SPARC scoring model

Abstract

Background: A big concern for physicians during the coronavirus disease 2019 (COVID-19) pandemic has been its uncertain prognosis and sudden progression into critical illness. No objective measures have yet been in use to predict complications and plan a management accordingly. Aims: The objective of this study was to establish a predictive model of disease outcome to facilitate early and effective decision-making. Materials and Methods: A cluster of (COVID-19) patient data was assessed retrospectively to build a binary logistic regression-based model that predicts mortality based on clinical, laboratory and radiological parameters. The model performance was characterised based on the receiver operating characteristic (ROC) curve. Results: During the study period, over 2000 patients infected with severe acute respiratory syndrome coronavirus 2 infection, of whom 200 were admitted to the intensive care unit. Eleven patients were omitted as per the entry criteria. Five predictors (determined within 24 h of admission) were identified as independent risk factors for worse outcomes in COVID-19-infected patients: serum lactate dehydrogenase, PaO2/FiO2, age, respiratory rate and computed tomography severity. These were then used as a cluster and a nomogram- COVID-19, critical care, infectious disease, outcome prediction, SPARC score (SPARC) was made. Intended to be used at or within 24 h of admission, it is intended for better triage and management. Higher score correlates with a better prognosis. The performance of the overall score was 80% as assessed by ROC. Conclusion: The risk score SPARC, based on five factors, aids in the early prediction of disease outcome and hence facilitates triage, individualised and efficacious treatment and improved results.