Does intention to recommend HPV vaccines impact HPV vaccination rates?

Abstract

Despite recommendations for routine vaccination, HPV vaccination rates among adolescent females have remained low. The objective of this prospective cohort study was to determine whether clinician intention to recommend HPV vaccines predicts HPV vaccine series initiation among previously unvaccinated 11 to 18 year-old girls (N = 18,083) who were seen by a pediatric clinician (N = 105) from a large primary care network within 3 years of vaccine introduction. We used multivariable logistic regression with generalized estimating equations, Cox Regression and standardized survival curves to measure the association between clinician intention and time to and rate of first HPV vaccine receipt among eligible females. All models adjusted for patient age, race / ethnicity, payor category, visit type, and practice location. Eighty-5 percent of eligible 11 to 12 year-old and 95% of 13 to 18 year-old girls were seen by a provider reporting high intention to recommend HPV vaccines. However, only 30% of the cohort initiated the HPV vaccine series and the mean number of days from first eligible visit to series initiation was 190 (95% C.I. 184.2, 195.4). After adjusting for covariates, high clinician intention was modestly associated with girls’ likelihood of HPV vaccine series initiation (OR 1.36; 95 % C.I. 1.07, 1.71) and time to first HPV vaccination (HR 1.22; 95% 1.06, 1.40). Despite high intention to vaccinate among this cohort of pediatric clinicians, overall vaccination rates for adolescent girls remained low. These findings support ongoing efforts to develop effective strategies to translate clinician intention into timely HPV vaccine receipt.

Abbreviations

HPV

Human papillomavirus

EMR

Electronic Medical Record

Introduction

In June 2006, the US. Food and Drug Administration approved a quadrivalent human papillomavirus (HPV) vaccine for females 9–26 years of age.¹ The vaccine is highly effective in preventing persistent HPV infection, cervical, anal, vulvar and vaginal precancers and cancers and anogenital warts associated with the 4 HPV types targeted by the vaccine - HPV 6, 11, 16 and 18.^2-5 A bivalent vaccine for HPV 16 and 18 was approved in October 2009 for the prevention of cervical cancers and pre-cancers (MMWR).⁶ The bivalent vaccine is approved for females and the quadrivalent HPV vaccine is now approved for both females and males. Despite recommendations for routine vaccination of all 11–12 year olds with catch-up vaccination of 13–26 year old girls and 13–21 year old males, immunization rates have remained low compared to other adolescent vaccines and may have plateaued.^7,8

Low vaccination rates have prompted many studies to evaluate facilitators and barriers to vaccination among parents and clinicians.^9-26 These studies have shown overall positive attitudes toward the vaccine. Still, concerns remain including vaccine safety, long-term efficacy, belief that vaccination is not necessary for young adolescents who are not sexually active and the perception that adolescents are at low risk for HPV infection. Surveillance data have also shown socioeconomic and racial disparities in completion of the vaccine series, suggesting potential barriers related to vaccine cost and access.^8,27-31

Studies of compliance with vaccination guidelines consistently show that physician recommendation is one of the strongest predictors of vaccine receipt, and positive recommendations have the potential to significantly improve vaccination rates. Well-established behavioral change theories would also predict that clinicians’ intention to recommend HPV vaccines is one of the most powerful predictors of vaccine recommendation.^32,33 However, this association has not been widely evaluated for HPV vaccination.³⁴ Clinician support of vaccination is generally high, especially for older adolescents,^17,35,36 yet rates remain well below Healthy People 2020 goals. Especially concerning are recent National Immunization Survey data that demonstrate lack of provider recommendation among parents of 13–17 males and declining parental intention to vaccinate for HPV despite increasing frequency of provider recommendation among parents of 13 to 17 year old girls. This suggests that other factors are impacting vaccine decision making.⁷ The objective of this study was to investigate the relationship between pediatric clinicians’ intention to recommend HPV vaccines and receipt of an HPV vaccine among adolescent girls. We hypothesized that likelihood of vaccine series initiation would be higher and time to series initiation would be shorter among eligible girls who were cared for by clinicians who reported high intention to recommend HPV vaccines. We also hypothesized that this effect would be most pronounced among 11–12 year old compared to 13–18 year-old girls.

