Longitudinal Predictors of Human Papillomavirus Vaccine Initiation Among Adolescent Girls in a High-Risk Geographic Area : Sexually Transmitted Diseases

Secondary Logo

Journal Logo

Original Study

Longitudinal Predictors of Human Papillomavirus Vaccine Initiation Among Adolescent Girls in a High-Risk Geographic Area

Brewer, Noel T. PhD*†; Gottlieb, Sami L. MD; Reiter, Paul L. PhD*†; McRee, Annie-Laurie MPH*; Liddon, Nicole PhD; Markowitz, Lauri MD; Smith, Jennifer S. PhD*†

Author Information
Sexually Transmitted Diseases 38(3):p 197-204, March 2011. | DOI: 10.1097/OLQ.0b013e3181f12dbf
  • Free

Each year, an estimated 3700 women in the United States die from cervical cancer,1 even though it is largely preventable through screening and early treatment. Blacks and women living in rural areas have higher rates of cervical cancer than white and urban-dwelling women.1,2 Promising new prophylactic human papillomavirus (HPV) vaccines prevent viral infections associated with 70% of cervical cancers.3,4 The US Advisory Committee on Immunization Practices recommends HPV vaccine for adolescent girls aged 11 or 12, with catch up for girls and women aged 13 through 26.5

HPV vaccine may be more likely to achieve its potential to reduce cervical cancer, genital warts and other HPV-related diseases if it is widely disseminated to girls before first sexual intercourse, especially among populations at high risk.6 While uptake has increased since the US Food and Drug Administration approved quadrivalent HPV vaccine in June 2006, the proportion of adolescent girls initiating the first dose of the 3 dose HPV vaccine series has remained relatively low, at 37% in 2008.7 Uptake is markedly higher (∼80%) in countries, like Australia and the United Kingdom, where school-based vaccine programs are more centralized and extensive than in the United States.8,9

Reasons for low HPV vaccine uptake remain unclear. Parent and physician behavior may be more important to vaccine initiation among adolescent girls in the United States, where healthcare delivery is less centralized. Studies of willingness to receive HPV vaccine, conducted primarily before vaccine licensure among relatively homogeneous populations, found that most parents and young adults approved of HPV vaccination.10 However, since vaccine licensing, only a few cross-sectional studies of HPV vaccine uptake have examined facilitators and barriers identified in the prelicensing acceptability studies.11–13 Furthermore, no longitudinal studies have examined our previous assertion that the health belief model, which was developed in part to address vaccination behavior, is a useful rubric for thinking about predictors of HPV vaccine uptake.10 The model suggests that people get vaccines they think are safe, effective, doctor recommended, with few barriers to uptake, and addressing common and severe hazards.

We sought to characterize HPV vaccine initiation by a racially diverse sample of adolescent girls from both rural and urban areas with elevated rates of cervical cancer. We also sought to identify reasons for low HPV vaccine initiation rates, using a longitudinal study design that could yield results to inform future vaccine interventions and shed light on theoretical assumptions about correlates of uptake.


Settings and Participants

As part of the Carolina HPV Immunization Measurement and Evaluation Project,14 trained staff conducted interviews using computer-assisted telephone interviewing equipment. Baseline interviews occurred between July and October 2007, and follow-up interviews occurred in October and November 2008. Participants received 10 dollars for completing each survey. The Institutional Review Board at the University of North Carolina approved the study.

We selected North Carolina counties that had (1) high cervical cancer rates (defined as annual cervical cancer incidence >10 cases and annual cervical cancer mortality >4 deaths; cervical cancer statistics are per 100,000 women); (2) >20% black residents; and (3) >1500 girls in the targeted age range of 10 to 18 years. One county was urban (Cumberland) and 4 others were rural (Duplin, Harnett, Sampson, Wayne). Annual cervical cancer rates in these counties ranged from 10.2 to 13.9 incident cases during 1993–2003 and 4.2 to 6.5 deaths during 1994–2004, substantially higher than corresponding US rates (8.6 incident cases and 2.9 deaths).15

Interviewers contacted a probability sample of households with telephone line access in these 5 counties using a dual-frame approach. Five percent of the sample came from a list-assisted random digit dialing telephone frame, while the remainder came from a nonoverlapping targeted-list frame of directory-listed residential telephone numbers with available recent household demographic information. We stratified the samples at the telephone exchange level by concentration of black residents and rural versus urban status (based on US Census 2000 block-level classification).16 We oversampled households likely to contain an age-eligible female child or blacks, as well as rural telephone exchanges.

