Correlates of Vaccination for Hepatitis B Among Adolescents: Results From a Parent Survey

Objective  To identify correlates of hepatitis B vaccination status in sixth-grade students in the year prior to implementation of a requirement mandating immunization for seventh-grade entry.

Methods  A survey of parents of sixth-graders in 5 schools in San Diego County. Two logistic regression models were tested to predict the outcome variables: initiation and completion of the vaccination series.

Results  Factors associated with initiating the series included a recent nonacute medical visit, white race, hearing about the vaccination law from a health care provider, and the availability of a school-based vaccination clinic. Factors associated with completing the series included English as the primary language spoken at home, hearing about the law from a health care provider, a school-based vaccination clinic, and higher socioeconomic status. Health insurance was not significantly related to either outcome.

Conclusions  There was a clear benefit for hepatitis B immunization status to have had a recent nonacute medical visit, to have heard about the law from a health care provider, and to have a school-based vaccination clinic. The factors associated with starting vs completing the vaccination series were not identical. However, both health care provider and school characteristics were related to starting and completing the vaccination series. Thus, a multifaceted strategy may be most appropriate for successful coverage of an adolescent population with a vaccination series such as hepatitis B.

HEPATITIS B is one of the most prevalent infectious diseases in the world, with more than 300 million carriers of the virus worldwide.¹ Estimates suggest that between 100 000 to 300 000 new cases of hepatitis B occur in the United States each year.^2-4 In the United States, 63% of acute infections of hepatitis B occur in the older adolescent or young adult population^5,6 (although prevalence is lower for younger adolescents than for older adolescents and young adults) and most persons infected with hepatitis B acquired their infection as adults or adolescents.⁷ This high prevalence rate is especially disturbing given the existence of a readily available effective vaccine. Adolescents are at a potentially increased risk to the degree that they begin to experiment with high-risk behaviors such as injection drug use, unprotected sex, tattooing, and body piercing.

In 1995, the Advisory Committee on Immunization Practices recommended hepatitis B vaccinations for all 11- to 12-year-old children.⁸ Despite the recommendations, adolescent coverage rates have remained low. For example, a population-based sample of parents of students planning to enter fifth and sixth grade in San Diego County (California) in the fall of 1998 found that only 16% had completed the hepatitis B series.⁹ Immunization against hepatitis B is a 3-dose regimen and for "older children and adolescents, doses may be given in a schedule of 0, 1, and 6 months or of 0, 2, and 4 months."¹⁰

In 1997, the California state legislature passed AB 381, requiring proof of hepatitis B immunization for all students entering, advancing to, transferring into, and/or repeating the seventh grade on or after July 1, 1999.¹¹ This law and the accompanying regulations follow the current adolescent immunization recommendations approved by the Advisory Committee on Immunization Practices, the American Academy of Pediatrics, and the American Academy of Family Practitioners as condensed in the 1998 Schedule published by the Centers for Disease Control and Prevention (Atlanta, Ga).¹² The law permits that a child will be conditionally admitted to school if he or she is currently up-to-date with the series. Parents may claim a medical or personal belief exemption. Twenty other states (including the District of Columbia) have had some form of middle school hepatitis B vaccination law in effect since the end of 2000, with an additional 5 states having laws that will go into effect in 2001 or later.¹³ With many other states considering such legislation, information about the California experience will be relevant to health care providers and to state and federal policy makers.

AB 381 is largely an unfunded mandate, providing only limited funds to schools to check immunization records. With few additional resources made available, effectiveness is crucial in immunizing the target population. What roles should the provider and school communities play in implementing this mandate? Pediatricians and family practitioners will likely be active participants in any new vaccination requirement. As such, information about factors related to immunization status will enable the health care provider community to maximize the effectiveness of its response to laws such as those of California.

