When faced with logistical issues that preclude doing patient-level randomized trials, researchers should be looking to use at least cluster randomization as the next best method, rather than before-and-after observational studies. Selvaraj and Prasad attempt to provide empirical evidence of the superior validity of cluster randomized controlled trials (RCTs) by showing that such trials have comparable effect sizes to patient-level RCTs in published studies in select high-impact journals. One might quibble over whether this proves its comparable validity without direct comparison of methods on the same research question. However, we believe this work raises the importance of the underused method of cluster randomization in clinical research. Although cluster randomization would never be preferred to patient-level randomization, when this is not feasible researchers should be looking to cluster randomization as the next best design, rather than before-and-after observational studies.
O’Malley PG. In Support of More Clustered Randomized Trials. JAMA Intern Med. 2013;173(4):315. doi:10.1001/jamainternmed.2013.3665