Data were pooled separately by study design type using random-effects models; the inverse variance technique was used for pooling of measures of effect. NR indicates not reported.
Data were pooled separately by study design type using random-effects models; the inverse variance technique was used for pooling of measures of effect. NA indicates not available.
Data were pooled separately by study design type using random-effects models; the inverse variance technique was used for pooling of measures of effect.
eAppendix 1. Literature Search Strategy
eAppendix 2. Exclusion Following Full Text Review
eAppendix 3. Pooled Estimates of Studies Reporting Unadjusted Effect Estimates
eAppendix 4. Calculation of Estimates for Absolute Risk Increase, Number Needed to Harm, and Population-Attributable Fraction
eAppendix 5. Discussion of Hill’s Criteria of Causation
eFigure 1. Forest Plots for Meta-Analyses of the Adjusted Estimates for the Association Between Vasectomy and Any Prostate Cancer for Cohort Studies Reporting on Time-to-Event Analyses
eFigure 2. Meta-Analysis of Cohort Studies That Accounted for PSA Testing
eFigure 3. Forest Plots for Meta-Analyses of Unadjusted Estimates for the Association Between Vasectomy and Any Prostate Cancer by Study Design and Risk of Bias
eFigure 4. Funnel Plots
eTable 1. Risk Adjustment for Each Included Study
eTable 2. Newcastle-Ottawa Scale for Risk of Bias Assessment of Studies Included in the Meta-Analysis
eTable 3. Pooled Adjusted Estimates for Association Between Vasectomy and Prostate Cancer, Excluding Abstracts
eTable 4. Studies Reporting on the Impact Time Since Vasectomy on the Association Between Vasectomy and Prostate Cancer
eTable 5. Studies Reporting on the Impact of Age at Vasectomy on the Association Between Vasectomy and Prostate Cancer
Customize your JAMA Network experience by selecting one or more topics from the list below.
Bhindi B, Wallis CJD, Nayan M, et al. The Association Between Vasectomy and Prostate Cancer: A Systematic Review and Meta-analysis. JAMA Intern Med. 2017;177(9):1273–1286. doi:10.1001/jamainternmed.2017.2791
Is there an association between vasectomy and prostate cancer, high-grade prostate cancer, advanced prostate cancer, and/or fatal prostate cancer?
In this systematic review and meta-analysis including 53 studies, there was a weak, clinically insignificant association between vasectomy and prostate cancer. No association was found between vasectomy and risk of high-grade, advanced, or fatal prostate cancer.
At most, there is a trivial association between vasectomy and prostate cancer that is unlikely to be causal; therefore, concerns about prostate cancer should not preclude the use of vasectomy as an option for long-term contraception.
Despite 3 decades of study, there remains ongoing debate regarding whether vasectomy is associated with prostate cancer.
To determine if vasectomy is associated with prostate cancer.
The MEDLINE, EMBASE, Web of Science, and Scopus databases were searched for studies indexed from database inception to March 21, 2017, without language restriction.
Cohort, case-control, and cross-sectional studies reporting relative effect estimates for the association between vasectomy and prostate cancer were included.
Data Extraction and Synthesis
Two investigators performed study selection independently. Data were pooled separately by study design type using random-effects models. The Newcastle-Ottawa Scale was used to assess risk of bias.
Main Outcomes and Measures
The primary outcome was any diagnosis of prostate cancer. Secondary outcomes were high-grade, advanced, and fatal prostate cancer.