Eligible households had to have one or more female children age 10 to 18 years. If a household had more than 1 eligible child, software randomly selected 1 as the index child for questions. Participants were parents, grandparents, or any other individual who self-identified as being responsible for the index child's care. Interviewers attempted first to speak with female caregivers and then male caregivers if a female caregiver was unavailable. As most caregivers interviewed (97%) reported being the parent or guardian of the index child, for the sake of simplicity, we refer to caregivers as parents hereafter.

Interviewers completed baseline surveys with 889 (73%) of 1220 eligible parents (flow diagram reported previously).14 Interviewers completed follow-up surveys with 650 (74%) of 873 baseline respondents, after excluding 16 ineligible respondents (Fig. 1). The mean age of parents was 41 (range, 21–73), and their daughters' mean age was 14 (range, 10–18). Half lived in rural areas. The majority of parents completing follow-up interviews were non-Hispanic white (56%) or black (35%), had attended college (76%), and lived in households with incomes greater than $50,000 per year (57%). Respondents and non-respondents during the follow-up interview had similar demographic characteristics and had similar proportions of daughters who had initiated HPV vaccine at baseline (Table 1).

Figure 1.:
Flow diagram for follow-up interviews. Diagram for baseline interviews reported by Hughes et al.14
Characteristics of Participants and Nonrespondents in Follow-Up Interview

Outcomes and Follow-Up

We assessed parents' report of daughters' receipt of one or more doses of HPV vaccine at baseline and follow-up. The primary outcome was HPV vaccine initiation at follow-up among daughters not vaccinated at baseline. We also assessed timely vaccine completion at follow-up, defined as having received all 3 doses or having received 1 or 2 doses but being not more than 2 months past the time recommended for receiving the next dose. The recommended HPV vaccine regimen at the time of the study was to receive the second dose 2 months after the initial dose, and the final dose at 6 months.5

Potential predictors of vaccine initiation included health belief model constructs (beliefs about likelihood and severity of cervical cancer; beliefs about HPV vaccine benefits, safety, and perceived barriers assessed using the Carolina HPV Immunization Attitudes and Beliefs Scale17; and doctor's recommendation) as well as demographic characteristics of parents and their daughters. The survey used an open-ended question to assess parents' main reason for not having gotten their daughters HPV vaccine.

Statistical Analyses

We applied sampling weights to analyses (including percents, means, and risk ratios [RRs]) to account for the study design. Frequencies are not weighted. We used Poisson regressions to calculate RRs and 95% confidence intervals (CIs) for predictors of reported vaccine initiation (defined as one or more doses). Models examined predictors assessed during the baseline interview (except for political affiliation and being a born-again Christian, which we assessed at follow-up) and vaccine initiation assessed at follow-up, for daughters not vaccinated at baseline. We entered statistically significant bivariate predictors (P < 0.05) into a multivariate Poisson regression model. To identify items in multi-item scales that predicted vaccine initiation, we conducted post hoc multivariate regressions controlling for other statistically significant multivariate predictors. Statistical analyses were conducted using Stata 10.0 (College Station, TX), 2-tailed tests, and critical α = 0.05.


