Abstract
OBJECTIVES. Prophylactic vaccines against human papillomavirus (HPV) are expected to be available for public use by 2007 and likely will be targeted to preadolescent children. Parental acceptance of these vaccines will be critical for their success. The objectives of this study were (1) to determine the overall acceptance of HPV vaccines for preadolescent children by parents, (2) to evaluate the influence of written educational information about HPV on parental acceptability of HPV vaccines, and (3) to identify independent predictors associated with HPV vaccine acceptability by parents.
METHODS. A randomized intervention study within a cross-sectional survey was conducted. Parental HPV vaccine acceptability was measured under 3 different hypothetical scenarios. A self-administered survey on the knowledge, attitudes, and beliefs about HPV and HPV vaccines was sent to 1600 parents of 8- to 12-year-old children. In addition to a baseline paragraph about HPV that was received by all study participants, a random half of the study participants received a detailed “HPV Information Sheet” outlining the epidemiology and potential clinical sequelae of HPV infection. Independent predictors of parental HPV vaccine acceptability were determined using multivariate linear regression models.
RESULTS. Parents who received the HPV information sheet had higher mean scores on the HPV knowledge assessment tool than the control group. However, despite this apparent improvement in knowledge, there was not a statistically significant difference in HPV vaccine acceptability between the 2 groups.
CONCLUSIONS. Providing parents with an HPV information sheet did seem to improve knowledge about HPV, but this increased knowledge had little effect on the acceptability of these vaccines by parents for their children. Instead, attitudes and life experiences seemed to be more important factors influencing HPV vaccine acceptability among parents.
Human papillomavirus (HPV) infection is believed to be the most common sexually transmitted infection both in the United States and worldwide.1,2 Current estimates suggest that 50% to 80% of sexually active US adults will be exposed to these viruses during their lifetime, resulting in a national prevalence of >20 million infections.1,3,4 Infection with HPV generally is asymptomatic and self-limited but can result in a variety of HPV type-specific clinical sequelae, including cervical neoplasia, genital warts, and possibly other anogenital and head and neck cancers. Although there are >80 types of HPV currently identified, only a handful of these result in clinically significant disease, making the development of prophylactic HPV vaccines feasible. These vaccines have been shown in recent phase III trials to be highly effective in preventing prolonged infection with HPV and in reducing cervical neoplasia.5,6 It is expected that prophylactic HPV vaccines will be licensed for public use by 2007.
Preadolescent children (aged 8–12) are a proposed target population for future HPV vaccines. Vaccinating children of this age range would provide, for the most part, immunity before the onset of sexual debut. Although current immunogenicity data are available only for ∼4.5 years after vaccination, anti-HPV antibody titers persist during this interval, indicating that adequate antibody levels are likely to be maintained during the high-risk periods of adolescence and early adulthood, when exposure to oncogenic HPV types is highest.7–9 Because HPV vaccines will be targeted to children, parental consent likely will be required for children to receive these vaccines, and parental opinions about these vaccines will have a critical impact on the vaccines' success in preventing HPV-associated illnesses. Although some studies have begun to explore the attitudes of parents about vaccines against sexually transmitted diseases in general,10–12 little is known about the parental acceptability of HPV vaccines specifically.13–15
Parental vaccine acceptability for other diseases has been shown to be influenced by specific attitudes and beliefs about disease processes and vaccination.10,16–19 Validated models of health protective behavior, such as the Health Belief Model and the Theory of Reasoned Action, provide a useful framework in which to analyze the relative influences of various psychological processes on complex tasks such as parental decision-making. In our study, psychological constructs from these 2 models of health behavior, as well as a variety of demographic and experiential factors, were assessed for their associations with parental acceptability of HPV vaccines for children.
In addition to understanding the predictors of parental HPV vaccine acceptability, we sought to determine the influence of written educational material about HPV on parental HPV vaccine acceptability. In clinical practice, parents frequently receive “Vaccine Information Sheets” during well-child checkups as a systematic means by which to address questions and concerns about specific vaccines and the diseases that they prevent. Although this practice is a convenient way to convey information to parents about vaccines, few studies have assessed whether this type of intervention influences vaccine acceptability. To test the hypothesis that written educational information about HPV would improve parental HPV vaccine acceptability, we conducted a randomized intervention study whereby half of the study participants received an “HPV Information Sheet” as an educational intervention, and measured parental vaccine acceptability as the primary outcome.
