Article Text

other Versions

PDF
Demographic and socioeconomic determinants of influenza vaccination disparities among university students
  1. M Uddin1,2,
  2. G C Cherkowski1,2,
  3. G Liu1,2,
  4. J Zhang1,2,
  5. A S Monto1,
  6. A E Aiello1,2
  1. 1Department of Epidemiology, University of Michigan School of Public Health, Ann Arbor, Michigan, USA
  2. 2Center for Social Epidemiology and Population Health, University of Michigan-School of Public Health, Ann Arbor, Michigan, USA
  1. Correspondence to Allison E Aiello or John G Searle, Center for Social Epidemiology and Population Health, 1415 Washington Heights, Ann Arbor, MI 48109-2029, USA; aielloa{at}umich.edu

Abstract

Background The Advisory Committee on Immunization Practices encourages dormitory residents to receive influenza vaccination. To our knowledge, there are no studies that have directly examined factors associated with vaccination uptake among university students residing in dormitories. We therefore sought to examine the influence of demographic, social and health behaviours on influenza vaccination coverage among college dormitory students.

Methods Cross-sectional analysis of baseline questionnaire data obtained from 845 eligible participants in a non-pharmaceutical intervention study for reducing influenza during the 2007–2008 influenza season. Significant predictors were identified through logistic regression analysis with generalised estimating equations to account for resident clustering.

Results Increasing parental educational attainment was significantly associated with a trend in higher vaccination uptake among students: college graduate versus some college or less (OR 3.48, 95% CI 1.33 to 9.12) and some postgraduate education versus some college or less (OR 5.89, 95% CI 2.35 to 14.80) (trend test p<0.001). Adjusting for covariates, reported influenza vaccination for the 2007–2008 influenza season was strongly associated with reported influenza vaccination for the 2006–2007 influenza season (OR 16.38, 95% CI 9.28 to 28.91) and with speaking to a health professional about precautions to take against influenza (OR 2.95, 95% CI 1.42 to 6.13).

Conclusions The effect of parental educational status on vaccination rates can carry over to offspring, even among those who attain college student status. Programs targeting students who are employed on campus and who have never been vaccinated may be an especially effective way to increase vaccination rates, as both of these factors were significantly related to parental socioeconomic status in this study.

  • Influenza
  • vaccination
  • immunization
  • health disparities
  • socioeconomic status
  • social inequalities

Statistics from Altmetric.com

The Advisory Committee on Immunization Practices (ACIP) currently does not consider healthy individuals between 18 and 50 years of age to be a key group for influenza vaccination,1 unless they belong to specifically defined at-risk groups. Nevertheless, within the general population in this age range, the ACIP recommends that dormitory residents be encouraged to receive vaccination and, in the case of influenza epidemics, strongly encourages vaccination of dormitory students to reduce disruption of routine activities.1 While college students are not at high risk for hospitalisation, the impact of influenza on this population is not inappreciable.

Influenza has the ability to spread rapidly within college student populations. Although influenza illness rates among adults less than age 65 years generally stay below 10% during non-pandemic years,2 attack rates well above 25% have previously been reported among college students in a single influenza season,3–5 with the disease being significantly more prevalent among those living on versus off campus.4 Illness due to influenza can put students at risk for reduced academic performance, absence from class and/or work, increased antibiotic use and increased healthcare utilisation.5–7 In short, the disease has the capacity to disrupt an entire campus, leading in some cases to whole-class suspension.5 To date, there is no formal estimate of the overall burden of influenza within colleges nationally, although with over 14 million undergraduate students enrolled in college8 the collective burden has the potential to be quite high. As such, reducing influenza infection within the college student population, particularly among dormitory students, should be a health priority.

