Tool for acute kidney injury prediction in hospitalised patients with COVID-19

Abstract

Background. The study focuses on acute kidney injury (AKI) in hospitalised COVID-19 patients. AKI is a significant medical issue often linked to severe conditions like pneumonia and sepsis. Understanding the predictors of AKI in COVID-19 is crucial for improving patient outcomes. Materials and methods. In Kyiv, a retrospective, case-control study was conducted at the KAPITAL Ltd. (Medical Centre “Universal Clinic “Oberig”). The study included 129 patients hospitalised with moderate to severe COVID-19 from 2020 to 2021. They were divided into those who developed AKI (n = 19) and those who did not (n = 110). We used various statistical logistic regression models to predict AKI. Results. Significant predictors of AKI included age, C-reactive protein levels, absolute lymphocyte count, Padua Prediction Score, and respiratory insufficiency. No significant differences were observed in gender distribution, estimated glomerular filtration rate on admission, prevalence of hypertension, diabetes, or body mass index between the two groups. Multivariate logistic regression incorporating the Padua Prediction Score showed strong predictive performance with an AUC of 0.803. Conclusions. The study highlights the critical need for accurate predictive models tailored to specific populations. It aims to develop a novel, region-specific predictive model for AKI in COVID-19 patients by focusing on the Ukrainian population. The model leverages local data to improve risk predictions and patient outcomes, emphasising the importance of early identification and stratification of high-risk individuals.