Controlling for Differential Regression-To-The-Mean via Propensity Scores: A Simulation Study

Regression-to-the-mean is a well-known source of bias in before-after treatment comparisons, but whether propensity score methods can effectively control for differential RTM when groups differ in baseline risk had not been rigorously evaluated through simulation.

Target RWE researchers conducted a simulation study examining the performance of propensity score methods in controlling for differential regression-to-the-mean, characterizing scenarios where standard PS adjustment succeeds or fails to remove this form of bias.

Establishing the conditions under which propensity scores can and cannot control for RTM bias directly informs the design of safer comparative analyses at Target RWE, reducing the risk of biased estimates in studies comparing treatment groups with different baseline risk profiles.

Propensity score bias control simulation for regression-to-the-mean in comparative effectiveness design, Methodological evidence for RWE study design best practices in pharma partner programs