Influenza vaccine uptake is low among healthy adult populations in the USA. Between 2000 and 2007, the National Health Interview Survey reported an annual coverage ranging from 10% to 18% among all respondents aged 18–49 years.9 Identifying factors that are associated with influenza vaccine uptake among college students may reveal information useful to improving low uptake rates. Although numerous studies have investigated predictors that are associated with influenza vaccine uptake among high-risk or ACIP-targeted groups (eg, 10–15), very few examine predictors among healthy adults.16 17 To our knowledge, there are no studies that have directly examined factors associated with vaccination uptake among university students. The purpose of this study was therefore to identify predictors of influenza vaccination uptake among university students residing in a dormitory setting. Candidate predictors included demographic factors, perceived risk and knowledge about influenza, and high-risk health status. We also investigated the role of behavioural factors, such as smoking and drinking, which are often considered to be characteristic of a university student population and might influence influenza vaccination uptake.

Methods

Study design and eligibility

A cross-sectional analysis was performed using baseline questionnaire data obtained from the M-FLU study for the 2007–2008 influenza season. The M-FLU study was a randomised, controlled, non-pharmaceutical intervention conducted to test the feasibility and efficacy of facemasks and hand hygiene for reducing influenza-like illness in the university community setting (http://clinicaltrials.gov identifier: NCT00490633) . The baseline questionnaire was administered between 15 January 2008 and 19 February 2008. This study was approved by the institutional review board at the University of Michigan.

For this analysis, eligible subjects were participants 18 years and older residing in campus dormitories at the University of Michigan who had not participated in the M-FLU study during the previous (2006–2007) influenza season. Before the intervention began, consenting individuals were asked to electronically complete a baseline survey. Our study focused exclusively on the baseline questionnaire data. There were 845 eligible participants, representing 76.2 of the M-FLU respondent population that completed the baseline questionnaire during the 2007–2008 influenza season. Of those, 19 were excluded for not providing information on vaccination status. Analysis was performed on the remaining 826 participants.

Variables

Influenza vaccination status for the 2007–2008 influenza season was the primary outcome variable for this study. Age data were available but were excluded from analysis due to the narrow age range (mean 18.8 (SD 0.925)). Race/ethnic categories that had limited participant representation (eg, Native Hawaiian, Pacific Islander, American Indian, Alaskan Native, and multiethnic) were combined and categorised as “other”. As a measure of socioeconomic status (SES), the highest level of education attained by the participant's father and mother were combined to create a variable representing highest level of education among either parent and coded as follows: some college or less, college graduate and some postgraduate education. Influenza vaccine status for the previous 2006–2007 season was derived from a variable containing the self-reported date of the most recent influenza vaccine before the 2007–2008 season. Reported soap and water hand-washing time was divided between those reporting times between 0 and <20 s and those reporting times of 20 s or greater, based on recommendations for the community setting.18 Average daily use of hand sanitiser was categorised into greater than three times per day versus less than three times per day, based on reported average use in the study population. A high-risk condition status variable was created to indicate whether the participant responded “yes” to conditions known to increase the risk for medical complication due to influenza (ie, asthma, reactive airway disorder, diabetes, treatment for cancer/HIV/AIDS/organ transplant, impaired immunity) for either themselves or for others with whom they were in contact. Finally, analysis variables that did not require re-categorisation included gender, US citizenship, smoking, drinking, physical activity, employment status, concerns about influenza sickness in self or family, viewing information in the media about precautions to take to avoid contracting influenza and speaking with a health professional about precautions to take to avoid getting influenza.

Statistical analysis

Analyses were conducted using the SAS software (SAS, Inc., V.9.1). χ2 analyses for dichotomous variables and Cochran–Armitage tests for ordinal variables were calculated to examine the relationship between each characteristic of interest and vaccine status. Predictor variables that produced p values ≤0.10 in univariate analyses were selected as variables for logistic regression models. Odds ratios (ORs) were generated from logistic regression using a genmod procedure in SAS. Genmod allows estimation of generalised estimating equations that accommodate the clustered nature of the data (eg, clustering by participants living in the same dormitory hall). Adjusted models were performed based only on respondents with data available for all variables included in the model (n=665). Bootstrapping analysis of unadjusted and adjusted models was conducted by taking a random sample of 826 observations and calculating the outcome 1000 times. Type 3 trend tests were used to test the association between predictors identified as significant in univariate tests and vaccination status for the 2007–2008 influenza season in the generalised estimating equation analyses.

