1. I am currently working on two different projects that attempt to use EMR data to understand patterns in the quality of care and identify possible systems-level improvements, specifically in colorectal cancer screening and the treatment of hypertension. As a result, I am interested in the superficial indicators of SES that we have available in the EMR and the balance between understanding what these measures are telling us, and ‘controlling for SES’ so that we can try to isolate system-level factors and try not to ‘penalize’ systems or clinics that treat a disproportionate number of patients with low SES.
As an example, I am using a data-set of Kaiser members to examine the follow-up of positive fecal immunochemical tests for colorectal cancer screening and stratifying follow-up rates by Kaiser regions in California. My data set includes race/ethnicity, age and gender as structural stratifiers; health status (using the Charlson Index), body mass index and smoking status are the only potential intermediary factors recorded. Each of these factors could contribute to why a patient may or may not follow-up with a positive screening test, but give nowhere near a complete picture of someone’s ‘socio-economic position’. I must therefore try to follow the advice of Braveman et al and assume that differences seen are likely to reflect unmeasured factors and interpret my results “thoughtfully in the context of plausible explanatory pathways”.
2. Someone’s past experiences with the healthcare system could affect their probability of having a follow-up colonoscopy after a positive FIT in many different ways. Race/ethnicity could be correlated with not only access to specialist care (even among Kaiser members), but also trust in the healthcare system and understanding the role of colonoscopy to potentially prevent cancer (and not simply confirm a diagnosis). Further, female gender and non-white race have been shown to be associated with increased fear of embarrassment and discomfort with colonoscopy in several qualitative studies. These findings are likely shaped by not only past experiences, but also fears transmitted by the other people in your social network or neighborhood.