Vasectomy and Prostate Cancer Risk: A 38-Year Nationwide Cohort Study

Abstract

Background

A man’s risk of prostate cancer has been linked to his prior reproductive history, with low sperm quality, low ejaculation frequency, and a low number of offspring being associated with increased prostate cancer risk. It is, however, highly controversial whether vasectomy, a common sterilization procedure for men, influences prostate cancer risk.

Methods

We established a cohort of all Danish men (born between 1937 and 1996) and linked information on vasectomy, doctor visits, socioeconomic factors, and cancer from nationwide registries using unique personal identification numbers. Incidence risk ratios for prostate cancer by time since vasectomy and age at vasectomy during the follow-up were estimated using log-linear Poisson regression.

Results

Overall, 26 238 cases of prostate cancer occurred among 2 150 162 Danish men during 53.4 million person-years of follow-up. Overall, vasectomized men had an increased risk of prostate cancer compared with nonvasectomized men (relative risk = 1.15, 95% confidence interval = 1.10 to 1.20). The increased risk of prostate cancer following vasectomy persisted for at least 30 years after the procedure and was observed regardless of age at vasectomy and cancer stage at diagnosis. Adjustment for the number of visits to the doctor and socioeconomic factors did not explain the association.

Conclusions

Vasectomy is associated with a statistically significantly increased long-term risk of prostate cancer. The absolute increased risk following vasectomy is nevertheless small, but our finding supports a relationship between reproductive factors and prostate cancer risk.

Male reproductive factors are suggested to influence the risk of prostate cancer later in life, as cohort studies have found statistically significantly associations between having a low ejaculation frequency (1), a low sperm quality (2–5), or a low number of offspring (6) and a subsequent increased risk of prostate cancer. However, studies investigating the association between vasectomy and prostate cancer have found inconsistent results (7–11). Such studies may have been challenged by short length of follow-up (7,9), limited statistical power (7–11), and imprecise classification of vasectomy (10). To overcome these restrictions, we took advantage of the unique Danish national health registries to establish a prospective nationwide cohort of 2 150 162 Danish men whom we followed for 53.4 million person-years of follow-up. The follow-up included information on vasectomies and incident cases of prostate cancer as well as detailed individual-level socioeconomic and health registry data.

Methods

Population Cohort

We established a population-based cohort of all Danish men based on information from the Danish Civil Registration System (CRS). All residents in Denmark who were alive April 1, 1968, or born since that date have been given a person-identifiable number, the CRS number. This number is used as a person identifier in all national registries and enables linkages between registries.

The cohort included all Danish men born between January 1, 1937, and December 31, 1996. Each man was followed from January 1, 1977, or from his 18th birthday, whichever came later, until prostate cancer, other cancer, death, emigration, or December 31, 2014.

Study Variables

Information on vasectomy procedures was obtained from the National Patient Register, which contains information on all surgical procedures conducted in Danish hospitals since 1977. From January 1, 1977, to December 31, 1980, a vasectomy procedure was classified based on operation code “vasectomy”; from January 1, 1981, to December 31, 1995, by the operation code “bilateral vasectomy”; and from January 1, 1996, by the operation code “bilateral vasectomy or ligature of vas deferens”. Vasectomies included in the study were restricted to vasectomies performed for sterilization purposes as defined by diagnostic code “Prophylactic sterilization” (International Classification of Disease [ICD]-8 code Y43.99) from January 1, 1977, to December 31, 1993; and by diagnostic code “Sterilization. Admission for interruption of vasa deferentia” (ICD-10 code DZ30.2) from January 1, 1994.

