Predicting COVID-19 Outcomes in Patients at Advanced Stages of HIV Infection: A Model-based Approach

Abstract

Abstract Background: Today clinicians and their decisions extensively rely on specific treatment algorithms. These algorithms include prognostic models to identify high-risk patients requiring hospital admission and clinical monitoring. Our paper suggests a prognostic model to forecast COVID-19 outcomes in patients with advanced HIV disease, considering the high risk for unfavorable outcome and the need of a specialized approach [2]. Objective: The objective is to develop a prognostic model combining predictors of unfavorable COVID-19 outcome in patients with advanced HIV disease. Materials and methods: The study is based on 500 medical records of patients with advanced HIV disease admitted for confirmed COVID-19 from March 1, 2020 through December 31, 2022 for inpatient treatment at the Infectious Diseases Hospital in Moscow. Results: Each of the 500 patients was evaluated for 167 predictive markers for unfavorable COVID-19 outcome, outlining 50 indicators that varied significantly across the subgroups of advanced HIV disease patients with COVID-19 depending on favorable or poor outcome. Oxygen therapy was the most significant factor showing strong correlation with poor outcome in advanced HIV patients with COVID-19. Subsequently, predictors were selected stepwise in order to enhance predictive accuracy of the resulting model by adding more factors. The resulting model included 7 factors, e.g. oxygen therapy requirements, CD4+ count under 50 cells/μL, manifest CMV infection characterized by lung damage, elevated lactate dehydrogenase, urea, and fibrinogen levels, as well as unspecified encephalitis. using available data in calculations, a prognostic scenario and a ROC curve were built to assess practical significance of the obtained prognostic model. The area under the ROC curve was 90.9%, thus confirming prediction accuracy and overall practical significance of the model. Conclusions: The suggested prognostic model enables to assess potential outcomes in patients with HIV and COVID-19 co-infection admitted to hospital at advanced stages of disease, as well as to plan adequate therapies based on the obtained results. Keywords: HIV, COVID-19, SARS-CoV-2