Objective: To evaluate the possible association between vasectomy and prostate cancer.
Design: Systematic review of the literature.
Patient(s): Fourteen original studies published between January 1985 and December 1996 that addressed the association between vasectomy and prostate cancer.
Main outcome measure(s): The strength of the association was estimated with the use of a meta-analysis (DerSimonian and Laird method). A sensitivity analysis was conducted to assess the impact of different sources of heterogeneity.
Result(s): Fourteen original papers were reviewed (5 cohort and 9 case-control studies). Relative risks ranged between 0.44 (95% confidence interval [CI] = 0.1-4.0) and 6.70 (95% CI = 2.1-21.6). The overall relative risk (DerSimonian and Laird estimate) was 1.23 (95% CI = 1.01-1.49). The sensitivity analysis showed that this measure was very sensitive to the study base, the type of design used, and the possibility of bias. Further, the funnel plot demonstrated the probable existence of publication bias.
Conclusion(s): No causal association was found between vasectomy and prostate cancer. Individuals who have undergone vasectomy are not at high risk for the development of prostate cancer.
PIP: The absence of a plausible biologic model, methodologic problems, and lack of consistency between the results of different studies have created skepticism about any association between vasectomy and prostate cancer. To clarify further the possibility of such a link, a systematic review of the empirical literature published in 1988-96 was conducted. 14 such studies (5 cohort and 9 case-control) were identified, 11 of which found an excess risk of prostate cancer in men who had undergone vasectomy. Relative risks ranged from 0.44 (95% confidence interval (CI), 0.1-4.0) and 6.70 (95% CI, 2.1-21.6). The risk was statistically significant in 6 studies. The weighted relative risk obtained using the age-adjusted results of the individual studies was 1.23 (95% CI, 1.01-1.49). However, both the statistical tests and the qualitative analysis detected heterogeneity between the studies. Possible sources of this heterogeneity include type of design, study base, presence of detection bias, and inadequate selection of controls. Moreover, the sensitivity analysis indicated that the detected effect depends to a great extent on studies that are more vulnerable to bias (i.e., case-control and hospital-based studies) and those that have internal validity problems. Further, the funnel plot demonstrated the possible existence of publication bias. Finally, when the relative risk was recalculated to exclude early stage tumors (located by active detection in vasectomy cases), the previously found association between vasectomy and prostate cancer disappeared. It is concluded that the available empiric evidence is of low quality because of multiple sources of bias that favor the overestimation of the effect of vasectomy on prostate cancer risk. These validity problems, along with the lack of a biologic model to explain the association, strongly suggest the association is not causal.