1. What are 3 different ways to account for SES in an analytic model when investigating racial/ethnic health disparities? (describe a fourth for extra bonus points!). Briefly discuss the interpretations/implications of each approach as it relates to the interest in understand health disparities by race/ethnicity.
We can account for SES in three types of analytic models: mediation analysis, generalized estimating equations, and principal component analysis.
#1: Mediation Analysis: As used by the Lorch paper, mediation analysis is useful when the predictor (race/ethnicity) is associated with an outcome (fetal mortality), but there are mediating factors (including SES) that are associated with both the predictor and the outcome. Moreover, the presence of those mediating factors in the analytic model changes the relationship between the predictor and outcome variables, verified by sequential logistic regression models.
#2: Generalized Estimating Equations: Generalized estimating equations is an analytic model that clusters data points to prioritize the population-averaged effects and is a good alternative to generalized linear models when the outcome measure is discrete (rather than continuous) and the covariance structure is unknown. The Headen article used GEE to estimate risk ratios for race (main predictor variable) and gestational weight gain (discrete outcome variable) that accounted for clustering of data points (pregnancies) and then to adjust those RRs for relevant covariates (including SES measures).
#3: Principal Component Analysis: The Coley article uses principal component analysis to create a “neighborhood risk” index for her retrospective cross-sectional study to test associations between race and low birth weight. Principal component analysis allows researchers to convert multiple variables (such as median household income and unemployment rate) into linearly uncorrelated variables called principal components, the first of which displays the most variance. In this way, multidimensional data sets can have a linear expression and variation among multiple variables is emphasized.
2. Describe a potential effect modifier, mediator, or contextual variable (for definition of contextual variable, see first page of option Merlo reading) for an association of interest to you and relevant to health disparities. For example, for investigating the association between SES and maternal mortality, I might be interested in the contextual variable of exposure to violence in the neighborhood. Describe how you would study whether this relationship exists.
Modifier: When investigating the relationship between country World Bank income classification (high, upper middle, lower middle, low) and ischemic stroke mortality, I might be interested in health expenditure per capita. I could use mediation analysis to account for the interactions between the predictor and outcome, the predictor and covariate, and the covariate and outcome, and ultimately observe the impact considering the covariate on the relationship between the predictor and outcome.