This study sample is from a region of Appalachia that contains a high number of distressed counties (42 of 54 Kentucky Appalachian designated counties), which are designated by the Appalachian Regional Commission (ARC) based on varying economic and employment indicators.³⁵ Many of the poorest counties in the nation can be found in this region.³⁶ As posited by the Diffusion of Innovations Theory, knowledge of new innovations, such as the HPV vaccine, is often lower in populations that have limited formal education and lower socioeconomic status (SES) and are less urban.³⁷ Kentucky as a whole has lower adolescent HPV vaccination rates compared to the United States (40% at least dose one),⁴ and Crosby et al.⁶ found higher vaccination rates among young adult women in urban Kentucky compared to rural Kentucky. Therefore, it is reasonable to extrapolate that our study targeted a population that was clearly undervaccinated compared to national and state data.

From March 2008 through September 2009, a research assistant recruited female patients in any of five regional health clinics located in five rural counties of Southeastern Kentucky. Age-eligible women were approached by a clinic nurse to first determine interest in participating in the study. If the women were interested, a research assistant then privately spoke with each woman to explain the study purpose and to obtain consent if she chose to participate. During that same time period, a second research assistant recruited age-eligible women attending a local community college (with buildings located in four of the same five counties used for the clinic sample). Recruitment at the community college used varying approaches, including emails, fliers, and presentations in the classroom and at college health fairs. Community college women were selected to offset what would have otherwise been a purely clinic-based sample of young women. Women were eligible if they were not pregnant, were 18–26 years old, and had not been vaccinated with Gardasil (the only HPV vaccine approved for use at the time). Of 505 eligible women, 495 (98%) agreed to participate in the study and provided written informed consent. The Institutional Review Board at the University of Kentucky approved the study protocol.

Because the HPV vaccine was not universally covered by insurance plans, it was determined that the study design would have to remove the barrier of cost to effectively identify sexual behaviors predicting uptake. However, the fact that the HPV vaccine would be provided at no cost was not advertised or disclosed until after the questionnaire was completed. To avoid self-selection bias, the project was called the Women’s Health Study. Volunteers were told that “the purpose of this survey is to learn more about why women would or would not accept the HPV vaccine if it was made available to them.” After providing consent, women recruited from the community college completed a self-administered questionnaire, and women recruited from the clinics completed the same questionnaire, but it was administered in an interview-assisted format. It should be mentioned that overall literacy rates in the study area are low. Although literacy was somewhat assured with the college sample, there was concern that the clinic sample might need one-on-one assistance to complete the written questionnaire. Assessments were conducted in a private or semiprivate area, with a research assistant being available to read words or entire questions on the self-administered questionnaire if needed. Women were compensated with a $25 gift card for their time, and the research assistant provided women with an HPV vaccine educational pamphlet and a voucher to receive all three doses of the HPV vaccine free of charge at the clinic they were recruited from or, in the case of college women, at a centrally located clinic. These vouchers were coded with an ID number that matched the ID number recorded on women’s questionnaires, and women had 30 days to receive dose one, after which they became ineligible to receive the free vaccine. To ensure the accuracy of those receiving dose one, a database of those eligible to receive the HPV vaccine was created, and clinic nurses were given access to ensure the correct study participants were redeeming vaccine vouchers. The number of women redeeming the voucher for the initial dose of vaccine served as the study outcome variable.

The questionnaire was refined based on experiences in a previous study.²⁵ The survey began with a brief paragraph about HPV, its relationship with cervical cancer, and the HPV vaccine. There were three questions that specifically assessed women’s sexual behaviors. The first question asked, “In the past 12 months, have you had sex?” with response options of yes or no. If women answered yes to this question, they were then asked, “With how many men?” and given an opportunity to answer with a number. Respondents who answered yes to this first question were also asked, “Please think about the last 5 times you have had sex (penis in vagina) with a man. Of these 5 times, how many times was a condom used from start to finish of sex?” with response options of 0–5 to select from. The second sexual behavior question asked, “HPV can be spread by skin to skin contact; therefore, in the past 12 months, have you had sex with women?” with response options of yes or no. If participants answered yes, they were asked, “With how many women?” The third question asked, “Because HPV can be spread by skin to skin contact, sex is not the only way to get the virus. In the past 12 months have you ever engaged in mutual masturbation with a partner? (this means that you each used fingers and hands to touch and arouse the other person’s genitals)” with response options of yes or no. If participants answered yes, they were then asked, “About how often in a typical month did you engage in mutual masturbation?” with response options of once a month or less, about twice each month, 3 to 5 times each month, and at least 6 times each month. The survey included one question pertaining to relationship status: “Which statement best describes your current or most recent relationship status?” The response options included: “We have a deep commitment to each other,” “We see each other exclusively, but the commitment is not necessarily long term,” and “We have agreed that our relationship is only temporary.”

The next set of questions assessed women’s past gynecologic history, and each had response options of yes or no. (1)”Have you ever had a family member or close friend tell you that she had cervical cancer?” (2) “Have you ever been told by a doctor or nurse that you have cancer?” (3) “Have you ever had a Pap test?—If yes, about how many times have you had a Pap test in your life?” with response options of 1 time up to five or more times. (4) “Have you ever been told by a medical professional that you have an abnormal result on a Pap test?” (5) “Are you currently using hormonal contraceptives?” (6) “Are you currently using an intrauterine device (IUD) to prevent pregnancy?” (7) “Have you ever been told by a medical professional that you have a sexually transmitted disease?” (8) “Have you ever been told by a medical professional that you have genital warts?” (9) Have you ever been told by a medical professional that you have human papillomavirus (HPV)?” The survey contained one question that served as a measurement of interaction with the healthcare system: “In the past 12 months, about how many times have you gone to see a medical doctor?” with response options ranging from zero through five or more times.

