The strong predictive role and their day-dependent behavior of blood urea nitrogen and complete blood count in covid-19’s inpatients prognosis

Abstract

Background: The outcome of hospitalized COVID-19 patients is predictable according to demographic, clinical, laboratory, and imaging risk factors. We aimed to determine the best outcome predictors and their trends during 30 days of hospitalization. Methods: This retrospective study was conducted on moderate to severe hospitalized COVID-19 patients from 26 January 2020 to 13 January 2021. The length of stay in the hospital was considered as the time interval between admission and discharge, and the patient's final condition was defined as either dead or alive. Demographic, clinical, and laboratory data were collected from the hospital information system. The generalized additive model and the Cox regression model were used to model data. Results: Of the 1520 hospitalized COVID-19 patients, 232 (15.26%) died and 1288 survived or reached the end of 30 days of hospitalization. We selected demographic, clinical, and 131 independent laboratory variables. Blood urea nitrogen (BUN) had a nearly double average in the dead group (44.603 [± 25.408] mg/dL) than the survived group (21.304 [± 13.318] mg/dL), and the lymphocyte (Lymph) count showed the opposite trend. The estimated hazard ratio (HR) of these 2 factors was higher than 1 and was statistically significant. In daily stay trends, the hazard function of them also increased rapidly after 15 days. Conclusion: Blood urea nitrogen and complete blood count provide strong predictive clues about the prognosis of hospitalized COVID-19 patients, and rapid dynamic changes in the second week can predict a poor outcome in these patients.

Keywords

• covid-19
• survival analysis
• statistical models
• blood urea nitrogen
• blood cells