All Times EDT
Keywords: Real World Evidence, Causal Estimation, Propensity Score Matching
Adequate and well-controlled studies, which are usually interpreted as randomized controlled trials, are required for the demonstration of substantial evidence of effectiveness of an investigational medicine. With the passage of the 21st Centuries Cures Act2, the FDA is mandated to expand the role of real-world evidence (RWE) in support of medicine approval. This mandate further broadens the scope of scientific discretion in interpreting the substantial evidence requirement for the approval of new medicines to include data collected outside clinical trials. One way to reduce the sample size is to leverage information on the control. For this purpose, propensity score matching methods based on pre-treatment characteristics are generally employed to minimize selection bias. Because there are essentially three arms – investigational treatment, concurrent control and external control, the propensity score matching requires modification and operationalized heuristically. This talk talks about different heuristic approaches to modify propensity score matching and implications on causal estimation.