Fifty-three studies (16 cohort studies including 2 563 519 participants, 33 case-control studies including 44 536 participants, and 4 cross-sectional studies including 12 098 221 participants) were included. Of these, 7 cohort studies (44%), 26 case-control studies (79%), and all 4 cross-sectional studies were deemed to have a moderate to high risk of bias. Among studies deemed to have a low risk of bias, a weak association was found among cohort studies (7 studies; adjusted rate ratio, 1.05; 95% CI, 1.02-1.09; P < .001; I2 = 9%) and a similar but nonsignificant association was found among case-control studies (6 studies; adjusted odds ratio, 1.06; 95% CI, 0.88-1.29; P = .54; I2 = 37%). Effect estimates were further from the null when studies with a moderate to high risk of bias were included. Associations between vasectomy and high-grade prostate cancer (6 studies; adjusted rate ratio, 1.03; 95% CI, 0.89-1.21; P = .67; I2 = 55%), advanced prostate cancer (6 studies; adjusted rate ratio, 1.08; 95% CI, 0.98-1.20; P = .11; I2 = 18%), and fatal prostate cancer (5 studies; adjusted rate ratio, 1.02; 95% CI, 0.92-1.14; P = .68; I2 = 26%) were not significant (all cohort studies). Based on these data, a 0.6% (95% CI, 0.3%-1.2%) absolute increase in lifetime risk of prostate cancer associated with vasectomy and a population-attributable fraction of 0.5% (95% CI, 0.2%-0.9%) were calculated.
Conclusions and Relevance
This review found no association between vasectomy and high-grade, advanced-stage, or fatal prostate cancer. There was a weak association between vasectomy and any prostate cancer that was closer to the null with increasingly robust study design. This association is unlikely to be causal and should not preclude the use of vasectomy as a long-term contraceptive option.
Vasectomy is a highly efficacious long-term contraceptive method that involves a simple outpatient procedure under local anesthetic. It is less expensive and has a lower risk of complications compared with tubal ligation, the analogous female surgical sterilization procedure.1,2Quiz Ref ID Although 43 million women worldwide rely on their partner’s vasectomy for contraception,3,4 vasectomy is still considered to be underused in the United States, with only 8% to 12% of couples using vasectomy for birth control.4-6
In the late 1980s and early 1990s, several reports began to emerge of an epidemiologic association between vasectomy and the risk of prostate cancer.7-11 This finding ignited a controversy captured in numerous editorials, reviews, and original research articles supporting or refuting the association. Several meta-analyses of this association have been performed, but they did not bring closure to the debate. One older meta-analysis pooling 5 cohort studies and 17 case-control studies found a significant association between vasectomy and prostate cancer.12 Meanwhile, 2 more recent meta-analyses of cohort studies found no statistically significant association, although the CIs of pooled effect estimates precluded definitive conclusions.13-15 Moreover, none of these meta-analyses included a sufficient number of studies with a low risk of bias to analyze them as a separate subset, nor were the meta-analyses able to evaluate the association between vasectomy and risk of high-grade, advanced-stage, and fatal prostate cancer.
Recently, several large, high-quality analyses demonstrating either an association16,17 or no association18-20 between vasectomy and prostate cancer have reignited the controversy. Quiz Ref IDWith the aim of shedding some light onto a debate that is 3 decades old, we conducted a systematic review of the literature and performed a meta-analysis, with particular attention to study quality, to determine if there is an association between vasectomy and any prostate cancer, high-risk prostate cancer, advanced prostate cancer, and lethal prostate cancer.
Is there an association between vasectomy and a subsequent diagnosis of prostate cancer? More specifically, is vasectomy associated with a diagnosis of any prostate cancer, high-risk prostate cancer, advanced prostate cancer, and/or fatal prostate cancer? The Mayo Clinic Institutional Review Board waived the need for review of this study.
We included cohort, case-control, and cross-sectional studies. Case series lacking comparator groups were excluded. Other publications, including editorials, commentaries, review articles, and those not subject to peer review (ie, reports of data from Vital Statistics and dissertations or theses), were excluded. When there was more than 1 publication resulting from the same patient cohort, we selected a single representative study, with a preference for more contemporary publications and publications with a larger number of patients and more reliable methods of exposure and outcome ascertainment.
We reviewed studies reporting on men of any age who underwent vasectomy compared with those who did not undergo vasectomy. Vasectomy exposure was determined by administrative and clinical health records, survey results, and/or patient recall.
The primary outcome was any subsequent diagnosis of prostate cancer. Secondary outcomes included the diagnoses of high-grade prostate cancer (based on individual study definition, typically Gleason score ≥8), advanced prostate cancer (based on individual study definition, typically T3/4, N+, or M+), and fatal prostate cancer.
