Development and external validation of models for predicting mortality, sepsis and septic shock in COVID-19 patients

Objective. To develop and externally validate the models for predicting in-hospital mortality and septic complications in patients with COVID-19. Material and methods. This study was conducted in accordance with the TRIPOD guidelines. A retrospective derivation cohort included patients hospitalized between April 2020 and December 2021. External validation was performed in a cohort of patients hospitalized during the period of Omicron variant predominance (January—February 2022). Results. The final model included age, SpO2, respiratory rate, and Glasgow Coma Scale score. The derivation cohort included 488 patients, while the validation cohort comprised 382 patients. During external validation, AUROC for in-hospital mortality was 0.798 (95% CI 0.752—0.845), while AUROC for sepsis and septic shock were 0.680 and 0.704, respectively. Calibration was good for mortality and less pronounced for other outcomes. Conclusion. Original model demonstrates stable discriminative performance and acceptable calibration for predicting in-hospital mortality in patients with COVID-19.