Development and validation of a dynamic 48-hour in-hospital mortality risk stratification for COVID-19 in a UK teaching hospital: a retrospective cohort study

Abstract

AbstractObjectivesTo develop a disease stratification model for COVID-19 that updates according to changes in a patient’s condition while in hospital to facilitate patient management and resource allocation.DesignIn this retrospective cohort study we adopted a landmarking approach to dynamic prediction of all cause in-hospital mortality over the next 48 hours. We accounted for informative predictor missingness, and selected predictors using penalised regression.SettingAll data used in this study was obtained from a single UK teaching hospital.ParticipantsWe developed the model using 473 consecutive patients with COVID-19 presenting to a UK hospital between March 1 and September 12, 2020; and temporally validated using data on 1119 patients presenting between September 13, 2020 and March 17, 2021.Primary and secondary OutcomesThe primary outcome is all-cause in-hospital mortality within 48 hours of the prediction time. We accounted for the competing risks of discharge from hospital alive and transfer to a tertiary Intensive Care Unit for extracorporeal membrane oxygenation.ResultsOur final model includes age, Clinical Frailty Scale score, heart rate, respiratory rate, SpO2/FiO2 ratio, white cell count, presence of acidosis (pH < 7.35) and Interleukin-6. Internal validation achieved an AUROC of 0.90 (95% CI 0.87–0.93) and temporal validation gave an AUROC of 0.86 (95% CI 0.83-0.88).ConclusionOur model incorporates both static risk factors (e.g. age) and evolving clinical and laboratory data, to provide a dynamic risk prediction model that adapts to both sudden and gradual changes in an individual patient’s clinical condition. Upon successful external validation, the model has the potential to be a powerful clinical risk assessment tool.Trial RegistrationThe study is registered as “researchregistry5464” on the Research Registry (www.researchregistry.com).Article Summary-Our dynamic prediction model is able to incorporate patient data as it accumulates throughout a hospital visit.-We use the established statistical landmarking approach to dynamic prediction; account for competing risks for the primary outcome of in-hospital mortality; and the potentially-informative availability of clinical and laboratory data.-The sample size of the first wave of patients admitted with severe COVID-19 was relatively low, due to the lower incidence in Cambridgeshire, but increased significantly during the winter months of 2020/21, providing the opportunity to temporally validate the model.-As a single centre study, the presented model will require external validation to assess its performance in other cohorts; and also if there are significant changes in the characteristics of new variants or the management thereof.-Our work also highlights the adaptability of the statistical landmarking framework to be used to model individual patient outcomes using densely-collected hospital data.