Hi Folks! I’m an assistant professor at Columbia DBMI, and I’m working with some others on a grant to understand site-specific data differences that impact analyses – e.g., what makes an analysis that is valid at one site invalid at another. Specifically, we want to extend beyond a static list of data validity and quality checks, and instead do a general discovery effort of differences between a given site and a reference site, framed through the lens of “Over a broad family of clinically meaningful questions, how would the local EHR dataset differ in its distribution of answers relative to a reference EHR dataset?”
If you’d be interested in possibly collaborating on this topic – concretely through the realm of participating in a possible multi-site study where we would assess how different OHDSI datasets differ from one another through this method and the extent to which such differences can be surfaced to analysts proactively during analytic workflows – please don’t hesitate to reach out and I can share more details!
2 Likes