Information on diagnosis of incident prostate cancer and all other cancers (excluding nonmelanoma skin cancer) was acquired from the Danish Cancer Registry, which contains information on all cancer diagnoses in Denmark since 1943 and is considered close to complete (12). The Cancer Registry also holds information on cancer stage at diagnosis, based on TNM classification, from 2004. Prostate cancer was categorized as either low-stage (tumor, node, metastasis [TNM] stage was T < T2b and N = N0 and M = M0) or intemediate- to advanced-stage (TNM was ≥T2b or N≥N1 or M = M1), in accordance with current risk group classification (13). For further analyses, prostate cancer was grouped into advanced or nonadvanced prostate cancer based on whether the tumor was metastatic and/or extracapsular (stage ≥ T3, or N+, or M+), or not (stage <T3 and N0, M0). Other cancers were defined as smoking and/or alcohol-related if more than approximately 40% of the attributable risk fraction was due to smoking and/or alcohol (14,15) and included lung cancer, oral cancer, laryngeal cancer, pharyngeal cancer, esophageal cancer, urinary tract cancers, and bladder cancer.

Information on number of visits to the general practitioner was acquired from the National Health Insurance Service Registry. Information was available from 1990 and classified as a time-varying variable with categories of 0–1, 2–4, 5–9, 10–14, or at least 15 visits to the general practitioner within the last 5 years, excluding the current year.

Fatherhood status was defined by linkage to offspring in the CRS and categorized as 0, 1, 2, 3, 4, or 5+ children. Additionally, we categorized marital status as married, divorced, widowed, or unmarried from information in the CRS.

From Statistics Denmark, we acquired individual-level socioeconomic data on educational attainment, employment status, disposable household income, and urbanicity to address confounders potentially associated with vasectomy and prostate cancer incidence. Disposable household income was grouped in 20% percentiles according to the national 5-year disposable household income distribution, and urbanicity was grouped according to municipality definition used by the Danish Ministry for Economic Affairs and the Interior. Information on level of educational attainment, employment status, and urbanicity was available throughout the follow-up, and information on disposable household income was available from January 1, 1990.

A checklist of the STrengthening the Reporting of OBservational studies in Epidemiology statement is reported in Supplementary Materials (available online). Because the study was based on deidentified information from the Danish national registries and study participants were never contacted, consent from the Danish research bioethics committees was not required. The study’s use of register data was covered by the approval from the Danish Data Protection Agency for registry-based studies conducted by Statens Serum Institut (approval No. 2015–57-0102).

Statistical Analyses

Incidence rate ratio (in the following termed relative risk) of prostate cancer or other cancer by time since and age at vasectomy was estimated by log-linear Poisson regression, in a competing- risks analysis, as described previously (16). All analyses were adjusted for effects of age in the categories younger than 40 years, 40–44 years, 45–49 years, 50–54 years, 55–59 years, 60–64 years, 65–69 years, 70–74 years, and 75 years and older, and period in the categories 1977–1984, 1985–1989, 1990–1994, 1995–1999, 2000–2004, 2005–2009, and 2010–2014. The analyses were furthermore adjusted for interaction between period and age for age categories younger than 50 years, 50–59 years, 60–69 years, and 70 years and older.

Additionally, we analyzed the relative risk of prostate cancer 10 years or more after vasectomy, stratified by potential confounding factors. For marital status, fatherhood (number of children), number of visits to the doctor, and attained period, we tested for interaction between strata, whereas for age of vasectomy and vasectomy period, we tested for a differential effect among strata of vasectomized men. For fatherhood and number of visits to the doctor, we tested for trend by examining the interaction effect including the variables as continuous variables, with number of visits to the doctor defined by the median number of visits within each group.

All tests were likelihood ratio tests, two-sided, and a P value of less than .05 was considered statistically significant. All analyses were performed using SAS procedure GENMOD (SAS Institute Inc, NC, USA). Graphics were generated in R version 3.3.1 (https://cran.r-project.org/bin/windows/base/old/3.3.1/) using package ggplot2.

