Usefulness of CURB-65, PSI and MuLBSTA in predicting COVID-19 mortality

Abstract

Objective: The aim is to evaluate the accuracy of CURB-65, Pneumonia Severity Index (PSI) and MuLBSTA, a viral pneumonia, in predicting mortality in Coronavirus Disease 2019 (COVID-19) pneumonia.

Methods: We retrospectively analyzed consecutive patients with laboratory-confirmed COVID-19 pneumonia admitted at Papa Giovanni XXIII Hospital from February 23rd to March 14th, 2020. We calculated CURB-65, PSI and MuLBSTA comparing their performances in discriminating between survivors and non-survivors at 28 days.

Results: Among 431 hospitalized patients, the majority presented with hypoxic respiratory failure: median (IQR) PaO2/FiO2 ratio at admission was 228.6 (142.0-278.1). In the first 24 hours, 111 (27%) patients were administered low-flow oxygen cannula, 50 (12%) Venturi Mask, 95 (23%) non-rebreather mask, 106 (26%) non-invasive ventilation, 12 (3%) mechanical ventilation and 41 (9%) were not administered oxygen therapy. Mortality rate at 28-day was 35% (150/431). Between survivors and non-survivors, median (IQR) scores were 1.0 (1.0-2.0) and 2.0 (2.0-3.0) (p<0.001) for CURB-65; 90.5 (76.0-105.5) and 115.0 (100.0-129.0) for PSI (p<0.001); 7.0 (5.0-10.0) and 11.0 (9.0-13.0) for MuLBSTA (p<0.001). Areas under the receiver operating characteristic curve (AUC) were, respectively, 0.725 (0.662-0.787), 0.776 (0.693-0.859) and 0.743 (0.680-0.806) (p>0,05). PSI and MuLBSTA did not show a better performance when compared to CURB-65.

Conclusions: CURB-65, PSI and MuLBSTA scores are useful to discriminate between survivors and non-survivors in COVID-19 pneumonia even if diagnostic accuracy in discriminating 28-day mortality is moderate with potential underestimation in the low-risk classes of the scores.