Parents reported increasing frequency of HPV vaccine initiation and completion over time (Fig. 2), with most who initiated vaccine for their daughters doing so in 2008. At baseline, 12% (83/650) of parents said their daughters had initiated the 3-dose HPV vaccine series. Of parents whose daughters had not initiated vaccine at baseline, 27% (149/567) reported their daughters had received vaccine by follow-up. Most parents of daughters who received any doses by follow-up reported their daughters had received all 3 doses (58%, 137/232) or had received 1 or 2 doses but were on schedule to complete the regimen within recommended time guidelines (28%, 60/232). Few parents reported vaccination dates that indicated their daughters were more than 2 months past the date recommended for the second dose (9%, 20/232) or final dose (6%, 15/232). Parents whose daughters had not initiated HPV vaccine at baseline (n = 567) are the analytic sample for the rest of the paper.

Figure 2.:
Dates of HPV vaccine receipt by dose. 232 parents reported their daughters received 553 doses. Parents reported information about year of receipt for 321 of these doses.

Reasons for Not Vaccinating

At baseline, the most common reasons for not getting daughters HPV vaccine were that parents needed more information about the vaccine (21%, 110/567) and that their daughters were too young to get it (18%, 82/567) (Table 2). Only 3 parents said the main reason for not vaccinating was that it would cause their daughters to become sexually active. At follow-up, the main reason reported for not having vaccinated adolescent daughters was that they had not been to the doctor or gotten around to it yet (18%, 74/418).

Reasons Parents Gave for Their Daughters Not Receiving Human Papillomavirus (HPV) Vaccine

Predictors of HPV Vaccine Initiation

Parents' beliefs about HPV vaccine and their daughters were the main predictors of vaccine initiation (Tables 3–5). Parents' intentions to vaccinate their daughters were associated with initiating HPV vaccination for their daughters in multivariate analyses (RR = 2.04, 95% CI = 1.16–3.59). During the baseline interview, 62% (348/567) of parents who had not obtained HPV vaccine for their daughters said that they intended to in the next year. However, only 38% (126/348) reported having done so by follow-up. This rate is higher than the 10% (23/219) vaccination rate among parents who did not intend to vaccinate or were unsure, but it represents a substantial number of motivated parents who were unable to fulfill their intentions. At follow-up, most parents who reported that they had not yet obtained HPV vaccine for their daughters intended to vaccinate their daughters within the next year (57%, 225/418).

Longitudinal Predictors of Parent-Reported HPV Vaccine Initiation (Categorical Variables in Unadjusted Analyses)
Longitudinal Predictors of Parent-Reported HPV Vaccine Initiation (Continuous Variables in Unadjusted Analyses)
Longitudinal Predictors of Reported HPV Vaccine Initiation, Adjusted Analysis

Doctor's recommendation was associated with vaccine initiation, but only in bivariate analyses (Table 3). At baseline, few parents revealed that a health care provider had recommended they get HPV vaccine for their daughters (22%, 153/650). About half of parents who had received a doctor's recommendation but had not acted on it before the baseline interview did so during follow-up (51%, 46/94) compared to 21% (103/473) of parents who had not received a recommendation before the baseline interview. Parents of all but a few daughters who had not initiated HPV vaccination at baseline reported they had been to their doctor between baseline and follow-up (97%, 533/563; data not available for 4 parents). Of these parents, 28% (145/533) had initiated vaccination for their daughters by follow-up. Of the remaining parents, only 1 in 3 reported receiving doctors' recommendations for HPV vaccine (37%, 122/388), suggesting missed opportunities.

In multivariate analyses, parents were more likely to have gotten the vaccine for their daughters if they anticipated greater regret if not vaccinating did not protect their daughters from HPV infections (RR = 1.85, 95% CI = 1.13–3.02) (Table 5). Not initiating the vaccine was associated with being a born again Christian (RR = 0.53, 95% CI = 0.39–0.74), perceiving greater barriers to getting the vaccine (RR = 0.57, 95% CI = 0.39–0.83) or reporting that needing more information was the main reason for not having vaccinated at baseline (RR = 0.41, 95% CI = 0.22–0.76). Removing intentions from the multivariate model did not affect any variable's level of statistical significance. Perceived likelihood did not interact with perceived severity to predict vaccine initiation (data not shown, P = 0.96).

