Geographic Factors and Human Papillomavirus (HPV) Vaccination Initiation among Adolescent Girls in the United States

The human papillomavirus (HPV) represents the most common sexually transmitted infection in the United States (1). HPV infection is associated with cancers in women, including cervical, vulvar, anal, oropharyngeal, and vaginal (2, 3). In the United States, approximately 20,589 HPV-associated cancers are diagnosed each year among females, many of which may be preventable with the current HPV vaccine (2, 4). Cervical cancer is the most common HPV-associated cancer found in women, accounting for about 55.5% (2).

Racial/ethnic and socioeconomic disparities in HPV-associated cancers have been documented in the United States (5–7). Hispanic women have higher rates of cervical cancer compared with non-Hispanic women. White and Black women had higher rates of oropharyngeal cancer compared with women of other races, and rates of vulvar and anal cancers were higher for Whites and non-Hispanics than they were for other racial/ethnic groups. Rates of vaginal cancer were highest among Blacks (3). Furthermore, in the United States, incidence rates of each HPV-associated cancer were highest among women living in neighborhoods where 20% or more of the population is living below the poverty line compared with women living in neighborhoods where <5% of the population is living below poverty line (8).

Since 2006, The Centers for Disease Control and Prevention (CDC) Advisory Committee on Immunization Practices (ACIP) has recommended the safe and effective HPV vaccine for girls aged 11 or 12 years, and for females aged 13 through 26 years not previously vaccinated (9). ACIP expanded its recommendation in 2011 to include boys. Receipt of the vaccine at ages 11 and 12 builds an effective immune response before the onset of sexual activity (10, 11). Despite ACIP recommendations, HPV vaccination rates in the United States remain low. The Healthy People's 2020 goal for completing the recommended HPV vaccine three-dose series is 80% for girls (12); yet in 2013, only 57.3% (±1.9) of U.S. teen girls had received at least one dose of HPV vaccine, and only 37.6% (±2.0) had three or more doses of the vaccine.

To date, most HPV vaccination studies have examined individual- and provider-based factors as possible explanations for the low vaccination rates (13). These studies found that higher rates of HPV vaccine initiation were associated with older age, health insurance type, urban residence, physician recommendation for the vaccine, receipt of other childhood vaccines, provider reminders, and knowledge of HPV (13). The most recent data from the National Immunization Survey-Teen (NIS-Teen) indicate that girls whose families live below the poverty line had higher rates of both HPV vaccine initiation and receipt of the three-dose series compared with girls whose families live above the poverty line (14). Survey data also indicate that non-Hispanic (NH) Whites and Asians had lower rates of HPV vaccine initiation and receipt of the three-dose series compared with Hispanics, NH Blacks, and American Indian/Alaska Native (3).

In addition to provider- and individual-level factors, it is also possible that geographic or neighborhood factors, like area-based socioeconomic status (SES), could independently influence vaccination uptake. In other areas of cancer prevention research, geographic factors such as area-based SES, racial–ethnic composition, and rural versus urban residence have been shown to influence health status (e.g., cancer stage at diagnosis; refs. 15, 16) and health behaviors (e.g., up-to-date on mammogram) independent of an individual's SES or other individual factors such as health insurance (17–19). Similarly, area-based SES and geographic factors could influence vaccination uptake through pathways linked to the material resources of the neighborhood, availability and ease of access to healthcare facilities, and social capital or networks (e.g., social contagion, similar norms of behavior; refs. 20–22).

Numerous possible outcomes related to HPV vaccination uptake exist, given the different ways geographic factors can influence health service utilization. For example, economic burdens and barriers to care among lower income ethnic minority or rural communities could result in lower HPV vaccination initiation because of fewer health resources and limited financial resources (23–25). Similarly, language barriers (13) and a lack of awareness in ethnic minority communities (26–28) could result in lower uptake (29). In contrast, HPV vaccination initiation could be higher among lower income ethnic minority communities due to social support networks and resource availability associated with living among co-ethnics or with certain groups in segregated communities, similar cultural norms related to vaccination and other health behaviors, or in part to long-term targeted interventions or safety-net services, which provide free or subsidized vaccines (30, 31).

