Prognostic usefulness of severity scales for the initial assessment of elderly people with a diagnosis of pneumonia due to COVID-19. Design of a multivariate predictive model of mortality

Abstract

Introduction: The elderly are the population group most vulnerable to COVID-19. To identify risk factors associated with poor evolution is important. The aim of this study is to analyze the prognostic value of three severity scales (Pneumonia Severity Index: PSI; CURB-65 scale; Severity Community Acquired Pneumonia Score: SCAP) and to design a novel predictive model for intra-hospital mortality in patients ageing ≥75 admitted due to pneumonia linked to COVID-19.

Methods: Retrospective study. Demographic, clinical and blood test variables were obtained. The PSI, CURB-65, and SCAP severity scales were assessed as predictors of mortality. A multivariate predictive model was composed using stepwise logistic regression.

Results: A total of 186 patients were involved. Mortality reached 47.3%. PSI, CURB-65 and SCAP severity scales achieved moderate performance, with area under the curve (AUC) equal to 0.74 (95%CI 0.64-0.82), 0.71 (95%CI 0.62-0.79) and 0.72 (95%CI 0.63-0.81), respectively. The predictive model composed of the absence or presence of symptoms (cough, dyspnea), comorbidity (Charlson index) and analytics (levels of aspartate-aminotransferase, potassium, urea, dehydrogenase-lactate) achieved an AUC of 0.81 (95%CI 0.73-0.88). The decision curves showed that the predictive model had higher net benefit than the severity scales for all probability theresholds.

Conclusion: PSI, CURB65, SCAP scales can help professionals to asses the prognosis of elderly patients diagnosed with COVID-19 pneumonia. However, the novel predictive model of mortality significantly improves the performance of those scales.