 to 0.71) |
0.1 | −0.61 (−1.9 to 0.71) |
0.4 | 0.14 (−3.8 to 4.1) |
0.9 | 1.1 (−0.98 to 3.1) |
0.3 | 0.14 (−1.9 to 2.1) |
0.9 |
| Trend change | 10.6 (−10.1 to 31.2) |
0.3 | 6.1 (−8.4 to 20.8) |
0.4 | −1.7 (−12.2 to 8.8) |
0.8 | 1.7 (−2.9 to 6.2) |
0.5 | 1.7 (−12.0 to 15.5) |
0.8 | −4.7 (−11.8 to 2.5) |
0.2 | −0.84 (−7.8 to 6.1) |
0.8 |
| ERSPC/PLCO | ||||||||||||||
| Level change | −3.9 (−12 to 4.1) |
0.3 | −0.24 (−5.6 to 5.1) |
0.9 | 1.9 (−2.1 to 5.8) |
0.4 | −0.32 (−2.0 to 1.4) |
0.7 | −0.99 (−6.0 to 4.0) |
0.7 | 1.8 (−0.84 to 4.4) |
0.2 | 0.87 (−1.7 to 3.4) |
0.5 |
| Trend change | −16.2 (−37.1 to 4.7) |
0.1 | −6.8 (−21.4 to 7.8) |
0.4 | 0.36 (−10.2 to 10.9) |
0.95 | −1.9 (−6.5 to 2.8) |
0.4 | −1.6 (−15.2 to 12.2) |
0.8 | 4.7 (−2.4 to 11.9) |
0.2 | 0.48 (−6.4 to 7.4) |
0.9 |
| 2012 USPSTF | ||||||||||||||
| Level change | −11.0 (−18.1 to −4.0) |
0.001 | −2.8 (−6.0 to 0.38) |
0.09 | 0.53 (−1.9 to 3.0) |
0.7 | −0.02 (−0.95 to 0.91) |
0.97 | −1.8 (−4.6 to 0.96) |
0.2 | −1.5 (−3.1 to 0.17) |
0.08 | 0.16 (−1.3 to 1.6) |
0.8 |
| Trend change | 7.0 (−7.2 to 21.1) |
0.3 | 1.7 (−3.6 to 7.0) |
0.5 | 0.12 (−4.1 to 4.2) |
0.96 | 0.12 (−1.4 to 1.8) |
0.9 | −0.48 (−5.3 to 4.3) |
0.9 | 1.6 (−1.3 to 4.3) |
0.3 | 0.12 (−2.4 to 2.6) |
0.9 |
| 2013 AUA | ||||||||||||||
| Level change | −8.2 (−15.9 to −0.38) |
0.04 | −1.0 (−4.8 to 2.8) |
0.6 | −0.44 (−3.2 to 2.3) |
0.8 | −0.15 (−1.3 to 0.99) |
0.8 | 0.28 (−3.2 to 3.7) |
0.9 | −0.50 (−2.7 to 1.7) |
0.7 | 0.47 (−1.3 to 2.3) |
0.6 |
| Trend change | −4.2 (−24.5 to 16.1) |
0.7 | −0.7 (−7.6 to 6.2) |
0.9 | 1.1 (−4.2 to 6.2) |
0.7 | −0.12 (−1.8 to 1.7) |
0.9 | −0.60 (−5.9 to 4.8) |
0.8 | −1.8 (−4.9 to 1.2) |
0.3 | −0.48 (−3.1 to 2.2) |
0.7 |
* Model 1: Base model with constant, time trend, intervention indicator variables. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
† Model 2: Adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, and anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 000 person-months. Baseline trend in rate and trend change per 100 000 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPTF = US Preventive Services Task Force.
