Development and diagnostic performance of a novel scoring system for predicting COVID-19 severity in a national referral hospital

Abstract

Coronavirus disease 2019 (COVID-19) shows a wide spectrum of clinical manifestations, ranging from mild illness to severe and critical disease. Early identification of patients at risk of deterioration remains important for timely clinical decision-making. The aim of this study was to develop a novel scoring system for predicting COVID-19 severity in a national referral hospital. This prospective cohort study included patients with confirmed COVID-19 admitted to Dr. M. Djamil Hospital, Padang, Indonesia, from May to December 2021. Demographic, clinical, and laboratory data were collected, and disease severity was classified according to the World Health Organization criteria. Univariate analysis was performed to identify candidate variables, followed by multivariable logistic regression analysis to determine independent predictors of severity. A scoring system was then constructed based on the retained predictors, and its diagnostic performance was evaluated using receiver operating characteristic analysis and the Youden index. Five variables were independently associated with COVID-19 severity: white blood cell count (WBC) ≥10,000/mm³, absolute lymphocyte count (ALC) <1,500/mm³, platelet large cell ratio (PLCR) ≥30%, interleukin-6 (IL-6) ≥7 pg/mL, and D-dimer ≥500 ng/mL. These variables were incorporated into a scoring system with a maximum total score of 6. Using a cut-off score of 3.5, the model showed a sensitivity of 96.15% and a specificity of 79.03% for differentiating severe from non-severe COVID-19. This study highlights that a novel scoring system based on WBC, ALC, PLCR, IL-6, and D-dimer performed well in predicting COVID-19 severity and may support early risk stratification in hospitalized patients.