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. This is the approach used in Lorch et al to investigate racial and ethnic disparities in fetal death rates. It is based on the framework described by Baron and Kenny as outlined in the paper. Here we want to assess the extent to which an exposure is by SES. So, we define the total effect of exposure, effect explained by SES and the effect unexplained by SES. This approach can be susceptible to potential biases due to mediator-outcome confounding and exposure-mediator interaction.
Multilevel General Linear Model Analysis. Used by Coley el al to investigate neighborhood risk to explain low birth weight infants to adolescent mothers. In this approach, a series of hierarchical models are constructed to determine if a SES is associated with the outcome in the context of race/ethnicity and other confounding variable. Using this hierarchical approach can provide improved estimation and to examine cross-level effects – effect of variables at one level on variables at another level.
Generalized Estimating Equations. This method is used by Headen et al to examine racial/ethnic disparities in inadequate gestational weight gain in the context of pre-pregnancy weight. It is an extension of GLM model. Here we estimate the parameters where there are dependent SES variables that are correlated and there are independent variables. In the context of race/ethnicity disparity, this correlation is due to sharing of a common SES.
Stratification. In certain applications, one can simply use stratification. So, we stratify the data with and without a SES variable and perform regression analysis to determine the effect of the variable.
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.
Outcome: lung cancer
Exposure: cigarette smoking
Effect Modifier: bars in the neighborhood.
Smoking is a well known cause of lung cancer. The hypothesis is that the amount of smoking is influenced by the number of bars in the neighborhood a person resides. One can use the multilevel generalized linear model to analyze the data. More simply, one can dichotomize the number of bars into two groups and do a stratification analysis.