1 What are the different ways to account for SES in an analytic model when investigating racial/ethnic health disparities? (Hint: you should have three options). Discuss the interpretations/implications of each approach as it relates to the interest in understand health disparities by race/ethnicity. Draw a DAG for each option and reference it in your response (you do not have to post this!).
When creating an analytic model for investigating racial/ethnic health disparities SES can act as a confounder, a mediator, or clustering.
In the Lorch et al paper about fetal death, SES acts as a mediator for Hispanic women. This means that the effect of Hispanic race/ethnicity on fetal death is mediated through SES. This means that SES must be controlled for in the analysis in order to estimate the direct effect of race/ethnicity on fetal death through SES. If you do not control for SES, you are only estimating the indirect effect of race/ethnicity on fetal death. You have to decide or determine whether the relationship between race/ethnicity and your outcome is via the mediator.
In the Headen et al paper, SES is described as a confounder. Since, SES is highly associated with race/ethnicity, and likely is a risk factor for pregnancy weight gain, the authors control for it in the analysis as a confounder. This differs from mediation because in this model SES impacts the relationship between race/ethnicity and pregnancy weight gain, but it is not on the direct pathway.
In Merlo et al, clustering using a multi-level epidemiology approach is used.
2 Think about multilevel influences on a health outcome of interest to you. Discuss how you would study this, including measurement and analytic approaches you would use to account for exposures across multiple levels.
One topic of interest tat multi-level analysis could be applied is infant mortality, where disparities in rates between Blacks and Whites are significant, and vary by location. The Lorch article looked at fetal death, which is similar, but not quite the same. The differences in rate are not explained by clinical factors alone. We know Black women have many risk factors for infant mortality, but the complexity in how they interact is poorly understood. MLRA could be used to understand the variances in rates in infant morality in different regions, accounting for race, SES, other sociodemographic factors, as well as regional differences in access to care, care style and clinical practices, availability of reliable infant care products, childcare, as well as differences in racial experience, provider attitudes, etc. Other policy level factors could be examined as well. This would allow for interventions designed to target areas of the most variance between black as and whites to be identified.