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.
Mediation Analysis is one way to account for SES in an analytic model when investigating racial/ethnic health disparities. A mediator is a variable that falls in the pathway of a particular association between a predictor/exposure and outcome. In this case, when understanding health disparities by race/ethnicity—we could account for SES as a potential mediator that is along the causal pathway between race/ethnicity and a particular outcome of interest. Said differently, the model would illustrate whether or not race/ethnicity influences SES, then SES influences the outcome of interest. Such model helps clarify or expand on the nature of the relationship between the predictor/exposure and outcome.
Generalized Estimating Equations (GEE) is another way to account for SES in an analytic model when investigating racial/ethnic health disparities. GEEs essentially extend the generalized linear model to conduct the analysis of repeated measurements or correlated observations. For example, it could be beneficial when dealing with data sets arising from clustering, in which measurements are taken on participants/subjects who share a common characteristic, such as SES.
Effect Measure Modification (EMM) [or Effect Modification] is a third way to account for SES in an analytic model when investigating racial/ethnic health disparities. EMM is when the measure of association used in a study (e.g., risk ratios) alters over values of some other variable. In contrast with confounding, EMM is associated with the outcome but not the exposure, and it is all about stratification, thus we can stratify by SES for racial/ethnic health disparities research to assess for a statistical interaction. EMM helps identify vulnerable populations (e.g., low SES).
Multilevel Modeling is a fourth way to account for SES in an analytic model when investigating racial/ethnic health disparities. It is helpful when analyzing data with repeated measurements or data organized in nested levels. This statistical technique can be used, as shown in the Coley article, to investigate variations in racial/ethnic differences in an outcome of interest across varying levels of for example SES.
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.
Mediator: I am interested in studying the association between race/ethnicity and time between referral to appropriate mental health services and seeing a specialist for evaluation among children and adolescents with behavioral/developmental concerns. I would want to assess for whether neighborhood is a potential mediator between the predictor and outcome. Said differently, whether race/ethnicity influences where individuals reside and whether residing in a particular neighborhood influences time between referral and evaluation. This study could be assessed via a cross-sectional study where I could obtain data for both predictor and outcome, and then stratify on neighborhood to assess for a statistical interaction—whether neighborhood is a mediating variable.