Post hoc analyses of the 5 items in the perceived barriers scale found that parents who reported their daughters had received at least 1 dose of HPV vaccine said it would be somewhat or very hard to find a provider or clinic that stocked the vaccine (4%, 11/149) less often than parents who reported their daughters had received no vaccine doses (14%, 45/418, P < 0.05). Vaccinators also said it would be hard to find a provider or clinic where the vaccine would be affordable (20%, 41/149) less often than parents who reported their daughters had received no vaccine doses (35%, 143/418, P < 0.05). Vaccine uptake was not associated with the remaining perceived barriers items (finding a provider or clinic that is easy to get to, finding a provider or clinic that has convenient appointments, and being able to pay for the vaccine).


In this longitudinal study, predictors of HPV vaccine initiation among adolescent girls differed from those suggested by HPV vaccine acceptability studies conducted primarily before vaccine licensure.10 Many health belief model constructs that we expected would be associated with uptake were not. Instead, key predictors of initiation included anticipated regret if their daughters got HPV that the vaccine could have prevented as well as not being a born-again Christian. While many studies have examined interest in a hypothetical HPV vaccine or cross-sectional correlates of vaccine initiation,10–13,18 this study adds important new information on predictors over time.

Over 2 years after HPV vaccine became available to the public, only about 1 in 3 eligible girls in an area with elevated cervical cancer rates had initiated the vaccine. This vaccine uptake is similar to that found in other US studies from the same time period.7 It is also similar to coverage for another adolescent vaccine (tetanus, diphtheria, acellular pertussis vaccine) about 2 years after its introduction in the United States (30% in 2007).7,19 Our findings suggest missed opportunities to increase HPV vaccine uptake, including a large percentage of daughters who had been seen by their doctors but had not received an HPV vaccine recommendation, and two-thirds of parents who intended to vaccinate their daughters but had not followed through with these plans a year later. One positive finding was parents' report that most girls who had received the vaccine were on schedule to receive their next dose or had received all 3 doses.

In contrast to what we previously found in cross-sectional analyses of our baseline data on uptake11 and in our systematic review of acceptability,10,12 HPV vaccine initiation was not longitudinally associated with key health belief model constructs (perceived risk, perceived severity, and physician recommendation as a cue to action). The small and not statistically significant associations with risk beliefs were within the range we observed in our previous meta-analysis of vaccine use among adults.20 However, the observed associations were much smaller than associations with anticipated regret from not vaccinating. While the special predictive power of anticipated regret from not vaccinating mirrors the findings of Weinstein et al.,21 we additionally show that anticipated regret from vaccinating, at least about sexual disinhibition, played essentially no role in vaccine decisions. This finding may be due to the different outcomes (cervical disease vs. sexual disinhibition), or it may reflect different beliefs about harms caused by action and inaction.22

Doctor's recommendation predicted reported HPV vaccine initiation in bivariate analyses, which is consistent with previous findings that doctors are uniquely credible and persuasive on issues related to medical care.23,24 The nonsignificant multivariate association may underestimate the importance of doctors' recommendations, because our longitudinal analyses evaluated vaccine initiation over time according to doctors' recommendations assessed at the baseline interview. As analyses only included those not yet vaccinated at baseline, daughters with doctor's recommendations were those who had gotten a recommendation before baseline, but had not acted on it, and thus were perhaps less likely to be vaccinated later. Physicians could play a larger role in encouraging HPV vaccine initiation by adolescent girls and in providing information about the vaccine to parents.

Uptake was lower among parents who said they needed more information about the vaccine. Messages from doctors or other respected professionals that focus on helping parents who already plan to act and reducing perceived barriers may be especially effective in increasing HPV vaccine uptake. Conversations about consequences of vaccination, such as potential side effects, may be better redirected to focus on the consequences of not vaccinating (e.g., anticipated regret). Simply imagining—and anticipating regretting—a future in which their daughters had HPV was a powerful motivator for parents.

