Two human papillomavirus (HPV) vaccines have been developed that are highly effective in preventing type-specific HPV infection. Both prevent infection with HPV-16 and HPV-18, the two high-risk HPV types that cause approximately 70% of cervical cancers, and one also prevents infection with HPV-6 and HPV-11, which cause more than 90% of anogenital warts and respiratory papillomas. It is widely accepted that universal HPV vaccination could reduce substantially the burden of cervical cancer and other HPV-related diseases.1–3
In addition to reducing the overall burden of cervical cancer, widespread HPV vaccination has the potential to narrow existing racial, ethnic, and socioeconomic disparities in cancer incidence and mortality.4–7 However, if poor and minority women have lower rates of HPV vaccination than other women, vaccination could have the unintended effect of widening even further current cervical cancer disparities among U.S. women. Because the HPV vaccine is highly effective, even small differences in access to vaccination could lead to significant worsening of disparities.8
Although sociodemographic predictors of cervical cancer are well understood, predictors of infection with the high-risk HPV types that are the biologic cause of cervical cancer are not. Predictors of HPV infection may differ from those of cervical cancer because development of cervical cancer is influenced by many variables including cofactors involved in carcinogenesis and adherence to Pap screening. Representative data that provides information about high-risk HPV infection by race and ethnicity, income, education, age, and marital status have not been available previously but may be useful for the design of outreach or tailored public health strategies to prevent cervical cancer and to ensure decreased disparities in cervical cancer mortality in the vaccination era.
The National Health and Nutrition Examination Survey (NHANES) offered type-specific HPV DNA testing to female study participants in 2003–2004, resulting in the first reliable estimates for type-specific HPV infection in U.S. women. Because this survey also collects detailed information on income and oversamples vulnerable populations, it provides an exceptional opportunity to examine sociodemographic factors linked to HPV infection. The primary aims of this study were to determine the prevalence of high-risk HPV infection and identify sociodemographic factors associated with high-risk HPV infection in U.S. women. A secondary aim was to explore in depth the interrelationships between race and ethnicity, income, and high-risk HPV infection.
MATERIALS AND METHODS
Data for these analyses were derived from the 2003–2004 NHANES, which is a national study conducted by the National Center for Health Statistics (NCHS) of the Centers for Disease Control and Prevention (CDC). Selected objectives of NHANES are to estimate the number and percent of persons in the U.S. population and designated subgroups with selected diseases and risk factors, to analyze risk factors for selected diseases, to explore emerging public health issues and new technologies, and to establish and maintain a national probability sample of baseline information on health and nutritional status. The NHANES survey design is a stratified, multistage probability sample of the civilian noninstitutionalized U.S. population. Detailed information about methods can be found on the NHANES Web site (http://www.cdc.gov/nchs/about/major/nhanes/nhanes2003–2004/nhanes03_04.htm).
All participants interviewed for the study were also asked to complete a health examination component, which was conducted separately in mobile examination centers. Participants eligible for this study were the subsample of 14- to 59-year-old women who completed the health examination component and agreed to HPV DNA testing. Of 2,387 women who were eligible, 2,026 (85%) submitted a swab, and 1,921 (95%) of these were adequate for analysis. Those who did not submit a swab did not differ from those who did in terms of age, race and ethnicity, education, income, or marital status.
If a participant agreed to HPV DNA testing, a testing center physician counseled her and explained the vaginal collection technique and specimen collection kit. The collection kit was designed specifically for women in the target age group and included illustrated collection instructions and one Dacron swab (Epicentre, Madison, WI) with foam tip. Each participant collected a vaginal swab herself in the mobile examination center bathroom, placed the swab back into its original container, and placed the container into a plastic bag. The plastic bag was then delivered to the laboratory, where the laboratory staff verified the participant and vessel identification numbers and stored the swab at room temperature. The laboratory staff shipped the specimens to the CDC laboratory weekly, following a specific storage and shipping protocol.
