Quasi-random reflections on randomized controlled trials and comparative effectiveness research

Abstract

Comparative effectiveness research (CER) is still an evolving framework for which much needs to be done to improve the ability of randomized controlled trials (RCTs) to supply the necessary evidence. Perhaps, most important is to start with a clearly specified decision and decision maker in mind when the RCTs are designed. Second is to initiate RCTs with clinically relevant outcomes and comparators earlier in the evaluation process. Third is to specify and measure factors that might modify the intervention's effect, subject to logistical constraints of complexity and cost, so the trial is maximally informative, about how and to whom the intervention should be administered. It will be necessary to borrow observational methodologies and approaches to extract meaningful causal and subgroup inferences from such trials. Process variables should be seen as potentially part of that framework of effect-modifying factors, perhaps amenable to embedded experimental assessment with a trial. Perhaps most importantly, we need to improve the nationwide CER infrastructure to allow for rapid initiation and accrual for CER trials to reduce the trade-off that often exists between the speed of evidence development and its quality.