Risk Factors Analysis Of Cognitive Dysfunction In Hemodialysis Patients And Construction Of Prediction Model

ObjectiveThe purpose of this study is to explore the status of cognitive function in hemodialysis patients,identify the risk factors that may affect cognitive function of hemodialysis patients,and establish a logistic regression prediction model for early prediction of cognitive dysfunction.The probability p value formula of cognitive dysfunction may be obtained,so as to quickly screen high-risk patients with cognitive dysfunction in HD patients,to predict the occurrence and development of Cognitive impairment in patients,and to provide basis for making clinical decisions.In addition,the methods of constructing prediction model and testing are introduced in detail,which can be used for reference to other diseases in order to effectively monitor and prevent diseases.Methods226 patients who received hemodialysis in blood purification center of two third-class hospitals in Qingdao from 2017.2-2017.11,and randomly divided into 70%（159 cases）listed as modeling cohort 30%（67 cases）as validation queue.The data collected from the modeling queue were analyzed by single factor analysis,and the significant factors were analyzed by the backward Wald stepwise selection method for multivariate binomial logistic regression analysis,and the variables corresponding to the minimum wald value were gradually eliminated.Finally,all variables reached significant level,and logistic regression prediction model was established.Hosmer-lemeshow test was used to verify the goodness of fit test of the model.The logistic regression prediction model was applied to the validation queue,and the AUC value（under the curve area）was obtained by bootstrap resampling（1000 times）.The accuracy,sensitivity and specificity of the model were further judged by the operating characteristic curve of the subjects.Results1.According to the MMSE criteria,99 of the 226 patients（43.8%）had cognitive impairment.2.Logistic single factor associated with HD in patients with Cognitive impairment regression analysis:gender（p<0.05,x²=14.496）,education（p<0.05,x²=18.354）,age（p<0.05,x²=22.858）,the number of disease（p<0.05,x²=42.667）,blood potassium levels（p<0.05,x²=10.697）,serum phosphorus level（p<0.05,x²=7.129）,serum calcium the level of（p<0.05,x2=39.529）,hemoglobin（p<0.05,x²=67.319）,albumin（p<0.05,x²=15.621）,creatinine（p<0.05,x²=13.453）,TC（p<0.05,x²=7.385）.3.Further multivariate two-classification logistic regression analysis showed the risk factors of cognitive dysfunction in HD patients:Level of education X₃（P<0.001,OR=0.174,95%CI:0.097-0.313）,Age X₁（P<0.001,OR=1.133,95%CI:1.062-1.209）,Number of combined diseases X₄（P<0.001,OR=3.293,95%CI:1.885-5.755）,Hemoglobin level X₈（P<0.001,OR=0.164,95%CI:0.087-0.308）.R² was 0.766,which could explain76.6%of the variation.4.Final logistic regression model:Logit（p）=ln（p/1-p）=-2.698-1.750*educational level+0.125*age+1.192*the number of complicated diseases-1.808*hemoglobin level.Probability formula for possible cognitive impairment:P=Exp（-2.698-1.750*educational level+0.125*age+1.192*the number of complicated diseases-1.808*hemoglobin level）/[1+Exp（-2.698-1.750*educational level+0.125*age+1.192*the number of complicated diseases-1.808*hemoglobin level）].5.modelling verification:Hosmer-lemeshow test its goodness of fit X²=14.689,P=0.65,The area under the ROC curve of Modeling queue model was 89.2%,the area under the ROC curve of Model queue model was 86.1%,the sensitivity was 80.8%（21/26）,the specificity Was 82.9%（34/41）,and the accuracy was 82.1%（55/67）.ConclusionIn this study,99 out of 226 HD patients had cognitive dysfunction and the incidence was high.The predictive model of cognitive dysfunction in hemodialysis patients was as follows:P=Exp（-2.698-1.750*educational level+0.125*age+1.192*the number of complicated diseases-1.808*hemoglobin level）/[1+Exp（-2.698-1.750*educational level+0.125*age+1.192*the number of complicated diseases-1.808*hemoglobin level）].The educational level,age,the number of complicated diseases and hemoglobin level were independent predictors of cognitive dysfunction,and the educational level and hemoglobin level were the protective factors of cognitive function.Age and the number of associated diseases are risk factors for cognitive function.It is proved that this model has a good value in predicting the occurrence of Cognitive impairment in HD patients.It is worth popularizing and applying in clinical nursing work to use this model method to predict the occurrence and development of disease.