To date, no studies in the United States have examined the potential influence of the racial–ethnic composition of a community on HPV vaccination, and the three U.S. studies that examined the influence of area-based SES were inconsistent (32–34). Furthermore, these studies (32, 34) were limited to specific states or populations (33). To advance research in this area, investigators must explore area-based SES and other relevant geographic factors including the community racial–ethnic composition that could impact HPV vaccination rates using national data representing a wide variety of demographic and economic groups.

In this study, we analyzed data from the NIS-Teen to examine associations between HPV vaccine initiation among female adolescents and community-level factors including area-based poverty, rural/urban residence, and residential racial–ethnic composition. To our knowledge, this is the first study examining associations between geographic factors and HPV vaccine initiation among a nationally representative sample of teen girls in the United States while also accounting for individual-level characteristics.

All individual data were obtained from the restricted-use data files of the 2011 and 2012 NIS-Teen, an annual survey conducted by the CDC to monitor vaccination uptake in the United States. The NIS-Teen includes a nationally representative stratified sample of girls and boys aged 13 to 17 years in all 50 states and the District of Columbia. The survey is based on random-digit dialing of both landline and cellular telephone numbers to identify eligible households. It includes two parts: a survey of parents or guardians of teens 13 to 17 years old to collect information about demographic and socioeconomic characteristics, and a survey mailed to all vaccination providers whom the parents identified and consented to verify their teen's immunization histories. The survey sampling procedures have been described elsewhere (35).

The 2011 and 2012 NIS-Teen included 18,573 and 15,137 girls aged 13 to 17 years, respectively, with completed surveys in the United States. Of the total 33,710 girls for 2011 and 2012, 20,565 (61%) had provider-verified vaccination records. The present study was limited to the girls with provider-verified vaccination records and complete records for all the variables included in the analysis.

We examined HPV vaccine initiation (receipt of at least one dose of the HPV vaccine, either quadrivalent or bivalent) as the primary outcome variable. The outcome measure was dichotomized as zero doses or one or more doses of the HPV vaccine. We studied several individual-level variables from the survey: (i) the teen's current age in years; (ii) race/ethnicity [non-Hispanic (NH) White, NH Black, NH Other, and Hispanic]; (iii) health insurance type (employer or union, Medicaid, or the State Children's Health Insurance, military or Indian Health Service, and no insurance); (iv) poverty status [categorized as above poverty, high income (annual income > $75,000); above poverty, moderate income (annual income ≤ $75,000), below poverty, based on the U.S. Census family poverty thresholds; ref. 36; and unknown]; (v) receipt of a provider recommendation for HPV vaccination (yes, no, or don't know); (vi) mother's age (≤34 years, 35–44, or ≥45); (vii) marital status (currently married or not currently married); and (viii) education level (<12 years; 12 years with high school diploma or general equivalency diploma; >12 years without college degree; or college degree or higher). We also included one healthcare system-level factor: facility type where vaccines were administered.

Using the ZIP codes of participants' current residences, several geographic measures based on U.S. census ZIP Code Tabulation Areas (ZCTAs) were merged with the NIS-Teen at the CDC Research Data Center (RDC). The ZCTAs are generalized areal representations of U.S. Postal Service areas, and in most instances, the ZCTA is the same as the ZIP code (37). Ninety-eight percent of the ZCTAs matched residential ZIP codes from the NIS-Teen participants included in this study.

