Forecasting COVID-19 Inpatient Mortality using Fundamental Parameters in Resource-Constrained Settings: a Countrywide Multi-Center Cohort Study

Abstract

During the COVID-19 pandemic, resource constraints necessitated effective mortality prediction tools to guide decision-making. Tailoring these tools to diverse healthcare settings, particularly those with sparse resources, remains an unmet need. Addressing this challenge, our nationwide multicenter study from Syria introduces LR-COMPAK, a simplified scoring system utilizing six easily obtainable variables: age, comorbidities (kidney disease, malignancy), and vital signs (pulse rate, oxygen saturation, consciousness) to predict COVID-19 mortality during hospitalization. LR-COMPAK exhibited superior performance compared to established scores (AUC 0.88), explaining 52% of mortality variability in our sample (n = 3199), and demonstrated applicability extending to non-hospitalized patients. Regional and temporal disparities in severity scores and mortality rates underscored healthcare capacity variations. Furthermore, incorporating two blood tests (lactate dehydrogenase and bicarbonate), LR-ALBO-ICU, a modified ICU-specific score, effectively predicted ICU mortality. The practical implications of LR-COMPAK and LR-ALBO-ICU include aiding informed hospitalization decisions, optimizing resource allocation in resource-limited settings, and enhancing patient outcomes globally.