Construction and efficacy of fall risk prediction model for hemodialysis patients
Objective To explore the risk factors for falls in hemodialysis patients,construct a predic-tive model,and conduct internal and external validation.Methods A total of 200 patients who received outpa-tient hemodialysis treatment in Department of Nephrology in the People's Hospital of Nanchuan Chongqing from January 2022 to January 2023 were selected as the study subjects,and this group of patients was used as the training set.Another 50 patients who underwent hemodialysis in the same hospital from February 2023 to February 2024 were selected as the validation set for the study.The general and clinical data of patients were collected and followed up for one year.The risk factors of falls in hemodialysis patients were analyzed,and a nomogram prediction model was constructed.The discrimination and goodness of fit of the model were verified by ROC curve and Hosmer-Lemeshow test.Results According to whether the subjects fell,they were divided into the fall group(n=79)and the control group(n=121),and the incidence of patients falling in the blood purification center of our hospital was 39.5％.Univariate logistic regression analysis showed that age,im-paired vision,use of walking AIDS,diastolic blood pressure before dialysis,potassium,self-care ability score,diabetes,stroke,use of hypoglycemic agents and sleeping pills were all risk factors for falls of hemodialysis pa-tients(P＜0.05).Multivariate Logistic regression analysis showed that the use of walking AIDS,potassium(≥4.65 mmol/L),self-care ability score(≤83),stroke and sleeping pills were all independent risk factors for the fall of hemodialysis patients(P＜0.05).The area under ROC curve was 0.796,and its sensitivity and speci-ficity were 64.8％and 89.9％respectively.(Hosmer-Lemeshow test:x2=5.041,P=0.283).Conclusion The risk prediction model constructed in this study can effectively predict the risk of falls in hemodialysis patients,and provide evidence for clinical formulation of fall prevention programs.
万方数据
维普
