Early recognition of the severe illness is critical in coronavirus disease-19 (COVID-19) to provide best care and optimize the use of limited resources.

We aimed to determine the predictive properties of common community-acquired pneumonia (CAP) severity scores and COVID-19 specific indices.

In this retrospective cohort, COVID-19 patients hospitalized in a teaching hospital between 18 March-20 May 2020 were included. Demographic, clinical, and laboratory characteristics related to severity and mortality were measured and CURB-65, PSI, A-DROP, CALL, and COVID-GRAM scores were calculated as defined previously in the literature. Progression to severe disease and in-hospital/overall mortality during the follow-up of the patients were determined from electronic records. Kaplan-Meier, log-rank test, and Cox proportional hazard regression model was used. The discrimination capability of pneumonia severity indices was evaluated by receiver-operating-characteristic (ROC) analysis.

Two hundred ninety-eight patients were included in the study. Sixty-two patients (20.8%) presented with severe COVID-19 while thirty-one (10.4%) developed severe COVID-19 at any time from the admission. In-hospital mortality was 39 (13.1%) while the overall mortality was 44 (14.8%). The mortality in low-risk groups that were identified to manage outside the hospital was 0 in CALL Class A, 1.67% in PSI low risk, and 2.68% in CURB-65 low-risk. However, the AUCs for the mortality prediction in COVID-19 were 0.875, 0.873, 0.859, 0.855, and 0.828 for A-DROP, PSI, CURB-65, COVID-GRAM, and CALL scores respectively. The AUCs for the prediction of progression to severe disease was 0.739, 0.711, 0,697, 0.673, and 0.668 for CURB-65, CALL, PSI, COVID-GRAM, A-DROP respectively. The hazard ratios (HR) for the tested pneumonia severity indices demonstrated that A-DROP and CURB-65 scores had the strongest association with mortality, and PSI, and COVID-GRAM scores predicted mortality independent from age and comorbidity.

Community-acquired pneumonia (CAP) scores can predict in COVID-19. The indices proposed specifically to COVID-19 work less than nonspecific scoring systems surprisingly. The CALL score may be used to decide outpatient management in COVID-19.