Results

Clinician characteristics

Of 178 eligible clinicians, 105 (59%) consented to join the study and 101 fully completed the questionnaire (96% of those who consented) and were followed for the entire study period. Seventy-6 percent of the respondents were female, 86% were white, approximately 50% practiced in urban clinics and 46.8% reported >50% patients receiving Medicaid / SCHIP (State Children's Health Insurance Program) insurance in their patient population. Seventy-8 percent of all respondents reported that they were extremely likely to recommend the vaccine to 11–12 year-old girls and 92% reported that they were extremely likely to recommend the vaccine to 13–18 year-olds girls.

Child characteristics

There were a total of 6,125 (33%) 11-12 and 12,250 (67%) 13-18 year-old girls (total N = 18,375) who were eligible to initiate the HPV vaccine series and had an encounter with one of the enrolled clinicians during the study period (Table 1). Approximately half of the encounters were for a well visit and 79% took place at a suburban practice. Sixty percent of the study population was white and 16% of girls were insured through Medicaid. These characteristics did not differ significantly between age groups (p > 0.05). Eighty-3 percent of 11–12 year-olds and 96% of 13–18 year-olds had an encounter with a provider who reported high intention to recommend HPV vaccines. Thirty percent of the cohort (23% of 11–12 year-olds and 34% of 13–18 year-olds) initiated the HPV vaccine series during the study period and the mean number of days from first eligible visit to receipt of the first HPV vaccine dose was 190 (95% CI: 184, 195). The mean number of days to vaccine receipt was 225 (95% C.I. 213.1, 236.9) among 11–12 year-olds and 177 (95% C.I. 171.3, 183.8) among 13–18 year-olds (p < 0.0001).

Table 1.

Characteristics of 11–18 year old girls eligible for HPV vaccine receipt who had an encounter with a participating clinician, stratified by age group, N (%)

		11–12 years, N = 6,125 N (%)		13–18 years, N = 12,250 N (%)
	TOTAL N = 18375	HPV 1437 (23)	No HPV 4688 (77)	HPV 4137 (34)	No HPV 8113 (66)
Race/Ethnicity		P <0.0001		P < 0.0001
Black	5132 (28)	567 (36)	993 (64)	1491 (42)	2081 (58)
Hispanic a	289 (2)	34 (30)	80 (70)	63 (36)	112 (64)
White	10717 (58)	711 (19)	2,995 (81)	2199 (31)	4812 (69)
Other b	2237 (12)	125 (17)	620 (83)	384 (26)	1108 (74)
Location		P < 0.0001		P < 0.0001
Suburban	14376 (78)	907 (19)	3992 (81)	2876 (30)	6601 (70)
Urban	3999 (22)	530 (43)	696 (57)	1261 (45)	1512 (55)
Insurance Type		P < 0.0001		P < 0.0001
Medicaid	2945 (16)	385 (40)	566 (60)	888 (45)	1,106 (55)
Other	15093 (84)	1,023 (20)	4,040 (80)	3,190 (32)	6,840 (68)
Doctor Intention		P < 0.0089		P < 0.0001
High	16773 (91)	1220 (24)	3840 (76)	3900 (33)	7813 (67)
Low	1602 (9)	217 (20)	848 (80)	237 (44)	300 (56)
Encounter Type		P < 0.0001		P < 0.0001
Preventive Visit	8901 (48)	1167 (37)	1951 (64)	3018 (52)	2765 (48)
Sick Visit	8839 (48)	233 (8)	2609 (92)	972 (16)	5025 (84)
Other	632 (4)	36 (22)	127 (78)	146 (31)	323 (69)

^aIncludes Hispanic or Latino Black, Hispanic or Latino Other, and Hispanic or Latino White ^bIncludes American Indian/Eskimo/Alaska Native, Asian, Native American/Pacific Islander, Other and Refused to Identify Race

