1. Give an example of a research question for investigating racial/ethnic health disparities where: [1] SES is a confounder; [2] SES is an effect modifier; [3] SES is a mediator. Briefly discuss the interpretations/implications of each approach as it relates to understanding health disparities by race/ethnicity.
I think that this question highlights the importance of designing your study and analysis plan for the question you are trying to address, as I can imagine for the same study data you could interpret SES to be a confounder, mediator, or effect modifier. I will be discussing a study I completing looking at use and outcomes of alternative access thyroidectomy. This is a rarely used method for thyroid surgery that employs robotic/endoscopic techniques to perform the surgery through incisions away from the anterior neck, with the primary goal of eliminating a scar on the anterior neck.
[1] SES as a confounder
Imagine a study investigating the rates of alternative access versus traditional open thyroidectomy in black vs white patients with thyroid cancer. If your question is whether there are different preferences for alternative access surgery between races, factors related to socioeconomic status (income, education) could be seen as confounders because they affect access to alternative access techniques and are associated with race. (I struggle a little here because to be a true confounder the variable must have a CAUSAL effect on both race and outcome, and I’m not sure that there are relevant variables that actually CAUSE race). In this scenario I would control for SES in my analysis. Additionally, patient reported data would be important in fully understanding this question.
[2] SES as an effect modifier
Here, we could imagine a scenario where income can be an effect modifier on the relationship between race and receiving an alternative access operation. It may be that income differentially affects access to or preference for alternative access techniques by race, ie low income blacks have relatively lower access to these techniques compared to low income whites. In this case it is important to investigate this relationship in your statistical analysis.
[3] SES as a mediator
Now, if the question is: is there a difference in the use of alternative access techniques for thyroid cancer surgery in blacks compared to whites? (any maybe even WHY is there a difference) it would be interesting to look at SES (income, education) as a mediator. In this case you wouldn’t want to simply control for SES and be done with it, but rather to first perform an analysis that does control for SES, and then one that does not. Both are important results in answering this question.
2. Describe a potential effect modifier, mediator, or contextual variable (for definition of contextual variable, see Diez-Roux reading) for an association of interest to you and relevant to health disparities. For example, for investigating the association between education and hypertension, I might be interested in evaluating whether the association between years of education and hypertension is different for Black men than for White men. Describe how you would study whether this relationship exists.
One of my current research interests is the differences in presentation, treatment, and outcomes in women compared to men with peripheral artery disease. Based on previous data that shows that both women and black patients have worse graft patency that whites following lower extremity bypass operations, I hypothesize that race could be an effect modifier on the relationship between sex and long term graft patency. To evaluate this I will use the vascular quality initiative database, which contains information on long term graft patency to first examine the relationship between sex and patency. Then, I will examine this relationship by race to see if there is effect modification present.