We used Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)21 and Meta-analysis of Observational Studies in Epidemiology (MOOSE)22 guidelines for reporting of this systematic review and meta-analysis.
The MEDLINE, EMBASE, Web of Science, and Scopus databases were searched by a professional librarian using the OvidSP platform for studies indexed from database inception to March 21, 2017. We used both subject headings and text word terms for vasectomy, ductus deferens, prostatic neoplasms, and related and exploded terms including MeSH terms in combination with key word searching. A full search strategy is presented in eAppendix 1 in the Supplement. No limitations were placed with respect to publication language or publication year. Following the literature search, all duplicates were excluded. References from review articles, commentaries, editorials, included studies, and conference publications of relevant medical societies were hand searched and cross-referenced to ensure completeness. Conference abstracts were included when sufficient information could be obtained from the corresponding authors.
Two of us performed study selection independently (B.B. and C.J.D.W.). Disagreements were resolved by consensus. Titles and abstracts were used to screen for initial study inclusion. Full-text review was used when abstracts were insufficient to determine if the study met inclusion or exclusion criteria. Studies were considered relevant if they reported an effect estimate for an association between vasectomy and any prostate cancer outcome, or provided sufficient data for this estimate to be calculated. One of us (B.B.) performed all data abstraction, including evaluation of study characteristics, risk of bias, and outcome measures, with independent verification performed by another one of us (C.J.D.W.).
We used the Newcastle-Ottawa Scale for assessment of risk of bias. This scale assesses risk of bias in the following 3 domains23: selection of the study groups, comparability of groups, and ascertainment of exposure and outcome.24 Studies with scores of less than 4 were considered to have a high risk of bias, those with scores of 4 to 6 an intermediate risk of bias, and those with scores of 7 or more a low risk of bias.
Measures of treatment effect varied among study designs. Among cross-sectional and case-control studies, we pooled odds ratios (ORs). In each case, we examined unadjusted and adjusted measures of effect separately. Among cohort studies, we first pooled ratio measures of effect (including hazard ratios [HRs], rate ratios [RRs], and ORs).25,26 Subsequently, we performed subgroup analysis using only studies that reported time-to-event data (HRs).
We identified heterogeneity using the Q test, estimated it using the DerSimonian-Laird method, and quantified it using I2 values.27 Furthermore, we used random-effects models for each of our analyses, given the identified clinical heterogeneity.
We assessed publication bias for outcomes with more than 10 included studies using funnel plots.
Meta-analysis was performed using Review Manager, version 5.3 (The Nordic Cochrane Centre, The Cochrane Collaboration). We used the inverse variance technique for pooling of measures of effect. Owing to the clinical heterogeneity inherent in our data, random-effects models were used for all meta-analyses. A sensitivity analysis was performed to determine if excluding any individual study altered our results.
We performed a priori subgroup analyses. First, as already mentioned, we examined cohort studies that used a time-to-event analysis separately. Second, we examined studies identified as having a low risk of bias. Third, we evaluated the studies that provided effect estimates stratified by time since vasectomy and age at which vasectomy was performed.
We also performed post hoc subgroup analyses. First, we examined cohort studies in which all patients were following a prostate-specific antigen (PSA) screening protocol or in which PSA screening was accounted for in multivariable models. Second, we examined cohort studies in which PSA screening was not performed or was very uncommon.
Finally, to facilitate interpretation of relative effect estimates for clinicians and policymakers, the pooled effect estimate from the meta-analysis was used to estimate the absolute risk increase, the number needed to harm, and the population attributable fraction. For these calculations, the lifetime risk of prostate cancer was considered to be 12.9%28 and the probability of undergoing vasectomy was 10%.4,5
Our literature search identified 433 unique references (Figure 1). After a full-text review of 115 manuscripts, 53 were selected for inclusion. The reasons for exclusion are summarized in Figure 1 and eAppendix 2 in the Supplement.
