Risk Factors for Non-initiation of the Human Papillomavirus Vaccine among Adolescent Survivors of Childhood Cancer

Genital human papillomavirus (HPV) is the most common sexually transmitted infection (1) and has a causal role in the expression of cervical and other cancers (2). Approximately, 80% of sexually active women are exposed to HPV during their lifetime (3), and HPV is most prevalent among females of ages 20 to 24 years (4). Rates increase sharply after the median age of sexual debut, 16.6 years for females in the United States (5). Recent efforts to reduce cervical cancer have led to the development of vaccines to protect against HPV, which are currently available and have been shown to be safe and effective (6–10). Quadrivalent HPV vaccination, approved in 2006 for females between 9 and 26 years of age (11) protects against HPV types 16 and 18 (which account for 70% of cervical cancers) and 6 and 11 (which account for 90% of genital warts; ref. 12). In 2009, HPV vaccination was also approved for males (13).

Routine HPV vaccination is currently recommended by the Advisory Committee on Immunization Practices for adolescent girls of ages 11 and 12 years, with catch-up vaccination for women through 26 years of age (14). It is recommended that the vaccine be administered before sexual debut due to the mechanism of HPV transmission (11). With appropriate use of the vaccine, the American Cancer Society estimates a potential reduction of cervical cancer risk by more than 70% over the next decade (15, 16). HPV vaccine uptake is particularly important for females surviving childhood cancer, many of whom are at increased risk for HPV-related complications secondary to the direct and indirect effects of cancer treatment. Survivors at increased risk for HPV persistence and complications include those with a history of hematopoietic stem cell transplantation (17), Hodgkin lymphoma (18, 19), treatment with pelvic irradiation (20, 21), and those receiving other cancer treatments resulting in sustained immunosuppression (22–26). Survivors of childhood cancer seem to also be at increased risk for HPV infection/complication/escalation given the unique behavioral, cognitive, and educational consequences of treatment. Specifically, survivors of childhood cancer are less likely than their healthy siblings to have undergone a Papanicolaou (Pap) smear within the previous 3 years (27). Survivors are also more likely to experience neurocognitive deficits such as impulsivity and inattention resembling attention deficit hyperactivity disorder, which have been associated with increased risky and sexual behaviors (28–32). In addition, survivors of childhood cancer are more likely to report unemployment, lower educational attainment, and lower annual incomes (33), factors independently associated with HPV infection. As such, the Children's Oncology Group's (COG) Long-Term Follow-Up Guidelines for Survivors of Childhood, Adolescent and Young Adult Cancer Version 3.0, which is the template for screening late effects of cancer treatment, has recommended HPV vaccination for all eligible females surviving childhood cancer (34).

Over the last 2 years, the National Immunization Survey for Teens (using clinic-validated repots) found that initiation rates among adolescent females in the general population ranged from 48.7% to 53.0%, whereas completion rates have ranged from 32.0% to 34.8% (35, 36). Both rates of initiation and completion are significantly lower than the 80% target established by the Healthy People 2020 initiative (37). Because the HPV vaccine was only approved recently by the U.S. Food and Drug Administration (FDA; 2006), little is known about the complexity of vaccination uptake among those surviving cancer. To date, no rigorous examination of HPV vaccination among survivors of childhood cancer has been reported. The current study serves as the first prevalence estimation of HPV vaccination initiation and completion among a large cohort of childhood cancer survivors, whereas also identifying factors that are most influential in HPV vaccination initiation and completion in this high-risk group.

Maternal caregivers who have daughters with a history of childhood cancer were recruited from the After Completion of Therapy (ACT) Clinic at St. Jude Children's Research Hospital (Memphis, TN). ACT is a long-term follow-up clinic for childhood cancer survivors who are more than 5 years postdiagnosis, and 2 years disease-free. Following completion of the study questionnaire, each mother was asked to provide contact information for up to five acquaintances to obtain a control sample demographically similar to the cancer group. This type of sampling allowed for the evaluation of HPV vaccination rates across cancer/no cancer groups while controlling for key demographic variables (daughter's age, socioeconomic status, maternal education, and region of the country), which have been found to influence vaccination uptake (36, 38–41). By controlling for these specific demographics, the distinguishing feature between groups was the presence/absence of daughter cancer history. Eligibility criteria for participants included: (i) the mother or female primary caregiver of a female 9 to 17 years of age, (ii) proficient in reading and writing English, (iii) cognitively able to understand and complete the study questionnaire, and (iv) willing and able to provide informed consent per Institutional Review Board (IRB) guidelines. Over an 18-month interval, a total of 235 mothers with daughters with a history of childhood cancer (“cancer survivor” group; daughter M_age = 13.2 years, SD = 2.69) and 70 mothers with daughters without cancer (“acquaintance control”; daughter M_age = 13.3 years, SD = 2.47) were enrolled in the study and returned questionnaires (Fig. 1). Cancer survivors and acquaintance controls did not significantly differ on any of the measured demographic variables (Table 1).

