Assessment of risk scores in Covid-19

Abstract

Objective: To analyse the accuracy of commonly used risk scores (PSI and CURB-65) in predicting mortality and need for ICU admission in Covid-19.

Material and methods: Prospective study of patients diagnosed with Covid-19 pneumonia. Patients were followed until home discharge or death. PSI, CURB-65, SMART-COP and MuLBSTA severity scores were assessed on admission. Risk scores were related to mortality and ICU admission.

Results: About 249 patients, 143 males (57.4%) were included. The mean age was 65.6 + 16.1 years. Factors associates with mortality in the multivariate analysis were age > 80 years (OR: 13.9; 95% CI 3.8-51.1) (P = .000), lymphocytes < 800 (OR: 2.9; CI 95% 1.1-7-9) (P = .040), confusion (OR: 6.3; 95% CI 1.6-24.7) (P = .008) and NT-proBNP > 500 pg/mL (OR: 10.1; 95% CI 1.1-63.1) (P = .039). In predicting mortality, the PSI score: AUC 0.874 (95% CI 0.808-0.939) and the CURB-65 score: AUC 0.852 (95% CI 0.794-0.909) were the ones that obtained the best results. In the need for ICU admission, the SMART-COP score: AUC 0.749 (95% CI 0.695-0.820) and the MuLBSTA score: AUC 0.777 (95% CI 0.713-0.840) were the ones that obtained better results, with significant differences with PSI and CURB-65. The scores with the lowest value for ICU admission prediction were PSI with AUC of 0.620 (95% CI 0.549-0.690) and CURB-65 with AUC of 0.604 (95% CI 0.528-0.680).

Conclusions: Prognosis scores routinely used for CAP (PSI and CURB-65) were good predictors for mortality in patients with Covid-19 CAP but not for need of hospitalisation or ICU admission. In the evaluation of Covid-19 pneumonia, we need scores that allow to decide the appropriate level of care.