I'm trying to use propensity score matching (PSM) to find matched controls. The data is not cross-sectional and it has time dependent information (monthly, quarterly, etc.). Let us say I have information from t_0 to t_N for both control units and treated units, and the event (treatment) for the treated units can happen at any time between t_0 and t_N. Is there an easy way to use propensity score matching using R or SAS to find the matched controls for each of the treated units?

  • $\begingroup$ I don't know about easy, but if you search at Mayo Clinic website they have SAS macros for propensity score matching. $\endgroup$ – Reeza Feb 27 '17 at 23:30
  • $\begingroup$ I understand a typical propensity score matching, but I'm looking for one in the context where we have time-dependent covariates. $\endgroup$ – Ram Feb 27 '17 at 23:33

Not matching, but you might want to look up marginal structural models, which essentially use propensity score based weights to balance treated/untreated cases on time invariant and time varying confounders.

For reference you can see:

Robins, Hernan, Brumback (2000) Marginal structural models and causal inference in epidemiology, Epidemiology, 11 (5): 550-560


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.