To organize the myriad of factors that might influence immunization coverage, we used the widely accepted conceptual model, first proposed by Donabedian,¹⁴ of structure, process, and outcome. This model views health and health care as the outcomes of a production system with both structural and process components.¹⁵ Structure refers to those factors necessary for health care services to occur, such as adequate facilities and licensure of physicians, and to characteristics of the patient population, such as sociodemographic variables or severity of illness. In this study, we considered variables such as the presence of a school-based vaccination clinic as well as insurance status, percentage of the school population eligible for free and reduced-price lunch, race, and primary language spoken at home to be structure variables. Process refers to interactions within the health services system. We considered factors over which providers have some control, such as the recency of the last nonacute medical visit and whether the parent had heard about the immunization requirements from a physician, clinic, or health plan, to be process variables. Outcomes were the results of structure and process variables, specifically, the parents' report of their child's immunization status.

A voluntary, anonymous survey for parents was conducted at 5 middle schools participating in an ongoing examination of the effects of a seventh-grade hepatitis B vaccination school entry requirement. The research protocol and informed consent were reviewed and approved by the institutional review board at San Diego State University.

We identified and selected 5 middle schools to participate in the survey. Our criteria in assembling this group of schools were the following: (1) Schools that planned to conduct on-site immunization clinics and schools that did not plan such clinics. (2) Schools within the City of San Diego Unified School District and in other districts. (3) Student populations generally representative of the county (race and socioeconomic diversity [based on the percentage of students eligible for federal free/reduced lunch programs]). (4) School staff able to devote the time and effort necessary to administer and collect surveys. Project staff interviewed a total of 11 schools and 5 schools met our criteria and agreed to participate. These 5 included 3 schools in the San Diego Unified School District, all of which had on-campus vaccination programs, and 2 outside of the city school district, neither of which planned to have on-campus vaccination programs. The 2 schools from outside the city school district were from 1 suburban and 1 rural district. The 5 schools had varied racial and socioeconomic demographics as presented in Table 1.

The survey was distributed to every sixth-grade student at each of the 5 schools during November 1998. Data used in this analysis were collected from English-language surveys. Although a Spanish-language survey was fielded, an error in translation resulted in omission of the main outcome variable. Thus, the surveys from the 60 Spanish-speaking respondents were dropped from the analysis. A project staff member or the classroom teacher handed out the surveys during home room or "core" class time. Students were asked to take the survey home, have a parent complete it, and return it to the classroom teacher by the end of the week. Brightly colored baskets with signs reading "Return Surveys Here" were placed in prominent locations in each classroom. An incentive was offered to both the students and the teachers in an effort to increase the response rate for returned surveys. All students were given a thank-you item regardless of whether they had returned a survey. Teachers received $1 for each completed and returned survey, with the money being used to benefit classroom activities.

Schools were categorized based on the presence of a school-based vaccination clinic. Three of the 5 schools had such a clinic prior to or during the survey period. Schools were further categorized into higher vs lower socioeconomic status, based on the percentage of the student body eligible for free or reduced-price lunch. Because the percentage eligible ranged from 25% to 65%, the median percentage of students eligible from the 5 schools (45.1%) was used to split the sample. This resulted in 2 "lower SES [socioeconomic status]" schools (percentage eligible greater than the median of the 5 schools) and 3 "higher SES" schools (percentage eligible less than or equal to the median of the 5 schools). Higher SES was a relative term since in this category of schools, 25%, 38.7%, and 45.1% of the students were eligible for free and reduced-price federal lunch programs.

Respondents were categorized based on primary language spoken at home (English vs other), race (white vs other), and insurance status (any health insurance vs none). Additionally, respondents were categorized based on whether they reported having heard about the hepatitis B vaccination requirement from their physician, clinic, or health plan (yes vs no) and on the recency of nonacute medical visits (within the last 6 months vs 6-12 months ago vs more than 1 year ago).

The outcome variable of hepatitis B vaccination status was based on parents' recall of their child's immunization status. This variable was categorized as no vaccinations, started the vaccination series (1 or 2 vaccinations), and completed the vaccination series.

Additional data to assess potential bias were gathered from existing data. The race distribution of the student body at each school was based on data reported to the state.