METHODS
This study combined a cross-sectional survey of parental attitudes with a randomized intervention study to evaluate the effect of written educational information about HPV on parental HPV vaccine acceptability. A mail-based, self-administered, written survey was used to assess overall parental acceptability of HPV vaccines for children and to determine predictors of vaccine acceptability. Within the context of the survey, a randomized intervention study was performed to test the hypothesis that receiving written educational information about HPV would result in increased acceptability of HPV vaccines by parents. All study activities were approved by the Group Health Cooperative Center for Health Studies institutional review board. Because completion of the survey implied consent, a consent waiver was obtained before study implementation.
Study Sample
Participants in the study were self-identified parents or primary caregivers of children aged 8 to 12 years. The Group Health Cooperative Center for Health Studies database of medical records was used initially to identify children who were aged 8 to 12 years, were enrolled continuously for at least 1 year in the Group Health Cooperative Health Plan, and resided in 1 of 3 Seattle-area counties. A letter that was addressed to the “parent or primary caretaker” of a random sample of 1600 children who fulfilled these criteria was used to recruit potential participants for the study. For simplicity, all study participants are referred to as “parents” in this article. Additional eligibility criteria for participation in the study included ability to read English and age ≥18 years. Three mailings were performed, with $2 compensation included in the first mailing.
Outcome Measure
The primary outcome measured in this study was parental acceptability of HPV vaccines. Participants were asked to rate the likelihood of allowing their child to receive a hypothetical HPV vaccine at 3 different proposed ages of vaccination: as an infant, as an 8- to 12-year-old child, and as an older teenager. Consistent with other studies of vaccine acceptability,11,19,20 responses were quantified using an 11-point response format that consisted of a series of integers that ranged from 0 (corresponding to the description “would definitely not allow” the child to receive the vaccine) to 10 (corresponding to the description “would definitely allow” the child to receive the vaccine), and the 3 related questions of vaccine acceptability were averaged to generate an “overall” vaccine acceptability scale score (coefficient α = .83).
Intervention and Randomization Process
All surveys contained a 3-sentence paragraph on the inside front cover outlining basic information about HPV that would allow those who were completely unfamiliar with the subject to answer the questionnaire. This paragraph mentioned that HPV was a sexually transmitted infections (STI), was common, was often without symptoms, and that a “vaccine against HPV” consisting of 3 shots likely would be available in the near future. A computer randomization program was used to assign half of the potential survey participants to receive an additional 2-page “HPV Information Sheet” as an educational intervention to be included with their survey. This information sheet, modeled on vaccine information sheets that are available from the Centers for Disease Control and Prevention for other childhood vaccinations, contained detailed information about the incidence, transmission, and potential clinical sequelae of infection with HPV, as well as additional information about HPV vaccines. The information sheet (available from the authors on request) was placed inside the front cover of the survey booklet to be seen before the survey questions, but no specific instructions as to its use were provided and participants were not queried directly about whether they had read it. Participants were blinded to the fact that they were participating in a study to evaluate the effect of an educational intervention on their opinions about HPV vaccines.
Survey Instrument
The 67-item survey assessed attitudes about vaccines, HPV and STIs, sociodemographic characteristics, knowledge about HPV, and experience with STIs and HPV-associated illnesses. Sociodemographic factors that were assessed included age and gender of the study participant, ages and genders of children in the family, number of children in the family, marital status, race/ethnicity, education level, and whether any children in the family had received at least 1 hepatitis B vaccination. Knowledge about HPV was assessed using a 7-item scale of true–false questions. A knowledge assessment “score” was calculated for each individual by adding the number of correct responses, dividing by the number of true–false questions answered, and multiplying by a factor to convert all scores to the number correct out of 7 questions. Experience with HPV-associated illnesses and STIs was assessed with 4-items that asked whether “you or anyone close to you” had experienced abnormal pap smears, cervical cancer, genital warts, or an STI.