Results

The study cohort consisted of 2 150 162 Danish men. The men were followed for 53.4 million person-years (on average 24.8 years of follow-up per man), and 139 550 men were registered during follow-up as having a vasectomy procedure for sterilization purposes (6.5%). The median age at vasectomy was 38.3 years (interquartile range [IQR] = 34.6–42.1 years), which increased slightly during follow-up (median age was 35.9 and 39.2 years in 1977–1989 and 1990–2014, respectively). During follow-up, 26 238 men developed prostate cancer (1.2%). Median age of incident prostate cancer diagnosis was 65.4 years (IQR = 61.1–69.1 years). Of the incident prostate cancer cases diagnosed from 2004 with information on cancer stage, 56.8% were low-stage prostate cancer and 43.2% were intemediate- to advanced-stage cancer. Table 1 shows follow-up time and number of prostate cancer events according to marital status, fatherhood (number of children), and socioeconomic variables by vasectomy status.

Table 1.

Prostate cancer events and person-years by vasectomy status, based on follow-up from 1977 to 2014*

Characteristic	Person-years (in 1000s)				Prostate cancer events
	Vasectomized				Vasectomized
	Yes (%)		No (%)		Yes (%)		No (%)
Overall	2216 (4.1)		51 190 (95.9)		2137 (8.1)		24 101 (91.9)
Marital status
Married	1804 (81.4)		23 043 (45.0)		1779 (83.3)		18 272 (75.8)
Divorced	272 (12.3)		3744 (7.3)		255 (11.9)		2881 (12.0)
Widowed	25 (1.1)		379 (0.7)		74 (3.5)		1090 (4.5)
Unmarried	115 (5.2)		24 024 (46.9)		28 (1.3)		1858 (7.7)
Fatherhood, No. of children
0	48 (2.2)		22 030 (43.0)		26 (1.2)		2916 (12.1)
1	189 (8.5)		8064 (15.8)		173 (8.1)		3420 (14.2)
2	1119 (50.5)		14 044 (27.4)		1077 (50.4)		10 899 (45.2)
3	632 (28.5)		5368 (10.5)		606 (28.4)		5092 (21.1)
4	179 (8.1)		1300 (2.5)		193 (9.0)		1346 (5.6)
≥5	49 (2.2)		385 (0.8)		62 (2.9)		428 (1.8)
Educational attainment
Primary schooling	1545 (69.7)		40 055 (78.2)		969 (45.3)		10 056 (41.7)
Short basic education	520 (23.5)		7846 (15.3)		874 (40.9)		9479 (39.3)
Higher education	150 (6.8)		3289 (6.4)		294 (13.8)		4566 (18.9)
Employment status
Job requiring basic skills	443 (20.0)		6117 (12.0)		516 (24.1)		4833 (20.1)
Job requiring intermediate skills	225 (10.2)		3230 (6.3)		350 (16.4)		3018 (12.5)
Job requiring advanced skills	242 (10.9)		4187 (8.2)		476 (22.3)		6264 (26.0)
Job with undescribed skill level	453 (20.4)		8139 (15.9)		409 (19.1)		4628 (19.2)
Business owner	118 (5.3)		2675 (5.2)		223 (10.4)		3725 (15.5)
Other†	734 (33.1)		26 841 (52.4)		163 (7.6)		1633 (6.8)
Disposable household income*
0–19th percentile	100 (5.0)		4036 (10.7)		60 (2.8)		1224 (5.1)
20–39th percentile	142 (7.2)		3975 (10.6)		80 (3.7)		1635 (6.8)
40–59th percentile	240 (12.1)		5437 (14.5)		219 (10.3)		2713 (11.3)
60–79th percentile	846 (42.8)		12 287 (32.7)		738 (34.6)		7235 (30.1)
80–100th percentile	648 (32.8)		11 891 (31.6)		1039 (48.6)		11 261 (46.8)
Urbanicity
Rural	1003 (45.3)		19 356 (37.8)		934 (43.7)		9731 (40.4)
Urban	1213 (54.7)		31 835 (62.2)		1203 (56.3)		14 370 (59.6)
No. of visits to the doctor*,‡
0–1	240 (12.1)		8522 (22.6)		96 (4.5)		1802 (7.5)
2–4	429 (21.7)		9444 (25.1)		282 (13.2)		3315 (13.8)
5–9	609 (30.8)		10 133 (26.9)		547 (25.6)		5727 (23.8)
10–14	330 (16.7)		4717 (12.5)		458 (21.4)		4674 (19.4)
≥15	367 (18.6)		4809 (12.8)		753 (35.3)		8569 (35.6)

*

Information on disposable household income and number of visits to the doctor is based on follow-up from January 1, 1990, to December 31, 2014.