Born-again Christian parents in our study were half as likely as other parents to get their daughters HPV vaccine. This is especially concerning, given that as many as 34% of the US population identify themselves in this way.25 A study in California found that born-again and evangelical Christian parents were less likely than other parents to prefer to vaccinate their daughters before age 13 than at older ages,26 but the study did not address the more fundamental question of willingness to get the vaccine at all or vaccine uptake. Our findings suggest public health programs to increase HPV vaccine uptake should make special efforts to reach born-again Christian parents.

Importantly, we found no differences in vaccine initiation by race, urbanicity, or age group. Given that the risk of dying from cervical cancer in the US is about twice as high for black women as for white women, equivalent but low uptake by blacks and whites may not be sufficient to reduce racial disparities.1 Additional efforts should focus on ensuring that HPV vaccine uptake is high across all groups, but these efforts should especially focus on those at highest risk. We previously reported higher uptake among older teens in analyses of our baseline data.27 The different findings might reflect a previous desire to catch up older teens to recommended vaccine guidelines or parents now being more comfortable with recommendations to vaccinate younger adolescent girls. Because HPV vaccine is likely to be most effective if given before adolescent girls initiate sexual activity, it may even be preferable to have higher rates of uptake for younger aged adolescent girls.

Study strengths include a diverse, population-based sample of parents from high-risk areas. The study's longitudinal design is also a strength, although we cannot completely rule out confounding by unmeasured variables. Limitations include the use of parent self-report to assess HPV vaccination of their female daughters. While studies have not yet examined accuracy of self-reported HPV vaccine initiation, adults' self-report of having received influenza vaccine is reasonably sensitive and specific.28,29 The generalizability of the findings to other populations is not yet known.

HPV vaccine research is important in groups at highest risk for cervical cancer, including blacks and Hispanics who have high cervical cancer rates and people living in rural areas with diminished access to care. In this high risk population, we found that many parents' intentions to vaccinate were unrealized, and doctors frequently missed opportunities to recommend HPV vaccine at clinic visits. Our study's findings suggest correlates of uptake (for example, anticipated regret) that offer novel opportunities for interventions designed to increase vaccine uptake in these and likely other populations.