Among the cancer survivor group, maternal caregivers were recruited during their daughter's regularly scheduled ACT Clinic visit. A trained member of the research team approached mothers, explained the purpose of the study, and obtained informed consent as approved by the IRB. After consent was obtained, participants completed paper-and-pencil questionnaires, which took approximately 15 to 30 minutes. Potential acquaintance control participants were contacted via telephone based on contact information provided by the mothers with daughters with a cancer history (typically by accessing information stored on their cellular phones). Controls who verbally consented via telephone were provided the option of completing either an online questionnaire or a mailed paper-and-pencil questionnaire. Questionnaires were identical across the two groups, aside from identifying the St. Jude daughters as “patients.” After completing questionnaires, all participants were provided with an information sheet on HPV and HPV vaccination. Questionnaires were collected in 2010 and 2011.

HPV vaccine initiation/non-initiation was defined as a binary outcome variable such that mothers who reported their daughters have received one or more HPV vaccine doses were categorized as “initiated,” and those having received zero doses were categorized as “non-initiated.” HPV vaccine completion/noncompletion was defined as a binary outcome variable such that mothers who reported their daughters have received all three doses of HPV vaccine were categorized as complete, and those who have received at least one dose but less than three doses of HPV vaccine were categorized as incomplete. Participants whose daughters have received zero doses of the HPV vaccine were excluded in the modeling of HPV vaccination completion.

All participants completed questionnaires about their daughters' sociodemographic and medical history, HPV-specific knowledge and communication, and health beliefs.

Mothers provided familial demographic information, including maternal and child age, race/ethnicity, marital status, education level, and annual household income, along with medical history of gynecologic care and cervical cancer screening. Items were adapted from instruments previously used in the HPV vaccine literature (refs. 42–44; Table 1). Items measuring maternal perceptions of daughter's sexual activity and relationship status were also adapted from previous self-report questionnaires (41).

Knowledge of HPV, cervical cancer, and HPV vaccination was measured by a scale adapted from Brabin and colleagues (42). Correct responses to 10 multiple-choice items were summed for a total knowledge score, with higher scores representing greater knowledge. The questionnaire content was abstracted from the Centers for Disease Control and Prevention (CDC) HPV vaccination information website as well as other sources (42, 45). Familial communication about the messages and purpose of HPV vaccination was assessed via a four-item scale also adapted from Brabin and colleagues (42). The 18-item Mother–Adolescent Sexual Communication Instrument assessed maternal–adolescent sexual behavior and development communication (46). Internal reliability in our sample was high (α = 0.92) and convergent and discriminant validity have been previously established and described (46). Communication scores were recoded into binary variables (median splits) before model inclusion: HPV communication (Mdn = 14; range, 4–16), and sexual communication (Mdn = 68; range, 18–90).

The HPV Vaccine Health Beliefs Questionnaire (47) is a validated instrument designed to measure maternal perceptions of daughters' vulnerability to HPV, severity of HPV, barriers to, benefits of, and self-efficacy for initiating/completing the vaccine. Internal reliability was acceptable for all subscales in our sample: vulnerability (α = 0.95), severity (α = 0.87), barriers (α = 0.81), benefits (α = 0.82), and self-efficacy (α = 0.91). Cox and colleagues (47) also found the internal reliabilities of these factors to be robust, which contributed to establishing the predictive validity of health belief factors as it relates to HPV vaccination acceptability among mothers of girls of ages 11 to 16 years. Additional measures of vaccine-related Cues to Action and Social Environmental Influence were also considered with scales adapted from previously validated surveys (42–44, 47). Health belief scores were recoded into binary variables (median splits) before model inclusion: vulnerability (Mdn = 12.0; range, 5–25), severity (Mdn = 32; range, 8–40), barriers (Mdn = 25; range, 12–52), benefits (Mdn = 23; range, 8–35), and self-efficacy (Mdn = 24; range, 6–30).

Univariate analyses were used to examine differences between groups (cancer history/no history) in HPV vaccine initiation and completion. Univariate differences were also assessed as a function of sociodemographic, medical, knowledge and communication, as well as health belief factors. Comparisons with P values less than 0.10 were included in each of the two multivariate models (vaccine initiation and completion). Differences for continuous variables were assessed using univariate one-way ANOVA. Differences in categorical variables were assessed using χ² and Fisher exact tests. Multivariable logistic regression models were used to calculate OR and 95% confidence intervals (CI) for vaccine outcomes. Given that no differences emerged between groups on vaccination outcomes, cancer survivors and acquaintance control participants were combined in the presented multivariate models. Participant status (cancer vs. control) was also retained as a factor in both models.