The data were analyzed using SPSS version 8.0 (SPSS Inc, Chicago, Il). Logistic regression was used to assess the effect of structure and process variables on the main outcome: parent-reported adolescent hepatitis B immunization status. Structure variables—both school factors (school-based vaccination clinics, percentage of students eligible for free lunch) and personal factors (primary language spoken at home, insurance status, race)—were entered simultaneously as the first step (first block) in the regression equation. Process variables (recency of last medical visit, whether parents had heard about the law from a physician, clinic, or health plan) were entered simultaneously as the second step (second block). All variables were forced to stay in the model, using the ENTER option in SPSS. Two separate logistic regression equations were constructed: 1 to predict beginning the vaccination series (any vaccinations vs no vaccinations reported), and 1 to predict completing the vaccination series (all vaccinations completed vs not all vaccinations completed). Recognizing the possibility of a clustering effect that would cause the responses within schools to be correlated, we also modeled school as a random effect in SAS Proc Glimmix (SAS Institute, Cary, NC).¹⁶ The quasi-likelihood estimates of the covariance parameters associated with school were 0, so we reported the odds ratios (ORs) based on fixed effects only. We report the χ² for each step of the regression equation, the OR, and the confidence interval (CI) for the OR for each predictor.

A total of 1322 surveys were distributed at the 5 schools. Eight hundred were returned, yielding an overall response rate of 60.5%. School response rates ranged from a low of 46% to a high of 65%. Of the 800 returned surveys, 156 (19.5%) had missing data on at least one of the variables in the regression equation, resulting in their being dropped from the regression analysis. We examined these 2 potential threats to internal validity.

We examined the issue of nonresponse bias by comparing the sample with available data on race and immunization status. We compared the race distribution of the sample with that of the overall student body at each of the 5 participating schools. Overall, the sample was fairly representative of the racial makeup of the schools, with the exception of Latino students, who were underrepresented by more than 10% in 3 of the 5 schools (Table 1). We compared immunization status of the sample with the overall student body for 3 of the 5 schools from which data were available. The overall sample showed 31% of students having no vaccinations, 39% having 1 or 2 vaccinations, and 29% having all 3 vaccinations. The data from the schools showed that for the entire sixth grade, the respective percentages were 47%, 30%, and 23%.

We examined the issue of bias owing to missing data by comparing the analysis sample with those dropped because of missing data. The respondents not included in the logistic regression analysis because of missing data were more likely to attend a school with an on-campus vaccination program (82.8% vs 70.5%; χ²₁ = 7.5, P = .006) and to be from a lower SES school (56.4% vs 42.4%; χ²₁ = 10.0, P = .002). They were less likely to speak English as the primary language at home (53.2% vs 76.9%; χ²₁ 35.1, P<.001), to be white (22.2% vs 47.4%; χ²₁ = 27.1, P<.001), to have health insurance (78.0% vs 89.4%; χ²₁ 13.0, P<.001), and to report hearing about the hepatitis B law from their health care provider (16.7% vs 29.0%; χ²₁ 9.8; P = .002). There was no difference between the included and excluded cases in the proportion with a nonacute medical visit in the past 6 months (Table 2).

Descriptive data were reported based on the entire sample. As can be seen, the overall sample is diverse, likely to be insured, and to speak English as the primary language. Almost three quarters attended schools with a school-based vaccination program, and the sample is equally split into higher and lower SES schools. A little more than one third of the overall sample had been to their health care provider for a nonacute visit in the past 6 months and about one quarter reported that they had heard about the hepatitis B vaccination requirement from a health care provider.

When asked whether their children had received any hepatitis B vaccinations (n = 800), 29.1% of parents reported that their child had not been vaccinated, 2.5% reported that their child had received 1 or more vaccinations but were not sure of the exact number, 16.1% reported that their child had received the first vaccination, 17.9% reported 2 vaccinations, and 27.0% reported all 3 vaccinations (7.4% were missing or not sure). Of the parents who reported that their child still needed 1 or more vaccinations (n = 679), the most common reasons given for their child still needing vaccinations were not knowing that their child needed vaccinations (20%), vaccinations were not yet medically due (19%), and other (18%). The "other" category included a range of reasons, including "against personal beliefs," "waiting for the school to give it to him," and "procrastination."

