Human papillomavirus (HPV) is the most common sexually transmitted infection in the United States.(1) Most infections will become undetectable within 1–2 years;(2) however, persistent infection may lead to HPV-attributable disease, including anogenital warts and penile, anal, and oropharyngeal cancers in males.(3, 4) In 2013–2016, approximately 23% of males in the United States had at least one high-risk, or oncogenic, penile HPV type.(1)

In the United States, HPV vaccination is routinely recommended for all adolescents at age 11–12 years with catch-up through age 26 years. Since the end of 2016, only a 9-valent vaccine has been available in the United States; this vaccine protects against the two non-high-risk types that cause nearly all cases of anogenital warts and seven high-risk types that cause most HPV-attributable cancers.(5)

In addition to vaccination, the role of male circumcision (hereafter referred to as circumcision) as an HPV prevention tool has been explored.(6–8) Circumcision is a recognized prevention strategy for other viral sexually transmitted infections, including human immunodeficiency virus (HIV) and herpes simplex II.(9) Clinical trials initially designed to determine the efficacy of circumcision against incident HIV infection have evaluated efficacy against HPV acquisition as a secondary objective.(6–8) These trials, conducted in Africa among adult men, showed rates of incident high-risk HPV (HR-HPV) infection were 30–40% lower among males who received adult male circumcision than those who remained uncircumcised. Observational studies exploring the association between circumcision and prevalent HPV infection have included persons likely circumcised at a variety of ages (e.g., infancy, adulthood) and have been less consistent.(10–12) To further explore this association in the United States, we assessed HR-HPV penile prevalence by circumcision status and evaluated circumcision as an effect modifier on the association between sexual activity and HR-HPV among sexually experienced males aged 18–59 years from a nationally representative sample.

The overall prevalence of circumcision in 2013–2016 was 77.7% among US adult males. (Table 1) Prevalence of circumcision differed significantly by age group, race/ethnicity, country of birth, poverty status, insurance status, and number of lifetime partners (all P ≤ 0.01); the largest absolute differences were seen by race/ethnicity (non-Hispanic White, 90.6%; other non-Hispanic race/ethnicity, 81.6%; non-Hispanic Black, 75.2%; Other Hispanic, 47.7%; non-Hispanic Asian, 44.1%; Mexican American, 37.3%) and country of birth (United States, 86.0%; other, 39.1%).

HR-HPV penile prevalence was 24.5% overall. Prevalence was significantly associated with race/ethnicity, country of birth, poverty status, insurance status, number of lifetime partners, and new partner in the past year (all P ≤ 0.01). (Table 1) Large absolute differences were seen by race/ethnicity (range: 13.6%–35.7%) and number of lifetime partners (range: 7.8%–38.8%).

HR-HPV penile prevalence was higher among circumcised (25.6%, 95% CI: 22.9–28.5) than uncircumcised (20.5%, 95% CI: 17.1–24.3) males prior to adjustment for sociodemographic or behavioral characteristics (PR: 1.25, 95% CI: 1.05–1.49). (Table 2) However, this association was attenuated after adjusting for race/ethnicity, country of birth, or number of lifetime partners alone or adjusting for either of the minimal sufficient sets (set 1: number of lifetime partners and new partner in the past year, aPR (95% CI) = 1.10 (0.92–1.32); set 2: number of lifetime partners, age group, race/ethnicity, insurance status, and country of birth, aPR (95% CI) = 1.07 (0.88–1.30)). When considering HPV 16/18 and non-HR-HPV prevalence, point estimates were further from the null for HPV 16/18 and closer to the null for non-HR-HPV. However, after adjustment for race/ethnicity, lifetime partners, or either minimal sufficient set, prevalence did not statistically significantly differ between circumcised and uncircumcised males (data not shown).

To assess circumcision as a modifier of the effect of HPV acquisition through sex, HR-HPV penile prevalence was estimated by number of lifetime partners, stratified by circumcision status (Figure 2 and Table 3). Among circumcised and uncircumcised males, point estimates for HR-HPV prevalence were higher with increasing number of lifetime partners (Figure 2). Using prevalence among uncircumcised males reporting 1–2 lifetime sex partners as the reference, unadjusted PR point estimates were higher with higher numbers of lifetime partners from 3–5 to ≥16 partners for both circumcised (range of PRs: 1.31–4.04) and uncircumcised (range of PRs: 1.84–3.14) males. Prevalence was significantly higher among both circumcised and uncircumcised males with 6–9, 10–15, and ≥16 lifetime partners than uncircumcised males with 1–2 lifetime partners. Adjusted PRs followed a similar pattern to unadjusted results, but they were somewhat attenuated.

Among persons born in the United States, overall, HR-HPV prevalence was comparable between circumcised (26.2%, 95% CI: 23.2–29.4) and uncircumcised males (25.0%, 95% CI: 20.3–30.4; aPR: 1.02, 95% CI: 0.80–1.29). Results from the effect modification analysis were similar to the main analysis (Supplemental Table 3).

In 2013–2016 in the United States, HR-HPV penile prevalence was not statistically associated with circumcision status among adult males after adjusting for number of lifetime partners and either new partner in the past year or demographic characteristics. Prior to adjustment, HR-HPV prevalence was 25% higher among circumcised males than uncircumcised males. However, there were substantial differences in prevalence of circumcision by race/ethnicity and country of birth, and we provide evidence these variables may have confounded the association between circumcision status and HR-HPV positivity. We also provide evidence circumcision may be an effect modifier on the relationship between number of lifetime partners and HR-HPV positivity. Among those with fewer lifetime partners, HR-HPV prevalence was lower among circumcised than uncircumcised males, but for those with more lifetime partners, prevalence was higher among circumcised males. However, HR-HPV prevalence was substantially higher in males with greater numbers of lifetime partners regardless of circumcision status.

