Abstract:
|
Randomized controlled trial is considered the gold standard for treatment comparisons. In circumstances, however, comparisons across studies or subgroups are needed to evaluate the benefits of new treatment compared to historical control treatments. This comparison can be crucial for better characterizing the clinical benefits to patients and help to accelerate drug development. Propensity score methods are widely used to eliminate potential confounding factors due to lack of randomization. In this paper, three propensity score methods including stratification, inverse probability weighting (IPW) and matching are assessed and compared. An application on hematology studies will be presented.
|