The overall percentage of parents (n = 679) reporting that their child had health care coverage of some type is similar to the 85% reported in San Diego County overall.¹⁷ Those speaking primarily English at home were more likely to report that they had health insurance (91%) than those who reported speaking primarily another language at home (73%) (χ²₁ = 49.8; P<.001). Those with health insurance were more likely to report that their child had received a regular medical checkup in the last 6 months when compared with those without insurance (39% and 22% respectively; χ²₁ = 15.4; P<.001).

Three of the 5 schools in the sample administered vaccinations on campus. Children attending these 3 schools were more likely than those at schools not offering vaccinations to be nonwhite (63.7% vs 44.3%; χ²₁ = 24.6; P<.001), to primarily speak a language other than English at home (31.0% vs 19.2%; χ²₁ = 11.1; P<.001), and to be a student at a higher SES school (61.6% vs 37.0%; χ²₁ = 39.0; P<.001).

Two logistic regression models were generated, one to identify factors associated with an increased likelihood of having started the vaccination series (Table 3) and one to identify factors associated with an increased likelihood of having completed the vaccination series (Table 4).

In the regression model predicting those who started the vaccination series (n = 644), the largest OR was related to having had a nonacute medical visit in the last 6 months (OR, 2.2; 95% CI, 1.39-3.51 compared with a visit 6-12 months ago; OR, 3.36; 95% CI, 2.18-5.16 compared with a visit more than a year ago). Other significant predictors were white race (OR, 1.96; 95% CI, 1.27-3.00), attending a school with an on-campus vaccination program (OR, 1.56; 95% CI, 1.04-2.34), and having heard about the law from a health care provider (OR, 1.74; 95% CI, 1.14-2.65).

In the regression model predicting those who completed the vaccination series (n = 644), significant ORs were associated with attending a school with a vaccination program (OR, 1.80; 95% CI, 1.15-2.80), attending a higher SES school (OR, 1.52; 95% CI, 1.00-2.30), primarily speaking English at home (OR, 2.22; 95% CI, 1.38-3.57), and having heard about the law from a health care provider (OR, 1.81; 95% CI, 1.23-2.65). Health insurance status was not significantly related to either of the outcomes measured.

This study examined the correlates of adolescent hepatitis B vaccination coverage in the year prior to an impending vaccination mandate. Two factors emerged as significant in starting and in completing the vaccination series: attending a school with an on-campus vaccination clinic and having heard about the law from a health care provider (eg, a physician, clinic, or health plan). The factor found to be the strongest determinant of starting the series—a well-child visit in the last 6 months—did not predict completion of the series. This may be because many adolescents had started their series at a physician's office but had not yet returned to complete the series. Surprisingly, our analysis showed that health insurance status, a traditional measure of access, was not directly related to the outcomes of interest and may therefore be less important than health services–seeking behavior (eg, recent health care visit) in immunization coverage.

These findings highlight the roles of health care providers and of schools in raising hepatitis B coverage rates. Provider behavior was related to increased coverage rates in that having heard about the new vaccination requirement from a physician, clinic, or health plan was a consistent predictor of both starting and finishing the vaccination series. This finding suggests the crucial importance of efforts on the part of the health care provider community to educate parents and patients. The benefit of having a recent nonacute medical visit is clear in terms of starting the hepatitis B series. Follow-up of this cohort might have revealed a benefit of a recent nonacute medical visit in terms of completing the series.

Because more than 99% of children attend school until at least age 13 years,¹⁸ school-based vaccination campaigns would seem to be an important part of an effective strategy for reaching large numbers of young adolescents. Previous studies have indicated that school-based vaccination programs can efficiently provide vaccination services¹⁹ and that most parents are accepting of such programs.^19,20 Our findings support this idea.