Bivariate associations were measured by prevalence ratios, 95% confidence intervals (CI), and p values. The variables that achieved significance at the α = 0.05 level were included in the logistic regression model. At the bivariate level, categorical variables were assessed using chi-square and, when appropriate for small cell sizes, Fisher’s exact test. Continuous variables were assessed using t tests.

A hierarchical logistic regression with a forward Wald approach was used for the multivariate analysis. This model controlled for (1) age, (2) ever having a Pap test, (3) ever having an abnormal Pap test, and (4) doctor visits. Sexually related behaviors included in the model were mutual masturbation, intrauterine contraception use, hormonal contraception use, and condom use. All statistical analysis was performed with SPSS version 18.

Table 1 displays the characteristics of the sample. The average age of the sample (n = 495) was 21.56 years (standard deviation [SD] 2.50). The majority of the women identified themselves as Caucasian (98.0%). A large portion of the sample (88%) reported their home county as rural as determined by the Rural-Urban Commuting Area Codes (RUCA).³⁸ The majority of the women reported having participated in sex with men in the past 12 months (86%); of these women, 16.7% reported having two or more male partners. Almost half of the sample (47%) did not use condoms for the past five sexual encounters with males. The majority of the sample (85%) identified their most recent love relationship as a deep commitment to one another; the rest described their relationship as either exclusive, yet not long term, or temporary. Over 50% of the women reported participating in mutual masturbation in the past 12 months. Almost half of the sample reported hormonal birth control (49.2%) use, and only 19% of the sample reported IUD use for birth control. Furthermore, 24% reported an abnormal Pap test in their gynecologic history. Only 128 (26%) of the women received dose one of the HPV vaccine within the allotted 30-day window. Adherence to doses two and three was extremely low, limiting subsequent analysis; results have been presented elsewhere.⁶

Table 2 displays the significant bivariate findings. After examining the data for interaction by study site, three variables were found to have interaction at the bivariate level: condom use and hormonal and intrauterine birth control use. Condom use was significantly associated with vaccine uptake ( p = 0.011). Women who did not use condoms were more likely to refuse dose one of the HPV vaccine. However, this association was moderated by recruitment site; the association applied only to clinic-recruited women. Among these women, 62.4% of those not using condoms declined the free vaccine compared to 44.0% of those using condoms ( p = 0.009). In contrast, among college-recruited women, 95.1% of those not using condoms declined the free vaccine compared to 92.0% of those using condoms ( p = 0.36).

Ever having a Pap test was also significantly associated with uptake of the vaccine ( p = 0.005). Women who had never had a Pap test were more likely to refuse dose one of the HPV vaccine. Specifically, among those never having a Pap test, 88.1% declined the free vaccine compared to 71.9% among those ever having a Pap test. Because of an empty cell in the contingency table analysis, a test for moderating effects could not be conducted for this variable.

Hormonal birth control was significantly associated with uptake ( p = 0.010). Women who did not use hormonal birth control were more likely to decline the vaccine compared to those currently using a hormonal method. Again, this association was moderated by recruitment site and applied only to clinic-recruited women. Among these women, 63.1% of those not using hormonal contraception declined compared to 47.6% of those using condoms ( p = 0.014). Among college-recruited women, 94.5% of those not using hormonal contraception declined compared to 91.5% of those using hormonal contraception ( p = 0.35).

Intrauterine birth control was also significantly associated with vaccine uptake ( p = 0.030). Women who did not use an IUD were more likely to decline dose one of the HPV vaccine. Once again, the association was moderated by site; however, in this case, it applied only to college-recruited women. Among these women, 96.9% of those not using an IUD declined the free vaccine compared to 78.4% of those using an IUD ( p = 0.0001). In contrast, among clinic-recruited women, 56.6% of those not using an IUD declined the free vaccine compared to 48.8% of those using an IUD ( p = 0.36).

Mutual masturbation was significantly associated with uptake ( p = 0.006). Women who engaged in mutual masturbation were more likely to decline regardless of recruitment site. Doctor visits was significantly associated with uptake ( p = 0.036). Women who had not been to the doctor in the past 12 months were more likely to decline regardless of recruitment site. Finally, ever having an abnormal Pap test result was significantly associated with lower uptake ( p = 0.001). Women reporting they had an abnormal result were more likely to decline regardless of recruitment site.

Table 3 displays the four predictor variables that retained significance at the multivariate level. The model controlled for the following variables: age, ever having a Pap test, ever having an abnormal Pap test, and frequency of doctor visits in the past 12 months. To control for confounding due to recruitment site, those variables found to be significant with interaction at the bivariate level were used in the model with an interaction term (recruitment site). Women with a history of an abnormal Pap test result were nearly two times more likely to decline the vaccine (adjusted odds ratio [AOR] 1.91, 95% CI 1.14–3.20, p = 0.015), and women reporting they had never had a Pap test were four times more likely to decline the vaccine (AOR 4.02, 95% CI 1.13–14.32, p = 0.032). Women engaging in mutual masturbation were nearly two times more likely to decline the free vaccine (AOR 1.91, 95% CI 1.17–3.10, p = 0.009). Use of hormonal birth control showed a protective effect against refusal of the free HPV vaccine (AOR 0.593, 95% CI 0.44–0.80, p = 0.001). The following variables were not significant predictors of HPV vaccination refusal: intrauterine birth control use (AOR 0.702, 95% CI 0.49–1.01, p = 0.056) and doctor visits (AOR 0.260, 95% CI 0.06–1.22, p = 0.088).