Results

The characteristics of study participants are presented in table 1. For the 2007–2008 influenza season, 17.3% of the sampled students reported receiving an influenza vaccination. Participants were between 18 and 23 years of age (mean 18.8 years, range 18–23 years). The ratio of women to men was approximately 1:1 with slightly more female subjects (54%) than male subjects (46%). Women had a slightly higher level of reported vaccine uptake during 2007–2008 (18.6%) compared to men (15.7%) (p=0.275). There were no significant differences in vaccination reports for 2007–2008 across race/ethnic groups: 19.0% among whites, 15.6% among Asians, 10.8% among blacks and 14.3% among others (p=0.302). Reported vaccination was slightly higher among US citizens compared to non-US citizens (17.7% vs 12.5%, respectively) but was not statistically significant (p=0.323). For employed students, reported vaccine use was significantly lower (13.2%) compared to non-employed students 20.0% (p=0.013). In addition, there were significant differences in reported vaccination uptake across levels of parental education: vaccine coverage was 4.6% for participants whose parents had some college education or less, 14.5% for those whose parents were college graduates and 22.3% for those whose parents had some postgraduate education (p<0.001). Those who were considered at higher risk for medical complication due to influenza had a significantly higher vaccination coverage (26.4%) compared to those considered not at risk (15.0%) (p<0.001).

Table 1

Characteristics of study participants

A subset of study participants reported characteristics associated with increased risk of medical complications due to influenza as outlined in ACIP recommendations (table 2). In total, 163 participants, or 20.2% of the sample, responded yes to at least one increased risk factor. Results from the univariate analysis between health behaviour information and current vaccine status appear in table 3. Influenza vaccination during the 2007–2008 season was strongly associated with influenza vaccination reported in the 2006–2007 season (p<0.001). The remaining health behaviour and infection precaution variables were not significantly associated with vaccination for the current influenza season. Table 4 shows the results of univariate analyses examining perception and knowledge information with current vaccination status. There was a significant positive association between speaking with a health professional and reported vaccination status (p<0.001). The remaining variables showed some variation across current vaccine status although these associations were not statistically significant (table 4).

Table 2

Risk for medical complication due to influenza characteristics

Table 3

Univariate analysis of health behaviour information with current vaccination status

Table 4

Univariate analyses of perception and knowledge information with current vaccination status

Using logistic regression, we assessed the relationship between bivariate highest level of parental education (ie, college or greater vs less than college) and other variables that may be related to parental SES but also significantly related to vaccination uptake. These included student employment, higher risk status, previous vaccination and speaking with a health professional. Students who were employed were less likely to have a parent with a college degree or more than those who were not employed (OR 0.46, 95% CI 0.31 to 0.70); therefore, we did not include student employment when examining parental SES in multivariable models to avoid overadjustment. Among the remaining three variables, only previous vaccination showed a positive and significant association with parental SES, with those reporting vaccination during the prior 2006–2007 influenza season showing a greater likelihood of having a parent with a college degree or more (OR 2.80, 95% CI 1.26 to 6.21). However, speaking with a health professional was positively associated with parental SES (OR 1.74, 95% CI 0.85 to 3.56), and results showed near significance when education was modelled using three rather than two levels (OR 1.47, 95% CI 0.96 to 2.25). Higher risk status was not significantly associated with parental SES (OR 0.97, 95% CI 0.58 to 1.60).