These findings have policy implications for other states considering implementation of a law such as AB 381. That different determinants emerged for starting vs finishing the vaccination series implies that a multifaceted approach may be necessary for successful coverage of an adolescent population with a vaccination series such as hepatitis B. Given the potential costs associated with such an initiative, state policy makers may see tension between the health care provider and school communities in the implementation of this mandate. Our findings suggest, however, that both providers and schools must play active roles in the implementation. In this sample, there was a clear benefit to immunization status both in being a student where school-based vaccination clinics were held, as well as in having heard about the law from a health care provider and having had a recent health care visit. The fact that the hepatitis B vaccine consists of a series, rather than a single vaccination, is a key consideration when designing effective strategies for implementing this requirement. Communication between providers and schools regarding immunization status will be essential to capitalize on the strengths of both entities.

This study has 2 main limitations. The first concerns issues of potential sampling bias and the resultant limitations on generalizability. Forty percent of the surveys distributed were not returned, raising the issue of potential nonresponse bias. Latino students were underrepresented in our sample, as were students who had not begun the immunization series. The extent to which the underrepresentation of these 2 groups affects the findings is unknown. Potential bias owing to missing data in the regression analysis was also examined. Respondents with missing data seemed to be a somewhat more vulnerable group than those with complete data. More studies are needed to investigate strategies for immunizing this higher-risk group. Another source of potential bias is that the schools participating in the study were a convenience sample and not randomly selected. A final potential source of bias is that surveys were administered only in English. These potential biases raise cautionary flags regarding the extent to which the findings from this study can be generalized beyond the study population. Any generalization should only be made with the caveats outlined above and with the knowledge that these sources of potential bias may alter the results for other samples.

The second main limitation is that of parent recall as a measure of vaccination status. The written vaccination record is often held as the gold standard for vaccination status since researchers have found that parent recall can inflate vaccination rates. However, we chose to measure parent recall because we were concerned that relying on written records in this case would actually underestimate the true vaccination rate. Given that this was a voluntary, anonymous, self-administered survey, we were concerned that respondents, while willing to complete a brief survey, would not be willing to extend that effort to finding their immunization records and copying this information onto the survey.

This study raises additional research questions. One issue bearing further scrutiny is that of the potential collateral benefits of immunization. Adolescents typically receive far fewer well-child or preventive services than do either children or adults, interacting with the health care system instead largely through acute care.^17,20 Hepatitis B vaccination mandates such as California's represent an opportunity to redesign adolescent health services to be centered around primary prevention rather than crisis intervention.²¹ Ideally, administration of vaccinations should be integrated with other preventive services to adolescents,^7,21 such as those outlined in the Guidelines for Adolescent Preventive Services.²² Further research is necessary to assess whether adolescents receiving their immunizations in health care provider offices actually receive these collateral benefits and what effect, if any, receiving immunizations at schools has on receipt of these services.

Another issue needing further scrutiny is the costs to the public and private sectors of such an initiative, and how those costs are apportioned. Research is needed to answer questions, such as the following: How will commercial and publicly funded health plans, schools, and communities share the cost of such an initiative? To what extent are redundant costs incurred, for example, children receiving a publicly funded vaccine at school despite being insured? How can policy makers craft the law to ensure equitable and efficient cost sharing?

Hepatitis B is one of the most prevalent infectious diseases worldwide, despite the availability of an effective vaccine. Adolescents are at increased risk for hepatitis B infection and several states have begun passing laws requiring proof of immunization prior to middle school enrollment. By examining the correlates of hepatitis B vaccination status in the year prior to implementation of California's mandate, this article provides insight into the role the medical and school communities may play in ensuring coverage. Logistic regression analysis shows that both medical (recency of a well-child visit and hearing about the law from a health care provider) and school (having an on-campus vaccination clinic) factors were important in vaccination status, while insurance status was not. A multifaceted strategy may be most appropriate for successful coverage of an adolescent population with a vaccination series such as hepatitis B.

Accepted for publication March 27, 2001.

This research was supported by a Cooperative Agreement between the Centers for Disease Control and Prevention, and the Association of Schools of Public Health, Washington, DC, Project S0534-17.

Corresponding author and reprints: Michael Seid, PhD, Center for Child Health Outcomes, Children's Hospital and Health Center, 3020 Children's Way, MC505, San Diego, CA 92123 (e-mail: mseid@chsd.org).