Comparing Drug Effectiveness at Health Plans: THE ETHICS OF CLUSTER RANDOMIZED TRIALS

摘要

The methodology has been widely used to evaluate the impact of public health interventions, like community health education, and medical management strategies, like early treatment of sexually transmitted diseases to reduce HIV transmission in Africa.4 The size of the clusters that are randomized typically ranges from single practices to entire communities.5 Although cluster randomization has been applied to the study of prescription drugs-as in comparing two different forms of potassium supplements by randomizing pharmacies to dispense one or another formulation-the methodology has not been widely used for this purpose.6 To our knowledge, the reasons for this have not been addressed systematically. In principle, CRTs could provide information about comparative effectiveness for two commonly used agents-drug A and drug B-if one cluster (which might consist of a practice, a delivery system, a health plan, a payor, a purchaser, or other entity that affects care for a group of patients) gave preference to A, while another cluster gave preference to B. CRTs could be especially advantageous if implemented in the context of health plans that have extensive information on members, treatments, and clinical outcomes, together with an existing research infrastructure.