NHANES 2009-2012 Findings: Association of Sexual Behaviors with Higher Prevalence of Oral Oncogenic Human Papillomavirus Infections in U.S. Men

Abstract

The incidence of human papillomavirus (HPV)-positive oropharyngeal cancers is higher and increasing more rapidly among men than women in the United States for unknown reasons. We compared the epidemiology of oral oncogenic HPV infection between men and women ages 14 to 69 years (N = 9,480) within the U.S. National Health and Nutritional Examination Surveys (NHANES) 2009-2012. HPV presence was detected in oral DNA by PCR. Analyses were stratified by gender and used NHANES sample weights. Oral oncogenic HPV prevalence was higher among men than women (6.6% vs. 1.5%, P < 0.001), corresponding to 7.07 million men versus 1.54 million women with prevalent infection at any point in time during 2009-2012. Prevalence increased significantly with age, current smoking, and lifetime number of sexual partners for both genders (adjusted Ptrend < 0.02). However, men had more partners than women (mean = 18 vs. 7, P < 0.001). Although oncogenic HPV prevalence was similar for men and women with 0 to 1 lifetime partners, the male-female difference in prevalence significantly increased with number of lifetime partners (adjusted prevalence differences for none, 1, 2-5, 6-10, 11-20, and 20+ partners = 1.0%, 0.5%, 3.0%, 5.7%, 4.6%, and 9.3%, respectively). Importantly, the per-sexual partner increase in prevalence was significantly stronger among men than among women (adjusted synergy index = 3.3; 95% confidence interval, 1.1-9.7), and this increase plateaued at 25 lifetime partners among men versus 10 partners among women. Our data suggest that the higher burden of oral oncogenic HPV infections and HPV-positive oropharyngeal cancers among men than women arises in part from higher number of lifetime sexual partners and stronger associations with sexual behaviors among men.

Figure 1

Shown is prevalence of any HPV infection, oncogenic HPV infection, non-oncogenic HPV infection, and HPV type 16 infection among men (black bars) and women (grey bars). The p-values comparing HPV prevalence between men and women are also shown.

Figure 2

Shown are male-female prevalence differences for oral oncogenic HPV infection (circles) and 95% confidence intervals (horizontal lines) stratified by lifetime number of any, oral, and vaginal sex partners. Male-female prevalence differences were calculated in separate models for each sexual behavior, and included age (modeled as splines with 5 knots), gender, race, marital status, education, cigarette smoking, lifetime number of partners, and interaction between gender and lifetime number of partners. Data on lifetime number of any sex partners were available for individuals aged 14-69 years, while data on number of lifetime oral or vaginal sex partners were available for individuals aged 14-59 years.

Figure 3

Shown is the modeled oral oncogenic human papillomavirus (HPV) prevalence and 95% confidence intervals (shaded area) among men (solid line) and women (broken line) by individual number of lifetime any sex partners (panel A), number of lifetime oral sex partners (panel B), and number of lifetime vaginal sex partners (panel C). All sexual behaviors were modeled with restricted cubic splines with 3 knots. Models were stratified by gender and incorporated adjustment for age (as splines with 5 knots), race, marital status, education, and cigarette smoking. The presentation of plots was truncated at 50 lifetime sex partners for visual comparison between men and women. The p-values for the association of each sexual behavior (one linear term + one spline term) with oncogenic HPV prevalence are also shown in each panel. The population percentiles for the number of sex partners for men and women are shown below the x-axis in each panel. Given the standardization for covariates included in the multivariate model, the oncogenic HPV prevalence curve obtained from the adjusted model is presented at the mean levels of the covariates (age, race, education, marital status, and smoking, as appropriate)

Figure 4

Shown is the burden of oral oncogenic HPV infection among men in the U.S. population, as measured by oncogenic HPV prevalence (panels A, B, and C) and the number of infected individuals (panels D, E, and F). Estimates for prevalence and number of infected individuals for men are stratified by age (14-40, 40-59, and 60-69 years), cigarette smoking status (never/former smokers, current smokers who smoked 1-10 cigarettes per day, current smokers who smoked 11-20 cigarettes per day, and current smokers who smoked 20+ cigarettes per day), and lifetime number of any sex partners (0, 1, 2-5, 6-10, 11-20, and 20+ partners). The predicted probability of infection with oral oncogenic HPV infection was estimated from multivariable models that were adjusted for age (modeled as splines with 5 knots), race, marital status, education, smoking, and number of lifetime any sex partners. These predicted probabilities were then summed across subgroups defined by age, smoking, and number of lifetime sex partners to calculate the number of infected individuals in the U.S. population. Prevalence was calculated as the ratio of number of infected individuals over the total number of individuals within a subgroup.

Figure 5

Shown is the burden of oral oncogenic HPV infection among women in the U.S. population, as measured by oncogenic HPV prevalence (panels A, B, and C) and the number of infected individuals (panels D, E, and F). Estimates for prevalence and number of infected individuals for women are stratified by age (14-40, 40-59, and 60-69 years), cigarette smoking status (never/former smokers, current smokers who smoked 1-10 cigarettes per day, current smokers who smoked 11-20 cigarettes per day, and current smokers who smoked 20+ cigarettes per day), and lifetime number of any sex partners (0, 1, 2-5, 6-10, 11-20, and 20+ partners). The predicted probability of infection with oral oncogenic HPV infection was estimated from multivariable models that were adjusted for age (modeled as splines with 5 knots), race, marital status, education, smoking, and number of lifetime any sex partners. These predicted probabilities were then summed across subgroups defined by age, smoking, and number of lifetime sex partners to calculate the number of infected individuals in the U.S. population. Prevalence was calculated as the ratio of number of infected individuals over the total number of individuals within a subgroup.