P value is for the χ² test of association for categorical variables

Receipt of first HPV vaccine dose

Intention was significantly associated with receipt of the first HPV dose for each age group in univariable analyses (Table 1). Among 11–12 year-old girls, there was a positive association between clinician intention and patient vaccination: 24% of girls whose clinician reported high intention to vaccinate received the HPV vaccine, compared to 20.4% of girls whose clinician reported low intention to vaccinate (p=0.009). However, among 13-18 year-old girls, there was an inverse association between clinician intention and patient vaccination: 33% of girls whose clinician reported high intention to vaccinate received the HPV vaccine, compared to 44% of those whose clinician reported low intention (p < 0.0001). Sociodemographic and health care utilization characteristics were also significantly associated with receipt of the first HPV dose across the total population and within each age group (Table 1).

In multivariable models (Table 3), intention was significantly associated with vaccine receipt even after adjusting for patient age, race / ethnicity, payor category, practice location, and visit type (OR 1.36; 95 % CI 1.07, 1.71). When stratified by age group, intention was significantly associated with vaccine receipt among 11–12 year-olds (OR 1.58, 95% CI 1.05, 2.38) but not among 13–18 year-olds (OR 1.06 95% C.I. 0.85, 1.32). Likelihood of vaccination was also significantly associated with Medicaid coverage (OR 1.39, 95% CI 1.19, 1.61), presenting for a preventive visit at time of vaccination or first eligible visit (O.R. 5.70, 95% C.I. 4.64, 7.01) and older age (OR 1.56, 95% CI 1.35, 1.97 for ages 13–18 compared to 11–12 year-olds). Girls who received care in suburban practices and who were defined as Other race were less likely to be vaccinated (OR 0.55, 95% CI 0.41, 0.73 and OR 0.72, 95% C.I. 0.63, 0.82).

Table 3.

Adjusted odds ratios for likelihood of HPV vaccine receipt in the total population and by age group

	Total Population	11–12 year olds	13–18 year olds
	A.O.R.a (95% C.I.)b	A.O.R. (95% C.I.)	A.O.R. (95% C.I.)
Intention
Low	Ref	Ref	Ref
High	1.36 (1.07, 1.71)e	1.58 (1.05, 2.38)e	1.06 (0.85, 1.32)
Insurance
Medicaid	1.39 (1.19, 1.61)f	1.62 (1.24, 2.11)g	1.29 (1.11, 1.50)g
Other	Ref	Ref	Ref
Race/Ethnicity
Whitec	Ref	Ref	Ref
Black	1.01 (0.85, 1.22)	1.09 (0.86, 1.38)	0.99 (0.81, 1.20)
Hispanicd	1.21 (0.85, 1.73)	1.73 (0.98, 3.03)	0.99 (0.65, 1.51)
Other	0.72 (0.63, 0.82)f	0.72 (0.60, 0.88)f	0.70 (0.60, 0.83)f
Location
Urban	Ref	Ref	Ref
Suburban	0.61 (0.48, 0.78)f	0.46 (0.32, 0.65)f	0.71 (0.54, 0.93)e
Visit Type
Preventive	5.70 (4.64, 7.01)f	6.63 (5.07, 8.80)f	5.55 (1.56, 4.09)f
Sick	Ref	Ref	Ref
Age
11–12 years	Ref
13–18 years	1.56 (1.35, 1.97)

^a Adjusted Odds Ratio- Logistic Regression with Generalized Estimating Equations

^b 95% Confidence Interval

^c Includes Hispanic or Latino Black, Hispanic or Latino Other, and Hispanic or Latino White

^d Includes American Indian/Eskimo/Alaska Native, Asian, Native American/Pacific Islander, Other and Refused to Identify Race

^e <0.01

^f <0.0001

^g <0.001

Time to receipt of first HPV vaccine dose

In univariable analysis, time to receipt of first vaccine dose was significantly shorter among girls seen by clinicians reporting high intention compared to clinicians reporting low intention (187.3 days vs. 217.8 days, p < 0.004) across the whole cohort and among 11–12 year-old girls (220 days vs. 253.3 days, p < 0.05) (Table 2). There was no significant association between time to first HPV vaccine dose and clinician intention among older girls. Conversely, sociodemographic characteristics were significantly associated with time to receipt of first vaccine dose among 13–18 year-old but not 11–12 year-old adolescents (Table 2).

