1. It is often mentioned that racial/ethnic differences in health should not be investigated without consideration of socioeconomic position/status. Describe two ways to account for SES in an analytic model and the interpretations/implications of each approach.
Single variables of SES should not be used as the sole predictor of health outcomes due to the dynamic network of factors influencing SES.(1) Conclusions about the role of certain factors are found to vary depending on how SES is measured.(2, 3)
SES can be better approximated by using both education and income and, if possible, other dimensions of SES such as occupation and wealth, measured in a variety of ways over varying temporal and spatial scales. This implies that proxies of SES, means of measurement, time period of analysis, and levels of analysis are not equivalent. In this account, SES is a multidimensional concept, rather than a variable that can be represented by any single socioeconomic factor.(2)
Another means to account for SES is by neighborhood socioeconomic status (NSES). NSES is geared to better estimate the multidimensional characteristics of an individual’s SES by capturing more generalizable, neighborhood characteristics such as access to nutritious food, health care, transportation, well-maintained infrastructure, public utilities, and services. NSES indices are often created by compiling a number of neighborhood-level variables that are assumed to be indicators of SES. This method accepts the limitation that by hand-picking variables that are generalizable (oftentimes inappropriately) to a group of people, incorrect causalities and distributions can result. Another implication is that the level of data that is analyzed, often census-level-tract, is assumed to match the neighborhood of that area. However, neighborhoods are often loosely and diversely defined, and they change over time. Lastly, by NSES variables are often restricted temporally and do not necessarily capture the dynamism of SES of the neighborhood’s constituents. For example, access to affordable and nutritious food will evolve over time as a city grows, shrinks, gentrifies, as policy is implemented, etc.(4)
2. Select a research question investigating associations between multi-level social factors (operating at least two levels) and a health outcome. State the exposures and outcomes, the additional study covariates that would be included in the analytic model, and a discussion of the analytic considerations in an multi-level investigations.
Braveman et al investigate associations between socioeconomic factors (several measures each of income and education), self-reported race/ethnicity, and four indicators of maternal and infant health (low birthweight, delayed prenatal care, unintended pregnancy, and lack of breastfeeding intention). Additionally, the researchers included other variables such as maternal age and parity.
The researchers looked at correlations among the variety of SES measures collected (education and income at both the individual and household level). They then estimated the distribution of health indicators in focus by each of the SES measures. Additionally, they examined the associations between SES measure and health indicator by each self-reported racial/ethnic group. Lastly, they developed multivariate models with different measures of SES (both income and education) to look at the association between race/ethnicity and the health outcomes. The intention with these multivariate models was to control for the confounding effects of SES, age, and parity. By using these multivariate models, the researchers were able to compare correlations between and across different measures of income and education and how these SES dimensions may affect the magnitude, significance, and direction of their associations.
The researchers chose not to include maternal occupation as a factor of SES when they learned of the temporal limitations of occupation – about 50% of their sample indicated maternal occupation as “homemaker” at the time of the survey, despite that many of the subjects had different occupations prior to and after the survey period.
By performing a multivariate investigation, Braveman et al were able to demonstrate that findings on the role of SES factors on racial/ethnic health disparities depend on the SES measure utilized. This is instrumental to recognize and account for when designing research that will inform health policy.
1. Holmes JH, Lehman A, Hade E, Ferketich AK, Gehlert S, Rauscher GH, et al. Challenges for multilevel health disparities research in a transdisciplinary environment. Am J Prev Med. 2008;35(2 Suppl):S182-92. PMCID: 2580051.
2. Braveman P, Cubbin C, Marchi K, Egerter S, Chavez G. Measuring socioeconomic status/position in studies of racial/ethnic disparities: maternal and infant health. Public Health Rep. 2001;116(5):449-63.
3. Bilheimer LT, Klein RJ. Data and Measurement Issues in the Analysis of Health Disparities. Health Services Research. 2010;45(5p2):1489-507.
4. Dubowitz T, Heron M, Bird CE, Lurie N, Finch BK, Basurto-Dávila R, et al. Neighborhood socioeconomic status and fruit and vegetable intake among Whites, Blacks, and Mexican-Americans in the United States. Am J Clin Nutr. 2008;87(6):1883-91.