• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of jgimedspringer.comThis journalToc AlertsSubmit OnlineOpen Choice
J Gen Intern Med. Aug 2006; 21(8): 857–861.
PMCID: PMC1831584

Impact of Health Literacy on Socioeconomic and Racial Differences in Health in an Elderly Population

Abstract

BACKGROUND

Differences in health literacy levels by race and education are widely hypothesized to contribute to health disparities, but there is little direct evidence.

OBJECTIVE

To examine the extent to which low health literacy exacerbates differences between racial and socioeconomic groups in terms of health status and receipt of vaccinations.

DESIGN

Retrospective cohort study.

PARTICIPANTS (OR PATIENTS OR SUBJECTS)

Three thousand two hundred and sixty noninstitutionalized elderly persons enrolling in a Medicare managed care plan in 1997 in Cleveland, OH; Houston, TX; South Florida; and Tampa, FL.

MEASUREMENTS

Dependent variables were physical health SF-12 score, mental health SF-12 score, self-reported health status, receipt of influenza vaccine, and receipt of pneumococcal vaccine. Independent variables included health literacy, educational attainment, race, income, age, sex, chronic health conditions, and smoking status.

RESULTS

After adjusting for demographic and health-related variables, individuals without a high school education had worse physical and mental health and worse self-reported health status than those with a high school degree. Accounting for health literacy reduced these differences by 22% to 41%. Black individuals had worse self-reported health status and lower influenza and pneumococcal vaccination rates. Accounting for health literacy reduced the observed difference in self-reported health by 25% but did not affect differences in vaccination rates.

CONCLUSIONS

We found that health literacy explained a small to moderate fraction of the differences in health status and, to a lesser degree, receipt of vaccinations that would normally be attributed to educational attainment and/or race if literacy was not considered.

Keywords: educational status, health literacy, health status, minority groups

Differences in health literacy levels by race, income, and education are widely hypothesized to contribute to health disparities. Healthy People 2010 states that “Equitably distributed health communication resources and skills, and a robust communication infrastructure can contribute to the closing of the digital divide and the overarching goal of Healthy People 2010 to eliminate health disparities.”1 Despite the intuitive connection between low health literacy and disparities, a recent review of the literature on health literacy2 found only 1 study3 documenting the link statistically. In this study, we used one of the only large datasets containing measures of health literacy, demographic characteristics, and health outcomes to explore the impact of health literacy on differences in health status and vaccination by educational attainment and race.

METHODS

Study Sample

Patient enrollment and data collection, which were conducted by the Prudential Center for Healthcare Research (now the Emory Center on Health Outcomes and Quality), have been described in detail previously.4 Individuals newly enrolling in the Medicare managed care plans of Prudential Healthcare in 4 locations (Cleveland, OH; Houston, TX; South Florida, and Tampa, FL) between December 1996 and August 1997 were eligible to participate. New members were contacted 3 months after enrollment, and those meeting the eligibility criteria were asked to complete an in-person survey. To be included in the study, members had to be comfortable speaking either English or Spanish, living in the community, and possess adequate visual and cognitive function. Spanish-speaking patients were interviewed in Spanish. Of the 7,471 enrollees who were originally contacted, 3,247 refused to participate. The sample size for the analysis of differences by education level was 3,260. The sample size for the analysis for differences by race, which included only blacks and whites, was 2,850.

Data

Health literacy and selected demographic and health characteristics of the population were obtained from the baseline in-person survey. Health literacy was measured using the Short Test of Functional Health Literacy in Adults.5,6 Based on their responses, subjects were classified as having either “adequate,”“marginal,” or “inadequate” health literacy.6 The survey also included questions about respondents' demographics, socioeconomic status, chronic conditions (“Have you ever had…”), and health-related behaviors.

Our dependent variables, measures of health status and receipt of vaccination, were as follows: physical health SF-12, mental health SF-12, self-reported health status (fair or poor vs good, very good, or excellent), receipt of influenza vaccination, and receipt of pneumococcal vaccination. Survey questions on vaccination asked respondents whether they had ever received the vaccination. Independent variables were age (65 to 75, 75 to 84, 85+), gender, race/ethnicity (white, black, Spanish speaking, other), education (<8 years, 9 to 11 years, high school degree, some college, college degree), health literacy (inadequate, marginal, adequate), income (<$10,000, $10,000 to $25,000, >$25,000, no response), tobacco (never, former, current) and alcohol (none, light to moderate, heavy) consumption, and self-reported chronic health conditions. For purposes of studying the impact of health literacy on differences by education level, we dichotomized the education variable into 2 categories: high school degree and no high school degree.