†

Other employment status includes pensioners, unemployed, and individuals in educational programs.

‡

Number of visits to the doctor is defined by number of visits to the general practitioner within the 5 latest years excluding the current year.

Overall, we found vasectomized men had an increased relative risk of prostate cancer of 1.15 (95% confidence interval [CI] = 1.10 to 1.20) compared with nonvasectomized men, whereas they had a diminished relative risk of all other cancers of 0.91 (95% CI = 0.89 to 0.93) (Table 2). The reduced risk of other cancers was partially explained by a substantially lower risk of smoking- and/or alcohol-related cancers.

Table 2.

Relative risk of cancer among vasectomized men compared with nonvasectomized men

Cancer type	Events		RR (95% CI)*	Adj. RR (95% CI)†	Adj. RR by time since vasectomy (95% CI)†‡
	Exposed	Unexposed			<10 years	≥10 years
Prostate cancer	2137	24 101	1.15 (1.10 to 1.20)	1.14 (1.09 to 1.19)	0.98 (0.82 to 1.16)	1.15 (1.10 to 1.21)
Advanced§	576	6775	1.11 (1.02 to 1.21)	1.10 (1.01 to 1.20)	0.89 (0.61 to 1.30)	1.11 (1.02 to 1.22)
Nonadvanced§	1436	15 231	1.17 (1.11 to 1.23)	1.16 (1.10 to 1.23)	0.94 (0.75 to 1.18)	1.18 (1.12 to 1.25)
Other cancers	7868	122 111	0.91 (0.89 to 0.93)	0.91 (0.89 to 0.93)	0.85 (0.81 to 0.89)	0.93 (0.90 to 0.95)
Smoking/alcohol-related ‖	2547	37 938	0.87 (0.84 to 0.91)	0.87 (0.83 to 0.90)	0.71 (0.64 to 0.79)	0.90 (0.86 to 0.94)
Nonsmoking/alcohol-related	5321	84 173	0.94 (0.91 to 0.96)	0.93 (0.91 to 0.96)	0.91 (0.86 to 0.96)	0.94 (0.91 to 0.97)

*

Adjusted for age, period, and interaction between age and period. CI = confidence interval; RR = relative risk.

†

Adjusted for age, period, educational attainment, employment status, urbanicity, and interaction between period and age, educational attainment, employment status, and urbanicity.

‡

Of the 2137 prostate cancer-exposed events, 127 were within 10 years from vasectomy, whereas 2010 were from 10 years after vasectomy. Of the 7868 other cancer-exposed events, 1773 were within 10 years from vasectomy, whereas 6095 were from 10 years after vasectomy.

§

Advanced prostate cancer was defined as metastatic and extracapsular prostate cancer (stage ≥T3, or N+, or M+) at diagnosis, and nonadvanced prostate cancer was defined as nonmetastatic and intracapsular prostate cancer (stage <T3 and N0, M0) at diagnosis. Information on stage was only available only from 2004, thus there are fewer events in these subcategories. Further sensitivity analysis on vasectomy and risk of prostate cancer with advanced prostate cancer stage restricted to stage no less than T3b, or N+, or M+ are presented in Supplementary Table 4 (available online).

‖

Composed of cancers of which more than approximately 40% of the attributable risk fraction is due to smoking and/or alcohol (ie, lung cancer, oral cancer, laryngeal cancer, pharyngeal cancer, esophageal cancer, urinary tract cancers, and bladder cancer).