Table 2.

Mean number of days from first eligible office visit to first HPV dose during the study period, (95% C.I.)

	Total Population 189.8 days (184,195)	11–12 years old 225 days (213,237)	13–18 years old 177.6 days (171,184)
Race/Ethnicity	p = .06	p = .60	p = .03
Black a	185.4 (171,200)	222.7 (193,253)	171.2 (155,188)
White	195.3 (183, 208)	231.3 (205,258)	183.7 (170,197)
Hispanic b	144.5 (78, 211)	195.1 (73,317)	117.3 (38,197)
Other	184.8 (156,214)	207.7 (144,272)	177.3 (145,210)
Insurance Type	p = .14	p = .27	p = .12
Medicaid	181.6 (170,193)	212.2 (189,235)	168.4 (155,182)
Other	191.8 (185,198)	227.7 (213,242)	180.4 (173,188)
Location	p = .09	p = .63	p = .003
Suburban	193.1(186, 200)	222.8 (208,233)	183.8 (176,191)
Urban	182.8 (173,193)	228.8 (209,249)	164.4 (152,175)
Doctor Intention	p = .004	p = .05	p = .56
High	187.3 (182,193)	220.0 (207,233)	177.1 (171,184)
Low	217.8 (201, 235)	253.3 (222,284)	185.2 (158,213)
Encounter Type	p = < .0001	p = .16	p = < .0001
Preventive Visit	182.1 (173,191)	219.2 (200,238)	167.8 (157,178)
Sick Visit	218.3 (201,236)	250.1 (208,292)	210.7 (192,229)
Other	176.1(132,220)	243.3 (135,351)	159.5 (112,207)

^a Includes Hispanic or Latino Black, Hispanic or Latino Other, and Hispanic or Latino White

^b Includes American Indian/Eskimo/Alaska Native, Asian, Native American/Pacific Islander, Other and Refused to Identify Race

In multivariable models, intention was significantly associated with time to receipt of the first HPV vaccine dose across the entire study population even after adjusting for patient race / ethnicity, patient age, insurance, visit type and practice location (HR 1.29; 95% 1.16, 1.42). When stratified by age group, intention was significantly associated with time to HPV vaccine series initiation only among 11–12 year-olds (HR 1.44, 94% C.I. 1.21, 1.72). This is illustrated by standardized survival curves (Fig. 1).

Figure 1.

Survival Curve- Time to Receipt of First HPV Vaccine Dose Standardized for Race / Ethnicity, Age, Insurance, Visit Type and Practice Location.

Discussion

In this prospective cohort study of clinicians in a pediatric primary care network, we observed that higher intention to recommend HPV vaccines was associated with a shorter time to receipt of the first vaccine, particularly among 11–12 year-olds for whom the vaccine is recommended for routine administration. In our cohort, 11–18 year-olds who were seen by clinicians reporting high intention received their first vaccine dose about one month earlier than girls seen by clinicians reporting low intention. We also observed a higher rate of HPV vaccine initiation among 11–12 year-old girls cared for by clinicians with higher intention to vaccinate after adjustment for demographic covariates. Presuming that reported intention to recommend vaccination is a reliable proxy for recommendation behavior, these results confirm that provider recommendation is an important predictor of vaccine acceptance, particularly among younger adolescents. However, our results also suggest that other factors mediate the relationship between clinician intention to recommend vaccination and actual vaccine receipt.

Provider recommendation is recognized as one of the most powerful predictors of vaccine acceptance. Recent studies illustrate a significant relationship between provider recommendation and HPV vaccine receipt.^25-27,37,38 However, there are also multiple barriers to HPV vaccine receipt identified in previous studies that may influence this relationship. Notably, recent National Immunization Survey data shows that a smaller proportion of parents intend to vaccinate for HPV compared to the previous year, despite increased frequency of provider recommendation, and a higher proportion of parents report vaccine safety concerns.⁷ This suggests that provider recommendation may not be enough to assuage concerns regarding HPV vaccines. Indeed, a very recent study shows that informational content specifically targeting misconceptions about the safety of MMR vaccines may increase concerns about side effects.³⁹ It is unclear if a similar relationship exists for HPV vaccines.

