Abstract: Objective Screening the dialysis risk factors of hemorrhagic fever with renal syndrome (HFRS) from the commonly used clinical indicators and constructing an early warning model, so as to guide the clinic. Method The clinical data of 112 HFRS patients in the first hospital of Changsha from December 2013 to April 2023 were retrospectively analyzed, divided into dialysis group (38 cases) and non dialysis group (74 cases) based on whether hemodialysis is performed or not. To compare and analyze the differences between the basic data of patients in dialysis group and those in non-dialysis group and the indexes of routine laboratory examination. Logstic regression was used to build an early prediction model, and receiver operating characteristic curve (ROC) was used to warn the dialysis ability of each index. Result In 112 HFRS patients, 38 in dialysis group (33.928%) with (24±8.95) days average hospitalization time and 74 in non-dialysis group (66.07%), and the average hospitalization time was (16±6.10) days. Sex and underlying diseases were no difference in two groups (P>0.05). Blood transfusion, shock, pulmonary infection and proteinuria incidence rates in dialysis group was higher than non-dialysis group, and the differences have significant (P<0.05). Age in dialysis group was less than non-dialysis group(P=0.015） and hospitalization time was longer than non-dialysis group(P<0.001）, and ALB in two groups were no differences (P=0.393). WBC、PCT、CR and CK-MB were higher in dialysis group and PLT was lower in dialysis group. ALT, AST, CRP and LDH were no differences in two groups. The AUC of Age, WBC, PLT, PCT, CR and CKMB is 0.650, 0.712, 0.801, 0.671, 0.700 and 0.712, respectively, and the optimal thresholds for prediction are 48.50 years old, 11.92×10⁹/L, 22.50×10⁹/L,1.63μg/L,181.75μmol/L and 25.15U/L, using Logstic regression model, P=1(1+e-γ), γ=1.745-0.084×PLT+0.006×CR (P is the probability value for predicting patient requiring dialysis, γ is the prediction index), and it is concluded that the AUC of the combined detection of PLT and CR is 0.888, the sensitivity is 81.10%, and the specificity is 86.80%. Conclusion sAge, WBC, PLT, PCT, CR, CKMB can be used as early predictors of whether hemorrhagic fever with renal syndrome needs hemodialysis, and combined indicators (PLT and CR) detection is more helpful to find out whether hemorrhagic fever with renal syndrome needs dialysis than single indicators detection.

Key words: Hemorrhagic fever with renal syndrome, Hantavirus, Laboratory index, Dialysis, Early predict