Figure 1 shows the relative risk of prostate cancer according to time since vasectomy compared with nonvasectomized men, with the number of person-years and events attributable to each time category. The relative risk of prostate cancer was increased in the first year following vasectomy. However, this estimate was not statistically significantly and was based on only six exposed men diagnosed with prostate cancer. The six men had a median age of 49.5 years at vasectomy, a median number offspring of two, and a median number of visits to the doctor of seven, thereby being markedly older than other vasectomized men, but otherwise having similar characteristics. From 1 to 4 years after the procedure, we observed that the risk of prostate cancer was similar to the overall level of other cancer. However, from 10 years after vasectomy, we found that vasectomized men had an increased relative risk of prostate cancer of 1.15 (95% CI = 1.10 to 1.21) compared with nonvasectomized men (Table 2). We further investigated the risk of prostate cancer by age at vasectomy (Supplementary Figure 1, available online) and found a similar increased risk of prostate cancer after vasectomy for men in all age groups.

In Figure 2, we show the association between time since vasectomy and risk of prostate cancer by cancer stage at diagnosis. We divided prostate cancer by TNM stage at diagnosis into low- or intemediate- to advanced-stage prostate cancer. We found a similar association between time since vasectomy and risk of low-stage and intemediate- to advanced-stage tumors, except for the first year following vasectomy. Here, we found a statistically significantly increased risk of low-stage prostate cancers for vasectomized compared with nonvasectomized men (relative risk [RR] = 3.57, 95% CI = 1.60 to 7.96), whereas no cases of intermediate- to advanced-stage prostate cancers were found among vasectomized men within the first year after vasectomy. Furthermore, when restricting the analysis to metastatic and extracapsular prostate cancer, we still found a statistically significantly increased long-term prostate risk associated with vasectomy (RR = 1.11, 95% CI = 1.02 to 1.22), despite a lower number of cases (Table 2).

Figure 1.

Relative risk of prostate cancer by time since vasectomy compared with nonvasectomized men, based on follow-up from 1977 to 2014. Estimates were calculated using log-linear Poisson regression. The error bars represent 95% confidence intervals.

Figure 2.

Relative risk of low-stage prostate cancer and intemediate- to advanced-stage prostate cancer by time since vasectomy compared with nonvasectomized men, based on follow-up from 2004 to 2014. Estimates were calculated using log-linear Poisson regression. The error bars represent 95% confidence intervals. Adv = advanced; Int = intermediate.

To explore potential effects of reproductive factors, health-seeking behavior, and diagnostic practices on the association, we further investigated if the effect of vasectomy on prostate cancer risk was modified by marital status, fatherhood (number of children), number of visits to the doctor, age at vasectomy, vasectomy period, attained age, and attained period (Table 3) and found only the number of children modified the association between vasectomy and prostate cancer risk. The association was stronger with increasing number of children fathered (P_trend = .01), with the estimated relative risk of prostate cancer being 1.16 (95% CI = 1.06 to 1.26), 1.25 (95% CI = 1.08 to 1.46), and 1.31 (95% CI = 0.99 to 1.72) increased within groups of men with three, four, and five or more children, respectively.

Table 3.

Relative risk of prostate cancer 10 or more years after vasectomy compared with nonvasectomized men based on follow-up from 1977 to 2014

Stratification variable	Relative risk (95% CI)*		P
Marital status
Married	1.12 (1.06 to 1.17)		.91†
Divorced	1.11 (0.97 to 1.27)
Widowed	1.17 (0.92 to 1.49)
Unmarried	0.98 (0.62 to 1.53)
Fatherhood, No. of children
0	0.95 (0.63 to 1.43)		.01‡
1	1.04 (0.89 to 1.22)
2	1.08 (1.02 to 1.16)
3	1.16 (1.06 to 1.26)
4	1.25 (1.08 to 1.46)
≥5	1.31 (0.99 to 1.72)
Number of visits to the doctor§
0-1	1.11 (0.89 to 1.40)		.37‡
2–4	1.18 (1.04 to 1.34)
5–9	1.14 (1.04 to 1.25)
10–14	1.12 (1.01 to 1.23)
≥15	1.10 (1.01 to 1.18)
Attained period
1987–1994	1.10 (0.41 to 2.97)		.16†
1995–1999	0.83 (0.54 to 1.29)
2000–2004	1.09 (0.91 to 1.29)
2005–2009	1.10 (1.01 to 1.19)
2010–2014	1.21 (1.14 to 1.28)
Attained age, y
<55	1.33 (1.11 to 1.58)		.45†
55–59	1.12 (1.00 to 1.25)
60–64	1.15 (1.06 to 1.24)
≥65	1.16 (1.08 to 1.24)
Age at vasectomy, y
18–29	1.21 (0.96 to 1.53)		.92†
30–34	1.10 (0.99 to 1.23)
35–39	1.16 (1.08 to 1.25)
40–44	1.17 (1.07 to 1.27)
45–49	1.20 (1.06 to 1.37)
≥50	1.12 (0.89 to 1.42)
Vasectomy period
1977–1979	1.12 (1.01 to 1.23)		.30†
1980–1984	1.12 (1.04 to 1.22)
1985–1989	1.13 (1.03 to 1.24)
1990–1994	1.28 (1.12 to 1.46)
1995–1999	1.28 (1.13 to 1.45)
2000–	1.20 (0.92 to 1.55)