Among the 53 included studies, 16 were cohort studies (2 563 519 participants),11,16-20,29-38 33 were case-control studies (44 536 participants),8,9,39-68 and 4 were cross-sectional studies (12 098 221 participants).69-72 Although most studies were population- or registry-based analyses, several articles presented single- or multiple-institution analyses (Table).8,9,11,16-20,29-73 Duration of follow-up varied widely among cohort studies, with a minimum of 1.8 years and a maximum of 24 years. In general, the patients with and the patients without a history of vasectomy were similar in age among cohort studies. Among the case-control studies, the patients with and the patients without a diagnosis of cancer were generally similar in age, although, in some analyses, patients who received a diagnosis of prostate cancer were older.39,40 The risk-adjustment approach varied across studies (eTable 1 in the Supplement).
Assessment of risk of bias was performed according to study methods (eTable 2 in the Supplement). Nine of the cohort studies (56%) were assessed as having a low risk of bias.16-20,29,32,34,38 Exposure was ascertained by self-report or questionnaire results in several of these studies. Eight of 16 cohort studies (50%) accounted for PSA testing and/or health-seeking behavior.16-19,32,33,37,38 Most case-control studies (26 [79%]) were considered to have a moderate to high risk of bias.8,9,39,40,42,43,45-48,51,52,54-56,58-68 Although cases were typically well defined, the use of hospital-based instead of community-based controls was considered a potential source of bias among many case-control studies. Ascertainment of exposure was also a common source of bias among case-control studies. All 4 cross-sectional studies were thought to have a moderate to high risk of bias.69-72
To assess the association between vasectomy and any diagnosis of prostate cancer, we first pooled results from cohort studies (Figure 2). Among the 13 of 16 cohort studies reporting on 2 563 519 patients that used adjusted measures of effect,11,16-20,29,31-33,35-37 we found a small but statistically significant increased risk of any diagnosis of prostate cancer among patients with a history of vasectomy (adjusted RR, 1.08; 95% CI, 1.02-1.14; P = .006; I2 = 63%) (Figure 2A). A similar outcome was observed among cohort studies using a time-to-event analysis (9 studies; adjusted HR, 1.09; 95% CI, 1.03-1.15; P = .004; I2 = 70%) (eFigure 1 in the Supplement).11,16-20,31,32,35 The effect estimate among studies deemed at low risk of bias remained statistically significant but was closer to the null (7 studies; adjusted RR, 1.05; 95% CI, 1.02-1.09; P < .001; I2 = 9%) (Figure 2B).16-20,29,32 Results were not meaningfully different when abstracts were excluded (eTable 3 in the Supplement).
Among cohort studies in which all patients explicitly underwent PSA screening (as in the screening arm of the PLCO [Prostate, Lung, Colorectal and Ovarian] Cancer Screening trial)17 or in which PSA screening was accounted for in multivariable models, the association between vasectomy and prostate cancer was consistent with our overall findings (6 studies; adjusted RR, 1.06; 95% CI, 1.02-1.09; P < .001; I2 = 16%) (eFigure 2 in the Supplement).16,17,19,32,33,37 Furthermore, among studies that reported on populations in which PSA screening was not performed or very uncommonly performed, we observed no association between vasectomy and prostate cancer, although this finding was limited by few studies and significant heterogeneity (2 studies; adjusted RR, 1.26; 95% CI, 0.51-3.07; P = .62; I2 = 84%).10,36
We then separately assessed the association between vasectomy and any diagnosis of prostate cancer among case-control studies (Figure 3). Meta-analysis of studies reporting adjusted ORs demonstrated a statistically significant association between vasectomy and prostate cancer (17 studies; adjusted OR, 1.31; 95% CI, 1.12-1.53; P < .001; I2 = 66%) (Figure 3A).9,39-44,48,49,51,53,56-58,61,63,73 When we restricted analysis to 6 studies deemed to have low risk of bias, there was no significant association (adjusted OR, 1.06; 95% CI, 0.88-1.29; P = .54; I2 = 37%) (Figure 3B).41,44,49,53,57,73
There were no cross-sectional studies reporting adjusted measures of effect. Analyses of cohort, case-control, and cross-sectional studies reporting unadjusted measures of effect were performed for completeness and are summarized in eAppendix 3 and eFigure 3 in the Supplement. These results did not meaningfully alter conclusions. In the sensitivity analysis, excluding any individual study did not alter the results in each of the respective meta-analyses.