Our findings may also reflect differences in the strength of recommendations employed by providers in our cohort. When faced with a reluctant parent, clinicians who have made an initial recommendation may be reluctant to push a parent to accept vaccination at that time.⁴⁰ Hughes et al showed that some clinicians presented HPV vaccine as part of the routine immunization schedule while other presented it as an optional vaccine.⁴⁰ While both clinicians could have similar intention to make a recommendation regarding HPV vaccines, these approaches communicate very different messages to parents and may have very different results. Opel et al. have observed that a presumptive recommendation (stating that vaccines need to be given) is associated with a higher likelihood of vaccine acceptance compared to a participatory recommendation (asking what the family would like to do about vaccines), providing further evidence that the recommendation approach can significantly influence outcome.⁴¹ In addition, clinicians may have their own beliefs regarding HPV vaccines that may impact the strength of their recommendation, including their own concerns regarding vaccine safety and efficacy or perceived risk for HPV infection among their patient population.^42,43

A smaller proportion of 11–12 year-olds compared to 13–18 year-olds were vaccinated during the study period (23% vs. 34%) and had a longer time to receipt of first HPV vaccine dose (225 days vs. 177 days). We therefore stratified our analyses by age to determine whether intention may differentially impact vaccine receipt in younger compared to older girls. We hypothesized that there may be effect modification by age since multiple studies have shown that both clinicians and parents are more likely to recommend or accept vaccination for older compared to younger adolescents.^{17,22,23,36,43} Indeed, in our cohort, almost all clinicians reported high intention to recommend vaccination to girls ≥13 years old; therefore it is not surprising that we did not see a significant association between intention and vaccine receipt in this age group.

We also found an association between sociodemographic characteristics and healthcare seeking behavior, i.e. having a preventive health visit. Adolescent girls who presented for a preventive visit were 5 times more likely to initiation the HPV vaccine series compared to girls who presented for a sick visit. These results support those noted in previous studies and highlight the importance of other factors that may influence the likelihood of vaccine receipt regardless of provider recommendation. The majority of both 11–12 and 13–18 year-olds girls in our study were vaccinated during a well-child visit. Previous research has shown that having a preventive healthcare visit is associated with a higher likelihood of vaccine receipt among adolescents.^25,44-47 A preventive care visit may provide more opportunity for anticipatory guidance regarding vaccination and parents may be more likely to accept discussion about vaccines at a well visit. Adolescent girls and their parents may be more likely to see their regular healthcare provider with whom they have an established, trusting relationship at a preventive rather than sick visit. Trust has been shown to positively impact vaccine acceptance.⁴⁸ A provider may also be more comfortable making a strong recommendation to a family with whom there is an established relationship.⁴⁹ Conversely, this association also points to the potential for missed opportunities for vaccination when adolescents present for sick or other non-preventive visit types. Any encounter may provide an opportunity for vaccination.⁴⁷

Across the entire cohort, girls seen by clinicians who reported high intention to recommend HPV vaccination initiated the HPV vaccine series one month earlier than girls followed by clinicians with low intention, even when adjusting for location and insurance type. While other studies have investigated associations between insurance type, sociodemographic characteristics and time to series initiation and completion, to our knowledge, this is the first study to identify an association between intention and time to series initiation.^44,50 This suggests that clinicians with higher intention may be more proactive in overcoming hesitancy and emphasizing the importance of vaccination. Additionally, baseline data from this cohort showed that high intention clinicians were also more likely to be early adopters of new technologies and recommendations.⁵¹ High intention clinicians may have been more comfortable endorsing vaccination and making a strong recommendation at the first opportunity. Early endorsement may therefore accelerate vaccination for parents who were open to eventually vaccinating. However, the modest association with overall vaccination rates suggests that even providers with high intention may not be effective in addressing other factors that impact parental and teen decision-making about vaccines. This would be consistent with literature suggesting that strongly held beliefs about vaccines may be difficult to address and modify with brief provider informational interactions and may instead require an ongoing dialog that targets parental concerns.^52-58

