Early Initiation of HPV Vaccination and Series Completion in Early and Mid-Adolescence

This study uses public-use data from the 2017–2020 waves of the NIS-Teen survey. NIS-Teen (https://www.cdc.gov/nchs/nis/data_files_teen.htm) is conducted by the Centers for Disease Control and Prevention and includes noninstitutionalized 13- to 17-year-old subjects from all 50 states and the District of Columbia. Although adolescents from Guam and the US Virgin Islands are included, these data are not available in the public use data files. Furthermore, data from Puerto Rico was not available in 2017 and 2018 because of hurricanes. The public-use data files contain data for all adolescents who have a completed household interview. Inclusion criteria for this study were (1) 15- to 17-year-old subjects in NIS-Teen 2017–2020 living in continental United States, (2) parent/guardian gave permission for NIS-Teen to contact their health care provider, (3) provider-reported HPV immunization history, and (4) provider-reported subject’s age of first HPV vaccine was between 9 and 12 years. Finally, because reasons why children who initiated between the ages of 9 and 12 years had more than the recommended number of HPV vaccinations were not known, the specific dose for series completion could not be determined for those who had 4 to 6 HPV immunizations (N = 340); as such, analyses were further restricted to those whose providers reported the subject received no more than 3 doses of HPV vaccine. There were 19 575 adolescents in NIS-Teen 2017–2020 who met these 5 inclusion criteria to create the study sample.

Figure 1 presents the study design along with key HPV vaccination program milestones. To create a retrospective cohort study from NIS-Teen 2017–2020, an accelerated longitudinal design approach was used. The accelerated longitudinal design uses a subject’s age rather than study year to organize data into birth cohorts. Six birth cohorts (2000–2005) were developed from the 4 NIS-Teen study years to create a retrospective cohort study (See Supplemental Figure 3 and Supplemental Tables 5 and 6 for details).

NIS-Teen includes several provider-reported measures of the age of first HPV vaccination (years, months, days). The age of HPV vaccination in years was dichotomized (9–10 years vs 11–12 years) to create the early initiation variable.

For all outcome measures, HPV series completion was determined per ACIP-recommended dose schedule. The HPV vaccine was initially approved as a 3-dose series. In 2016, ACIP recommended that those who initiated the vaccine at younger than age 15 years could receive a 2-dose schedule (0, 6–12 months). NIS-Teen includes a variable noting for those who initiated HPV vaccination before age 15 years if a subject had the requisite number of doses per ACIP guidelines and, for those with 2 doses, that there was an interval of at least 5 months, 4 days, between the first and second dose. As described in the following section, this NIS-Teen variable was used to derive the time and developmental pathway outcome variables for this study.

To assess the time pathway, 2 measures of HPV vaccination series completion were used: series completion by age 13 years (early adolescence) and series completion by age 15 years (mid-adolescence). We constructed a variable to denote the age of the last HPV vaccination received based on the total number of HPV vaccines and the age at which the final dose was received. Age at last HPV dose was set to the age of the second HPV vaccination for those who received 2 HPV vaccines and equal to the age at the third HPV vaccination for those who received 3 HPV vaccines. For series completion by ages 13 and 15 years, we constructed binary variables based on the NIS-Teen variable noted previously describing completion per ACIP recommended guidelines and the age of the last HPV vaccine received. Subjects who were up to date per ACIP guidelines and who received their last HPV vaccination before age 13 years were considered up to date by age 13 years. Likewise, subjects who completed before age 15 years were considered up to date by age 15 years.

To assess the development pathway, we constructed a binary variable to denote series completion within 3 years of initiation. The time interval between first and last HPV vaccination was calculated. Subjects were considered complete within 3 years if this interval was less than 3 years and the subject was complete per ACIP guidelines.

Demographic, geographic, and provider covariates and their response categories from NIS-Teen included sex (female, male), race/ethnicity (non-Hispanic white, non-Hispanic Black, Hispanic, other, including other non-Hispanic race + multiple races), insurance status (private only, any Medicaid, other, uninsured), maternal education (less than 12 years, 12 years, greater than 12 year but less than college, college), Census region of residence (Northeast, Midwest, South, West), and type of facility where all of the subject’s providers practiced (all private = private, all public = public, all hospital = hospital, all providers from a Special Supplemental Nutrition Program for Women, Infants, and Children clinic, school-based health center, pharmacy, military health care or other facility = school/other, and mixed/unknown). The NIS-Teen variable for income to poverty ratio was dichotomized at 1.5 to represent living in poverty (income to poverty ratio less than 150% or at least 150%). To assess urbanicity/rurality, 2010 US Census data were used.⁹  The US Census defines an urbanized area as one with at least 50 000 people and a rural area as one with less than 2500 people. Data on the percent of the population living in urban and rural areas by state (and the District of Columbia) is available from the US Census (https://www.census.gov/programs-surveys/geography/guidance/geo-areas/urban-rural/2010-urban-rural.html; accessed March 16, 2022). The distribution of percent of the population living in an urban area was explored and the cut point representing the top quintile determined. States with the percent of their population living in an urban area above the cut point (N = 10) were categorized as “urban.” Similarly, distribution of percent of the population living in a rural area was explored and the cut point representing the top quintile determined. States with the percent of their population living in a rural area above that cut point (N = 9) were categorized as “rural.” The state-based categorization of urban and rural were merged into the NIS-Teen data by matching to the state variable within NIS-Teen, which also includes the District of Columbia.

