Factors associated with initiation and completion of the quadrivalent human papillomavirus vaccine series in an ontario cohort of grade 8 girls

In July 2006, Health Canada approved Gardasil^®, a quadrivalent human papillomavirus (HPV) vaccine designed to protect against strains of HPV that cause 70% of cervical cancers [1] and 90% of genital warts [2]. The following year, the Canadian government allocated $300 million to the provinces and territories on a per-capita basis to launch a national HPV immunization initiative for young girls [3]. Accordingly, the Ontario government received $117 million and, by September 2007, implemented a program offering free HPV vaccination to all Grade 8 girls in the province. The program is administered by each of Ontario's 36 local public health agencies (LPHAs) primarily through school-based immunization clinics and provides all three indicated doses of the quadrivalent HPV vaccine. Since HPV vaccination is not mandatory, the decision of whether or not to receive the vaccine is left to the parents/guardians of eligible girls [4].

While the initial goal of the Ontario government was to immunize 80% of eligible girls during the program's first two years [5], only 53% received the first dose in year 1 [3]. This level of vaccine coverage is not only considerably lower than that of similar school-based vaccination programs in the province [6], but also represents the lowest known level of HPV vaccine acceptance in Canada [7]. The reasons for such low coverage in Ontario remain unknown; however, surveys have consistently reported that parental concerns about the safety and long-terms effects of the HPV vaccine affect intent to vaccinate [8‐12]. As a result, clinical factors, such as medical history and health services utilization, may be important predictors of HPV vaccine receipt. For instance, parents of a girl with a serious health condition may be more reluctant to consent to having their daughter receive a new vaccine for fear that it might further impact their child's health. On the other hand, increased contact with healthcare providers (e.g., physicians) may lead to increased opportunities to discuss the benefits of vaccination, thereby increasing HPV vaccine acceptance and use.

Given that low vaccine coverage undermines public health efforts aimed at reducing the burden of HPV-related illnesses and threatens the program's cost-effectiveness [13, 14], it is important to monitor use of the HPV vaccine and identify factors that influence use in order to inform future HPV immunization programs. To this end, we conducted a population-based, retrospective cohort study of girls residing in Kingston, Frontenac, Lennox, and Addington (KFL&A) who were eligible for Ontario's Grade 8 HPV vaccination program in 2007-08 and 2008-09 to describe HPV vaccine use and to identify socio-demographic and clinical factors associated with this use.

This study was approved by the Research Ethics Boards of Queen's University and Sunnybrook Health Sciences Centre.

Based on birth year, we determined that 185,398 girls were eligible for Ontario's Grade 8 HPV vaccination program in its first two years (Figure 1). Of these, 2519 (1.36%) were residing in KFL&A and comprised our study cohort. Cohort members were between 12.7 and 13.6 years of age at cohort entry.

Overall, 56.6% (n = 1425) of eligible girls received at least one dose of the HPV vaccine and less than half (48.2%) received all three (Figure 2). Among vaccinated girls, 85.3% (n = 1215) completed the three dose series, while 11.2% (n = 160) received two doses and 3.5% (n = 50) received only the first.

Series initiation increased significantly across campaign years (OR = 1.29; 95% CI 1.08-1.53), but the increase corresponded to an absolute difference of only 4.0% (Table 1). While there was no significant change in series completion between program years (OR = 0.88; 95% CI 0.65-1.20), the proportion of non-completers receiving two doses was higher in the second year of the program compared to the first (13.1% vs. 9.4%; p = 0.03).

After adjustment, there were no statistically significant differences between vaccinated and unvaccinated girls with respect to health services utilization, medical history, or socio-demographics, except girls in the fourth income quintile were less likely to receive the HPV vaccine than those in the third quintile (Table 2). Conversely, compared to their unvaccinated counterparts, HPV vaccinated girls were significantly more likely to have received the MMR (OR = 3.21; 95% CI 2.25-4.57), meningococcal C (OR = 2.16; 95% CI 1.64-2.84), or hepatitis B (OR = 2.86; 95% CI 2.14-3.83) vaccine and were even more likely to have received all three vaccines (OR = 4.89; 95% CI 4.04-5.92).

