Details
The U.S. Food and Drug Administration (FDA) initiated this study within the Sentinel System to support its response to the coronavirus disease 2019 (COVID-19) pandemic. Early in the pandemic, the FDA recommended that the applicants categorize baseline disease severity in the clinical trial populations using a five-level severity framework. To facilitate regulatory decision-making and post-pandemic evaluations using real-world data, this study aimed to develop operational definitions of COVID-19 severity at diagnosis or at treatment initiation, in ambulatory settings, through code-based algorithms for each severity category based on the FDA’s conceptual guidance. These definitions were designed for use with administrative claims data, a key source of real-world evidence during the pandemic. At the time, no published studies had validated such algorithms for identifying COVID-19 cases or classifying severity within claims data.
The study found that the algorithm demonstrated low to moderate performance in classifying COVID-19 severity in ambulatory settings, with results varying depending on whether severity was assessed at the point of diagnosis or at treatment initiation. While limitations were noted, particularly among untreated populations, the algorithm may be suitable for select study designs, especially those focused on treated patients. This work underscores the Sentinel System’s role in developing new methodological tools and provides a foundation for future surveillance and monitoring leveraging administrative claims data to assess disease severity in emergent public health scenarios.