Table 5 presents unadjusted and adjusted results of vaccination status according to significant study characteristics. In unadjusted models, increasing parental education level was significantly associated with increased likelihood of vaccination: compared to students whose parent(s) had completed some college or less, the odds of vaccination were more than three times greater in students whose parent(s) were college graduates (OR 3.48, 95% CI 1.33 to 9.12) and more than five times greater for students whose parents had completed some postgraduate education (OR 5.89, 95% CI 2.35 to 14.80, p<0.001). This trend remained significant after adjusting for potential mediators of the SES–vaccination uptake relationship, including reported discussions of vaccination precautions with health professionals, high-risk status and prior season influenza vaccination (table 5). In unadjusted models, the odds of vaccination among those who spoke with a health professional within the past 3 months regarding precautions to take to avoid getting influenza were seven times greater than those who did not have the discussion (OR 7.19, 95% CI 4.56 to 11.34, p<0.001), and almost three times greater in the adjusted model (table 5). The odds of vaccination among students at increased risk for medical complication due to influenza were more than two times greater when compared to individuals not at risk in unadjusted models (OR 2.02, 95% CI 1.34 to 3.05, p=0.003); adjustment for other covariates, however, attenuated this association (table 5). Finally, the odds of vaccination for the 2007–2008 influenza season among individuals who were vaccinated during the prior 2006–2007 influenza season were 24 times greater when compared to those who were not vaccinated for the prior season (OR 24.11, 95% CI 14.47 to 40.19, p<0.001), an association that remained significant in the adjusted model (table 5). Bootstrapping analyses of unadjusted and adjusted models using a random resampling of 826 observations showed results that were similar in magnitude and direction to those presented in table 5 (Supplementary Table 6).

Table 5

Logistic regression analyses of vaccination status by study characteristics (unadjusted and adjusted models)*

Discussion

College students living in dormitories face an increased risk for influenza, yet have relatively low rates of influenza vaccine coverage. We conducted this study to identify predictors of vaccination uptake within a university dormitory student population to improve our understanding of influenza vaccination practice within this population. Our findings show that influenza vaccination coverage varies significantly according to vaccination status for the prior influenza season, recent influenza discussion with a health professional and SES as measured by parental educational attainment.

The strongest predictor of vaccination in our population was the report of a vaccination during the prior influenza season. The association between current and prior influenza vaccination has been previously reported,19 with sometimes similarly large observed ORs.19–22 This finding suggests that criteria necessary to overcome or negate barriers to influenza vaccination may carry over from year to year within an individual. Similarly, this study also confirms earlier work showing that encouragement from health professionals can improve beneficial health practices, including vaccination.23–27 Discussion with a health professional about precautions to take against influenza may effectively mitigate negative perceptions of vaccine effectiveness and side effects, both of which have previously been associated with influenza vaccination practice.28 29

Significant associations have previously been observed between SES and influenza vaccination uptake.30–32 Consistent with those studies, the present work detected a significant association between parental SES and vaccination, with students of parents with higher educational attainment having higher influenza vaccination coverage. However, our analysis also suggests an additional, important dimension to this association: SES gradients in influenza vaccination rates can persist, even across successive generations that have achieved university student status. In this study, the sample was comprised of young adults enrolled in a prestigious university; in other words, these students can be considered collectively as a high SES population. Given this relatively high educational attainment, we might expect the effects of parental SES to be non-significant or, at minimum, more dilute across vaccination status than what we observed. Instead, we found that an SES marker of adults in one generation—in this instance, parental education—carried over to influenza vaccination disparities among adults in the next generation. One possible mechanism behind this observation is that university students may adopt, in part, parental vaccination habits. More specifically, if low SES parents have lower influenza vaccination coverage, the characteristics that modulate this relationship may carry over to the offspring of low SES parents. Although the survey on which this study was based did not ask questions that would have enabled us to test this hypothesis, this mechanism remains plausible given that associations have been found between parent SES status in relation to health outcomes and health behaviours in offspring33–35 with effects that can persist into adulthood.33 Further life course studies should examine the existence of generational SES disparities in influenza vaccination rates.

In addition to SES as measured by parental education level, other factors included in this study showed significant associations with vaccination uptake and may be potential mediators of this outcome. For example, influenza vaccination in the prior influenza season and speaking with a healthcare professional about the influenza were positively and significantly (or nearly significantly) associated with higher SES. On the other hand, employment was negatively and significantly associated with higher SES. It is likely that at least some of these factors act as intervening variables in the pathway between SES and vaccination uptake. Programs aimed at improving influenza vaccination uptake rates should consider these factors as potential targets for intervention that, along with SES, may improve vaccination uptake rates in university student populations. For example, one potential way to reach low SES students and/or students with low vaccination coverage might be to develop university healthcare campaigns that target employed students. Since colleges typically employ their students as work study or temporary employees, school influenza intervention programs may be able to target these individuals on their own premises and, potentially, make inroads into reducing SES-related vaccination disparities that persist across generations.

