Clinical prediction rules for adverse outcomes in patients with SARS COV-2 infection by the omicron variant (preprint)

ABSTRACT

Background: Factors related to an adverse evolution in COVID19 infection are needed for proper decision making. We try to identify factors related to hospitalization, ICU admission, and mortality related to the infection. Methods. Retrospective cohort study of patients with SARS-CoV-2 infection from March 1st 2020 to January 9th 2022. The sample was randomly divided into two subsamples, for the purposes of derivation and validation of the prediction rule, until omicron variant appearance and afterwards, respectively. Data collected for this study included sociodemographic data, baseline comorbidities and treatments, and other background data. Multivariable logistic regression models using Lasso logistic regression were used . Results. In the multivariable models, older age, male, peripheral vascular disease, heart failure, heart disease, cerebrovascular, dementia, liver, kidney, diabetes, hemiplegia, interstitial pulmonary disease, cystic fibrosis, malignant tumors, as well as diuretics and the chronic systemic use of steroids were common predictive factors of death. Similar predictors, except liver disease, plus arterial hypertension, were also related to adverse evolution. Similar predictors to the previous, including liver disease, plus dyslipidemia, inflammatory bowel disease, respiratory diseases, and the basal prescription of NSAIDs, heparin, bronchodilators, or immunosuppressants were related to hospital admission. All risk scores developed had AUCs from 0.79 (hospital admission) to 0.94 (death) in the validation in the omicron sample. Conclusions. We propose three risk scales for adverse outcomes and hospital admission easy to calculate and with high predictive capacity, which also work with the current omicron variant, which can help manage patients in primary, emergency, and hospital care.