Analyses were conducted using SPSS version 26. Sample weights were used only in regression analyses because this study is looking at correlates of completion rather than estimating vaccination coverage across years in the NIS sample, uses a select subset of NIS-Teen participants, and has created a retrospective cohort from multiple years of NIS-Teen data to take advantage of the provider-reported dates of vaccination. Descriptive statistics for the HPV measures and covariates were provided first. Next, bivariate contingency table analyses using Pearson χ² tests were performed to explore the associations between demographic, geographic, and provider variables and early initiation. An a priori level of significance of α < 0.05 was set that, with Bonferroni adjustment led to P < .0125 to establish statistical significance. To determine if earlier age of initiation was an independent correlate of series completion, blocked stepwise logistic regression based on the literature and on bivariate analyses was performed. In step 0, a univariate model with just early initiation was run to determine the unadjusted odds ratio (OR) and 95% confidence intervals (CIs). Next, in step 1, a model without early initiation but including demographic, geographic, and provider covariates was run. Finally, early initiation was entered into the fully adjusted model to assess if early initiation was independently associated with time and development pathway outcomes and explore if early initiation influenced any disparities demonstrated in step 1. Adjusted ORs and 95% CIs are reported from both unweighted and weighted regression analyses.

To test the time pathway, this series of models was run separately for completion by age 13-year and completion by age 15-year outcomes. To test the development pathway, this series of models was run for the complete within 3-year outcome variables.

Table 1 describes the study population. More than one-half of these 19 575 adolescents were female (52.1%), non-Hispanic white (57.1%), or had private insurance only (57.3%). Early initiation was uncommon (7.5%). There was a more than twofold rise in series completion between the ages of 13 (34.5%) and 15 (83.4%) years. The majority (84.8%) were complete within 3 years.

Table 2 shows correlates of early initiation. All factors except living in a state with a high proportion living in an urban area were significantly associated with early initiation. Females were 1.5 times more likely to initiate at 9 to 10 years compared with males (8.9% vs 5.9%). Early initiation was also higher among socially disadvantaged populations, whether defined by race/ethnicity, socioeconomic status, or insurance. Geography was associated with early initiation: living in the South was associated with greater likelihood of early initiation, whereas living in a state with a high proportion living in a rural area decreased likelihood of early initiation. Early initiation was more common among those whose providers all practiced at public facilities or at school/other facilities compared with all hospitals, all private, or mixed/unknown facilities.

Table 3 shows demographic, geographic, and provider variables associated with the 3 measures of HPV vaccine series completion. Females were more likely to complete HPV vaccination by ages 13 and 15 years, but, although statistically significant, the magnitude of the differences were small. Completion within 3 years did not differ by sex. Table 3 reveals multiple disparities in series completion. Differences by Census region were found across the 3 outcomes, with the highest proportion completed in the Northeast for both time and the developmental pathway outcomes. Several markers of social disadvantage were also associated with lower likelihood of completion for all 3 outcome measures: non-Hispanic Black race/ethnicity, being uninsured, and having providers who all practiced in a public facility. However, lower maternal education was associated with increased likelihood of being up to date at age 13 years but decreased likelihood of being up to date at age 15 years or complete within 3 years compared with their peers with higher maternal education. Living in poverty was not associated with completion by age 13 years but was associated with the other 2 outcome measures. Although statistically significant, in each case, the magnitude of the disparities was small.

Bivariate analyses suggested that early initiation worked through both the time and development pathways, but in the opposite direction (Fig 2). For the time pathway, early initiators were 2.4 times more likely to complete the series by age 13 years compared with those who initiated HPV vaccination at 11 to 12 years (74.0% vs 31.3%; P < .001). This difference was maintained at age 15 years, although decreased in magnitude to a 1.1-fold difference (91.7% vs 84.9%; P < .001). In contrast, looking at the development pathway, early initiation was associated with a slightly lower likelihood of completion within 3 years (82.3% vs 84.9%; P = .007).

Multivariable logistic regression modeling demonstrated that the association of early initiation was independent of other correlates of series completion for time pathway outcomes (Table 4). However, early initiation was not an independent predictor for the development pathway outcome in unweighted analyses and had a negligible association in weighted analyses. Furthermore, logistic regression modeling did not suggest that early initiation altered sociodemographic, geographic, or provider-based disparities in series completion (see Supplemental Table 7 for details).