*

Adjusted for age, period, and interaction between age and period. CI = confidence interval.

†

For marital status, fatherhood, urbanicity, number of visits to doctor, attained age, and attained period, the P value represents a two-sided test for interaction.

‡

For age at vasectomy and vasectomy period, the P value represents a two-sided test for differential effect among vasectomized men.

§

Number of visits to the doctor is defined by number of visits to the general practitioner within the last 5 years excluding the current year. Information on number of visits to the doctor is based on follow-up from January 1, 1990, to December 31, 2014.

To estimate the effect of vasectomy on prostate cancer incidence, we calculated the population-attributable fraction of prostate cancer from vasectomy (Supplementary Box 1, available online). We estimated that approximately 326 incident prostate cancer cases (95% CI = 214 to 441) could be attributed to vasectomy during the follow-up. Moreover, one man was estimated to develop prostate cancer before age 75 years because of vasectomy for every 71 men vasectomized (95% CI = 52 to 107).

Finally, we performed sensitivity analyses to investigate the association between vasectomy and prostate cancer by exploring the effect of male sterilization performed in nonhospital specialist clinics (Supplementary Table 1, available online), the effect of different socioeconomic factors (Supplementary Figures 2 and 3, Supplementary Table 2, available online), tumor stage definition (Supplementary Table 3, available online), and birth cohort (Supplementary Table 4, available online). We found no indications of effect modification by place of sterilization, adjustment for socioeconomic factors, tumor stage definition, or birth cohort.

Discussion

The precise etiology of prostate cancer is unknown, but cohort studies suggest that reproductive factors influence disease development (1–6). Using Danish nationwide registries, we investigated the effect of vasectomy, a common sterilization procedure for men, on prostate cancer risk. We showed that vasectomy is associated with a long-term increased risk of prostate cancer, which manifests itself from 10 years after the procedure. The increased risk was not modified by age at vasectomy, vasectomy period, number of visits to the doctor, and socioeconomic factors, and was present regardless of tumor stage at diagnosis.

A recent meta-analysis on the association between vasectomy and prostate cancer reported a statistically significantly increased risk of prostate cancer following vasectomy (17). Of the 13 cohort studies included in the meta-analysis, 10 found vasectomy to be associated with an increased risk of prostate cancer, but only five of these found the association to be statistically significant, with their estimates being either similar to or more adverse than the estimate found in our study. The authors noted that the prostate cancer risk was lower among studies with stricter study design and that neither high-grade nor fatal prostate cancer was statistically significantly increased among vasectomized men and argued that these observations supported a null association. Nevertheless, their subanalysis of cohort studies accounting for prostate-specific antigen (PSA)-screening still found prostate cancer incidence statistically significantly increased following vasectomy. Furthermore, the similar findings of our study (26 238 events) and the strict meta-analysis of seven low-bias cohort studies (32 290 events) support the existence of a small increased prostate cancer risk following vasectomy. However, the meta-analysis counted only two studies with a follow-up period up to 20 years. With an average follow-up of 25 years, more than 26 000 incident cases of prostate cancer, and detailed information on socioeconomic status, health-care usage, and reproductive history, our study is by far the largest cohort study on vasectomy and prostate cancer to date.

