Evaluating the interaction between the therapy and the treatment in clinical trials by the propensity score weighting method.

In clinical trials for antihypertensive drugs, a combination therapy trial and a monotherapy trial are often conducted simultaneously. In this situation, it can be a clinical concern to know the difference of the safety or efficacy of the new drug between the two therapies, in other words, to investigate the interaction between the therapy (monotherapy or combination therapy) and the treatment (test or control). However, because patients are often registered in either of these trials on the basis of their background characteristics, specific patients may be selected to participate in the monotherapy trial or combination therapy trial and not chosen at random, whereas the treatment is assigned randomly in each trial after registration. If this fact is not considered, the statistical analysis of the interaction may be biased. In this paper, we aim to evaluate the interaction between the two aforementioned factors by adjusting for covariates that may affect registration in the two trials. For this purpose, we apply the propensity score weighting method to suit the problem. The propensity score in this case is decomposed into the usual propensity score for the registration and the assignment probability for the random treatment assignment on the basis of their two-stage structure. We also discuss the augmented estimator known as the doubly robust estimator. In addition, we apply this method to data of a clinical trial for an antihypertensive drug that was conducted in Japan and conduct a simulation study to evaluate the performance of our proposed method.