External validation of risk scores for the outcome of COVID-19 on a 2-year Swiss cohort – A retrospective study

<bold>Introduction:</bold> Several different scoring systems are available for COVID-19 risk stratification. This study aimed (1) to validate the performance of different scores for the prediction of COVID-19 outcome on a large, heterogeneous cohort and (2) to assess predictive power of a combination of their respective parameters. <bold>Methods:</bold> National Early Warning Score (NEWS), CURB-65 score, 4C mortality score (4C), Spanish Society of Infectious Diseases and Clinical Microbiology score (SEIMC), and COVID Intubation Risk Score (IRS) at time of admission were calculated for 1’056 patients consecutively hospitalized for COVID-19 at a Swiss hospital in 2020 and 2021. Predictive power for all-cause in-hospital death and intubation was compared by means of receiver operating characteristic curves and the respective area under the curve (AUC). A complementation of 4C with parameters from NEWS, CURB-65, SEIMC, and IRS (‘4Cplus’) underwent internal validation. <bold>Results:</bold> The 4C performed best in the prediction of in-hospital death (AUC 0.85), followed by SEIMC and CURB-65 (AUC 0.83 and 0.82). Intubation was best predicted with IRS and NEWS (AUC 0.77 and 0.75), while all other scores performed poorly (AUC ≤ 0.6). The new 4Cplus reached significantly higher AUC for both evaluated endpoints (in-hospital death: AUC 0.88, p = 0.015; intubation: AUC 0.79, p < 0.001). <bold>Conclusion:</bold> Predictive power of 4C, CURB-65 and SEIMC scores for in-hospital death in COVID-19 was good on this large, two-year cohort. IRS and NEWS performed acceptably in the prediction of intubation. The new, combined model improved predictive power for both endpoints, while its performance calls for external validation.