Association between parent attitudes and receipt of human papillomavirus vaccine in adolescents

The human papillomavirus (HPV) vaccine was introduced in 2006, but coverage has remained lower than for other recommended adolescent vaccines for tetanus, diphtheria, pertussis (Tdap), and meningococcal disease [1]. HPV vaccine series completion, which included three doses in 2015, was recently estimated at 35% in the U.S. Factors contributing to low population-level HPV vaccine coverage are not well understood. Parents are generally aware of the HPV vaccine [2], but specific knowledge about HPV vaccination (e.g., schedule, benefits) is weakly correlated with actual coverage [3]. While multiple causes likely exist, low coverage is hypothesized to be a function of hesitancies, ambivalence, or resistance by many parents who are deciding about their child’s healthcare. In considering their adolescent’s young age and (presumed) timeline until sexual exposure, parents typically underestimate their child’s susceptibility to acquiring sexually transmitted infections or cervical cancer [4, 5]. Parents also tend to overestimate the risks of HPV vaccine adverse events [4, 6, 7], some of which are real (e.g., syncope) [8], while causal links to others have been repudiated (e.g., thromboembolism [9], risky sexual behaviors [10]) despite widely available false claims.

The degree to which parents’ HPV vaccine attitudes influence their decisions to have their adolescent children get the HPV vaccine series has been examined in prior studies [11‐13]. Most prior investigations, however, were cross-sectional and assessed self-reported vaccination or vaccine intent. There are few longitudinal studies that have examined how HPV vaccine attitudes influence HPV vaccine series completion, particularly those that take advantage of more objective vaccination data sources such as medical records. The purpose of this study was to examine the association between changes in parents’ HPV vaccine attitudes and HPV vaccine uptake in their adolescents, using electronic health records (EHR) data from a north-central Wisconsin integrated healthcare system. The hypothesis was that positive changes in parental attitudes would be associated with greater adolescent HPV vaccine uptake.

At baseline, 1998 surveys were mailed to eligible parents, with 221 (11%) respondents. Among those, 187 returned responses at the 1-year follow-up, with 175 confirmed as the same parent who responded at baseline (cohort retention 79%). Eleven respondents were excluded from analyses due to missing covariate data, yielding a final analytical sample of 164 parents. Parent respondents were primarily Non-Hispanic White females (Table 1). Because surveys were mailed to the parents of study-eligible adolescent patients (and any parent/guardian could respond), the enumerated sample was not identifiable. Thus parent information (e.g., age, sex) only became available upon survey completion, which precluded comparisons between invited parents who did vs. did not respond to the survey. Comparisons were possible on some adolescent characteristics though. Relative to non-responders, study-eligible adolescents of parents who responded to the survey were significantly younger and more likely to have received Tdap and meningococcal vaccines. There was no difference in the proportion of adolescents who received a prior HPV vaccine.

At baseline, 32% of parents’ adolescent children had initiated the HPV vaccine, having received 1–2 prior HPV vaccine doses (none had completed the 3-dose series, per study eligibility criteria). Sixty percent of parents’ adolescent children (58% of males, 62% of females) had received at least one HPV vaccine dose and 38% (44% of males, 33% of females) had completed the 3-dose series by the end of the 18-month follow-up period. As outlined in Table 2, the highest baseline CHIAS sub-factor scores were for harms (mean ± SD = 2.2 ± 0.6 points) and uncertainties (2.2 ± 0.7 points), followed by ineffectiveness (2.1 ± 0.9 points), and barriers (1.4 ± 0.4 points). Barriers and harms scores remained stable over time. Scores for uncertainties and ineffectiveness improved, both decreasing by 0.2 points (~10% improvement; p’s < 0.001) over one year. At the 1-year follow-up survey, 77% of parents reported having received a physician’s recommendation for their adolescent child to get the HPV vaccine in the prior year.

Prior research has shown that parents’ CHIAS scores are significant predictors of their intent to get their child vaccinated against HPV, as well as their self-reported history of HPV vaccine receipt [15, 20, 21]. To our knowledge, this was the first study to examine how changes in parents’ attitudes were associated with improvements in objectively-measured HPV vaccine receipt. Despite relatively good uptake in HPV vaccine during the 18-month study timeframe, vaccine attitudes appeared to be less flexible in parents, as concerns about effectiveness and uncertainties waned modestly, while beliefs about barriers and harms did not change. In the multivariable model, only change in the CHIAS uncertainties score was significantly associated with receipt of the next dose of HPV vaccine. Parents who experienced a favorable shift in uncertainties had a greater likelihood of their adolescent child receiving at least one HPV vaccine dose during the follow-up period. No changes in CHIAS scores were associated with completion of the HPV vaccine series.

