Desenvolvimento de um escore de risco para mortalidade em pacientes com Covid-19 admitidos em Unidades de Terapia Intensiva

Abstract

Introduction: Several risk scores were tested during the COVID-19 pandemic in order to stratify those patients most at risk, properly allocating resources and giving better chances of a successful outcome. Despite that, most of them are bounded by methodological flaws and they use data before the emergence of new variants and the broad use of vaccination, which may impact scores’ performance. A reliable mortality risk score for COVID-19 patients admitted to intensive care units (ICUs) is still lacking. Objectives: To develop a mortality risk score for patients with covid-19 upon admission to ICU and compare it with other existing scores. Methods: This is a multicenter observational retrospective study, which included adult patients with COVID-19 admitted to the ICU of 18 hospitals in 9 Brazilian cities, from September 2021 to July 2022. Potential predictors were selected through review of the literature. Generalized Additive Models (GAM) were used to evaluate the relationship between the predictors and the outcome. LASSO (least absolute shrinkage and selection operator) regression was used to derive the score score. Results: Of the 558 patients included, the median age was 69 years (IQR 58-78), 56.3% were men, 19.7% required invasive mechanical ventilation (IMV), and overall mortality was 44.8%. The final model included 6 variables: Age, Breathing (pO2/FiO2, and respiratory function [respiratory rate and whether on IMV]), COPD (chronic obstructive pulmonary disease) and Obesity. AB2CO had an area under the the receiver operating characteristic curve (AUROC) of 0.781 (95% confidence interval 0.744-0.819), a good overall performance and excellent calibration curve (slope=1.063, intercept=0.015, p-value=0.834). The score was compared with other existing ones (ABC2-SPH, Atschul et al., Modified CHA2DS2-VASc, COVID-SOFA, CURB-65, SOARS, SOFA, NEWS2, and 4C Mortality Score) and demonstrated better discrimination power than most of them, and with a better calibration curve. Conclusions: AB2CO is a simple, quick and objective score, based on just 6 variables, ranging from 0 to 16 points, with better performance than other scores. It was developed using the most recent data from the pandemic, making it even more reliable to the current scenario. However, further studies are necessary to carry out external validation of the score.