Analysis

We assessed differences in independent variables between groups using χ2 tests. To determine the impact of health literacy on differences in the dependent variables by education and by race, we estimated 2 regression models for each dependent variable. The first included all of the independent variables listed above but omitted controls for health literacy; the second included controls for health literacy. We used ordinary least squares regression for continuous physical and mental SF-12 scores and logistic regression for the other dependent variables, which were dichotomous.

Because individual regression coefficients are difficult to interpret, we re-stated results from the regression models in terms of regression-adjusted, or predicted, values. Regression-adjusted values for physical and mental SF-12 scores are on the original scale. Regression-adjusted values for the other dependent variables, which are categorical, are probabilities. For each regression and dependent variable, we computed 2 adjusted values per respondent. We obtained the first value by recoding the education variables to indicate that the respondent had less than a high school education (regardless of the respondent's true level of education) and leaving the other independent variables as is. For the second value, we recoded the education variables to indicate that the respondent completed high school. We averaged these values across the sample and computed the difference, yielding the “sample average treatment effect” of education. This procedure effectively nets out the impact of the observed covariates. By comparing the magnitude of the average difference across regression models (one without health literacy controls, the other with), we determined the degree to which differences by education were mediated by health literacy. Because results are stated on the original scale, they are easier to interpret.7 We repeated this procedure for race, computing each outcome as if everyone in the sample was black and then as if everyone in the sample was white. We used bootstrapping to determine the confidence intervals around the differences in regression-adjusted values.8

RESULTS

Table 1 shows differences in health literacy, education, race, and other characteristics by education level (high school graduate vs not a high school graduate) and by race (white vs black). Differences in health literacy levels were substantial and significant (P<.001 in each case). For example, 78% of high school graduates, but only 40% of persons without high school degrees, had adequate health literacy.

Table 1
Sample Characteristics by Education and by Race

Table 2 displays results from the fully adjusted regression model for the sample used in our analysis of educational differences in health status and vaccination rates. For the sake of brevity, we display results only for regression models that included health literacy. Compared with persons with adequate health literacy, persons with inadequate health literacy had significantly worse health outcomes and were significantly less likely to receive influenza vaccine. The coefficient on marginal health literacy was statistically significant in the model for the physical SF-12 score, but only marginally significant or not significant in the other models. Having a high school degree was positively and significantly associated with physical and mental SF-12 scores (P=.013 and .004) and the likelihood that self-reported health was good or better (P<.001). It was not significantly associated with receipt of influenza and pneumococcal vaccines (P=.117 and .206). Blacks were significantly less likely to report good or better health status compared with whites (P=.012) and were less likely to report receipt of influenza and pneumococcal vaccines (P<.001 and P<.001). Differences in SF-12 scores between blacks and whites were insignificant.

Table 2
Results from the Regression Models Used to Examine the Impact of Health Literacy on Differences by Education Level

Actual and regression-adjusted health status scores and vaccination rates are displayed in Tables 3 and 4. The first column in Table 3 shows the unadjusted means of each dependent variable by education level. For dichotomous independent variables (for e.g., receipt of influenza vaccine), the mean is a proportion. The second column displays regression-adjusted values (or, equivalently, “predicted values”) from a regression model that excluded controls for health literacy. The third column displays the regression-adjusted values from a regression model that included controls for health literacy.

Table 3
Impact of Controlling for Health Literacy on Differences in Health Status and Vaccination by Education

The difference in regression-adjusted physical health SF-12 scores between high school graduates and nongraduates from the model that excluded health literacy controls was 1.7. The difference from the model that included health literacy controls was 1.0. Thus, controlling for health literacy decreased the adjusted difference by 0.7 (95% confidence interval [CI]: 0.4 to 0.9) or 41% (≈0.7÷1.7). For the other dependent variables, controlling for health literacy decreased the estimated difference by approximately one-fourth, although the magnitude of the effect was small. For example, a model with health literacy controls predicted that the probability that a survey respondent with a high school degree received influenza vaccination was 0.799 (or about 80%) versus 0.762 for a respondent without a high school degree. The difference was 0.037 versus 0.027 for a model that omitted health literacy controls, a difference between models of 0.010 (95% CI: 0.001 to 0.020) or, equivalently, 1 percentage point.