Human papillomavirus DNA was detected by using HPV L1 consensus polymerase chain reaction with biotinylated PGMY09/11 primer sets and β-globin as an internal control for sample amplification.9–11 The primer mix amplifies essentially all known HPV types found in the genital tract. The amplicons were evaluated by gel electrophoresis for the presence of the 450 bp HPV amplicon. Positive samples were typed by hybridization to the Roche prototype line probe typing strips followed by colorimetric detection. The strip was a linear array of probes specific for 37 HPV types (6, 11, 16, 18, 26, 31, 33, 35, 39, 40, 42, 45, 51, 52, 53, 54, 55, 56, 58, 59, 61, 62, 64, 66, 67, 68, 69, 70, 71, 72, 73, 81, 82, 83, 84, IS39, and 89) and for the positive β-globin control. Types were read by comparing the reaction pattern to the typing template. Samples that did not hybridize to the typing strip are sequenced to determine the HPV type. If there was no β-globin present in the sample and no HPV type was detected either by hybridization or sequencing, the sample was coded as inadequate. If any of the types on the strips (LBDH06-LBDHP1) were positive, or a type was identified by sequencing, the type was coded as positive.
We performed analyses to describe participants’ age, race and ethnicity, educational attainment, marital status, household income, and level of poverty. For these analyses, poverty was measured by using a ratio of household income to the appropriate poverty threshold (poverty/income ratio). Outcome variables included any HPV infection, high-risk HPV infection (defined as a positive test for at least one of 15 high-risk HPV types), and HPV infection with HPV types contained in current vaccines (HPV-6, -11, -16, and -18 and HPV-16 and -18). Univariable analyses using χ2 analyses and unadjusted logistic regression analysis were performed to examine whether sociodemographic characteristics were associated with any HPV infection and with high-risk HPV infection. Logistic regression models were then adjusted for variables associated with HPV infection in univariable logistic analyses at a significance level of P≤.1.
We also explored in more depth the associations between race and ethnicity, income, and HPV infection, because previous studies have shown that race and income are associated both with HPV infection and cervical cancer,7 and multivariable analyses using these data suggested that there were interactions between race and income. We first determined the prevalence of any HPV and high-risk HPV infection in women of different race and ethnicity (white, black, and Mexican-American) and income levels (poverty/income ratio less than 1.0 or below the poverty line, 1.0–1.99, 2.0–2.99, and more than 3.0). We then performed multivariable logistic regression analyses stratified by poverty status to determine whether factors associated with HPV infection differed in poor and nonpoor women and to explore whether differences in factors such as mean household income and marital status could explain racial differences in HPV prevalence rates within these strata.
The National Health and Nutrition Examination Survey is a complex probability sample. We used sample weights and accounted for the stratification and clustering of the design to produce unbiased national estimates and to prevent overstatement of significance levels. The sample weights used reflected the unequal probabilities of selection, nonresponse adjustments, and adjustments to independent population controls. All analyses were performed with SAS 9.1 (SAS Institute Inc, Cary, NC) using SAS procedures designed for survey data analysis. We obtained approval to perform this secondary data analysis from the Cincinnati Children’s Hospital Medical Center Institutional Review Board.
Sociodemographic characteristics of the participants (n=1,921) are shown in Table 1. The mean age of participants was 36.2 years (range 14–59 years). The majority of participants (69%) were non-Hispanic white. Almost 18% were living below the poverty line, as indicated by an income/poverty ratio less than 1.0. Approximately half of participants were married. The cross-sectional prevalence rates of HPV infection in this nationally representative sample were as follows: 26.8% (95% confidence interval [CI] 23.0–30.6%) of participants were positive for at least one HPV type, 15.6% (95% CI 12.6–18.6%) for at least one high-risk type, 3.4% (95% CI 2.0–4.7%) for HPV-6, 11, 16, and/or 18, and 2.2% (95% CI 1.2–3.2%) for HPV-16 and/or 18. Thus, 3.4% of women were positive for one of the four types that make up the licensed vaccine. Corresponding U.S. population estimates for women with prevalent infection were 20,722,668 (any HPV), 12,028,293 (at least one high-risk type), 2,619,309 (HPV-6, 11, 16, and/or 18), and 1,710,682 (HPV-16 and/or 18). Of women positive for high-risk HPV, 62% had one high-risk type, 24% two types, and 14% three or more types. Human papillomavirus-6 was identified in 1.3% of women, HPV-11 in 0.09%, HPV-16 in 1.5%, and HPV-18 in 0.8%. The most common HPV types, identified in more than 2% of women, were HPV types 62 (3.3%), 84 (3.3%), 53 (2.8%), 61 (2.4%), 89 (2.4%), 52 (2.3%), and 54 (2.2%).