Questions regarding attitudes about vaccines, HPV, and STIs were based on 5 psychological constructs from 2 validated models of health protective behavior: the Health Belief Model and the Theory of Reasoned Action. The normative beliefs construct was derived from the Theory of Reasoned Action and was assessed with 2 items that measured how much parents were motivated to comply with the preference of their peers and physicians. The remaining 4 constructs were derived from the Health Belief Model: (1) the perceived susceptibility of disease, which was assessed with 3 items that measured parental perceptions about the likelihood that their child would become infected with a sexually transmitted disease, including HPV specifically; (2) the perceived severity of disease, which was assessed with 3 items that measured parental beliefs about pain, mortality, and embarrassment associated with HPV-related illnesses in their children; (3) the perceived benefits to HPV vaccination, which was assessed with 3 items that measured societal and personal benefits from vaccination against HPV; and (4) the perceived barriers to vaccination, which was assessed with 3 items that measured the perceived discomfort and danger associated with receiving vaccines in general. Responses for questions from each of these 5 constructs were quantified using 5-point Likert scales (1 = strongly disagree to 5 = strongly agree). “Don't know” responses were recoded to be numerically equivalent to “neutral.” Internal reliability estimates were calculated for each construct and found to be adequate for the perceived susceptibility, perceived benefits, and perceived barrier constructs (coefficient α ranging from .68 to .78). For these 3 constructs, a composite scale score was generated by calculating the mean value of responses to the items that composed each construct, and this composite score was used in a multivariate linear regression model to determine independent predictors of parental HPV vaccine acceptability. For the normative beliefs and perceived severity constructs, internal reliability estimates were unacceptably low (coefficient α = .20 and .27 for perceived severity and normative beliefs constructs, respectively), indicating that each of the individual questions from these constructs actually measured multiple, unrelated psychological domains. Therefore, questions from these constructs were considered individually in the multivariate linear regression model of predictors of parental HPV vaccine acceptability.
Sample Size
Using a standard deviation of 25 (based on results from published studies of STI vaccine acceptability10,11,19) we calculated that to have an 80% chance of detecting as significant (at the 2-sided 5% level), a 1-point difference between the 2 groups on the vaccine acceptability scale, 393 participants would be needed for each arm of the study (786 total). We assumed an approximate response rate of 50% and therefore sampled 1600 parents.
Statistical Analysis
Mean vaccine acceptability scale scores were compared between the 2 experimental groups using unpaired t tests with unequal variance assumptions. Repeated-measures analysis of variance was used to make comparisons of vaccine acceptability among the 3 proposed ages of vaccination. Bivariate associations between vaccine acceptability and predictor variables were assessed using Pearson product-moment correlations for continuous predictors (child and parent age, number of children in family, knowledge assessment score, and behavioral model construct scale scores), and t tests for categorical and dichotomous predictor variables (parent and child gender, marital status, race/ethnicity, education, and past experience with HPV/STIs).
Multivariate linear regression analysis was used to determine significant independent predictors of vaccine acceptability. Variables that corresponded to the composite scale measurements from the 3 behavioral model constructs with adequate internal reliability estimates (perceived benefits, perceived barriers, and perceived susceptibility), scores from individual questions from the behavioral model constructs with low internal reliability estimates (perceived severity and normative beliefs), and group assignment (control versus educational intervention) were entered and locked in the multivariate linear regression model as a block because it was hypothesized a priori that these factors would be influential on parental vaccine acceptability. The remaining demographic and experiential variables were added into the model in a forward stepwise manner with an inclusion criteria of a P < .05 to enter the model and a P < .1 to remain in the final model. All data were analyzed using STATA8 (Stata Corp, College Station, TX) statistical software.
RESULTS
Of an eligible sample of 11644 children, a convenience sample of 1600 were randomly selected and their parents were approached for participation in the study (Fig 1). Of these 1600 potential participants, half were randomly assigned to the control group that received only the baseline survey, and the other half were randomly assigned to the intervention group that received both the baseline survey and the HPV information sheet. In both the control and the intervention groups, there were similar numbers of parents who indicated ineligibility for participation, refused to participate, or had surveys returned because of invalid addresses. Using parents with valid addresses as the denominator, response rates in the control and intervention groups were 56.0% and 58.8%, respectively, with an overall response rate of 57.5%.
Sample selection and randomization.
Baseline characteristics of the study participants were similar between the 2 groups (Table 1). The majority of study participants were married, white women with at least some college education. In each group, similar numbers of parents indicated that they or “someone close to them” had experienced an STI and/or the HPV-associated conditions of genital warts, abnormal Pap smears, and cervical cancer. This suggested that the level of “experiential knowledge” about HPV-associated conditions was equivalent between the 2 groups.