In the United States, circumcision is most commonly performed as an elective procedure on neonates. Prevalence of circumcision has been declining among more recent birth cohorts but remains higher in the United States than any other developed country.(18, 19) Within the United States, prevalence varies substantially by race/ethnicity, socioeconomic status, and insurance status; this was seen in our study as well as others.(14, 19, 20) Although circumcision has always ultimately been decided by parental preference, guidance from the American Academy of Pediatrics (AAP) has varied since the 1970s. Evidence and acknowledgment of potential medical benefits of circumcision has grown. In the 1970s, AAP concluded there was no medical benefit of circumcision; in 1989, there may be potential benefit; and in 1999, there was potential benefit.(21) In 2012, AAP convened a multiorganization task force to re-evaluate evidence regarding circumcision. Their evaluation determined there were health benefits for circumcision (e.g., HIV prevention) that outweighed risks associated with the procedure (e.g., bleeding, infection); parents had the right to factual information, physician counseling, and the ability to weigh the health benefits and risks relative to their own personal and cultural preferences; and there was enough evidence of benefit to justify the procedure should be covered by third-party reimbursement.(9) However, the benefits of circumcision were not deemed great enough to warrant a routine recommendation of circumcision for all male newborns.(22)

Multiple clinical trials have shown circumcision reduces risk of HR-HPV acquisition by 30–40% and increases the rate of clearance of infection by 35–75%.(6, 8, 23) Our analysis of circumcision as an effect modifier found HR-HPV prevalence was 33% lower (nonsignificant) in circumcised than uncircumcised males among those with 1‒2 lifetime sex partners. However, HR-HPV prevalence was higher among circumcised than uncircumcised males for those with >5 lifetime partners. There are several reasons our results may have differed from the results of these clinical trials. First, the trials primarily focused on HPV acquisition rather than prevalent infection. Second, in the trials, circumcision was performed on adult males as an intervention. Third, HPV prevalence varies by sampling site in the anogenital area, particularly among males.(24) In the clinical trials, the glans was sampled, whereas the entire penis was sampled in NHANES. One trial that sampled the glans and shaft separately found that circumcision was protective against HR-HPV at the glans, but not the shaft.(6)

Several systematic reviews have concluded circumcision is protective against prevalent HPV infection.(25–27) A recent review demonstrated that HR-HPV prevalence was statistically significantly lower among circumcised males than uncircumcised males when the glans was the anatomic site sampled. However, when the sample was from the shaft only or a combination of sites (e.g., glans and shaft in one sample), there is more ambiguity in the association between circumcision status and HR-HPV prevalence among studies included in the review. Among studies that reported an odds ratio, the odds ratios were less than one but had confidence intervals that included the null, suggesting HR-HPV prevalence was non-statistically significantly lower among circumcised than uncircumcised males. However, among studies that reported prevalence ratios, there was no difference in prevalence between circumcised and uncircumcised males, similar to our findings. (27) The strength and consistency of the association between circumcision status and HR-HPV prevalence at the glans may be due to differences in prevalence at that site by circumcision status. There is evidence that HR-HPV prevalence is higher at the glans/corona among uncircumcised than circumcised males, while prevalence at the shaft is similar.(12) About 60% of penile cancers are caused by HPV, and the glans is the most common subsite of penile cancer.(28) Lower HR-HPV prevalence at the glans among circumcised males may explain the lower rates of penile cancer among circumcised compared to uncircumcised males.

The biological mechanism for circumcision as an HPV prevention tool is unclear. It is theorized that HPV is less likely to infect the more keratinized glans of a circumcised penis than the less keratinized epithelium of the preputial space. However, the difference in thickness of keratin has not been conclusively shown. Current theories on the biologic mechanism of circumcision as a prevention tool are largely based on HIV transmission research and include differences in the microbiome and immune environment, which may also be affected by behavioral practices such as hygienic practices, between circumcised and uncircumcised penises.(29, 30)

This analysis has some limitations. First, NHANES participants are asked to sample the entire penis (including shaft, glans, and, if uncircumcised, prepuce) using one swab. It is not possible to determine whether the association between circumcision and HR-HPV status varies by sampling site in our study. Second, circumcision status was self-reported using medical illustrations and was not medically confirmed. Third, some groups had smaller sample sizes limiting our ability to perform stratified analyses, such as analyses by race/ethnicity, which could have provided some information on the presence of residual confounding as there are substantial differences between circumcised and uncircumcised males that may not be fully accounted for through model adjustment. Fourth, NHANES does not collect information on age at circumcision. Age at circumcision is infrequently collected in studies and likely varies across studies assessing the association between circumcision status and HPV, limiting the ability to compare findings. As a proxy for controlling for age at circumcision, we performed a sensitivity analysis restricted to participants born in the United States; results were similar when restricted to this population.

In this cross-sectional U.S. study, male circumcision was not statistically associated with lower HR-HPV prevalence among adult, sexually experienced male NHANES participants. While circumcision as an HPV prevention tool may be useful in some settings, national HPV prevention measures should continue to focus on vaccinating young males and females at the recommended ages.