Let be the treatment, be the covariates, and be the outcome, with the potential outcome. We then have the propensity model and the outcome model within

Unlike in other fields, I think it is more likely that one can satisfy no unmeasured confounders in medicine, since there is a provider recommending treatments. One can ask the provider what information they take into account. I therefore might tend toward a method like inverse-probability weighting (see this post), in this case. In other areas of study, it’s much harder to specify a propensity, since nature assigns the treatment /exposure.

Interesting discussion on Datamethods.