Baseline Characteristics of the Study Sample
Experience with HPV-associated illnesses did seem to modify attitudes toward HPV, as we found a general trend toward lower scores on the perceived severity scale and higher scores on the perceived benefits and perceived susceptibility scales among participants who had experience with HPV-associated illnesses, compared with participants who did not have experience with HPV (data not shown). Experience with HPV-associated illnesses did not seem to be associated with differences in the perceived barriers scale, presumably because questions in this scale pertained to barriers to vaccination in general and were not specific to HPV vaccines. Despite these differences in the mean scale scores for these constructs between respondents who were dichotomized by their experience with HPV-associated illnesses, there were no statistically significant differences in these mean scale scores (unadjusted for other factors) when respondents were dichotomized by experimental group.
Parental Knowledge About HPV and Vaccine Acceptability
Our hypothesis was that receipt of the HPV information sheet would lead to increased knowledge about HPV and subsequent increased parental acceptability of HPV vaccines. Parents who received the HPV information sheet had a higher mean score on the 7-item HPV knowledge assessment tool than those in the control group (point estimates: 5.57 vs 4.17, intervention and control groups, respectively; 95% confidence interval [CI]: 5.41–6.73 and 3.99–4.35 for intervention and control groups, respectively; P < .001). However, despite this apparent improvement in knowledge, there was no significant difference between the 2 groups with respect to the mean parental vaccine acceptability scale scores (intervention group score: 6.56 [95% CI: 6.28–6.84]; control group score: 6.28 [95% CI: 5.99–6.57]: P = .17), suggesting that receipt of the HPV information sheet did not substantially alter parental acceptability of HPV vaccines.
Proposed Age of Vaccination and Vaccine Acceptability
The primary outcome of interest was “overall” HPV vaccine acceptability, which was measured using the average score from a 3-item scale that assessed parental acceptability of HPV vaccines at 3 different proposed ages of vaccination: when their children were infants, preadolescents (8–12 years of age), or “older teenagers.” However, when vaccine acceptability at each of these 3 proposed ages was assessed separately, we found that parental vaccine acceptability increased as the proposed age of vaccination increased (Fig 2). Repeated measures analysis of variance indicated that there were statistically significant differences in vaccine acceptability between each age category and that this effect occurred in both the control and the intervention groups (intervention group F statistic: 86.40 [P < .001]; control group F statistic: 68.11 [P < .001]). Although point estimates for unadjusted vaccine acceptability was lower in the control group than in the intervention group in every age category, these differences were not statistically significant.
Vaccine acceptability at different proposed ages of vaccination.
Predictors of HPV Vaccine Acceptability
Multivariate linear regression analysis was used to identify independent demographic, experiential, and attitudinal predictors of parental HPV vaccine acceptability. Figure 3 shows the relative standardized β coefficients for variables with statistically significant associations (P < .05) with parental vaccine acceptability. Believing in the benefits of HPV vaccines to society and to their children (perceived benefits construct; β = .38, P < .001) was associated with the largest effect on parental vaccine acceptability. In addition to perceived benefits, 5 other characteristics were statistically associated with increased parental HPV vaccine acceptability. These included being influenced by peer groups (normative belief construct individual question 1: β = .08, P = .004), being influenced by physician recommendation (normative belief construct individual question 2: β = .19, P < .001), perceiving that their child was susceptible to STIs and/or HPV infection (perceived susceptibility construct: β = 0.17, P < .001), having had personal experience with genital warts (β = .05, P = .042), and answering questions in the survey about a female child (β = .06, P = .030). Believing that their child experienced significant discomfort or danger when receiving immunizations (perceived barriers construct: β = −.19 P < .001) was associated with decreased parental HPV vaccine acceptability. Three other variables remained in the final model because they had been “locked in” when added as a block with the other behavioral model constructs, but their associations with vaccine acceptability did not reach statistical significance. These included receiving the HPV information sheet (β = .03, P = .236) and believing that HPV infection leads to serious consequences (perceived severity construct, β = .006, P = .845; and β = .05, P = .078, respectively, for the 2 individual questions).
Statistically significant β coefficient from multivariate linear regression model of independent predictors of HPV vaccine acceptability among parents.