Similar to series initiation, series completion was not significantly associated with age, health services utilization, or medical history (Table 3). On the other hand, series completion was associated with certain socio-demographic characteristics. For one, girls living in the lowest income neighbourhoods were significantly less likely to complete the three-dose regimen than girls living in middle-income neighbourhoods (OR = 0.45; 95% CI 0.28-0.72). In addition, girls residing in rural areas were significantly more likely to complete their HPV vaccine series than those residing in urban areas (OR = 1.76; 95% CI 1.14-2.74). Although previous MMR, meningococcal C, or hepatitis B vaccination was not significantly associated with HPV vaccine completion, receipt of all three vaccines was (OR = 1.85; 95% CI 1.29-2.65).

Among eligible girls residing in KFL&A, HPV vaccine initiation was low, but completion of the three-dose regimen was high. While initiation increased significantly between the first and second year of the immunization program, the absolute increase was small and vaccine coverage remained far below that required to maximize the health benefits and cost-effectiveness of the publicly funded program. The factor most strongly associated with series initiation was a history of vaccine use. Income was associated with series completion, but not initiation.

Cost has been shown to be an important barrier to HPV vaccine use [28, 29]. Therefore, our finding that series initiation was not influenced by income demonstrates the benefit of providing the HPV vaccine through a publicly funded program. At the same time, we found that those in the lowest income quintile were the least likely to receive all three indicated doses of the vaccine, suggesting that inequities in access may exist despite the availability of free immunization. This is particularly important given that low income has also been associated with lower rates of cervical cancer screening [30, 31] and higher rates of cervical cancer incidence and mortality [32, 33]. Since non-completers may not acquire the full benefits of HPV immunization and girls of low income have increased need for protection against HPV-related infections, low series completion in this population could have important public health implications and requires further investigation.

Although series completion was lower among those of low income, overall, completion was high among vaccinated girls in our study population. This likely reflects the convenience of a school-based program, where a girl who received consent need only be present to receive her scheduled dose of the vaccine. Among non-completers, girls were three times more likely to receive two doses of the HPV vaccine as opposed to only one, suggesting that lack of completion may be due to factors additional to the withdrawal of consent. One reason for the observed pattern may be attributed to the fact that if a girl is absent on the day of a school immunization clinic she must go to her LPHA or family physician to obtain her missed dose. Low income has been consistently related to school absenteeism [34‐36] and it is possible that girls of lower income have fewer resources to follow-up on a missed dose at a location other than their school. This hypothesis is consistent with our findings that income is associated with series completion but not initiation.

As expected, a history of vaccine use, measured using both mandatory (MMR) and optional (meningococcal C and hepatitis B) vaccines, was the strongest predictor of HPV vaccine uptake. These findings are consistent with survey data that indicate that parental attitudes and beliefs about vaccines in general predict their attitudes toward the HPV vaccine [8, 37]. Nevertheless, in our study cohort, overall initiation of the HPV vaccine (56%) was still well below that of the meningococcal C (76%) and hepatitis B (78%) vaccines that were offered to the same girls just one year before (i.e., in Grade 7). The low uptake of the HPV vaccine compared to these other optional vaccines suggests that factors additional to "anti-vaccine attitudes" are contributing to poor acceptance of the HPV vaccine.

Low use of the HPV vaccine in Ontario has been partially attributed to the program's rapid implementation and to the ban on the promotion of provincially funded programs that was in place at the time of the program's launch because of Ontario's 2007 provincial election campaign [38]. However, since series initiation barely increased in the second year of the program, our results do not support this hypothesis. Although a number of other factors have also been proposed to explain HPV vaccine avoidance (e.g., fear of its negative influence on sexual behaviour, low perceived risk of HPV infection, lack of knowledge about HPV), surveys consistently report parental concerns about the safety and long-term effects of vaccines as important barriers to vaccination [8‐12, 39].