There are several limitations to our study. Although utilizing a cohort within a primary care network with an integrated EMR allowed for linkage between clinicians, practice and vaccination data and for enrollment of a diverse patient population, the clinician sample size may have been too small for sufficient power to detect small differences in vaccination rates. Our results were also highly skewed toward high clinician intention, particularly for older adolescents. This may have impeded our ability to detect a significant relationship between intention and vaccination. However, we did stratify by older and younger age groups and it is notable that despite high intention, vaccination rates remained low. Our results focus upon the first few years of vaccine introduction. The relationship between clinician intention and subsequent vaccination may have changed. Lastly, vaccine receipt is not a precise measure of actual recommendation - we were not able to determine whether a clinician actually recommended the vaccine at a given visit and it was refused or whether the clinician decided not to recommend vaccination.

Conclusions

This is one of the first studies to prospectively examine the association between reported clinician intention to recommend HPV vaccines and actual vaccine receipt. Such an analysis has important implications for improving HPV vaccination rates as provider recommendation is a key facilitator of vaccine acceptance, yet vaccination rates continue to lag behind rates for other adolescent vaccines. Despite high intention in our study and a significant shortening of the time to vaccine receipt, overall vaccination rates for the adolescent girls in this study remained low. These findings should motivate ongoing efforts to develop the most effective strategies to translate clinician intention into HPV vaccine receipt.

Building upon this work, future studies can better elucidate actual provider recommendation practices at the point of care. Potential interventions may include provider education efforts to promote strong recommendations, especially for young adolescents. Provider education may also enhance clinicians’ ability to address specific parental concerns. It will also be important to evaluate the influence of communication from other sources - social media outlets, for example, may provide a powerful mechanism for the dissemination of education about adolescent vaccines. Lastly, a growing body of literature has begun to evaluate the effectiveness of tailored messaging to target specific concerns - HPV vaccination provides an excellent platform for the continued evaluation of such interventions. Since the introduction of HPV vaccines, morbidity and mortality from cervical and other anogenital cancers among men and women are now readily preventable. It is imperative to improve HPV vaccination rates so that this potential can be fully realized.

Methods

Study design and study subjects

In this prospective cohort study, we analyzed data from pediatricians and nurse practitioners (N = 179) practicing in a primary care network affiliated with a large metropolitan children's hospital. To be eligible for the study, clinicians had to be employed either full-time or part-time at one of the 29 practices in the primary care network. We did not include fellows or residents in training. We then collected data regarding HPV vaccination for all girls 11–18 years who had an encounter with an enrolled clinician during the study period (November 2006 – February 2009). Patients were excluded if they had received a HPV vaccine prior to study initiation. We did not include males as HPV vaccination was not approved for boys during the study period. The protocol was approved by the Institutional Review Boards of The Children's Hospital of Philadelphia and the University of Pennsylvania.

Data sources

Between December 2006 and February 2007, during the time of first HPV vaccine availability within study site practices, enrolled clinicians were asked to complete a 31-item, on-line, theory-based and validated questionnaire measuring intention to recommend HPV vaccines and associated knowledge, attitudes and beliefs regarding HPV and HPV vaccines.^22,35 Data regarding intention to recommend HPV vaccination within this cohort have been previously published.⁵¹ For this study, the primary predictor of interest was clinician intention to recommend HPV vaccines. This was measured using the following item: “What is the likelihood that you will recommend the HPV vaccine to girls in the following age groups within the first 12 months after it becomes available for clinical use: 9–10 year-olds, 11–12 year-olds, 13–18 year-olds and 19–26 year-olds?” Responses for each age group ranged from extremely likely to extremely unlikely (5-point scale), however, based on the survey instrument results, clinician's responses were dichotomized into high intention vs. low intention. We analyzed data from responses addressing intention to recommend vaccines to 11–12 and 13–18 year-old girls, the age groups that comprise the majority of adolescent patients who present for regular care in the study network.