Next, we examined 30-d postbiopsy complications among men aged ≥40 yr as a measure of the absolute morbidity of biopsy (Fig. 1, Supplementary Fig. 3, and Table 2). Across the study periods, overall complication rates decreased 10% from 8.7 to 7.8 per 100 000 person-months. In unadjusted ITS analyses, there was a baseline growth in the overall complication rate (0.84 per 100 000 person-months per year; 95% CI, 0.48–1.3; p < 0.001), driven by growth in the rate of nonsepsis infections (0.72 per 100 000 person-months per year; 95% CI, 0.48–1.1; p < 0.001). Only the 2012 USPSTF recommendations were associated with a significant change in complication rates, following which there was an immediate decrease in the overall complication rate of −2.5 per 100 000 person-months (95% CI, −4.5 to −0.45; p = 0.02), driven by reductions in nonsepsis infections (−1.7 per 100 000 person-months; 95% CI, −3.1 to −0.31; p = 0.02) and urinary retention (−0.75 per 100 000 person-months;95% CI, −1.1 to −0.44; p < 0.001). Following the 2013 AUA guidelines there was also a reduction in the overall complication baseline growth rate (−4.1 per 100 000 person-months per year; 95% CI, −7.9 to −0.24; p = 0.04). In analyses adjusted for case mix (Table 2, model 2), effects were overall attenuated, and only the 2008 USPSTF recommendations were associated with a reduction in overall complication rates (−5.4 per 100 000 person-months; 95% CI, −9.6 to −1.2; p = 0.01).
To examine whether there was a change in the relative morbidity of biopsy, we performed a secondary analysis examining complication rates among men undergoing biopsy (Fig. 2). The proportion of men undergoing biopsy who experienced a complication increased from 14% in January 2005 to 18% in September 2014, an increase of 36%. There was again a baseline growth in overall complication rates (0.72% per year; 95% CI, 0.36–1.1; p < 0.001), driven by an increase in nonsepsis infections and sepsis (Supplementary Table 2). None of the interventions affected this growth. Similar results were obtained after adjusting for case mix (Table 3).
Fig. 2
Interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584).
AUA = American Urological Association; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; USPSTF = US Preventive Services Task Force.
Table 3
Adjusted* interrupted time series analysis of monthly rates of 30-d postbiopsy complications (January 2005 to September 2014) among men aged ≥40 yr undergoing biopsy (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Urinary retention | Hospitalization | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Baseline rate | 15.6 (14.3–16.8) |
<0.001 | 7.1 (6.5–7.7) |
<0.001 | 0.34 (0.11–0.57) |
<0.001 | 6.7 (5.9–7.5) |
<0.001 | 2.8 (2.5–3.2) |
<0.001 | 3.2 (2.8–3.6) |
<0.001 |
| Baseline trend in rate | 0.60 (0.12–1.1) |
0.02 | 0.72 (0.60–0.96) |
<0.001 | 0.12 (0.12–0.24) |
<0.001 | 0.12 (−0.12 to 0.36) |
0.3 | −0.12 (−0.24 to 0.00) |
0.3 | −0.12 (−0.24 to 0.00) |
0.0498 |
| 2008 USPSTF | ||||||||||||
| Level change | 1.1 (−2.0 to 4.1) |
0.5 | −1.3 (−3.2 to 0.58) |
0.2 | 0.13 (−0.45 to 0.72) |
0.7 | 1.9 (−0.11 to 3.9) |
0.07 | 0.46 (−0.50 to 1.4) |
0.4 | 0.27 (−0.69 to 1.2) |