There is no national registration of PSA testing for prostate cancer in Denmark, which could be associated both with vasectomy and prostate cancer incidence because men who are vasectomized might be more likely to be screened for prostate cancer. However, Denmark has historically had a low intensity of PSA screening compared with other countries (18) and routine PSA screening has never been recommended (19), which argues against a strong impact of a possibly higher rate of PSA screening among vasectomized men in Denmark. Nevertheless, the use of PSA screening seems to have increased from the mid-1990s, predominantly because of increased testing by general practitioners (20). Studying prostate cancer risk by time since vasectomy, we found a manifest increased prostate cancer risk from 10 years after vasectomy and also a tendency toward an increased risk of prostate cancer in the first year following the procedure. However, the analysis of prostate cancer risk by stage at diagnosis indicates detection bias as the driver of the increased risk the first year following vasectomy, because all prostate cancers detected in this interval were low-stage tumors, whereas the long-term association was similar for low- and intemediate- to advanced-stage tumors. Taken together, our finding that the long-term association is observed both for low-stage and intemediate- to advanced-stage prostate cancer argues against the possibility that our results could be explained by detection bias due to a possibly higher rate of PSA testing among vasectomized men.

Investigating the risk of prostate cancer and all other types of cancer among vasectomized men, we found that vasectomized men on average had a lower risk of other cancers compared with nonvasectomized men. Grouping other cancers based on whether they were primarily attributable to smoking and/or alcohol, we still found vasectomized men to have a reduced risk of nonprostate cancers, indicating that vasectomized men on average are healthier than the general population. However, studies have found smoking to be associated with an increased risk of advanced prostate cancer (21,22). The increased risk of prostate cancer observed following vasectomy therefore most likely reflects an underestimation of the true effect of vasectomy on prostate cancer risk, given that vasectomized men have a reduced risk of smoking- and/or alcohol-related cancers and an overall reduced cancer risk.

To further evaluate the association between vasectomy and long-term prostate cancer risk, we stratified the association by factors potentially explaining the relationship and additionally adjusted the association for socioeconomic factors. The association was not explained by differences in socioeconomic factors, doctor-visiting frequency, or attained period, which further argues against detection bias explaining the association. However, we found the association to be stronger for each additional child fathered. This finding is in line with currently emerging evidence on infertility and prostate cancer, finding infertility to be associated with an increased risk of prostate cancer (2–5). Thus, as nonvasectomized men include a higher proportion of infertile and subfertile men, given their reduced need for the procedure, a more unconfounded estimate of a vasectomy effect on prostate cancer risk would be found by focusing on men with high fertility levels. Focusing on the effect of vasectomy among men with five or more children, we found a 31% increase in long-term risk of prostate cancer, supporting a strong adverse effect of vasectomy on prostate cancer risk.

Other suggested prostate cancer risk factors, such as body mass index, physical activity, intake of protein from dairy sources, and family history of prostate cancer, could not be accounted for in our study. However, previous studies on vasectomy and prostate cancer that have been able to adjust for differences in these factors found no evidence of them having an important confounding effect (8,11), indicating a minimal role of these risk factors in the association between vasectomy and prostate cancer risk.

