Minimizing Bias Due to Confounding by Indication in Comparative Effectiveness Research

Abstract

Health care reform has invigorated the genre of comparative effectiveness. With new congressional funding to identify which clinical strategies may work best, this research is poised to inform clinical care, health care policy, and the funding of health care. Although comparative effectiveness studies also include randomized clinical trials and systematic reviews, the expanded availability of large administrative databases and electronic medical records has provided new opportunities to conduct observational studies without the traditional burden of actually having to collect data with purpose. Observational studies of the health outcomes associated with clinical strategies—drug therapies, surgical procedures, and chronic disease management—generally take 1 of 2 forms, comparative effectiveness or comparative safety, which pose different challenges.¹