0.6 |
| Trend change | −3.5 (−13.4 to 6.6) |
0.5 | −1.8 (−8.4 to 4.7) |
0.6 | −1.0 (−3.0 to 0.96) |
0.3 | −4.3 (−11.4 to 2.6) |
0.2 | −1.6 (−4.9 to 1.9) |
0.4 | −0.72 (−4.0 to 2.6) |
0.7 |
| ERSPC/PLCO | ||||||||||||
| Level change | −0.14 (−3.9 to 3.6) |
0.9 | 1.1 (−1.3 to 3.4) |
0.4 | 0.55 (−0.19 to 1.3) |
0.2 | 0.39 (−2.2 to 3.0) |
0.8 | 0.75 (−0.47 to 2.0) |
0.2 | 0.65 (−0.56 to 1.9) |
0.3 |
| Trend change | 2.9 (−7.1 to 13.0) |
0.6 | 0.72 (−5.8 to 7.2) |
0.8 | 0.84 (−1.2 to 2.9) |
0.4 | 4.2 (−2.8 to 11.3) |
0.2 | 1.7 (−1.8 to 5.0) |
0.4 | 0.72 (−2.6 to 4.0) |
0.7 |
| 2012 USPSTF | ||||||||||||
| Level change | −0.27 (−2.7 to 2.1) |
0.8 | −0.56 (−2.0 to 0.87) |
0.5 | 0.16 (−0.31 to 0.63) |
0.5 | 1.1 (−0.38 to 2.5) |
0.2 | −0.38 (−1.0 to 0.28) |
0.3 | 0.44 (−0.34 to 1.2) |
0.3 |
| Trend change | 1.3 (−2.6 to 5.3) |
0.5 | 2.9 (0.60–5.3) |
0.02 | −0.24 (−1.1 to 0.60) |
0.6 | −0.72 (−3.2 to 1.7) |
0.5 | 0.24 (−0.96 to 1.3) |
0.7 | −0.48 (−1.8 to 0.84) |
0.5 |
| 2013 AUA | ||||||||||||
| Level change | 0.25 (−2.6 to 3.1) |
0.9 | −0.64 (−2.3 to 1.1) |
0.5 | 0.22 (−0.40 to 0.83) |
0.5 | 1.1 (−0.68 to 2.8) |
0.2 | −0.29 (−1.1 to 0.48) |
0.5 | 0.47 (−0.55 to 1.5) |
0.4 |
| Trend change | −1.6 (−6.1 to 2.9) |
0.5 | −2.5 (−5.5 to 0.60) |
0.1 | 0.12 (−0.72 to 1.1) |
0.8 | −0.36 (−3.1 to 2.4) |
0.8 | 0.24 (−1.2 to 1.7) |
0.8 | 0.12 (−1.3 to 1.6) |
0.9 |
* Model adjusted (standardized) for case mix using age, race, five most prevalent comorbidities (diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, cardiovascular disease, and cancer), fluoroquinolone use in the preceding 12 mo, anticoagulant use in the preceding 30 d. Baseline rate and level change per 100 person-months. Baseline trend in rate and trend change per 100 person-months per year.
AUA = American Urological Association; CI = confidence interval; ERSPC = European Randomized Study of Screening for Prostate Cancer; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial]; USPTF = US Preventive Services Task Force.
Given the increase in the relative morbidity of biopsy, we performed logistic regression to identify predictors of each complication category (Supplementary Table 3 and Table 4). Due to the large sample size, we focused on those features associated with the greatest magnitude of effect. A prior diagnosis of cancer was associated with an increased risk of all complication categories including any complication (odds ratio [OR]: 1.30; 95% CI, 1.22–1.39; p < 0.001) and hospitalization (OR: 1.59; 95% CI, 1.39–1.82; p < 0.001). Similarly, prior fluoroquinolone use was associated with an increased risk of all complication categories including any complication (OR: 1.27; 95% CI, 1.22–1.32; p < 0.001) and sepsis (OR: 1.55; 95% CI, 1.33–1.81; p < 0.001). Age at biopsy ≥70 yr was associated with a nearly 4-fold increased risk of urinary retention. Although anticoagulant use was associated with increased risks of bleeding (OR: 1.28; 95% CI, 1.14–1.45; p < 0.001), sepsis (OR: 1.48; 95% CI, 1.06–2.07; p = 0.02), and hospitalization (OR: 1.39; 95% CI, 1.16–1.67; p < 0.001), the increase in the overall complication risk was modest (OR 1.14; 95% CI, 1.04–1.25; p = 0.004).
