Abstract ED01-02: Measurement of socioeconomic status in health disparities research

Socioeconomic status (SES) is often implicated as a contributor to disparities in health among Americans. Populations that experience the greatest disparities in health also experience the greatest socioeconomic disparities including lower education levels and incomes, higher unemployment rates and more frequent employment in lower SES occupations. In fact, more than 32% of African Americans, 28.4% of Hispanics and 19.4% of individuals from other race/ethnic groups lived in poverty in 2006–2007 compared to 11.5% of Non-Hispanic whites (1). Unemployment rates for the third quarter of 2008 were 10.6% for African Americans, 7.7% for Hispanics and 5.3% for Non-Hispanic whites (2).

Occupation is a frequently used measure of SES because of its role in defining how individuals are positioned within society which determines access to resources, exposure to psychological risks and physical hazards and through the influence of the occupation on lifestyle (3). Education has been called the most basic component of SES because its influence on future occupational opportunities and earning potential (4). Income represents of the flow of economic resources over a period of time. Persons with higher incomes are more likely to have the means to pay for health care and to afford better nutrition.

Traditional measures of SES (i.e. education, occupation and income), however; have a number of limitations that affect their usefulness when examining their association with health outcomes in comparative studies of race/ethnicity, gender and age. First, they may be difficult to collect from populations known to experience health disparities (i.e. particularly income), often do not correlate well with each other and do not capture the effect of SES over the life time due to the dynamic nature of SES. Furthermore, government statistics consistently show that education achievement level and occupational categories do not provide the same financial return for racial/ethnic minorities and women. Data from the Bureau of Labor Statistics for the third quarter of 2008 show that the usual weekly earnings for women in managerial positions is nearly 35% less than the salaries of men in managerial positions (5). The average monthly income among similarly educated individuals is lower for African Americans and Hispanics when compared to non-Hispanics whites at almost all educational levels.

More recently, researchers have increased use of community contextual measures of SES to examine disparities in health. This approach acknowledges that the SES context of the community in which an individual develops and resides may be more than or at least as important as the more traditional SES measures in understanding the unique contribution of SES to health disparities among the US population. The community is most often defined by geographic boundaries such as county, census tract or Zip Code, however; other definitions such as neighborhood have been used. A major criticism of these studies has been the failure to identify the specific pathways through which the community is hypothesized to affect the particular outcome under study and the general focus on only the negative contributions of the community to health.

A potential and likely more useful benefit of studying the community SES context is the ability to capture information regarding the effects of living in a community with a particular SES level (6) such as the prevalence of community stressors such as high crime, physical hazards, exposure to environmental toxins; the availability of goods and services such as transportation, presence and number of health care facilities, availability of fresh fruits and vegetables and needed medications in neighborhood pharmacies as well as those that may be deleterious to good health such as availability of alcohol, tobacco and illicit drugs. It can also capture the effects of positive influences on health outcomes including community stressbuffering mechanisms such as the presence or availability of walking or bicycle paths or other recreational facilities, churches and other community organizations. This information is likely to be more useful in terms of designing effective interventions to improve health outcomes.

There are also problems with the way in which these data are analyzed and interpreted (7). Many studies comparing the association of SES and health outcomes are at least in theory, based on the assumption that having a particular SES level confers the same level of resources and access to goods and services necessary for good health across racial/ethnic and gender groups. Research studies generally show that individuals in lower SES levels have poorer health outcomes for most demographic groups. However, the specific influence of SES when modeled as a covariate in multivariate models often provides little information on the relative importance of the covariate for racial/ethnic minorities, women or other population subgroups. This is due to the fact that results of multivariate statistical analyses of aggregate data are often driven by the patterns of association of covariates to health outcomes for the majority group, most frequently this is Non-Hispanic whites. Therefore, without stratification (e.g. by race/ethnicity, gender, age) it is difficult to determine whether variables found to be statistically significant in the aggregate multivariate analysis correlate with covariates that are most strongly associated with the health outcome for specific subgroups of the study population.

Another frequent problem is in the interpretation of the results of multivariate models in which SES is a covariate. In some published reports researchers conclude that although racial/ethnic groups statistically differed in unadjusted analyses, there were no racial/ethnic differences in the study outcomes after adjustment for SES in multivariate models. While technically true, it gives the erroneous impression that the race/ethnic groups do not statistically differ in terms of the study outcome or that being in a low SES somehow warrants the poor outcome without consideration or explanation as to how SES might have impacted the outcome. A more accurate and meaningful interpretation would be that covariates found to be significantly associated with the outcome in multivariate analysis contribute to the differences seen between the race/ethnic groups in the unadjusted analyses.

Choosing the best variable or approach for measuring SES is dependent in part on its relevance to the population and outcomes under study. The current research challenge is to go beyond attributing well documented variations in socioeconomic status, as measured by income, education or occupation. More detailed, better specified and properly conceptualized studies of the manner in which SES influences health can inform social policy and program design to effectively intervene to reduce health disparities in a socially and economically diverse society.

This presentation describes approaches to the measurement of SES, strengths and limitations of specific approaches, methodological issues related to the study of the influence of SES on disparities in health, and strategies for the analysis and interpretation of data examining SES and health disparities.