This retrospective study, which takes advantage of provider reported vaccination data in NIS-Teen, assessed HPV series completion in early and mid-adolescence among 15 to 17 year olds spanning 6 birth cohorts and explored pathways by which early initiation might influence completion. We demonstrated that early initiation was a strong predictor of completion at both ages 13 and 15 years but that 9- to 10-year-old initiators were no more likely to complete vaccination within 3 years than their counterparts who initiated vaccination at ages 11 to 12 years. These findings suggest that, by increasing the available time between initiation and target completion age, early initiation gives providers more touch points and therefore more opportunities to complete the series. The idea that early initiation could increase completion because the vaccine was given at an earlier developmental period was not supported, nor do our findings suggest that early initiation alters disparities in HPV series completion.

HPV immunization is most effective if given before sexual debut. In the United States, since the late 1990s, sexual activity rates have been relatively constant: 7% report sexual debut by age 13, and 30% of females and 34% of males are sexually active by age 16 years.¹⁰  Given the more than fourfold rise in initiation of sexual activity between ages 13 and 16 years, improving HPV series completion by the younger age of 13 years would have important public health benefits. Our findings indicate that moving routine HPV vaccination from age 11 to 12 years, the current ACIP recommended age, to age 9 to 10 years would improve coverage by age 13 years, thereby providing greater population-level protection from HPV-related disease.

Early initiation was uncommon and the reasons underlying early initiation in the 7.5% who did initiate at ages 9 to 10 years are unknown. However, our findings that early initiation was more common in females, non-Hispanic Black and Hispanic populations, and lower socioeconomic status teens may provide clues as to 1 potential factor underlying why these children initiated HPV vaccination before the current ACIP recommended ages. These factors are all associated with earlier entry into puberty. Girls enter into puberty about 1.5 years earlier than boys and markers of social disadvantage have been linked to pubertal timing.¹¹  At age 9 years, 62.6% of non-Hispanic Blacks have breast and pubic hair development at Tanner stage 2 compared with 32.1% of non-Hispanic whites for breast development and 20.0% for pubic hair.¹²  Median age of menarche, which occurs late in puberty, is also earlier in non-Hispanic Black (12.06) and Hispanic (12.25) girls compared with non-Hispanic white girls (12.55).¹³  Socioeconomic disadvantage has increased the rate of early puberty fourfold in boys and twofold in girls.¹¹  Providers may have recommended early HPV vaccination for early maturing patients, especially because early puberty is associated with increased sexual risk behaviors.¹⁴  Provider concern of the higher rates of cervical cancer among non-Hispanic Black and Hispanic women and practice patterns are other potential provider-related factors that may have led to the recommendation to initiate HPV vaccination at ages 9 to 10 years. Parental factors such as parental health seeking behaviors and attitudes toward vaccines in general and HPV-related knowledge could also influence early initiation. Further research is needed to assess this hypothesis and to understand both provider and parental reasons for early initiation.

Most adolescents in this study received a 3-dose series of HPV vaccine for completion. In late 2016, ACIP changed the dose recommendation from 3 to 2 for those who initiate vaccination before age 15 years.¹⁵  The impact of the change in recommended doses on vaccination coverage rates remains to be determined, but it is expected that fewer required doses would improve completion. In this retrospective study, we explore if early initiation affected series completion and found that early initiation was associated with increased completion at both ages 13 and 15 years. Further research will need to confirm our findings, particularly in this era of a 2-dose recommendation. Future research will also be needed to understand the acceptability and impact of moving the active recommendation for initiation of HPV vaccination to ages 9 to 10 years.

This study has several limitations. We assigned birth cohorts based on age at interview and survey year because date of birth was not available in the public-use data. This likely led to some misclassification of the birth cohort variable in that an individual interviewed before his or her birthday in any calendar year would be assigned to a 1-year younger birth cohort. However, any misclassification is distributed across all birth cohorts. Because this variable is used only as a marker for time in the study, such misclassification is likely to have minimal impact. Indeed, findings did not change in multivariable logistic regression analyses run without the birth cohort variable. As noted, reasons for offering early initiation and whether initiation was requested by the parent or recommended by the provider are unknown. Although national in scope, this study should not be considered nationally representative. Provider report of vaccination status relied on parental consent to contact providers and that those providers respond to the NIS-Teen questionnaire, both of which could introduce bias. Furthermore, this convenience sample constructed with only a subset of NIS-Teen subjects with adequate provider data through use of an accelerated longitudinal design, was not consistent with use of NIS-Teen sample weights throughout. Our urban/rural measure, although derived from Census data, was state-based and therefore not specific to area of residence. Balancing these limitations are the study’s significant strengths: the creative design, use of provider-reported HPV vaccination, careful attention to dose requirement, and novel exploration of the time and development pathways.

In conclusion, this study provides evidence that moving routine HPV vaccination from ages 11 to 12 years to ages 9 to 10 years may improve vaccination coverage rates in early and mid-adolescence, thereby increasing the public health benefit of vaccination. Providers should be vigilant to patient interactions after initiation of HPV vaccination to ensure series completion within the recommended time frame. Future research on acceptability of and barriers to routine vaccination at ages 9 to 10 years could promote uptake and completion before sexual debut, thereby maximizing HPV vaccine effectiveness.

ACIP

Advisory Committee on Immunization Practices

CI

confidence interval

HPV

human papillomavirus

NIS

national immunization surveys

OR

odds ratio