Table 4
Multivariable analysis* of 30-d postbiopsy complications among men aged ≥40 yr undergoing prostate biopsy (January 2005 to September 2014) (n = 104 584)
| Any complication | Nonsepsis infection | Sepsis | Bleeding | Hospitalization | Urinary retention | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | Estimate (95% CI) | p value | |
| Age at biopsy, yr | ||||||||||||
| 40–49 | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| 50–59 | 1.01 (0.94–1.09) | 0.8 | 0.94 (0.85–1.04) | 0.2 | 1.16 (0.83–1.63) | 0.4 | 1.01 (0.90–1.14) | 0.8 | 1.12 (0.93–1.36) | 0.2 | 1.54 (1.20–1.97) | <0.001 |
| 60–69 | 1.05 (0.97–1.13) | 0.2 | 0.94 (0.85–1.03) | 0.2 | 0.98 (0.70–1.37) | 0.9 | 1.05 (0.94–1.18) | 0.4 | 1.16 (0.95–1.40) | 0.1 | 2.16 (1.68–2.77) | <0.001 |
| ≥70 | 1.25 (1.15–1.36) | <0.001 | 1.03 (0.93–1.15) | 0.6 | 1.05 (0.74–1.49) | 0.8 | 1.29 (1.15–1.46) | <0.001 | 1.31 (1.07–1.60) | 0.008 | 3.98 (3.08–5.15) | <0.001 |
| Race | ||||||||||||
| White | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref | 1.00 | Ref |
| Asian | 1.27 (1.15–1.40) | <0.001 | 1.27 (1.11–1.44) | 0.0003 | 1.65 (1.19–2.29) | 0.003 | 1.36 (1.18–1.56) | <0.001 | 0.94 (0.76–1.17) | 0.6 | 1.15 (0.93–1.42) | 0.2 |
| African American | 1.09 (1.03–1.15) | 0.002 | 1.05 (0.97–1.13) | 0.2 | 1.34 (1.09–1.64) | 0.005 | 1.17 (1.09–1.27) | <0.001 | 1.13 (1.01–1.27) | 0.03 | 1.15 (1.03–1.28) | 0.02 |
| Hispanic | 1.32 (1.24–1.40) | <0.001 | 1.36 (1.26–1.47) | <0.001 | 1.52 (1.21–1.90) | <0.001 | 1.33 (1.22–1.45) | <0.001 | 1.12 (0.97–1.28) | 0.1 | 1.37 (1.21–1.56) | <0.001 |
| Unknown/missing | 0.99 (0.93–1.06) | 0.8 | 1.07 (0.99–1.17) | 0.1 | 1.23 (0.94–1.61) | 0.1 | 0.96 (0.88–1.05) | 0.4 | 0.94 (0.81–1.08) | 0.4 | 1.02 (0.88–1.18) | 0.8 |
| Diabetes mellitus | 1.03 (0.99–1.07) | 0.09 | 1.06 (1.00–1.11) | 0.045 | 1.08 (0.92–1.28) | 0.4 | 1.01 (0.96–1.07) | 0.7 | 1.20 (1.10–1.31) | <0.001 | 0.99 (0.91–1.08) | 0.8 |
| COPD | 1.14 (1.09–1.20) | <0.001 | 1.07 (1.00–1.14) | 0.04 | 1.23 (1.01–1.49) | 0.04 | 1.14 (1.06–1.22) | <0.001 | 1.36 (1.23–1.51) | <0.001 | 1.14 (1.03–1.25) | 0.01 |
| PVD | 1.24 (1.16–1.31) | <0.001 | 1.1 (1.02–1.2) | 0.02 | 1.19 (0.93–1.51) | 0.2 | 1.28 (1.18–1.40) | <0.001 | 1.31 (1.15–1.48) | <0.001 | 1.26 (1.12–1.42) | <0.001 |
| Cerebrovascular disease | 1.11 (1.04–1.19) | 0.001 | 1.12 (1.03–1.22) | 0.01 | 1.21 (0.93–1.57) | 0.2 | 1.03 (0.94–1.14) | 0.5 | 1.26 (1.10–1.44) | <0.001 | 1.30 (1.15–1.47) | <0.001 |
| Cancer | 1.30 (1.22–1.39) | <0.001 | 1.16 (1.06–1.27) | 0.001 | 1.37 (1.05–1.79) | 0.02 | 1.40 (1.27–1.53) | <0.001 | 1.59 (1.39–1.82) | <0.001 | 1.32 (1.16–1.52) | <0.001 |
| Anticoagulant use | 1.14 (1.04–1.25) | 0.004 | 1.01 (0.89–1.14) | 0.9 | 1.48 (1.06–2.07) | 0.02 | 1.28 (1.14–1.45) | <0.001 | 1.39 (1.16–1.67) | <0.001 | 1.11 (0.92–1.33) | 0.3 |
| Prior fluoroquinolones | 1.27 (1.22–1.32) | <0.001 | 1.37 (1.31–1.45) | <0.001 | 1.55 (1.33–1.81) | <0.001 | 1.21 (1.14–1.28) | <0.001 | 1.25 (1.14–1.36) | <0.001 | 1.27 (1.16–1.38) | <0.001 |
* Models adjusted for year of biopsy and the features listed in the table.