We were unable to directly assess whether safety concerns affected receipt of the HPV vaccine in our population, but we were able to explore the relationship between medical history and health services utilization and vaccine usage. Although these associations were not statistically significant, there were some interesting trends. First, it appeared that non-initiation was associated with an increasing number of hospitalizations. Since admission to a hospital is a medical rather than personal decision, a high number of hospitalizations likely indicates the presence of a serious health problem. Therefore, as hypothesized, our results suggest that parents of sicker girls may be less likely to consent to HPV vaccination than those of relatively healthy girls. Furthermore, in our study, a high number of outpatient visits was associated with an increased likelihood of receiving the HPV vaccine, suggesting that increased contact with healthcare providers may be related to increased vaccine acceptance. It is important to note that, given the general good health of the age group studied and the corresponding low event rates, we had limited power to assess the influence of these clinical factors on HPV vaccine receipt. As such, these potential associations should be re-examined in a larger sample.

Although a couple of studies have utilized administrative rather than survey data to study the correlates of HPV vaccine use in the United States [40, 41], to our knowledge, ours is the first to link a population-based immunization database with administrative health data in the context of a publicly funded program. This provided us with a unique opportunity to report on patterns of use of the HPV vaccine and examine the clinical factors associated with its acceptance in a situation in which cost of vaccination is not an issue. While this study also benefits from the use of validated HPV vaccination data [26], it has a number of important limitations. For one, because school grade was not available in the databases, we used birth year to identify eligible girls. While children are typically in Grade 8 thirteen years after they are born, this is not the case for those who are advanced or held back a school grade. Therefore, we may have included some ineligible girls in our cohort (e.g., girls born in 1995 who were held back a grade) and excluded some eligible girls (e.g., girls born in 1996 who were advanced a grade). Although these errors may have caused us to underestimate the number of girls vaccinated, given the small proportion of girls who are advanced or held back a grade, the proportion affected by this is expected to be small. In addition, although data were available to assess series completion during the Grade 8 school year, we did not have vaccination data after December 31, 2009. As such, series completion for any girl in the 1995 birth cohort who completed her regimen after December 31, 2009 of her Grade 9 year would have been misclassified. Another limitation is that, since the provincial health databases contain only information on services claimed under OHIP, the clinical history of girls who have not resided in Ontario throughout their lifetime or who have received privately funded care may be incomplete. Also, the accuracy of many of the diagnostic codes used to identify medical history has not yet been assessed, nor has the validity of using neighbourhood income as a proxy for household income in this population. Another disadvantage of these administrative health data is that we were unable to consider factors that have previously been identified as possible predictors of HPV vaccine acceptance, such as race/ethnicity, perceived vulnerability to HPV infection, and concerns about the vaccine's long-term effects and impact on sexual behavior [8‐11, 37]. Since these factors may have introduced residual confounding into our study, the influence of these factors on series initiation and completion now needs to be assessed taking clinical predictors into account. Finally, since KFL&A is a medium-sized health district of Ontario (N = 184,407, 1.5% of the province) with a slightly lower than average median family income ($66,466 vs. $69,156) and a low proportion of visible minorities (5.0% vs. 23.0%) [42], the generalizability of our results to other regions remains unclear.

The current low level of HPV vaccine use in KFL&A will likely have important implications in terms of the health benefits and cost-effectiveness of its publicly funded vaccination program. We identified important factors that were associated with series initiation and completion that should be considered in efforts to guide future HPV vaccine programming. For example, girls in the lowest income quintile were the least likely to complete the recommended three-dose regimen, suggesting that program delivery should be modified to improve series completion in vulnerable populations. Furthermore, since vaccine acceptance is known to be related to concerns about the safety and long-term effects of vaccines, our findings of possible associations between HPV vaccine coverage and medical history and health services utilization require further study.