Early Warning Models for Predicting Severity in Febrile and Nonfebrile Stages of Hemorrhagic Fever with Renal Syndrome

Abstract

：The treatment of severe hemorrhagic fever with renal syndrome (HFRS) cases is difficult.We lack an early warning model for severe HFRS patients today. We retrospectively collected the data of 235 HFRS patients from January 2013 to December 2019, as well as 394 laboratory indicators. The multivariate logistic regression model was used to construct an early warning model for severe disease.The accuracy of the model was evaluated based on the area under the receiver operating characteristic (ROC) curve. The area under curve (AUCs) of the early warning models both exceeded 0.9 for the two stages. In the febrile stage, there were significant differences between the severe and mild groups (P < 0.05) in the renal estimated glomerular filtration rate (eGFR), urinary leukocytes , electrolytes, urine conductivity and urinary epithelial cell count. In the non-febrile stage, there were significant differences between the severe and mild groups (P < 0.05) in the renal eGFR, electrolytes, urine conductivity and renal cystatin C. The two early warning models are well fitted and have excellent predictive performance. That can help clinicians gain time to provide appropriate preemptive treatment to avoid the further development of severe disease and reduce the mortality rate.