Among cohort studies, there was no statistically significant association noted between vasectomy and the diagnosis of high-grade prostate cancer (6 studies; adjusted HR, 1.03; 95% CI, 0.89-1.21; P = .67; I2 = 55%),16,18-20,31,38 advanced prostate cancer (6 studies; adjusted HR, 1.08; 95% CI, 0.98-1.20; P = .11; I2 = 18%),16,18-20,31,38 or fatal prostate cancer (5 studies; adjusted HR, 1.02; 95% CI, 0.92-1.14; P = .68; I2 = 26%) (Figure 4).16,18-20,38
Five case-control studies assessed the association between vasectomy and prostate cancer, stratified by stage.41-44,46 The varied definitions of advanced prostate cancer in these studies precluded meta-analysis of these results. Qualitatively, there was a greater association between vasectomy and low-risk prostate cancer than between vasectomy and advanced disease among these studies. Meanwhile, a single case-control study assessed the association between vasectomy and fatal prostate cancer and found no association (unadjusted OR, 1.3; 95% CI, 0.85-1.9).45 Similarly, a single cross-sectional study examined the association between vasectomy and high-grade prostate cancer and found no association (unadjusted OR, 0.69; 95% CI, 0.37-1.27).69
Six cohort studies,11,16,20,29-31 15 case-control studies,8,9,40-44,46-48,50-52,73 and 1 cross-sectional study70 reported on the association of time since vasectomy with the development of prostate cancer (eTable 4 in the Supplement). Of these, 1 cohort study11 and 6 case-control studies8,9,40,46-48 suggested a stronger association between vasectomy and prostate cancer with increasing time since vasectomy, although in several instances the CIs of the individual strata included unity and formal tests for trend were only performed in 3 studies.8,9,48 The remainder of these studies did not find any such biological gradient associated with time since vasectomy.16,20,29-31,41-44,49-52,70,73
Four cohort studies16,20,30,31 and 11 case-control studies40-44,46,48-51,73 reported effect estimates stratified by age at vasectomy (eTable 5 in the Supplement). Of these, 1 cohort study20 and 4 case-control studies42,46,48,49 found that the association between vasectomy and prostate cancer was stronger among men who had their vasectomy performed at a younger age. Conversely, 1 cohort study31 and 1 case-control study40 found that men who were older when they had their vasectomy were at greater risk for incident prostate cancer. The remainder of these studies did not find any association of age at vasectomy with development of prostate cancer.16,30,41,43,44,50,51,73 Effect estimates for time since vasectomy and age at vasectomy could not be pooled owing to the varying types of effect estimates and time intervals used by the individual studies.
We assessed publication bias using funnel plots comparing effect size and measure of precision of the effect size among case-control studies reporting unadjusted and adjusted measures of effect and cohort studies reporting adjusted measures of effect (eFigure 4 in the Supplement). There was potential publication bias among cohort studies, with a relative paucity of small studies demonstrating a large increased risk, and among case-control studies reporting adjusted measures of effect, with a relative paucity of small studies demonstrating no increased risk.
Using the pooled effect estimate from the meta-analysis of cohort studies with a low risk of bias, we found that the absolute increase in lifetime risk of prostate cancer is estimated to be 0.6% (95% CI, 0.3%-1.2%), that the number needed to harm (assuming a causal association, the number of men who would need to undergo vasectomy to result in 1 incident case of prostate cancer) is estimated to be 156, and that the population attributable fraction (assuming a causal association, the proportion the lifetime risk of prostate cancer would be reduced if no vasectomies were performed) is estimated to be 0.5% (95% CI, 0.2%-0.9%) (eAppendix 4 in the Supplement).