Though the baseline survey response rate was low and somewhat skewed toward parents of younger adolescents who were more apt to have received Tdap/meningococcal vaccines, parental concerns about the safety of the HPV vaccine seemed to be the most influential attitudinal barrier. On average, the CHIAS harms score remained stable between baseline and follow-up. But even among parents whose CHIAS harms score shifted favorably, their adolescents were not more likely to get the HPV vaccine after covariate adjustment. In contrast, the baseline CHIAS harms score was strongly associated with receipt of the next HPV vaccine dose and was the only significant attitudinal predictor of HPV vaccine series completion. Thus it seemed parents with high harms scores were not apt to change their minds, and even if they did, it had no impact on HPV vaccine receipt. This finding was consistent with some prior studies. Notably, the CHIAS harms score was also among the strongest predictors of HPV vaccine initiation after one year among parents of adolescent girls in North Carolina [21]. Also, fear of side effects was among the main reasons cited for avoiding HPV vaccination (or intent to vaccinate) in cross-sectional studies with Dutch [22] and Romanian parents [23]. This suggests that public awareness and patient education interventions should emphasize the HPV vaccine’s post-licensure safety profile [9]. How to best translate such messaging to a clinical environment, however, is yet unclear. Compared to brief/straightforward recommendations, more in-depth conversational approaches by medical providers may yield limited influence on parents’ decision-making regarding HPV vaccination [24, 25].

Several covariates were also significantly associated with HPV vaccine receipt. As expected, a prior HPV vaccine dose was the strongest predictor of both receipt of next dose and series completion. A reported physician’s recommendation to get the HPV vaccine in the prior year was also associated with receiving the next HPV vaccine dose, but adolescents’ age and gender were not associated with receipt of next dose or series completion. Interestingly, parents with publicly-assisted health insurance had six times greater odds of their adolescent child initiating or getting their next HPV vaccine dose during the follow-up period relative to parents with private health insurance. Reasons for this are obviously speculative, but it may be related to some Medicaid pay-for-performance initiatives targeting childhood immunizations [26], recognizing that such incentives have had limited impact in other settings [27, 28]. Why public-assisted health insurance did not translate into greater likelihood of HPV vaccine series completion though was unclear, but could be related to fewer total preventive care visits in adolescents from lower income households [29].

This study was strengthened by the systematic sampling of parents of adolescents who receive primary care within an integrated regional healthcare system, a level at which clinical care quality improvement initiatives are apt to occur. Also, adolescents’ clinical data on HPV vaccine coverage was linked to their parents’ CHIAS scores. In terms of limitations, the baseline survey response rate was of particular concern because it was very low, which might reflect sensitivities about the general topic of HPV or the lack of a baseline response incentive. Low baseline response does not negate the associations observed within the cohort and is not by itself indicative of nonresponse bias [30, 31], but it can increase the potential for selection biases if respondents were very different from non-respondents. Unfortunately, a comparison between parents who did vs. did not respond to the survey was impossible since only those who responded could be identified from the enumerated sample. The adolescent children of parent respondents vs. non-respondents, however, were significantly younger and more likely to have received Tdap/meningococcal vaccine, but not HPV vaccine. The assumption in this analysis was that changes in parents’ HPV vaccine attitudes affected their decision to have their adolescent child get the HPV vaccine. But the temporality of the attitudinal exposures (or time when measured) and HPV vaccination period overlapped some, thus reverse causation could be influential to a degree in that some parents’ HPV attitudes measured at the 1-year follow-up may have only shifted after (and as a function of) their adolescent child received the HPV vaccine. Other study limitations included the limited generalizability of our sample, which was relatively small and selected from a racially homogenous source population.

HPV vaccine coverage remains low in the U.S. [1], but improving parents’ HPV vaccine attitudes is considered a key factor in increasing these low rates [32]. How flexible some parents’ HPV vaccine attitudes are, or how responsive to targeted intervention they may be, remains uncertain. Though the baseline response rate was low in this sample of Wisconsin parents, attitudes about the HPV vaccine’s effectiveness and certainty in having their adolescent child get it improved modestly. Reduced uncertainties scores over one year was a significant predictor of adolescent children getting their next scheduled HPV vaccine dose, while baseline harms score (which, on average, did not improve over time) was the main attitudinal factor associated with HPV vaccine series completion. Given the positive influence of medical providers’ advice on parents’ HPV vaccine decisions in prior studies [17‐19], as well as our current study, the healthcare system appears to be an optimal setting to test future methods to improve HPV vaccine attitudes and coverage.