Between 2007–2010, 25–65 year old women were recruited through the Internet and enrolled into a longitudinal study of genital HPV infections. Recruitment advertisements were posted primarily on Craigslist.org (a community-based website offering free classified advertisements in over 700 cities worldwide). Advertisements were posted weekly in the “volunteer opportunities” section of one city on a rotating basis (37 major U.S. cities were targeted). Advertisements were also posted on University of Washington research volunteer recruitment websites and other national websites offering free classified advertisements. Advertisements targeted 25–65 year old women who reported using the Internet to search for romantic partners within the previous 12 months. Women who were pregnant or breastfeeding or planning a pregnancy in the next 6 months, hysterectomized, immunocompromized, or reported no history of sex with men were not eligible to participate. 1170 women responding to the recruitment advertisements were screened over the telephone, of whom 832 (71.1%) women were eligible. 743 (63.5% of women screened) returned a mailed consent form and were enrolled. () The protocol was reviewed and approved by the University of Washington Institutional Review Board.
Enrollment and status of subjects.
Every four months for up to one year, enrolled women were mailed demographic, sexual behavior, and health history questionnaires and materials for self-collecting vaginal samples for HPV DNA testing. Self-collection materials included illustrated instructions, two Dacron-tipped swabs (two sequential self-collected swabs increases sensitivity for HPV detection5
), a capped tube containing 1.5 ml of Specimen Transport Medium (Qiagen, Gaithersburg, MD), nitrile gloves, and packing and shipping materials. The kit also contained written instructions to refrain from douching, vaginal intercourse, and the use of vaginal medications or preparations for 48 hours prior to collecting the vaginal samples, and to collect the samples at least two days after the end of the last menses. Women were instructed to complete the questionnaire and collect the samples on the same day, and to return the materials in the provided pre-paid standard overnight Federal Express envelope.
Vaginal samples were tested for HPV DNA using polymerase chain reaction (PCR)-based methods. Exfoliated cell samples were digested with 20 μg/ml protease K at 37°C for one hour. DNA was isolated using the QIAamp DNA blood mini column (Qiagen, Inc., Valencia, CA) according to the manufacturer’s protocol. Two μl purified DNA (equivalent to one 250th of each sample) was amplified in 50 μl PCR reaction and 10 μL of PCR products were dotted onto nylon filters and probed with both a biotin-labeled HPV generic probe and a biotin-labeled β-globin probe. Samples negative for β-globin DNA were deemed insufficient. Specimens determined to be HPV positive by generic probe were typed using the Roche Linear Array HPV genotyping test (Roche Molecular Systems, Inc., Alameda, CA) for 37 HPV types.
The current analyses focused on the baseline visit only. We evaluated risk factors for prevalent hrHPV infection (positive for any of the following types classified as carcinogenic, probably carcinogenic, or possibly carcinogenic): 16/18/26/31/33/35/39/45/51/52/53/56/58/59/66/68/73/82/IS396, 7
) using Poisson regression models with robust variance estimates to obtain prevalence ratios (PRs). Potential risk factors included demographics, smoking history, women’s health history (e.g. pregnancy history, history of an abnormal Pap test or genital warts, menopausal status, and hormone use) and cumulative and recent sexual behaviors. Age was divided by 5 and included in the model as a continuous variable to reflect the relative change in the likelihood of hrHPV detection associated with a 5-year difference in age (we evaluated a linear trend using a likelihood ratio test to compare a categorical variable based on 5-year age groups to the linear model based on age divided by 5; the likelihood ratio test was not significant, indicating a linear trend). Cumulative sexual behavior variables included age at first intercourse with a male partner and lifetime number of male sex partners. Recent sexual behavior variables included characteristics of male sex partners and partnerships in the past 6 months. If multiple partners were reported in the past 6 months, the characteristic was summarized across partners/partnerships. () Variables found to be statistically significant (p<.10) in univariate analyses were entered into multivariate models. Forward stepwise regression was used to construct the final multivariate models, with p<.10 used as the criterion for entering and removing variables. Because the final multivariate model included partner/partnership variables restricted to women who reported sex with male partners in the past 6 months, women without recent male sex partners were excluded. While effect modification by recent sexual activity was not formally explored (due to a limited number of women reporting no male sex partners in the past 6 months; n=87), we compared the final multivariate model to a second multivariate model that excluded recent partner/partnership variables to explore whether the magnitude of association between hrHPV infection and any other variables in the model differed between recently sexually active women and the full dataset of women who were and were not recently sexually active.
Prevalence Ratios (PR) for the Associations Between Selected Risk Factors and High-Risk (hr) HPV Infection in 25- to 65-Year-Old Women Who Date Online (n = 518)
Separately, we evaluated a composite variable constructed to characterize additive effects of partner/partnership risk factors on the likelihood of hrHPV detection. Partner or partnership characteristics found to be significantly associated (p<.10) with hrHPV detection in the univariate analyses described above were selected as risk factors and incorporated into this variable (3 risk factors were selected: report of ≥1 casual male sex partner, report of ≥1 male sex partner with a concurrent partnership, and sex with ≥1 male partner whom the subject met online). The composite variable was categorized as a five-level variable: (1) not sexually active with a male partner in the past 6 months; or, sexually active with male partner(s) who (2) did not have any identified risk factors, had (3) 1 risk factor, (4) 2 risk factors, (5) or all 3 risk factors. We evaluated potential confounding by variables that were associated with hrHPV detection in the univariate analyses described above (excluding individual sexual behavior variables over the past 6 months). Each variable was tested individually with the composite variable; those that changed any of the composite variable point estimates by >10% were considered confounders and retained in the final model.
Finally, among women testing positive for hrHPV, we evaluated whether any of the variables described above were predictive of testing positive for multiple hr types versus a single hr type. All variables were tested univariately, and forward stepwise regression was used to construct one final multivariate model, with p<.10 used as the criterion for entering and removing variables.
Most analyses were pre-specified; we explicitly note post-hoc analyses.