Table 4 displays the effect of controlling for health literacy on differences between blacks and whites. The difference in mean physical health SF-12 scores between whites and blacks was 1.3. This difference was entirely explained by the observed respondent characteristics; regression-adjusted physical SF-12 scores from the full model, which included health literacy, were actually higher for blacks than for whites. The difference in regression-adjusted values from the regression model that excluded health literacy controls was minimal: 0.1. The difference between models was 0.60 (95% CI: 0.32 to 0.85).

Table 4
Impact of Controlling for Health Literacy on Differences in Health Status and Vaccination by Race

The difference between models examining the likelihood that self-reported health status was good or better was 0.02 (95% CI: 0.01 to 0.03), or 2 percentage points. The models for receipt of influenza and pneumococcal vaccinations indicated that black enrollees were substantially less likely to receive vaccinations than whites. For example, the probability that a black enrollee received influenza vaccine at any point during his or her life, adjusted for observable covariates (including health literacy), was 0.747 versus 0.820 for whites, a difference of 0.074. However, health literacy had only a small and nonsignificant effect on measured differences: 0.009 (95% CI: −0.001 to 0.20) and 0.003 (−0.007 to 0.013), respectively.

DISCUSSION

We found that health literacy explained a small fraction of the differences in health status and, to a lesser degree, receipt of vaccinations that would normally be attributed to educational attainment or race if literacy was not considered. Controlling for health literacy reduced adjusted differences by educational attainment in physical and mental health SF-12 scores by 25% to 41%, respectively. While there is no universally agreed upon standard for what constitutes a “clinically meaningful” difference, these differences fall below the thresholds for clinical significance commonly cited in the quality-of-life literature.9,10

The (unadjusted) difference that we observed in self-reported health status between high school graduates and enrollees without a high school degree was about the same as in the general population, but the difference by race was over twice as large.11 Our results indicate that if health literacy levels were similar, differences in self-reported health status by education and by race would be about 20% to 25% lower.

Observed differences in receipt of influenza and pneumococcal vaccinations were generally in line with those reported elsewhere.12,13 In contrast to the findings for self-reported health status, differences in health literacy did not appear to explain much if any of the differences in receipt of vaccinations. Our results should be interpreted cautiously as these data were drawn from a survey of managed care enrollees. Managed care plans encourage beneficiaries to use preventive services, and, in doing so, may diminish the differences in preventive service use by education and race attributable to differences in health literacy. Indeed, use of cancer screening tests was uniformly high among survey respondents (data not shown). Previous research has found that managed care differentially increases use of preventive care among beneficiaries with less than 12 years of education.14 Findings with respect to the impact of managed care on differences by race are mixed.1519

Although these data did not permit us to investigate possible mechanisms by which health literacy influences disparities, past research on the relationship between education and health provides indirect evidence. Goldman and Smith20 found that well-educated patients are better able to manage complicated self-care regimens in HIV/AIDS and diabetes. Other studies have found that education is linked to faster adoption of new medical technologies21 (although not all studies have found this result22) and that consumer knowledge is linked to increased use of preventive care.23 Of course, it is not altogether surprising that controlling for health literacy reduces observed differences by educational attainment, literacy being a more direct measure of ability than years of schooling.

Our study has a number of limitations. First, the response rate to the survey was less than 60%. An analysis of nonresponders' ZIP codes, which were obtained from Prudential Healthcare's enrollment file, suggests that high school graduates and whites were overrepresented among nonresponders.4 Second, the Short Test of Functional Health Literacy in Adults does not fully capture all dimensions of the concept of health literacy (for e.g., oral literacy). Third, health literacy measurements may also be correlated with unobserved variables, such as occupation and social class that, if included in regression models, might account for the effects attributed to literacy in our study. Fourth, external validity is limited by the fact that our study sample was comprised entirely of elderly Medicare managed care enrollees in the South and Midwest.

Practically, this research indicates that programs to improve health literacy have the potential to reduce health disparities, but probably only by a small to moderate amount. That said, interventions to improve health system access among persons with low health literacy are probably inexpensive compared with larger, structural changes to the health system, and thus ought to be considered as part of an overall strategy to reduce disparities.

Acknowledgments

This work was supported by a grant from the Healthcare Georgia Foundation.

