Clinical Risk Factors for COVID-19 related Severe Outcome

Abstract

AbstractBackgroundWe aimed to evaluate the risk factors for Coronavirus disease 2019 (COVID-19) related severe outcome in New York State (NYS) and proposed a method that could be used to inform future work to develop clinical algorithms and predict resource needs for COVID-19 patients.MethodsWe analyzed COVID-19 related hospital encounter and hospitalization in NYS from April 1st to November 17th, 2020, using Statewide Planning and Research Cooperative System (SPARCS) hospital discharge dataset. Logistic regression was performed to evaluate the risk factors for COVID-19 related in-hospital death using demographic variables, symptom, rapid clinical examination, and medical history of chronic co-morbid conditions. Receiver operating characteristic (ROC) curve was calculated, and cut-off points for predictors were selected to stage the risk of COVID-19 related fatal outcome.FindingsLogistic regression analysis showed age was the greatest risk factor for COVID-19 related fatal outcome, which by itself achieved the diagnostic accuracy of 0.78 represented by the area under the ROC curve. By adding other demographic variables, dyspnea or hypoxemia and multiple chronic co-morbid conditions, the diagnostic accuracy was improved to 0.85. We selected cut-off points for predictors and provided a general recommendation to categorize the levels of risk for COVID-19 related fatal outcome.InterpretationWe assessed risk factors associated with in-hospital COVID-19 mortality and identified cut-off points that might be used to categorize the level of risk. Further studies are warranted to evaluate laboratory tests and develop laboratory biomarkers to improve the diagnostic accuracy for early intervention.