1.American Cancer Society. Cancer Facts and Figures. Atlanta, GA: American Cancer Society, 2008.
2.Benard VB, Coughlin SS, Thompson T, et al. Cervical cancer incidence in the United States by area of residence, 1998–2001. Obstet Gynecol 2007; 110:681–686.
3.Smith JS, Lindsay L, Hoots B, et al. Human papillomavirus type distribution in invasive cervical cancer and high-grade cervical lesions: A meta-analysis update. Int J Cancer 2007; 121:621–632.
4.Villa LL, Costa RL, Petta CA, et al. High sustained efficacy of a prophylactic quadrivalent human papillomavirus types 6/11/16/18 L1 virus-like particle vaccine through 5 years of follow-up. Br J Cancer 2006; 95:1459–1466.
5.Markowitz LE, Dunne EF, Saraiya M, et al. Quadrivalent Human Papillomavirus Vaccine: Recommendations of the Advisory Committee on Immunization Practices (ACIP). MMWR Recomm Rep 2007; 56(RR-2):1–24.
6.Kahn JA. Maximizing the potential public health impact of HPV vaccines: A focus on parents. J Adolesc Health 2007; 40:101–103.
7.Centers for Disease Control and Prevention. National, state, and local area vaccination coverage among adolescents aged 13–17 Years—United States, 2008. MMWR Morb Mortal Wkly Rep 2009; 58:997–1001.
8.Shefer A, Markowitz L, Deeks S, et al. Early experience with human papillomavirus vaccine introduction in the United States, Canada and Australia. Vaccine 2008; 26(suppl 10):K68–K75.
9.Brabin L, Roberts SA, Stretch R, et al. Uptake of first two doses of human papillomavirus vaccine by adolescent schoolgirls in Manchester: Prospective cohort study. BMJ 2008; 336:1056–1058.
10.Brewer NT, Fazekas KI. Predictors of HPV vaccine acceptability: A theory-informed, systematic review. Prev Med 2007; 45:107–114.
11.Reiter PL, Brewer NT, Gottlieb SL, et al. Parents' health beliefs and HPV vaccination of their adolescent daughters. Soc Sci Med 2009; 69:475–480.
12.Ziarnowski KL, Brewer NT, Weber B. Present choices, future outcomes: Anticipated regret and HPV vaccination. Prev Med 2009; 48:411–414.
13.Rosenthal SL, Rupp R, Zimet GD, et al. Uptake of HPV vaccine: Demographics, sexual history and values, parenting style, and vaccine attitudes. J Adolesc Health 2008; 43:239–245.
14.Hughes J, Cates JR, Liddon N, et al. Disparities in how parents are learning about the human papillomavirus vaccine. Cancer Epidemiol Biomarkers Prev 2009; 18:363–372.
15.National Cancer Institute. Surveillance, Epidemiology, and End Results Cancer Statisitics. Updated 2008. Available at: http://seer.cancer.gov/faststats/index.php.
16.US Census Bureau. Census glossary. Updated 2008. Available at: http://factfinder.census.gov/home/en/epss/glossary_a.html.
17.McRee AL, Brewer NT, Reiter PL, et al. The Carolina HPV Immunization Attitudes and Beliefs Scale (CHIAS): Scale development and associations with intentions to vaccinate. Sex Transm Dis 2010; 37:234–239.
18.Zimet GD, Liddon N, Rosenthal SL, et al. Chapter 24: Psychosocial aspects of vaccine acceptability. Vaccine 2006; 24(suppl 3):S3/201–S3/209.
19.Broder KR, Cortese MM, Iskander JK, et al. Preventing tetanus, diphtheria, and pertussis among adolescents: Use of tetanus toxoid, reduced diphtheria toxoid and acellular pertussis vaccines recommendations of the Advisory Committee on Immunization Practices (ACIP). MMWR Recomm Rep 2006; 55(RR-3):1–34.
20.Brewer NT, Chapman GB, Gibbons FX, et al. Meta-analysis of the relationship between risk perception and health behavior: The example of vaccination. Health Psychol 2007; 26:136–145.
21.Weinstein ND, Kwitel A, McCaul KD, et al. Risk perceptions: Assessment and relationship to influenza vaccination. Health Psychol 2007; 26:146–151.
22.Ritov I, Baron J. Reluctance to vaccinate: Omission bias and ambiguity. J Behav Decis Mak 1990; 3:263–277.
23.Kahn JA, Rosenthal SL, Hamann T, et al. Attitudes about human papillomavirus vaccine in young women. Int J STD AIDS 2003; 14:300–306.
24.Boehner CW, Howe SR, Bernstein DI, et al. Viral sexually transmitted disease vaccine acceptability among college students. Sex Transm Dis 2003; 30:774–778.
25.Kosmin BA, Keysar A, eds. American Religious Identification Survey (ARIS) 2008. Hartford, CT: Institute for the Study of Secularism in Society and Culture, 2008.
26.Constantine NA, Jerman P. Acceptance of human papillomavirus vaccination among Californian parents of daughters: A representative statewide analysis. J Adolesc Health 2007; 40:108–115.
27.Gottlieb SL, Brewer NT, Sternberg MR, et al. Human papillomavirus vaccine initiation in an area with elevated rates of cervical cancer. J Adolesc Health 2009; 45:430–437.
28.Skull SA, Andrews RM, Byrnes GB, et al. Validity of self-reported influenza and pneumococcal vaccination status among a cohort of hospitalized elderly inpatients. Vaccine 2007; 25:4775–4783.
29.Zimmerman RK, Raymund M, Janosky JE, et al. Sensitivity and specificity of patient self-report of influenza and pneumococcal polysaccharide vaccinations among elderly outpatients in diverse patient care strata. Vaccine 2003; 21:1486–1491.
© Copyright 2011 American Sexually Transmitted Diseases Association