Regresar

Analysis of COVID-19 severity by multinomial regression in hospitalized patients in Riobamba, Ecuador

Abstract:

Introduction: Finding predictive factors of severity in a specific population helps to recognize sets of people at high risk, which allows directing preventive actions to avoid the severe development of COVID-19 and decrease the associated complications. Objective: The objective of the study was to analyze by multinomial regression the severity of COVID-19, according to vaccination, level of exposure to the virus, socioeconomic level and sex, in patients in a hospital in Riobamba, Ecuador, during the first quarter of 2023. Methods: The study was predictive, observational, analytical, retrospective and cross-sectional. The study sample consisted of 259 patients diagnosed with COVID-19 confirmed by laboratory tests. A multinomial regression design with likelihood ratio tests was performed. Results: Multinomial regression analysis showed that the majority were men with non-current vaccination and medium-high levels of exposure to the virus and socioeconomic status. The disease did not progress to extreme severity in most cases. The final model with all predictor variables fitted the data significantly better than the null model, although the interactions evaluated were not statistically significant. Measures of fit and pseudo R-squared indicated an acceptable model fit for predicting COVID-19 severity as a function of predictor variables. Conclusions: The results did not show a statistically significant relationship between COVID-19 severity and the predictor variables evaluated, Therefore, the null hypothesis cannot be rejected.