The prevalence of HPV and high-risk HPV infection by sociodemographic characteristics is shown in Table 1. Women 22–25 years of age, compared with women in other age groups, had the highest prevalence rates of HPV infection. Over 40% of women in this age group were positive for HPV, compared with approximately 20% of women 14–17 years of age or 40–59 years of age. Non-Hispanic black participants had the highest rates of any HPV and high-risk HPV infection, whereas Mexican-American and non-Hispanic white women had the lowest rates. For example, 39% of black women, compared with 24% of Mexican-American and 24% of white women, were positive for any HPV. Women in the lowest category of income (income/poverty ratio less than 1.0) were approximately twice as likely as those in the highest category (ratio 3.0 or more) to be infected with any or high-risk HPV (for high-risk HPV, 23% versus 12%, P=.03). Finally, married women had markedly lower rates of HPV infection than unmarried women, with the highest rates found in women who were widowed, divorced, separated, or living with a partner. There were no differences in the proportion of women in different racial or ethnic groups who were positive for certain HPV types (eg, 16, 18).
Odds ratios of HPV infection, unadjusted and adjusted for sociodemographic characteristics, are shown in Table 2. In adjusted logistic regression models, age, race and ethnicity, and marital status were significantly associated with HPV infection. Participants 22–25 years of age had approximately twice the odds of infection with any HPV type and three times the odds of infection with a high-risk type, compared with women 40–59 years of age. Non-Hispanic black women, compared with white women, had approximately 50% higher odds of infection with any HPV, whereas Mexican-American women had 38% lower odds of infection with high-risk HPV than white women. Unmarried women had higher odds of HPV infection than married women: those living with a partner had almost 3.5 times the odds of any HPV infection than married women, and those who were widowed, divorced, or separated had more than 2.5 times the odds of high-risk HPV infection than married women.
Prevalence of HPV by race and ethnicity in women living below and above the poverty line is shown in Figure 1. Generally, women living below the poverty line were at higher risk than women living above the poverty line for any HPV (37% versus 24%, P=.005) and high-risk HPV infection (23% versus 14%, P=.006). However, the association between race and HPV prevalence differed by income level. Among those living below the poverty line, rates of any HPV infection did not differ significantly by race (P=.22), although prevalence was lower among Mexican-American women (30%) than white women (40%) and black women (42%). Among women living above the poverty line, rates of HPV were significantly higher in black women than in white women (39% versus 22%, P<.001) and, similarly, were higher in black women than in Mexican-American women (39% versus 20%, P<.001). Findings generally were similar for high-risk HPV infection. Among women living below the poverty line, white women had the highest prevalence of high-risk HPV infection (28%), followed by black women (21%) and Mexican-American women (14%). Only the difference between white and Mexican-American women was statistically significant (P=.01). Among women living above the poverty line, the prevalence of high-risk HPV infection was higher among black women than white women (21% versus 12%, P<.001) and Mexican-American women (21% versus 11%, P=.004).
Additional univariable analyses suggested that differences in marital status or mean income might help to explain the higher prevalence of HPV infection in black women living above the poverty line compared with white and Mexican-American women living above the poverty line. Black women in this group were significantly less likely to be married than white or Mexican-American women: 29% of black women were married compared with 59% of white and 58% of Mexican-American women (P<.001). In addition, mean household income was significantly lower in the group of black women living above the poverty line than in white and Mexican-American women (P<.001). We therefore estimated logistic regression models stratified by poverty status, controlling sequentially for race, income, marital status, and age. Among women living below the poverty line (Table 3), only ethnicity and marital status were associated with HPV infection. Mexican-American women had lower odds of high-risk HPV than white women, both unadjusted and adjusted for mean income, marital status, and age. Unmarried women had higher odds of any HPV and high-risk HPV than married women in adjusted models. Among women living above the poverty line, findings differed. Black race, lower mean income, unmarried status, and younger age were associated with any HPV and with high-risk HPV infection. In these models, although black women had higher odds of any HPV and high-risk HPV than white women, these odds decreased in models controlling for income and decreased even further in models controlling for income, marital status, and age. The odds of any HPV infection among black women decreased 30%, and the odds of high-risk HPV infection decreased 26%, in the adjusted models. For high-risk HPV infection, the odds of HPV infection for black women became almost nonsignificant. These results indicate that the association between black race and HPV infection is explained in large part by the fact that black women living above the poverty line in this study sample had a lower mean income and were less likely to be married than white women (ie, lived closer to the poverty line).