We developed geographic factors using measures of community disadvantage previously included in health disparities research (38, 39) and research on vaccination uptake and use of cancer screening services (16, 33, 34, 40–42). Using the 2008–2012 U.S. Census American Community Survey, we developed the following community-level geographic factors: (i) percentage of a ZCTA's residents living below the federal poverty level (36); (ii) median household income of ZCTA; (iii) population density, and (iv) the racial–ethnic residential composition of the ZCTA. These variables were each divided into quartiles based on the distribution of these measures across ZCTAs. Percent living below poverty was categorized into (i) <5%, (ii) 5%–< 10%, (iii) 10%–< 20%, and (iv) ≥20%.

We operationalized the measure of racial-ethnic residential segregation using census data on racial–ethnic composition because of our focus on the majority racial/ethnic group in each ZIP code (43, 44). Numerous studies have shown that persistent residential racial/ethnic segregation in the United States serves as both a health promoter by facilitating stronger social support networks and as a health barrier by fostering conditions that limit resources (16, 39, 45, 46). Racial–ethnic composition in this study was based on ZCTAs having >50% Hispanic, >50% NHW, >50% NH Asian-Pacific Islander (NHAPI), >50% NHB, >50% NH American Indian/Native Alaskan, or mixed (NHM), if not any of the previous. Because of small numbers after merging the racial–ethnic composition variables with the NIS-Teen, we combined NH American Indian/Native Alaskan with NHAPI into a category called NH-other.

We also included ZCTA measures of rural and urban residence based on the rural-urban commuting area codes (RUCA; ref. 47). The RUCAs provided a definition of rural/urban residences based on criteria that included population density and population work-commuting patterns. We categorized this variable as urban, large rural city/town, small rural town, and isolated small rural town (47).

For analysis, we combined the NIS-Teen datasets for 2011 and 2012, using suggested methods (35), and included appropriate sampling weights for calculating the weighted percentages and effect estimates (48). We used weighted frequencies and 95% confidence intervals (CI) to describe the characteristics of the sample. We performed bivariate association tests between HPV vaccination initiation (receipt of one HPV vaccine dose) and individual-level and community-based geographic variables with Wald χ² tests.

The study used bivariate logistic regression models to estimate the crude ORs and 95% CIs of vaccination series initiation for each of the individual-level and geographic variables. Multivariable logistic regression models were used to calculate the adjusted odds ratios (AOR) and 95% CIs of vaccination initiation. One multivariable model included only individual and provider factors, and two separate multivariable models were run to examine the independent associations of ZIP code level poverty and racial–ethnic composition with vaccination outcomes. Joint contributions of individual poverty and area poverty and racial segregation were assessed with interaction terms between individual-level race/ethnicity and ZIP code-level poverty and racial–ethnic composition. The surveyed state was included as a random effect to account for dependency by state residence (e.g., state health programs that will influence the girls from that state).

Bivariate associations were analyzed using procedures for complex survey data in SAS statistical software (i.e., PROC SURVEYFREQ; ref. 49). Logistic regression analyses were conducted using SAS GLIMMIX, which implements the Generalized Linear Mixed Model and allows for the incorporation of stratum-specific weighted analysis (50). Statistical tests were two-tailed with a critical α of .05.

Sociodemographic and geographic characteristics of the study population are presented in Table 1. Sixty percent, or 20,565, of girls in the age range of 13 to 17 years had adequate provider data in the 2011 and 2012 NIS-Teen and were included in the study. Overall, 53.5% of girls received more than one dose of HPV vaccine; the proportion was slightly lower in 2011 (53.2%) than in 2012 (53.8%). Most girls were NHW (55.3%) living in urban communities (88.6%) with population densities greater than 651 persons per square mile (quartile 4; 59.5%), and in predominantly NHW (69.5%) ZIP codes. Most mothers were >35 years old (89.3%), without a college degree (33.8%) and had employer- or union-provided health insurance (46.3%).

Table 2 shows the bivariate (unadjusted) ORs and AORs from the multivariable logistic regression models. In the bivariate analysis, all variables were statistically significantly associated with HPV vaccine initiation. Girls living in high density, high poverty, or urban communities had higher odds of vaccine initiation than girls living in low density, low poverty, or rural communities, respectively. Girls living in predominantly minority communities also had higher odds of vaccine initiation compared with those in predominantly NHW communities, regardless of race/ethnicity.