HPV vaccination data was obtained from the CHOP electronic medical record (EMR) (EpicCare®, Epic Systems Corporation, Verona, WI). The EMR contains clinical and sociodemographic information including immunization history, family history and residential address. All administered vaccine doses are recorded in the EMR and are associated with a specific encounter. The primary outcomes of interest were (1) receipt of the first HPV vaccine dose and (2) time to receipt of the first HPV vaccine dose. Time to receipt of first HPV vaccine dose was measured as the number of days between the date of the first visit during the study period when the patient was eligible for HPV vaccine and the date of first vaccine receipt. Only the quadrivalent vaccine was available during the study period.

The following patient-level covariates were abstracted from the EMR: encounter type (nurse visit, preventive visit, sick visit or other), insurance type (Medicaid vs. Other), age and race/ethnicity. Additional clinician characteristics included practice location which was defined as either suburban or urban. These covariates were selected based upon their potential to either confound or modify the relationship between clinician intention to recommend HPV vaccines and HPV vaccination.

Statistical analysis

The goal of this analysis was to identify any significant association between intention to recommend the HPV vaccines and HPV vaccine receipt among 11-18 year-old girls eligible to initiate the HPV vaccine series during the study period, while controlling for potential confounders and effect modification. We stratified our analyses by age group (11-12 years and 13-18 years) as reported intention was significantly different in these age groups. The primary outcome was measured as both a categorical variable, receipt of the first HPV vaccine dose, and a continuous variable, time to receipt of the first HPV vaccine dose.

We performed univariable analyses using Chi-square tests of association for the categorical outcome (HPV vaccine receipt) and ANOVA for the continuous outcome (mean number of days to receipt of first vaccine dose) to compare proportions of girls who were vaccinated in each age group and across each categorical covariate to identify any significant associations. We stratified these results by practice type to evaluate for effect modification.

We then performed multivariable analysis to determine whether intention was an independent predictor of HPV vaccine receipt after adjusting for selected covariates. For the outcome of HPV vaccine receipt, we used logistic regression with a generalized estimating equation to evaluate the effect of clustering within individual practices. We included all covariates that were associated with the outcome at p < 0.05. These variables included practice location, race/ethnicity, payor category and visit type. Results of the multivariable logistic regression are presented as odds ratios with 95 % confidence intervals. For the continuous outcome (number of days to receipt of first HPV vaccine), we constructed a proportional hazard model and accounted for clustering by clinician clinic site using the Wei-Lin-Weissfeld method.⁵⁹ We then constructed a standardized Cox Proportional Hazards (PH) model to evaluate associations between time to first HPV immunization among female patients and clinician intention adjusted for race/ethnicity, practice type, and payor category. We presented hazard ratios with 95 % confidence intervals and a survival curve standardized for the same covariates to validate the assumptions of the proportional hazards model. All statistical analyses were conducted using SAS version 9.3 (SAS Institute Inc.).

Disclosure of Potential Conflicts of Interest

K.A. Feemster has received honoraria from Pfizer, Inc.. and Abbott, Inc.. for speaking and participating on an advisory board. She has received no funding from any commercial sources.

A. G. Fiks has received an independent research grant from Pfizer, Inc.. The funded work has no relationship to vaccination.

J.A. Kahn has co-chaired 2 clinical trials of an HPV vaccine; the trials were NIH funded but Merck, Inc.. provided vaccines and immunogenicity testing. She also chaired a grant review committee for the Society for Adolescent Health and Medicine which provided funding for projects to improve adolescent vaccination; funding for that program was from Merck, Inc.

Acknowledgments

The authors would like to thank the Pediatric Research Consortium (PeRC) and the Healthcare Analytics Unit at the Children's Hospital of Philadelphia for their assistance in recruitment and data management.

Funding

This work was supported in part by the Robert Wood Johnson Foundation through the Robert Wood Johnson Foundation Clinical Scholars Program, the University of Pennsylvania School of Medicine and an institutional training grant (T32) from the Children's Hospital of Philadelphia.