DISCUSSION
Our results suggest that although providing parents with an HPV information sheet did seem to improve knowledge level about HPV, at least in the short term, this increased knowledge had little effect on the acceptability of HPV vaccines by parents for their children. These results suggest that simply “educating” parents about HPV and HPV vaccines may not be sufficient to influence their attitudes toward HPV vaccination, as attitudes may be driven by other, non–information-based preferences.
Our analysis of the independent predictors of HPV vaccine acceptability indicated that parental beliefs and attitudes may be more influential than knowledge about HPV on the parental decision-making process. This has important implications for medical providers and public health practitioners when communicating with parents about HPV vaccines in the future. Focusing discussions on the benefits that are associated with giving HPV vaccines to children or on addressing general concerns that parents may have about the discomforts and/or dangers that are associated with vaccines may be useful strategies for promoting these vaccines. Furthermore, our finding that answering questions in the survey about female children was predictive of increased vaccine acceptability suggests that if policy makers adopt a universal HPV vaccination policy (male and female), then addressing the benefits to male vaccination specifically may be necessary for these vaccines to be widely used. Additional studies should be aimed at developing and testing specific “public health messages” to find those that have the greatest impact on improving the acceptability of these and other vaccines.
Exposure to HPV is ubiquitous in the sexually active adult population,1,3,4,21 and that HPV is a sexually transmitted disease may have a negative impact on the acceptance of these vaccines in society. Although several studies have found parental acceptance of STI vaccines in general to be high,11,12,15,22 sexually transmitted diseases are associated with significant social stigmata, and some parents may believe that having their child vaccinated against an STI would equate to condoning precocious sexual behavior. Indeed, a recent study on the predictors of parental and adolescent acceptability of STI vaccines did find that parental concern that STI vaccination promoted unsafe sexual behaviors was associated with a significant decrease in overall STI vaccine acceptability.12 Although specific questions to address this issue were not included in our study, some parents (∼1%) expressed similar sentiments in the “additional comments” section of the survey. Future studies should attempt to determine which populations of parents object to giving STI vaccines to children and the underlying reasons for these beliefs.
The social stigma of sexually transmitted diseases also may explain why having had experience with genital warts but not cervical cancer or abnormal Pap smears was found to be an independent predictor of increased vaccine acceptability in our model. Unlike cervical cancer and abnormal Pap smears, genital warts are a visually apparent condition with the potential to be recognized by the lay person as a sexually transmitted disease. Significant psychological and emotional distress has been associated with having overt symptoms of disease23–26; therefore 1 hypothesis to explain the discrepant effects of the various HPV-associated conditions on parental vaccine acceptability is that parents who have experience with genital warts may have a stronger desire to protect their children from the negative social effects that are associated with having a sexually transmitted disease than those who had not experienced this condition. Indeed, results from our survey indicate that among parents who would consider having their child vaccinated against HPV, the vast majority (94%; data not shown) would prefer that their child receive a vaccine that is protective against both cervical cancer and genital warts, rather than a vaccine that is protective against cervical cancer alone.
This study has several important limitations that may affect the interpretation of results. First, because participants all were drawn from the same Washington regional health maintenance organization, the study sample was relatively homogeneous, and similar results may not be seen in other populations of parents. A recent Georgia-based study presented by Davis et al13 found that written information about HPV did improve parental acceptability of HPV vaccines, although this effect was prominent only among specific subgroups of parents. Although differences in study methods could account for the contrast of these results with the findings of our study, it also is possible that attitudes about HPV vaccines vary among people from different geographic regions and different sociocultural groups. Understanding these differences will be important in the future as policy makers plan national HPV vaccine implementation strategies.
Another limitation of this study is that because the survey was self-administered, parents did not have the opportunity to ask medical providers unresolved questions about HPV that might influence their acceptability of HPV vaccines. Furthermore, parents who completed the survey may have had inherently higher motivation and interest in issues related to their child's health and therefore greater interest in preventive health behaviors for children such as vaccination. However, our results showed that even in this population of potentially highly interested and motivated parents, HPV vaccine acceptability varied widely. In future studies, it will be important to test interventions in both clinical and nonclinical settings to understand more completely the factors that affect implementation of HPV vaccines into our society.
