Regression discontinuity designs in healthcare research

Abstract

Clinical decisions are often driven by decision rules premised around specific thresholds. Specific laboratory measurements, dates, or policy eligibility criteria create cut-offs at which people become eligible for certain treatments or health services. The regression discontinuity design is a statistical approach that utilizes threshold based decision making to derive compelling causal estimates of different interventions. In this review, we argue that regression discontinuity is underutilized in healthcare research despite the ubiquity of threshold based decision making as well as the design's simplicity and transparency. Moreover, regression discontinuity provides evidence of "real world" therapeutic and policy effects, circumventing a major limitation of randomized controlled trials. We discuss the implementation, strengths, and weaknesses of regression discontinuity and review several examples from clinical medicine, public health, and health policy. We conclude by discussing the wide array of open research questions for which regression discontinuity stands to provide meaningful insights to clinicians and policymakers.