To date, the potential for bias in the studies on the association between vasectomy and prostate cancer has been a major focus of criticism of this body of literature. Accordingly, we found that the effect estimates of the association between vasectomy and prostate cancer were increasingly closer to the null when examining studies with increasingly robust study design and study quality. Quiz Ref IDIn our meta-analysis of cohort studies with a low risk of bias, we found a 5% increase in the risk of incident prostate cancer with vasectomy. This result was not driven by any single study alone. Meanwhile, the associations between vasectomy and high-grade, advanced, and fatal prostate cancer were not statistically significant, although the point estimates were similar to those for total diagnoses of prostate cancer. If assuming causality, for the individual patient, the effect estimate for overall prostate cancer corresponds to a 0.6% absolute increase in lifetime risk of incident prostate cancer, or a number needed to harm of 156. At the population level, only 0.5% of prostate cancers are estimated to be associated with vasectomy. It is questionable whether such a small increased risk is important to the public.
Although meta-analyses have been previously performed, an analysis of cohort studies with a low risk of bias has only recently been possible, with 6 of 7 such studies reported within the last 3 years.16-20,32 One meta-analysis published in 2002 pooled data from 5 cohort and 17 case-control studies and found a significant association between vasectomy and prostate cancer.12 Meanwhile, 2 more recent meta-analyses combined data from 9 cohort studies and found no significant association but had wide CIs.14,15 Another meta-analysis combined data from 10 cohort studies and found no association, but the lower limit of its CI barely crossed 1 (RR, 1.11; 95% CI, 0.98-1.27).13 Our meta-analysis, which included 53 studies, was able to separately analyze case-control, cross-sectional, and cohort studies and was well powered and achieved effect estimates with narrow CIs.
Quiz Ref IDIn our assessment, most of the 33 case-control studies, all 4 of the cross-sectional studies, and almost half of the 16 cohort studies had a moderate to high risk of bias. Accordingly, the point estimates for association between vasectomy and prostate cancer were furthest from the null among pooled analyses of case-control and cross-sectional studies. Publication bias might also have contributed, in part, based on funnel plots. In contrast, the observed association was smaller in magnitude in the pooled analysis of cohort studies, and even smaller when restricted to studies (cohort and case-control designs) deemed as having a low risk of bias. In addition, despite methodological rigor, failure to account for differential use of PSA screening may bias results because patients who have undergone vasectomy are more likely to undergo PSA screening and thus receive a diagnosis of prostate cancer.33 In our study, the association between vasectomy and prostate cancer held in a subset analysis restricted to studies that accounted for PSA screening. Our analysis illustrates the susceptibility of observational studies to bias and highlights the importance of meticulous study design.
More important, simply because a statistically significant association was detected, one cannot confirm with certainty that a causal association exists. Owing to the observational nature of pooled studies, residual unmeasured bias is still possible. Residual detection bias remains an ongoing concern, even though several studies accounted for serum PSA screening and/or contact with the health care system. A recent article by Tangen et al32 illustrated that detection bias influences the evaluation of several risk factors that have been described as associated with decreased or increased prostate cancer risk. The only way to address this source of bias would be a trial randomizing men to undergo vasectomy vs no vasectomy. However, a sufficiently powered trial with long enough follow-up would be neither practical nor ethical. Therefore, the present meta-analysis likely approaches the highest level of clinical evidence reasonably attainable in evaluating the association of vasectomy with risk of prostate cancer.
If applying the criteria of Hill,74 which have been widely accepted as aiding in making causal inferences, one cannot make a strong argument for a causal association between vasectomy and prostate cancer (eAppendix 5 in the Supplement). Moreover, the case for biological plausibility is tenuous. Although hormonal imbalances,75,76 immunologic effects,77 and cell proliferative changes78 have been suggested to play a role, the exact mechanisms remain to be described in animal models.76,79 Although 1 study found that serum testosterone levels are elevated in men who underwent a vasectomy more than 20 years ago relative to men who did not undergo a vasectomy,75 most studies have shown no changes in testosterone levels following vasectomy.76,79 Moreover, there is no established association between elevated serum testosterone level and risk of prostate cancer.80 Although vasectomy may lead to the development of antisperm antibodies,81 there is no evidence that these antibodies or the subsequent formation of immune complexes leads to prostate cancer.79 One study found increased cell proliferation 7 days after vasectomy in the ductal system of the rat ventral prostate,78 although the mechanism remains unclear.