CI = confidence interval; COPD = chronic obstructive pulmonary disease; PVD = peripheral vascular disease.
We conducted sensitivity analyses where the postintervention run-in period was varied from 2 to 4 mo (Supplementary Table 4). Results overall were similar, although effects were attenuated for the 2013 AUA guidelines on biopsy rates and the 2012 USPSTF recommendations on overall complication rates, likely due to fewer time points for analysis. In a separate sensitivity analysis, we examined bleeding complications excluding microhematuria from 2009 to 2014, following introduction of a separate code for microhematuria. Bleeding rates paralleled those of the primary bleeding end point (Supplementary Fig. 4), and results of both ITS and logistic regression analyses were similar (Supplementary Table 5 and 6). As an exploratory analysis, we examined PCa diagnosis rates within 60 d following biopsy (Supplementary Fig. 5). These appeared to increase steadily across the study period.
To our knowledge, this is the largest study to examine the impact of landmark PSA screening publications on rates of prostate biopsy and the first to examine their impact on postbiopsy complications. Overall, biopsy rates have declined by 33%, which appears attributable to the publication of the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines. Interestingly, the magnitude of reduction in biopsy rates exceeded previously reported decreases in PSA screening rates [6], [7], [8], [9], [10], [11], [12], [22], and [23], which suggests that either the threshold for referral to a urologist or the threshold for performing a biopsy has changed. More importantly, there was a concurrent 22% decrease in 30-d complication rates that appears largely attributable to the 2012 USPSTF recommendations. Still, the reduction in complication rates was less than the reduction in biopsy rates due to an increase in the relative morbidity of biopsy, driven by growth in infectious complications. This may in part be related to a shift to older patients with greater comorbidity undergoing biopsy, although similar results were obtained in analyses adjusted for case mix, which suggests additional factors may exist.
To this end, we evaluated predictors of postbiopsy complications to identify actionable targets for further risk reduction. Some of these features represent modifiable risk factors (eg, anticoagulant use), and others represent opportunities for altering periprocedural management. For instance, prior fluoroquinolone use may identify patients who would benefit from prebiopsy rectal swabs to guide periprocedural antibiotics [24] and [25]. Nonmodifiable risk factors, including specific comorbidities such as cancer and peripheral vascular disease, may encourage reconsideration of the risk–benefit ratio before proceeding with biopsy.
It is also important to consider the concurrent impact of landmark PSA screening publications on rates of PCa diagnosis to contextualize the observed changes in biopsy and its morbidity. An exploratory analysis suggested that cancer diagnosis rates have steadily increased, although further studies are necessary to examine this issue comprehensively.
A number of studies have examined the impact of the 2008 USPSTF recommendations [7], [8], [9], and [22], ERSPC/PLCO trials [6] and [9], and 2012 USPSTF recommendations [6], [10], [11], [12], and [23] on PSA screening rates, generally noting reductions ranging from 0% to 8%. However, much less data are available on the impact of these publications on biopsy rates. One single-center study noted a 31% reduction in the number of biopsies performed following the 2012 USPSTF recommendations [13]; another noted no change in the proportion of referred men who underwent biopsy [26]. A report from a Canadian academic health network noted a reduction in biopsy rates from 58.0 to 35.5 biopsies per month following the 2012 USPSTF recommendations [27]. Our findings therefore confirm a reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines.