In multivariable analyses that included only individual and provider factors (Table 2, Model 1), all variables associated with vaccine initiation from the bivariate analysis remained statistically significant (P ≤ .05). Findings were consistent after adjusting for ZIP code poverty and population density (Model 2), except for the individual NHB race/ethnicity, which was no longer significant.

Girls with health insurance through Medicaid or SCHIP had higher odds of HPV vaccine initiation than those with employer- or union-provided insurance (Model 1, AOR, 1.35; 95% CI, 1.24–1.48). Among girls who received a provider recommendation to vaccinate, the odds of vaccination initiation were 3.73 times greater than in girls without such a recommendation. Higher odds of vaccine initiation were also associated with girls whose mothers had <12 years of education compared with those with a college degree (Model 1, AOR, 1.73; 95% CI, 1.53–1.95); among girls with mothers younger than 35 years old compared with girls with mothers 45 years and older (Model 1, AOR, 1.45; 95% CI, 1.29–1.62); and among girls from households with income below the poverty threshold compared to households above the poverty threshold (Model 1, AOR, 1.24; 95% CI, 1.10–1.39).

On the basis of Model 2 results, which included individual and provider factors and ZIP code-level poverty and population density, girls living in communities with the highest poverty levels had higher odds of HPV vaccination initiation compared with girls living in communities with the lowest poverty levels (Model 2, AOR, 1.18; 95% CI, 1.04–1.33). Girls living in communities with the highest population density had higher odds of HPV vaccination initiation compared with girls from the lowest population density categories (Model 2, AOR, 1.49; 95% CI, 1.24–1.80).

Model 3 included individual factors and ZIP code-level racial–ethnic composition and population density. The odds of HPV vaccine initiation was higher among girls from communities where the majority race–ethnic group was Hispanic (AOR, 1.64; 95% CI, 1.43–1.87), mixed race (AOR 1.30; 95% CI, 1.17–1.44), NHB (AOR, 1.16; 95% CI, 1.0–1.34), or other NH race (AOR, 2.63; 95% CI, 1.87–3.71) compared with girls from communities where the majority race–ethnic group was NHW.

The interaction terms—individual race/ethnicity × ZIP code poverty and individual race/ethnicity × ZIP code racial–ethnic composition—were both overall statistically significant (results not shown). The results indicated that Hispanics living in high poverty ZIP codes (poverty ≥ 10%) had higher odds of HPV vaccine initiation compared with Hispanics living in lower poverty ZIP codes (poverty 0%–9.9%; AOR, 1.39; 95% CI, 1.16–1.65). A similar association was found for NH other/multiple races living in high poverty versus low poverty ZIP codes (AOR, 1.43; 95% CI, 1.16–1.76). No other race/ethnicity and ZIP code poverty groups were significant.

Figure 1 shows the model-adjusted percent of girls vaccinated based on race/ethnicity and ZIP code racial–ethnic composition. Initiation rates were lowest for NHW and NHB and highest for Hispanics regardless of the community's racial–ethnic composition. The odds of initiation among Hispanic girls living in Hispanic communities were 2.23 (95% CI, 1.87–2.65) and 1.90 (95% CI, 1.20–3.04) times higher than for those living in predominantly NHW or NHB communities, respectively (results not shown).

This study examined the relationship between HPV vaccine initiation and individual- and community-level geographic factors among teen girls in the United States. Overall the odds of initiation were highest among older girls, girls with SCHIP or Medicaid health insurance, girls with younger mothers, and those whose mothers had lower levels of education and incomes below the federal poverty level. Receipt of a provider recommendation was one of the factors most strongly associated with HPV vaccination initiation. These findings are consistent with previous studies that examined survey data (33, 34, 51–53). Also consistent with recent reports (54, 55), we found that the rates of HPV vaccination initiation were highest for Hispanic girls compared with NHW, NHB and other NH race groups. We also found that geographic factors including area-based poverty, residential racial–ethnic composition, and population density were significantly associated with HPV vaccination initiation even after adjusting for individual-level factors.