Factors that showed no significant association with influenza vaccine coverage included gender, US citizenship, smoking, drinking, physical activity, hand hygiene, concerns about influenza sickness in self or family and viewing information about precautions to avoid contracting influenza in the media. Nevertheless, the possible association between some of these factors and influenza vaccination coverage should not be completely rejected. Certain categories within these variables had small sample sizes, such as being a current or past daily smoker (table 2) and being a non-US citizen (table 1). Limited sample sizes in some categories could have prevented the analysis from capturing potentially significant differences among these variables. Many of these factors have been identified as important predictors in other samples. It is possible that our university sample may be too homogeneous with respect to these characteristics to identify relationships with vaccination rates.

This study has several limitations that are relevant to a thorough evaluation of its findings. Most notably, all results are based on data that were collected by a larger intervention study conducted at a single university; results from the cross-sectional baseline questionnaire presented here thus did not include all desired variables specific to testing our hypothesis and, furthermore, may not be generalisable to other institutions. Moreover, all data were self-reported by participants who were paid a nominal amount (US$40–100) to participate in the parent M-FLU intervention study, and there may be recall bias in their survey reports. In addition, adjustment for likely mediators of vaccination uptake could have attenuated the results obtained in our final multivariable models. Finally, since assessments of SES were based on the sole SES indicator included in the baseline M-FLU survey, namely parental education, our study may not have captured the complete context of the relationship between SES and influenza vaccination. Although parental education is an accepted measure of SES (eg, ref36), investigation of this relationship by other studies using other indicators, such as household income, would provide additional welcome confirmation of the trend observed in this study.

Conclusions and implications

In conclusion, we have shown that, among undergraduate students residing in dormitory halls at the University of Michigan, influenza vaccination coverage varies significantly according to prior influenza season influenza vaccine status, recent influenza discussion with a health professional and SES as measured by parental educational attainment. Health programs aimed at increasing influenza vaccination uptake among college dormitory students should account for the finding that, once a student is vaccinated, they are far more likely to be vaccinated the following year; targeting college students who have never been vaccinated may produce a more lasting impact on student vaccination rates over the longer term than targeting the entire student population. This strategy would also be a more cost-effective way to raise student vaccination rates in an era of increasingly scarce healthcare resources. Programs targeting students who are employed on campus and who have never been vaccinated may be an especially effective way to increase vaccination rates, as both of these variables were significantly related to parental SES in this study. Future research should work to identify differences between successive and non-successive influenza vaccinators to better guide intervention efforts geared towards creating successive vaccinators. Finally, given our finding of a significant association between SES and influenza vaccination coverage within this relatively high SES sample (ie, a sample in which a large majority had parents with a college degree or greater), it is possible that such an association may be even more profound in other settings with more SES variation. Determining the nature and scope of social correlates that contribute to influenza vaccination disparities in settings with more SES variation could serve as an important step towards improving the health of successive, overlapping generations, by identifying additional targets for intervention that may increase influenza vaccination uptake rates.

What this study adds

  • This is the first study to demonstrate that vaccination uptake in an adult population increases with increasing parental educational attainment, suggesting that socioeconomic status (SES) gradients can persist even through successive generations that have achieved university status.

  • This study also supports the importance of discussions with health professionals and prior vaccination as important predictors of continued vaccination uptake in a university age population.

  • Universities should consider developing programs that increase influenza vaccination rates among low SES students, including targeted discussions with university healthcare providers.

Acknowledgments

We gratefully acknowledge the hundreds of participants who made this study possible. Special thanks go to Stephen H Waterman and David Shay, of the Centers for Disease Control and Prevention, who provided helpful comments on this work.

References

View Abstract

Footnotes

  • Funding This work was supported by funding from the Centers for Disease Control (U01 C1000441 to AEA and ASM).

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the institutional review board at the University of Michigan.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.