Univariate differences emerged between cancer/no cancer groups on risk factors including vulnerability to HPV infection and complication (P = 0.04) and prediction of daughter's sexual activity (P = 0.09). Specifically, mothers of daughters with a cancer history perceived their child to be more susceptible to HPV infection and complication, but were less likely to predict that their daughters would be sexually active by high school graduation. No other significant cancer/control differences were found on any other sociodemographic and medical history variables, HPV-specific knowledge and communication variables, or health belief variables.

Overall, 32.6% (75 of 230) of cancer survivors and 34.3% (24 of 70) healthy controls had initiated the HPV vaccine series. Among those who had initiated, 56.0% (42 of 75) of cancer survivors and 58.3% (14 of 24) of healthy controls had completed the vaccine series. In the overall sample, 17.9% (42 of 230) of survivors and 20% (14 of 70) of controls had completed the vaccine. No significant differences emerged in the rates of vaccine initiation or completion between cancer survivors and healthy controls.

Univariate analyses revealed significant differences between those who have/have not initiated the HPV vaccine (see Tables 2 and 3). On the basis of the univariate findings, the multivariate model for HPV vaccine initiation included the following variables: race, daughter's age, adolescent history of gynecologic care, adolescent history of annual Pap smear, doctor recommendation for vaccine, parental permission to date socially, maternal perception of daughter's past and current relationship status, maternal perception of daughter's sexual activity status, maternal–adolescent sexual communication, and maternal health beliefs of vulnerability, barriers to vaccination, benefits of vaccinating, and self-efficacy about HPV vaccination (Table 4). The final multivariate logistic regression model predicting binary vaccine initiation outcome indicated that older daughter age and physician recommendation for vaccination were associated with an increased likelihood of HPV vaccine initiation. Furthermore, mothers who perceived greater barriers to having their daughter receive the HPV vaccine (e.g., financial or religious conflicts, concerns about vaccine promoting sexual activity in daughter, historical lack of vaccination endorsement, etc.) were less likely to have initiated the vaccine.

Univariate analyses for participants who had initiated the vaccine revealed significant differences between those who have/have not completed the vaccine series (Tables 2 and 3). On the basis of these differences, the final multivariate logistic regression model for HPV vaccine completion included the following variables: race, daughter's age, physician recommendation for HPV vaccination, parental permission for dating socially, and maternal health beliefs of vulnerability, severity, and barriers about HPV vaccination (Table 5). Among vaccine-initiated participants, those who were non-White, held perceptions of greater HPV severity, and who perceived greater barriers to vaccination were less likely to have completed the HPV vaccine series postinitiation.

Advances in the treatment of childhood cancer have resulted in the majority of survivors living into adulthood (48, 49). Given the reduction of mortality associated with cancer treatment, increased attention has been placed on promoting health and quality of life in survivorship (29, 50). HPV vaccination is one tool to assist in these efforts, and as such, a need exists to better understand vaccine prevalence and determinants in this vulnerable group.

On the basis of maternal report, the results of our study found that 32.6% of cancer survivors have initiated the vaccine series, whereas 17.9% have completed it. No differences in vaccine rates were identified between cancer survivors and acquaintance control groups, but univariate differences in known risk factors for vaccine initiation and completion did emerge. Specifically, mothers of survivors perceived greater vulnerability to HPV-related complication upon patient exposure but were less likely to believe that their daughters would engage in sexual activity before high school graduation. Although survivors are at increased risk for HPV-related complication, they did not engage in higher rates of vaccination. Cancer survivors and control participants were similar on many risk factors previously identified as being predictive of vaccination status, including age (39) physician recommendation (51–54) and race (36, 38, 55). The similarities between groups are consistent with previous research that identified no differences in risky sexual behavior between adolescent childhood cancer survivors and healthy siblings (56). Conceivably, interventions designed to increase vaccine uptake in the healthy population may be generalizable for use among childhood cancer survivor populations as well based on these similarities.

Among the entire sample, the modeling of determinants associated with vaccine initiation found that older daughter age and physician recommendation were both related to increased vaccine uptake, whereas perceptions of high vaccine barriers were associated with decreased initiation. Our study aligns with previous research showing that physician recommendation for HPV immunization is a robust predictor of vaccine uptake (52, 57). It is interesting to note that only half of all mothers endorsed physician recommendation for HPV vaccination. Amidst the nonsignificant cancer/control differences described in the results, a trend was seen in which a minority of survivor families received a physician recommendation for vaccination, whereas a majority of controls reported receiving one. This is discouraging given survivors' frequency of medical encounters and their increased risk for HPV-related complication (58). These data suggest potential confusion in vaccine management in that some primary care physicians may assume that oncologists are managing this aspect of care and vice versa. This lack of clarity may account for these less than optimal vaccine rates in the cancer group, and physician communication/recommendation may be targets of future intervention, particularly in light of physician recommendation being predictive of vaccine initiation. Physician endorsement of HPV vaccination, as well as problem-solving specific to perceived barriers to vaccine initiation or completion, may also be mechanisms to increase vaccine uptake in adolescents (59).