In this study, we examined sociodemographic factors associated with HPV infection in a representative sample of U.S. women. The overall cross-sectional prevalence of HPV in U.S. women was 26.8% and the prevalence of high-risk HPV was 15.6%. In previous studies, prevalence rates of HPV infection have varied widely, from approximately 15% to 90%12; this variation is likely due to the fact that most studies have not involved recruitment of nationally representative samples. The high prevalence of any HPV and high-risk HPV in all age groups supports current recommendations for universal vaccination, and the finding that 13% of girls 14–17 years of age were positive for high-risk HPV supports current national recommendations to target vaccination to 11- to 12-year-old girls, before sexual initiation and, therefore, before HPV exposure.13 However, the finding that the cross-sectional prevalence of HPV-16 and HPV-18, the types contained in the current vaccine, were quite low in a population targeted for vaccination suggests that additional subtypes should be included in future vaccines. It should be noted, however, that HPV-16 and HPV-18 are the types most likely to cause cancer. In addition, the cumulative prevalence of infection over several years will exceed the cross-sectional prevalence.14 Thus, widespread vaccination using the currently licensed vaccine should still decrease significantly the incidence of cervical dysplasia and cancer caused by HPV-16 and HPV-18.
We found that all sociodemographic factors examined, including age, race, education, income, and marital status, were associated significantly in unadjusted univariable logistic regression analyses with any HPV and with high-risk HPV infection. Although high-risk HPV infection was more common among younger, unmarried women, cervical cancer is primarily a disease of older, married (or previously married) women. These differences reflect the time interval between infection and development of cancer, as well as the many other variables that influence the development of cancer, such as Pap screening practices. Other sociodemographic risk factors for HPV infection in this study were similar to known risks for cervical cancer. These included race, poverty, and low educational level. The higher risk for both HPV infection and cervical cancer among minority, poor, and under-educated women is likely driven by health system inadequacies that disproportionately affect poor and other underserved women, such as insufficient access to medical care, lack of culturally competent care, and lack of effective health education about sexually transmitted infection (STI) prevention and Pap screening.6 Although socioeconomic realities will not be affected by the introduction of HPV vaccines, these findings suggest that HPV vaccination should be part of a larger, coordinated strategy to provide preventive services to underserved women.
Women 22–25 years of age and those who were unmarried had the highest odds of HPV infection: age and marital status likely serve as important risk markers for exposure to HPV and other STIs. Previous studies have demonstrated consistently that sexual behaviors, especially number of sexual partners, are important risk factors for HPV acquisition.15–18 We purposefully focused these analyses only on sociodemographic factors associated with HPV infection, because it is not feasible, appropriate, or effective to target interventions to prevent cervical cancer toward women based on their sexual history. However, in this sample, number of sexual partners also is associated with HPV infection.19
The findings of multivariable models predicting high-risk HPV infection that were stratified by poverty status demonstrated that HPV infection was substantially higher among women living below, compared with above, the poverty line. However, among women living below the poverty line, few factors distinguished those who were from those who were not infected with HPV. Married women were less likely than unmarried women, and Mexican-American women less likely than white women, to be infected. Interestingly, Latina women in the United States are at higher risk than white women for development of cervical cancer6; the reason for the difference in risk for cervical cancer compared with high-risk HPV infection is not clear but may be related to access to or acceptability of Pap testing. The finding that few variables were associated with HPV infection in women living below the poverty line implies that interventions designed to prevent HPV-related disease must ensure that all low-income women have enhanced access to HPV vaccines as well as education and other preventive services. To maximize access to and affordability of vaccination, ideally girls would be immunized while they are still covered by benefits for low-income children such as the Vaccines for Children Program or Medicaid. Legislative and community efforts will be critically important to ensure that vaccines are available at low or no cost to low-income young adult women who are not covered by such programs.
Among women living above the poverty line, black race, lower mean income, unmarried status, and younger age were associated with high-risk HPV infection. The higher rates of HPV infection in black women living above the poverty line were explained in large measure by racial differences in marital status and mean income. The higher odds of HPV infection in black women are probably also driven by unmeasured differences in social and economic assets not entirely reflected by household income, such as wealth, community resources, and access to health services and education. Further exploration of these reasons underlying the racial differences in HPV infection among women living above the poverty line may provide insight into the factors driving health disparities in HPV infection and in cervical cancer incidence and mortality.