On the other hand, the benefits of vasectomy as a method of contraception must be considered. More than 99% of women who have ever been sexually active have used a form of contraception at some point.82 However, according to the most recent data, 45% of pregnancies in 2011 in the United States were unintended83; 41% of these pregnancies occurred among women who used contraception inconsistently.82 Two-thirds of women using contraception rely on nonpermanent methods, many of which have a higher failure rate with typical use vs perfect use.84 Meanwhile, 25% of women rely on surgical sterilization for contraception and 8% rely on male surgical sterilization. Given the lower costs and lower risk of complications for vasectomy compared with tubal ligation,1,2 it is clear that vasectomy is underused and should be offered more routinely to couples seeking a long-term method of contraception.
There are several strengths to this study, including its size, its comprehensive search strategy in all languages, its careful review for study inclusion, its thorough assessment of study quality, and its use of a priori secondary analyses. This study is the first, to our knowledge, to separately evaluate studies with a low risk of bias and to evaluate high-grade, advanced, and fatal prostate cancers as secondary outcomes. Furthermore, we used relative effect estimates to calculate absolute effect estimates, which are more useful and more readily interpretable for clinicians, policymakers, and patients.85
There are also limitations to our study. Quiz Ref IDFirst, this meta-analysis is based on observational data because randomized trials are neither presently available nor likely to be performed in the future. As such, the unmeasured biases present in the individual studies must be taken into account. Our analytic approach addressed this issue in part by separately evaluating cross-sectional, case-control, and cohort studies and, additionally, separately evaluating the cohort and case-control studies with a low risk of bias. Second, publication bias cannot be ruled out. On the other hand, publication bias would have had outcomes in opposite directions in case-control and cohort meta-analyses, yet these analyses were consistent with each other. Finally, our analysis cannot definitively prove or disprove causality. However, as already outlined, a strong argument for a causal association between vasectomy and prostate cancer does not exist based on our data and on other existing literature.
Although it is tempting to consider potential avenues of further research on the link between vasectomy and prostate cancer, there are many other research topics that warrant greater priority in a system with finite health care research resources. We have demonstrated that any risk, if present, is sufficiently small that it is unlikely to be of clinical importance. We believe that this meta-analysis, drawing on 3 decades of epidemiologic literature, provides sufficiently robust data to inform clinical care and supports the current guidelines of the American Urological Association.1
Our meta-analysis found a weak association between vasectomy and the risk of prostate cancer among cohort studies with a low risk of bias and a similar but nonsignificant association among case-control studies with a low risk of bias. There was a similar nonsignificant association between vasectomy and high-grade, advanced-stage, or fatal prostate cancer. The association between vasectomy and prostate cancer was stronger when studies with moderate to high risk of bias were included. If assuming a causal association, which is unlikely based on our data and other existing literature, vasectomy would confer only a 5% relative increase or a 0.6% absolute increase in lifetime risk of prostate cancer, and would be responsible for only 0.5% of cases of prostate cancer in the population. Therefore, although patients should be appropriately counseled, concerns about the risk of prostate cancer should not preclude clinicians from offering vasectomy to couples seeking long-term contraception.
Corresponding Author: R. Jeffrey Karnes, MD, Department of Urology, Mayo Clinic, 200 First St SW, Rochester, MN 55905 (firstname.lastname@example.org).
Accepted for Publication: May 7, 2017.
Published Online: July 17, 2017. doi:10.1001/jamainternmed.2017.2791
Author Contributions: Drs Bhindi and Wallis contributed equally, should be considered co–first authors, had full access to all the data in the study, and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Bhindi, Wallis, Fleshner, Karnes.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Bhindi, Fleshner, Wallis.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Bhindi, Wallis.
Administrative, technical, or material support: Farrell, Fleshner.
Study supervision: Bhindi, Fleshner, Boorjian, Karnes.
Conflict of Interest Disclosures: None reported.