Once the vaccine has been initiated, the series must be completed to achieve maximum protection (6). Being non-White and perceiving high barriers and severity associated with HPV infection/complication are all risk factors decreasing the likelihood of vaccine completion. The interpretation of the finding about high perceptions of HPV severity and its association with noncompletion, while unexpected, is primarily a function/limitation of the cross-sectional study design. It seems that mothers whose daughters have completed the vaccine series are least concerned about the severity of future HPV complications. This is presumably due to the comfort associated with full vaccine protection. This also suggests that an awareness of HPV-related complications could act as a motivator for vaccine completion. Modifiable variables consistent with the Health Belief Model (60) such as perceived severity and barriers to vaccination (and to a lesser extent vulnerability to HPV infection), should be prioritized in vaccine promotion efforts. Furthermore, there may be other modifiable variables that, though significant at the univariate level, did not reach significance in the multivariate models. Given the opportunity, interventionists may consider targeting such variables, like HPV communication or health beliefs, as part of vaccine promotion efforts in the future.

This is the first study to report the prevalence and correlates of HPV vaccination in childhood cancer survivors; however, this work was limited by (i) inclusion of females from a single-site, (ii) survivors greater than 5 years postdiagnosis, and (iii) used maternal report of child vaccination only. Furthermore, with cross-sectional study designs, only associations (not causalities) can be determined between the considered risk factors and vaccine outcomes. This literature is in the early stages of quantifying HPV-specific risk profiles among survivors of childhood cancer (61), necessitating additional research in this population. In addition, our definition of incompletion did not differentiate pending versus intentionally incomplete participants. It is possible that our incomplete group is heterogeneous in their intentions to complete the vaccine series. Related to this point, although all adolescents in the study were within the age indication for receiving the vaccine, the time since vaccine initiation and questionnaire completion may have been less than 6 months. As such, it is possible that the reported rates of vaccine completion in this sample may vary as a function of time since vaccination initiation.

Despite the COG recommendation for vaccination, the immunogenicity of the HPV vaccine among childhood cancer survivors has not yet been shown. Among children living with HIV, for example, differences have been noted in quadrivalent HPV vaccine seroconversion, as titers against HPV subtypes 6 and 18 were shown to be 30% to 50% lower than age-matched controls (62). In response, a phase II study is warranted examining the safety, immunogenicity, and tolerability (as well as scheduling and dosing of vaccine) among survivors of childhood cancer. In addition, future research should examine cancer-specific factors (e.g., diagnosis, treatment, and etiology) and their associations with HPV vaccine outcomes among survivors. As the HPV vaccine is also approved for young adults and males, examining factors contributing to vaccinations in these groups are warranted as well.

In conclusion, cancer survivors are at increased risk for adverse late effects including second malignancies and organ compromise, and as such, would benefit from HPV vaccination. Findings of the current study establish the prevalence and identification of factors influencing HPV vaccination among adolescent females surviving childhood cancer and their peers. A minority of adolescents surviving childhood cancer have initiated or completed HPV vaccination despite their increased risk for HPV complication. Future interventions designed to increase vaccination among childhood cancer survivors may draw upon these study findings to enhance immunization rates and promote their daughters' health in the future.

R.H. Foster is employed as Pediatric Psychologist (part-time work providing therapy services) in Gundersen Health System. No potential conflicts of interest were disclosed by the other authors.

Conception and design: J.L. Klosky, D.M. Green, M.M. Hudson

Development of methodology: J.L. Klosky, M.M. Hudson

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.L. Klosky, J.R. Hodges, R.H. Foster, M.M. Hudson

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.L. Klosky, K.M. Russell, K.E. Canavera, J.R. Hodges, R.H. Foster, G.R. Parra, J.L. Simmons, M.M. Hudson

Writing, review, and/or revision of the manuscript: J.L. Klosky, K.M. Russell, K.E. Canavera, H.L. Gammel, J.R. Hodges, R.H. Foster, G.R. Parra, J.L. Simmons, D.M. Green, M.M. Hudson

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.L. Klosky, K.M. Russell, K.E. Canavera, J.R. Hodges, J.L. Simmons

Study supervision: J.L. Klosky, R.H. Foster, M.M. Hudson

This work was supported by the Cancer Center Support grant (CA21765; R. Gilbertson, Principal Investigator); and The American Lebanese Syrian Associated Charities (ALSAC).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.