Improving the efficiency and robustness of statistical inference for patient-oriented treatment effect in real-world clinical trials

The randomized clinical trial (RCT) is the preferred study design for assessing causal effects of medical interventions. A patient and their treatment decision makers are often interested in intervention efficacy that informs what to expect when the patient actually complies with treatment.

In many real-world RCTs, however, the patient-oriented intervention effect is often challenging to evaluate because of limited sample size, a small number of compliers due to low compliance rate and small to moderate effect size on outcome measures, which can significantly reduce the power to detect intervention efficacy.

Furthermore, in many RCTs, especially when evaluating multifaceted interventions for chronic diseases, such as arthritis, the endpoints often involve multiple outcomes to measure a complex trait. This raises the challenge of how to optimally pool treatment efficacy estimation across outcome measures. The “complier-average causal effect” (CACE) approaches have become popular in informing such patient-oriented treatment effects.