An additional limitation to this study is that using the vaccine acceptability scale score to predict parental decision-making about having a child vaccinated against HPV may be imprecise. This scale, which has been used in several studies on vaccine acceptability,11,19,20 provides an informative framework in which to evaluate the relative impact of different characteristics on decision-making. However, it is unknown to what extent 1-point differences in this scale reflect differences in the actual receipt of vaccines or whether scale scores for different vaccines are comparable. At the same time, a large body of marketing research and research on models of health protective behavior, such as the Theory of Reasoned Action and the Health Belief Model, indicate that measuring the intention to engage in a behavior is a significant and reliable predictor of subsequent engagement in the behavior.27
Finally, this study examined the influence of only a subset of parental sociodemographic, experiential, and attitudinal characteristics on the decision to have children vaccinated against HPV, and there likely are additional factors that also inform the opinions of parents about HPV vaccines. For example, studies on the acceptability of other vaccines indicate that vaccine efficacy and cost are important characteristics that affect the acceptability of vaccines among parents and adolescents.11,20 The influence of these factors was not assessed in our survey.
CONCLUSION
Providing parents with a written information sheet about HPV did lead to improvement in their knowledge about HPV but did not result in substantial increases in HPV vaccine acceptability. Increased HPV vaccine acceptability instead was associated with the belief that HPV vaccination was beneficial and protective to health and to a lesser extent belief in the susceptibility of children to HPV infection, being influenced by the opinions of peers and by doctor recommendation, vaccinating a female child, and parental experience with genital warts. Decreased vaccine acceptability was associated with general concern that children experience significant discomfort or danger from vaccination. If these factors are found consistently to be influential in diverse populations of parents, then future HPV vaccination campaigns should attempt to stress the benefits of HPV vaccination for children and address the fears of parents about dangers or discomfort associated with vaccines.
Footnotes
- Accepted October 18, 2005.
- Address correspondence to Amanda F. Dempsey, MD, PhD, MPH, 300 N Ingalls St, Room 6C13, Ann Arbor, MI 48109-0456. E-mail: adempsey{at}med.umich.edu
Financial Disclosure: Dr Zimet received speaking honoraria from an educational grant funded by Merck. Ms Koutsky receives partial research funding from Merck.
REFERENCES
- Copyright © 2006 by the American Academy of Pediatrics
In this issue
Jump to section
Related Articles
Cited By...
- Vaccine trust and the limits of information
- Awareness of HPV and Uptake of Vaccination in a High-Risk Population
- The Student Human Papillomavirus Survey: Nurse-Led Instrument Development and Psychometric Testing to Increase Human Papillomavirus Vaccine Series Completion in Young Adults
- Promoting HPV Vaccination in Safety-Net Clinics: A Randomized Trial
- Risk Factors for Non-initiation of the Human Papillomavirus Vaccine among Adolescent Survivors of Childhood Cancer
- Human Papillomavirus Vaccination Series Initiation and Completion, 2008-2009
- HPV and HPV Vaccine Education Intervention: Effects on Parents, Healthcare Staff, and School Staff
- Parental Decision Making about the HPV Vaccine
- Human Papillomavirus Vaccination Practices: A Survey of US Physicians 18 Months After Licensure
- Does Framing Human Papillomavirus Vaccine as Preventing Cancer in Men Increase Vaccine Acceptability?
- Mother-Daughter Communication and Human Papillomavirus Vaccine Uptake by College Students
- Behavioral Correlates of HPV Vaccine Acceptability in the 2007 Health Information National Trends Survey (HINTS)
- Mothers' Intention for Their Daughters and Themselves to Receive the Human Papillomavirus Vaccine: A National Study of Nurses
- "You Get What You Get": Unexpected Findings About Low-Income Parents' Negative Experiences With Community Resources
- A Critique of Criteria for Evaluating Vaccines for Inclusion in Mandatory School Immunization Programs
- Intention of parents to have male children vaccinated with the human papillomavirus vaccine
- Parental intention to have daughters receive the human papillomavirus vaccine
- The HPV vaccine: An analysis of the FUTURE II study
- HPV and Cervical Cancer Prevention Counseling With Younger Adolescents: Implications for Primary Care
- A National Survey of Pediatrician Knowledge and Attitudes Regarding Human Papillomavirus Vaccination
- Factors associated with parental acceptance of HPV vaccines