There are several limitations to this study. Survey and laboratory data were subject to sampling and nonsampling errors, including measurement error. Data were based on self-reports and were therefore potentially subject to inaccurate recall. Laboratory data were subject to measurement variation, although the Roche linear array test has been used in many studies and appears to be reliable in both self-collected and clinician-collected samples.9–11 Self-collected swabs, rather than clinician-obtained swabs, were used for HPV testing. A number of studies have demonstrated that the results of self-collected vaginal swabs are similar to those of clinician-collected cervical or cervicovaginal swabs.9,20–24 Compared with cervical swabs, vaginal swabs may be more likely to detect additional high-risk and low-risk HPV types, detect multiple HPV types, and contain adequate cellular material.14,24 Although this was a nationally representative study of U.S. women, the numbers in this first cross-sectional surveillance study were modest, and therefore the power to detect statistically significant differences in HPV infection in specific subgroups of women was limited. Finally, it should be noted that in this study we only examined the associations between sociodemographic factors and HPV infection. We were not able to examine factors that affect whether women who are infected with HPV ultimately develop cervical cancer. These include access to health care services, screening behavior history, adherence to follow-up care, and the underlying immune status of the participants, among other factors.
In conclusion, this study provides U.S. population estimates of high-risk HPV infection and detailed information about the sociodemographic factors associated with infection. In addition, the study provides novel data about the interrelationships between sociodemographic predictors of high-risk HPV infection; for example, that the association between black race and high-risk HPV infection in women who are not poor was explained in large part by marital status and mean income. The finding that more than one of ten 14- to 17-year-old girls has high-risk HPV infection supports the targeting of early adolescents for vaccination, and the finding that poverty is a strong predictor of high-risk HPV infection supports the need to ensure enhanced access to HPV infection in these women. Vaccines to prevent HPV acquisition have the potential to reduce health disparities in cervical cancer incidence and mortality, if poor and minority women find HPV vaccination to be acceptable, have universal access to HPV vaccines, and can afford vaccination. Interventions to prevent HPV-related disease must also ensure that low-income women have enhanced access to education and other preventive health services such as Pap screening. As Dr. Paul Wise points out, “Medical progress alone can never guarantee equity in health outcomes. Rather, growing efficacy merely provides an expanding substrate for disparity reduction or enhancement, depending on patterns of provision.”8 Thus, universal provision of the new, highly efficacious HPV vaccines will be a key component of the multi-pronged strategy to reduce disparities in cervical cancer mortality among women.
1. Goldie SJ, Grima D, Kohli M, Wright TC, Weinstein M, Franco E. A comprehensive natural history model of HPV infection and cervical cancer to estimate the clinical impact of a prophylactic HPV-16/18 vaccine. Int J Cancer 2003;106:896–904.
2. Villa LL, Costa RL, Petta CA, Andrade RP, Ault KA, Giuliano AR, et al. Prophylactic quadrivalent human papillomavirus (types 6, 11, 16, and 18) L1 virus-like particle vaccine in young women: a randomised double-blind placebo-controlled multicentre phase II efficacy trial. Lancet Oncol 2005;6:271–8.
3. Harper DM, Franco EL, Wheeler CM, Moscicki AB, Romanowski B, Roteli-Martins CM, et al. Sustained efficacy up to 4.5 years of a bivalent L1 virus-like particle vaccine against human papillomavirus types 16 and 18: follow-up from a randomised control trial. Lancet 2006;367:1247–55.
4. McWhorter WP, Schatzkin AG, Horm JW, Brown CC. Contribution of socioeconomic status to black/white differences in cancer incidence. Cancer 1989;63:982–7.
5. Baquet CR, Horm JW, Gibbs T, Greenwald P. Socioeconomic factors and cancer incidence among blacks and whites. J Natl Cancer Inst 1991;83:551–7.
6. Freeman HP, Wingrove BK. Excess cervical cancer mortality: a marker for low access to health care in poor communities. NIH Pub. No. 05-5282. Rockville (MD): National Cancer Institute, Center to Reduce Cancer Health Disparities; 2005.