Cohort studies on reproductive factors and prostate cancer point to a protective effect of a man’s fecundity (ie, his ability to produce offspring) on his prostate cancer risk (1–6). Four separate studies that investigated prostate cancer risk in men who had undergone evaluation for infertility found a statistically significantlyassociation between infertility and increased prostate cancer risk (2–5), with the association being especially pronounced for male factor infertility (2,3,5) and high-grade prostate cancer (2). A nationwide cohort study investigating the effect of fatherhood on prostate cancer risk reported that each additional offspring was associated with a statistically significantlydecrease in prostate cancer risk (6). The underlying biology linking reproductive factors and prostate cancer risk has, however, been unclear, and the lack of a plausible mechanism has led some authors to dismiss the association as spurious (11,17). However, two of the recent findings on reproductive factors and prostate cancer risk point to a plausible biological framework for understanding the effect of vasectomy on prostate cancer risk (1,5). In an analysis of prospectively collected data from the Health Professionals Follow-up Study cohort, high ejaculation frequency both in early and mid-life was found to be associated with a statistically significantly decreased risk of prostate cancer (1), even when adjusted for multiple potential confounders. Likewise, in a cohort study of men evaluated for infertility and followed for up to 33 years, sperm cell concentration was found to be inversely proportional with prostate cancer risk in a dose-response manner (5). Together with our finding of a statistically significantly adverse effect of vasectomy, this substantiates the evidence of a protective effect of high fecundity on prostate cancer risk, with the evidence pointing to a protective effect of active testicular function. It can therefore be hypothesized that the malignant potential of the prostate is reduced from either the passage of testicular secretions through the prostate gland during ejaculation or systemically through a hormonal mechanism, with the latter being in line with findings of long-term hormonal imbalances and testicular interstitial fibrosis following vasectomy (11,23–27). Further research is nevertheless needed to investigate how testicular function might reduce the proliferative potential of the prostate.

Our finding has important global public health implications because prostate cancer is the most common male cancer, and vasectomy is a widespread form of contraception. Taking a conservative approach, not taking into account vasectomized men’s generally reduced cancer risk and the increased risk of prostate cancer among infertile or subfertile nonvasectomized men, we estimate that more than 300 cases of prostate cancer were attributable to vasectomy during our follow-up. In addition, we estimate that one man would develop prostate cancer because of vasectomy for every 71 men vasectomized. However, this extrapolation may not be suitable outside a Danish context, and because of the observational nature of our study, we cannot claim causality. Nevertheless, as the possibility of conducting a randomized experiment on the subject is nonexistent, our findings are important to help quantify a potential adverse effect of vasectomy. Furthermore, seen in relation to the recent finding of an increased breast cancer risk following oral contraceptive use in women (28), our findings underscore that the most common contraceptive methods used today have disadvantages and highlight the need for ongoing research on alternatives to, or modifications of, the contraceptive methods currently in use.

In summary, we found vasectomy to be associated with a long-term increased risk of prostate cancer across cancer stages and irrespective of socioeconomic status and health-care usage. The absolute increased risk of prostate cancer following vasectomy is nevertheless small and similar to the increased breast cancer risk in women following oral contraceptive use. Furthermore, taken together with other recent findings of an association between male reproductive factors and prostate cancer, our results suggest that active testicular function is protective against the development of prostate cancer.

Funding

This work was supported by Helsefonden (grant number 16-B-0257 to AH), Anita og Tage Therkildsens Fond (grant number 100039 to AH), and Fonden til Lægevidenskabens Fremme (grant number 16–145 to AH).

Notes

The funders did not have any role in the design of the study, the collection, analysis, and interpretation of the data; the writing of the manuscript; and the decision to submit the manuscript for publication. All researchers acted independently from the study sponsors in all aspects of the study.

All authors have completed the International Committee of Medical Journal Editors uniform disclosure form at www.icmje.org/coi_disclosure.pdf and declare no support from any organization for the submitted work, no financial relationships with any organizations that might have an interest in the submitted work in the previous 3 years, and no other relationships or activities that could appear to have influenced the submitted work.

Author contributions: AH conceived the study, contributed to the study design, classified registry data and performed statistical analysis, interpreted the study results, drafted the manuscript, and is the guarantor of the study. JW contributed to the study design, planned statistical analysis, oversaw the conduct of the statistical analysis, interpreted the study results, and revised the manuscript. MM contributed to the study design, interpreted the study results, and revised the manuscript. All authors had access to all the data and take full responsibility for the integrity of the data, the accuracy of the data analysis, the finished article, and the decision to submit the manuscript for publication.

The authors thank Annemette B. Kristensen for editorial assistance with the manuscript.