Robust COVID-19 Mortality Risk Assessment: Validation of a Two-Step Algorithm From the National COVID Cohort Collaborative

• Published in: The Journal of infectious diseases
• Main Authors: Li, Bingnan, Ke, Yuan, Chen, Xianyan, Martinez, Leonardo, Shen, Ye
• Format: Journal Article
• Language: English
• Published: United States 26.07.2025
• Subjects: COVID-19; Model Validation; Risk Stratification; Mortality Risk Prediction; Missing Data Imputation; N3C
• ISSN: 0022-1899; 1537-6613; 1537-6613
• DOI: 10.1093/infdis/jiaf393

This study introduces and validates a 2-step algorithm for assessing coronavirus disease 2019 (COVID-19) mortality risk, leveraging data from over 7 million COVID-19 cases in the National COVID Cohort Collaborative (N3C). The original algorithm stratifies patients into risk categories based on routine clinical metrics and was initially tested across diverse cohorts from multiple institutions, demonstrating strong predictive performance. Further validation of this algorithm on 2.4 million valid N3C COVID-19 records, including a subset of 768 957 with complete data, yielded a C-statistic exceeding 0.85. The algorithm adapts effectively to evolving mortality trends, particularly during the Omicron variant surge. Comparative analyses of full and imputed datasets underscore the algorithm's robustness across varied clinical settings. Our work offers a scalable tool for pandemic management, highlighting the critical role of data-informed approaches in public health.