7. Krieger N, Quesenberry C Jr, Peng T, Horn-Ross P, Stewart S, Brown S, et al. Social class, race/ethnicity, and incidence of breast, cervix, colon, lung, and prostate cancer among Asian, Black, Hispanic, and White residents of the San Francisco Bay Area, 1988–92 (United States). Cancer Causes Control 1999;10:525–37.
8. Wise PH. The anatomy of a disparity in infant mortality. Annu Rev Public Health 2003;24:341–62.
9. Gravitt PE, Lacey JV Jr, Brinton LA, Barnes WA, Kornegay JR, Greenberg MD, et al. Evaluation of self-collected cervicovaginal cell samples for human papillomavirus testing by polymerase chain reaction. Cancer Epidemiol Biomarkers Prev 2001;10:95–100.
10. D’Souza G, Sugar E, Ruby W, Gravitt P, Gillison M. Analysis of the effect of DNA purification on detection of human papillomavirus in oral rinse samples by PCR. J Clin Microbiol 2005;43:5526–35.
11. Stevens MP, Garland SM, Tabrizi SN. Human papillomavirus genotyping using a modified linear array detection protocol. J Virol Methods 2006;135:124–6.
12. Revzina NV, Diclemente RJ. Prevalence and incidence of human papillomavirus infection in women in the USA: a systematic review. Int J STD AIDS 2005;16:528–37.
13. Markowitz LE, Dunne EF, Saraiya M, Lawson HW, Chesson H, Unger ER; Centers for Disease Control and Prevention. Quadrivalent human papillomavirus vaccine: recommendations of the Advisory Committee on Immunization Practices. MMWR Recomm Rep 2007;56 (RR-2):1–24.
14. Brown DR, Shew ML, Qadadri B, Neptune N, Vargas M, Tu W, et al. A longitudinal study of genital human papillomavirus infection in a cohort of closely followed adolescent women. J Infect Dis 2005;191:182–92.
15. Burk RD, Ho GY, Beardsley L, Lempa M, Peters M, Bierman R. Sexual behavior and partner characteristics are the predominant risk factors for genital human Papillomavirus infection in young women. J Infect Dis 1996;174:679–89.
16. Kjaer SK van den Brule AJ, Bock JE, Poll PA, Engholm G, Sherman ME, et al. Determinants for genital human papillomavirus (HPV) infection in 1000 randomly chosen young Danish women with normal Pap smear: are there different risk profiles for oncogenic and nononcogenic HPV types? Cancer Epidemiol Biomarkers Prev 1997;6:799–805.
17. Peyton CL, Gravitt PE, Hunt WC, Hundley RS, Zhao M, Apple RJ, et al. Determinants of genital human papillomavirus detection in a US population. J Infect Dis 2001;183:1554–64.
18. Kahn JA, Rosenthal SL, Succop PA, Ho GYF, Burk RD. Mediators of the association between age of first sexual intercourse and human papillomavirus infection. Pediatrics 2002;109:E5.
19. Dunne EF, Unger ER, Sternberg M, McQuillan G, Swan DC, Patel SS, et al. Prevalence of HPV infection among females in the United States. JAMA 2007;297:813–9.
20. Hillemanns P, Kimmig R, Huttemann U, Dannecker C, Thaler CJ. Screening for cervical neoplasia by self-assessment for human papillomavirus DNA. Lancet 1999;354:1970.
21. Sellors JW, Lorincz AT, Mahony JB, Mielzynska I, Lytwyn A, Roth P, et al. Comparison of self-collected vaginal, vulvar and urine samples with physician-collected cervical samples for human papillomavirus testing to detect high-grade squamous intraepithelial lesions. CMAJ 2000;163:513–8.
22. Wright TC, Denny L, Kuhn L, Pollack A, Lorincz A. HPV DNA testing of self-collected vaginal samples compared with cytologic screening to detect cervical cancer. JAMA 2000;283:81–6.
23. Harper DM, Noll WW, Belloni DR, Cole BF. Randomized clinical trial of PCR-determined human papillomavirus detection methods: self-sampling versus clinician-directed—biologic concordance and women’s preferences. Am J Obstet Gynecol 2002;186:365–73.
24. Kahn JA, Slap GB, Huang B, Rosenthal SL, Wanchick AM, Kollar LM, et al. Comparison of adolescent and young adult self-collected and clinician-collected samples for human papillomavirus. Obstet Gynecol 2004;103:952–9.