Limited data exist regarding secular trends in postbiopsy complications, and prior studies have focused exclusively on the relative morbidity among men undergoing biopsy rather than absolute complication rates. A Canadian population-based study observed that the 30-d hospitalization rate increased from 1.0% in 1996 to 4.1% in 2005, driven by an increase in infectious complications [28]. A Surveillance, Epidemiology and End Results–Medicare study also noted an increase in hospitalizations related to infectious complications over time [14]. Although we observed a similar increase in 30-d overall complications among men undergoing biopsy, likewise driven by infectious complications, there was an absolute 22% reduction in 30-d complications over the study period. Taken together, these findings support a reduction in the morbidity of biopsy following publication of revised PSA screening guidelines but highlight the continued need to reduce the relative morbidity of biopsy.
A number of studies have described complication rates following prostate biopsy [15], [16], [29], and [30]. Hematuria has been reported in 10–84% of biopsies, rectal bleeding in 1–45%, infectious complications in up to 6.3%, urinary retention in up to 1.7%, and hospitalization in up to 6.9% [14] and [29]. The wide range in complication rates is due to differences in definitions, ascertainment of end points, and variation in clinical practice. Given the heterogeneity in postbiopsy complication rates, our findings have two advantages. First, they represent a contemporary, geographically diverse population across the entire United States. Second, use of claims-based ascertainment limited complications to those cases in which patients sought medical care, and although this may underestimate the comprehensive morbidity of biopsy, it more accurately reflects the economic burden of serious complications.
This study has several limitations. Most importantly, a control population of men who were not exposed to the PSA screening publications was not available. This study is retrospective, and additional, unrecognized changes in policy or clinical practice may possibly have confounded the impact of the interventions examined. However, the ITS approach has been described as the strongest quasi-experimental design to evaluate policy changes when a randomized trial is not possible [19]. Because we relied on administrative claims for ascertainment of complications, minor complications that did not generate a claim were not captured, thus underestimating these end points. Although we adjusted for case mix, we were unable to adjust for PSA. However, population shifts in PSA may be confounded by changes in screening practices, and PSA adjustment may therefore under- or overestimate the impact of interventions. We were also unable to adjust for other factors that may be associated with biopsy or complication rates, such as prostate magnetic resonance imaging (MRI), ancillary biomarker testing, biopsy approach (eg, MRI-fusion biopsy, transperineal biopsy), indication for anticoagulation, use of periprocedural bridging, or operator biopsy volume/experience.
We observed an immediate reduction in biopsy rates following the 2008 USPSTF recommendations, 2012 USPSTF recommendations, and 2013 AUA guidelines, with a smaller reduction in absolute 30-d complication rates. Still, the relative morbidity of biopsy has increased over time, driven by infectious complications. Accordingly, if health policy aims to reduce the morbidity of PSA screening related to biopsy, further efforts should focus on improving patient selection and periprocedural processes of care.
Author contributions: Boris Gershman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Gershman, van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Acquisition of data: Van Houten, Shah.
Analysis and interpretation of data: Gershman, Van Houten, Shah, Karnes.
Drafting of the manuscript: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Critical revision of the manuscript for important intellectual content: Gershman, Van Houten, Herrin, Moreira, Kim, Shah, Karnes.
Statistical analysis: Van Houten.
Obtaining funding: Karnes, Shah.
Administrative, technical, or material support: Karnes, Shah.
Supervision: Karnes, Shah.
Other (specify): None.
Financial disclosures: Boris Gershman certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: This publication was made possible by funding from the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery.
Funding/Support and role of the sponsor: None.
Despite an absolute reduction in biopsy rates and 30-day complications following landmark publications and updated guidelines regarding PSA screening, the relative morbidity of prostate biopsy continues to increase.