Our findings indicating higher HPV vaccination initiation among girls living in the poorest communities may seem contrary to conventional beliefs that socioeconomic disadvantage may result in barriers to care and underutilization of health services. However, it is also plausible that the availability of safety-net immunization services for these populations and their access to HPV vaccines through the Vaccine for Children (VFC) program, which is a federally funded program providing free recommended vaccines to eligible children, may be contributing to the higher HPV vaccine uptake (56, 57). Independent of race/ethnicity, the lower HPV vaccine uptake in less disadvantaged or affluent communities could be attributable to barriers related to out-of-pocket expenses for girls who are underinsured, a lack of physician recommendation, and/or infrequent preventive medicine visits. Furthermore, parents in less disadvantaged or affluent communities may be less supportive of the HPV vaccine (58). Future research should consider changes in HPV vaccine initiation in these communities resulting from The Affordable Care Act (ACA), which offers opportunities to improve HPV vaccination coverage among teens. Presently, the ACA requires health insurers to cover the costs of all ACIP-recommended vaccinations at no cost to consumers when provided by in-network providers (59).

Our findings related to community-level poverty were similar to those of one of three previous studies examining area-based SES and HPV vaccination. Pruitt and colleagues (33) reported that girls in poorer counties in six U.S. states were more likely to initiate the vaccine than girls in wealthier counties. In contrast, Chao and colleagues (32) used data from a managed care organization in California and found that girls from high-income neighborhoods were more likely to initiate the vaccine than girls from low-income neighborhoods. The difference in the findings might result from Chao's inclusion of girls within a different age range (9–17) and only those who were members of the managed care organization, therefore not including underinsured and/or unemployed populations, who might have different vaccination rates (32). Our findings were also different from those of Tsui and colleagues' study that included adolescent girls from Los Angeles County. They found that neighborhood factors, including poverty, were not independently associated with vaccine initiation after adjusting for individual factors (34). The lack of a significant association in Tsui's study may be related to the unique, small, and homogeneous sample (n = 479) of immigrant, low-income mothers of adolescent girls in the study.

Our findings indicating an independent association of residential racial-ethnic segregation and HPV vaccination initiation suggest that geographic factors play an important role in HPV vaccinations beyond individual-level characteristics. We found that girls from geographic communities with a majority Hispanic, Black, or mixed-race population had significantly higher odds of vaccination initiation than those from majority NHW communities. Residential racial-ethnic segregation in the United States has been shown both to negatively influence health outcomes through social isolation and lack of access to health and preventative services (60–62) and to positively affect health through strong social support networks and the availability of more health-promoting resources (43, 63, 64). Racial-ethnic minorities are more likely to live in poorer, inner-city urban areas than are NHWs (16, 65) for reasons including racial discrimination, institutional racism, and immigrant settlement patterns (16, 65). In our study, among the girls living in majority Hispanic or NHB communities, 72% and 74% of them, respectively, lived in communities where ≥20% of the population was below the poverty line. In contrast, among the girls living in majority NHW or mixed race/ethnicity communities, only 13% and 35% of them, respectively, lived in similarly impoverished communities. The higher HPV vaccination rates in poor communities with a high concentration of racial-ethnic minorities could therefore be a result of healthcare practice and community-based interventions (e.g., educating key stakeholders influential to the adolescents' vaccination behaviors) and other intervention strategies (66–70) in support of HPV vaccination that have focused efforts on poor Hispanic and Black communities where rates of cervical cancer rates are highest (6).