REFERENCES

1. U.S. Department of Health and Human Services. 2. Washington, DC: U.S. Government Printing Office; 2000. Healthy People 2010: Understanding and Improving Health.
2. DeWalt DA, Berkman ND, Sheridan SL, Lohr KN, Pignone M. Literacy and health outcomes: a systematic review of the literature. J Gen Intern Med. 2004;19:1228–39. [PMC free article] [PubMed]
3. Bennett CL, Ferreira MR, Davis TC, et al. Relation between literacy, race, and stage of presentation among low-income patients with prostate cancer. J Clin Oncol. 1998;16:3101–4. [PubMed]
4. Gazmararian JA, Baker DW, Williams MV, et al. Health literacy among Medicare enrollees in a managed care organization. JAMA. 1999;281:545–51. [PubMed]
5. Parker RM, Baker DW, Williams MV, Nurss JR. The test of functional health literacy in adults a new instrument for measuring patients' literacy skills. J Gen Intern Med. 1995;10:537–45. [PubMed]
6. Baker DW, Williams MV, Parker RM, Gazmararian JA, Nurss J. Development of a brief test to measure functional health literacy. Patient Educ Couns. 1999;38:33–42. [PubMed]
7. King G, Tomz M, Wittenberg J. Making the most of statistical analyses improving interpretation and presentation. Am J Political Sci. 2000;44:341–55.
8. Efron B. Bootstrap methods another look at the jackknife. Ann Stat. 1979;7:1–26.
9. Hays RD, Wolley JM. The concept of clinically meaningful difference in health-related quality of life research how meaningful is it? PharmacoEconomics. 2000;18:419–23. [PubMed]
10. Samsa G, Edelman D, Rothman ML, Williams GR, Lipscomb J, Matchar D. Determining clinically important differences in health status measures a general approach with illustration to the Health Utilities Index Mark II. PharmacoEconomics. 1999;15:141–55. [PubMed]
11. Lucas JW, Schiller JS, Benson V. Summary health statistics for U.S. adults. 2004;218:1–134. 2001:Vital & Health Statistics—Series 10: Data from the National Health Survey. [PubMed]
12. Centers for Disease Control and Prevention (CDC). Racial/ethnic disparities in influenza and pneumococcal vaccination levels among persons aged > or =65 years—United States, 1989–2001. MMWR. 2003;52:958–62. [PubMed]
13. Sambamoorthi U, Findley PA. Who are the elderly who never receive influenza immunization? Prev Med. 2005;40:469–78. [PubMed]
14. Fiscella K, Franks P, Doescher MP, Saver BG. Do HMOs affect educational disparities in health care? Ann Fam Med. 2003;1:90–6. [PMC free article] [PubMed]
15. Morales LS, Rogowski J, Freedman VA, Wickstrom SL, Adams JL, Escarce JJ. Use of preventive services by men enrolled in Medicare+Choice plans. Am J Public Health. 2004;94:796–802. [PMC free article] [PubMed]
16. Haas JS, Phillips KA, Sonneborn D, McCulloch CE, Liang SY. Effect of managed care insurance on the use of preventive care for specific ethnic groups in the United States. Med Care. 2002;40:743–51. [PubMed]
17. Tai-Seale M, Freund D, LoSasso A. Racial disparities in service use among Medicaid beneficiaries after mandatory enrollment in managed care a difference-in-differences approach. Inquiry. 2001;38:49–59. [PubMed]
18. Fiscella K, Franks P. Is patient HMO insurance or physician HMO participation related to racial disparities in primary care? Am J Manag Care. 2005;11:397–402. [PubMed]
19. DeLaet DE, Shea S, Carrasquillo O. Receipt of preventive services among privately insured minorities in managed care versus fee-for-service insurance plans. J Gen Intern Med. 2002;17:451–7. [PMC free article] [PubMed]
20. Goldman DP, Smith JP. Can patient self-management help explain the SES health gradient? Proc Natl Acad Sci USA. 2002;99:10929–34. [PMC free article] [PubMed]
21. Lichtenberg FR, Lleras-Muney A. The effect of education on medical technology adoption: are the more educated more likely to use new drugs? 2002. National Bureau of Economic Research Working Paper No. 9185.
22. Goldman D, Smith J. Socioeconomic differences in the adoption of new medical technologies. Am Econ Rev. 2005;95:234–7.
23. Parente ST, Salkever S, DaVanzo J. The role of consumer knowledge of insurance benefits in the demand for preventive health care among the elderly. Health Econ. 2005;14:25–38. [PubMed]

Articles from Journal of General Internal Medicine are provided here courtesy of Society of General Internal Medicine
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

  • Cited in Books
    Cited in Books
    PubMed Central articles cited in books
  • MedGen
    MedGen
    Related information in MedGen
  • PubMed
    PubMed
    PubMed citations for these articles

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...