M Alan BrookhartNuvan RathnayakaUniversity of North Carolina at Chapel HillAbstract/ManuscriptKayla HendricksonCareYesKathleen E HurwitzR&D
Propensity Score Methods for Confounding Control in Observational Studies of Therapeutics for COVID-19 Infection.
Propensity Score Methods for Confounding Control in Observational Studies of Therapeutics for COVID-19 Infection
Challenge
Standard propensity score methods estimate treatment effects in populations defined by overlap between treatment groups, restricting inference to a subpopulation that may not reflect the full clinically relevant population—a critical limitation in COVID-19 observational research.
Solution
Target RWE researchers published a methodological analysis demonstrating how IPTW and other propensity score variants differ in the population to which they generalize, using COVID-19 therapeutic observational studies as illustrative examples.