Our findings also indicated that Hispanic girls living in predominantly Hispanic or mixed race/ethnicity communities had HPV vaccination initiation rates significantly higher than Hispanics living in predominantly NHW or NHB communities. This result could be due to positive or protective factors related to living in communities with a high concentration of ethnic minorities sharing the same health-related values, who provide critical social and instrumental support, and are targeted for public health safety-net services and interventions, (71). These findings are consistent with those of previous studies showing residential racial–ethnic composition can influence health behaviors (43, 45). Higher HPV vaccination initiation rates among Hispanics living in predominantly Hispanic communities could also be a result of low levels of acculturation. Although we did not directly measure acculturation in this study, Hispanic residential segregation in the United States has been shown to be correlated with lower levels of acculturation and low SES (44, 72–74); and, there is evidence from past studies that Hispanics with lower levels of acculturation are more supportive of vaccination for their children than are other demographic groups (75, 76).

For NHB girls, the community's racial–ethnic composition did not substantially change the HPV vaccination initiation rates. Why this study does not show the same protective effect among NHB girls living in poor, predominantly NHB communities as it does for Hispanics girls living in predominantly Hispanic communities is not known; however, it could be related to ease of accessing safety-net services and/or less effective intervention programs. Additional research is needed to better understand why these differences exist and to further understand the independent and/or joint contributions of race, segregation, and poverty on the uptake of HPV vaccination.

Our findings are subject to several limitations. First, the household response rates for 2011 and 2012 were 57.2% and 55.1%, respectively, and the cell phone response rates for the same years were 22.4% and 23.3%, respectively. Although sampling weights are designed to minimize nonresponse and noncoverage bias, some bias remains in the weighted estimates used for analysis. Second, the parents' role in the survey is subject to recall bias. Some parents may have incorrectly recalled whether they had ever received a provider recommendation for the HPV vaccine. Third, the NIS-Teen did not collect certain data potentially important to HPV vaccine initiation, including the parents' health beliefs and uptake of other childhood immunizations. Finally, community geographic measures were based on ZCTA data because these were the smallest geographic units available via NIS-Teen. ZCTA measures have been shown to be more heterogeneous than other preferred geographic units like census tracts and blocks (77). A change in geographic unit (ZIP code to census tract) may result in different conclusions than those that are based on another unit of analysis (often called the modifiable unit problem).

In conclusion, this study found that initiation rates of HPV vaccination among teen girls were highest in the poorest communities and among Hispanics living in communities where the racial–ethnic composition was predominantly Hispanic or mixed race. It also found that regardless of individual race/ethnicity, girls living in predominantly NHW and NHB communities had the lowest initiation rates. The reasons are unknown, but if public health officials wish to increase vaccination coverage in girls in these communities, they should obtain a better understanding of the individual characteristics and geographic or contextual factors that may lead to barriers to HPV vaccination and ineffective public health interventions.

No potential conflicts of interest were disclosed.

The findings and conclusions in this report are those of the authors and do not necessarily represent the official positions of the NIH, Huntsman Cancer Institute Foundation, Primary Children's Hospital Foundation, or the Beaumont Foundation.

Conception and design: K.A. Henry, E.L. Warner, D. Kepka

Development of methodology: K.A. Henry, D. Kepka

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K.A. Henry, E.L. Warner, D. Kepka

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K.A. Henry, D. Kepka

Writing, review, and/or revision of the manuscript: K.A. Henry, A.M. Stroup, E.L. Warner, D. Kepka

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K.A. Henry, E.L. Warner

Study supervision: K.A. Henry, D. Kepka

Other (completed analysis at Research Data Center): K.A. Henry

This work was supported in part by the Huntsman Cancer Institute Foundation (to D. Kepka), Primary Children's Hospital Foundation (to D. Kepka), and the Beaumont Foundation (to D. Kepka). D. Kepka was also supported by the University of Utah, School of Medicine, Clinical and Translational Science Center (National Center for Advancing Translational Sciences, NIH Award KL2TR001065).

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.