From cutoff to causality: regression discontinuity design for evaluating acute kidney injury e-alerts

In 2014, acute kidney injury electronic alerts were implemented across Wales based on a predefined increase in creatinine levels. In their recent paper, Xie et al. used a regression discontinuity design to evaluate the effectiveness of these electronic alerts. Threshold-based eligibility mimics randomization, and when all assumptions are met, regression discontinuity design can estimate causal effects from observational data. Regression discontinuity design can play an important role in evaluating health and medical interventions, particularly where trials are not feasible or do not reflect real-world clinical practice.