OHDSI Home | Forums | Wiki | Github

IPTW in CohortMethod

Hi! I wonder that can we conduct IPTW in CohortMethod instead ps matching? For instance, let’s say I want to make a characteristic table Before weighting and After weighting. Is this possible in here? Thank you! :slight_smile:

IPTWis fully supported in the upcoming CohortMethod 5.0.0 release. It is ready in the develop branch, I’m just waiting for the Shiny app to be finished. V5.0.0 introduces quite a lot of changes throughout CohortMethod, so may not be worth the effort.

In general I recommend against using IPTW. We’ve found it to produce biased estimates compared to PS matching or stratification, as for example demonstrated here. I know IPTW is the fashionable thing to do, but be sure to include plenty of negative controls to verify it doesn’t mislead you.

Oh I see. Well then I can’t wait to release CohortMethod 5.0.0!

Thank you for everything :slight_smile: