| ObjectiveTo investigate the status quo of post-dialysis fatigue(PDF)in patients with mai-ntenance hemodialysis,explore the influencing factors of PDF,establish and vali-dated the PDF nomogram prediction model,which is convenient for the early sc-reening and intervention of clinical medical staff.Method1.217 MHD patients who were in the hemodialysis room of a tertiary hospital i n Anhui province from February 2021 to January 2022 were selected in accorda-nce with the exclusive criterion.The Functional Assessment of Chronic Illness T hrea Py-Fatigue(FACIT-F)was used for investigation and evaluation.217 patients were divided into PDF group and non-PDF group according to the occurrence of PDF.R 4.1.2 software was used to randomly divide the database into model gr-oup(80%,173 cases)and validation group(20%,44 cases).2.The research variables and measurement tools were determined according to t-he previous literature reviews,systematic evaluation and clinical investigation.217 MHD patients were investigated with the Pittsburgh Sleep Quality Index(PSQ I),Social Support Rate Scale(SSRS),Simplified Nutritional Appetite Questionnair e(SNAQ),and Connor-Davidson Resilience Scale(CD-RISC-10)were used to c ollect data.3.SPSS25.0 was used to conduct single factor logistic analysis in model group and the variables with statistical significance(P<0.1)were included in the multi variate logistic analysis.R 4.1.2 software was used to perform LASSO regression analysis and the results of screening variables were included in the multivariate l ogistic analysis.The rms program package of R software was used to draw the prediction factors of the model group into the nomogram prediction model.4.Apply the prediction model of the model group to the validation group,and e valuate the prediction performance of the model through the area under the rece-iver operating characteristic(AUC),calibration curve and the Homster-Lemeshow(H-L)fitting test value.Decision curve analysis(DCA)of the nomograph model was to determine the clinical applicability of the model.5.The prediction efficiency of the two models was judged by AUC,and Bayesi an Information Criterion(BIC).ResultA total of 217 MHD patients were investigated in this study,173 in the model group and 44 in the validation group.There was no statistically significant diffe-rence between the two groups in terms of general data,psychosocial factors and clinical disease factors(P>0.05)which revealed that the database randomization was pretty good.1.The PDF univariate factor logistic analysis of MHD patients showed that ther-e were statistically significant differences in age,marital status,educational level,the frequency of intradialytic hypotension,cardiovascular and cerebrovascu-lar dis eases,the number of other diseases,the number of complications,physical exerci ses,constipation,PSQI score,SNAQ,blood magnesium level,albumin leve-l,blo od creatinine,psychological elasticity,etc.(P<0.1).Using lambda=lambda.min as the standard,LASSO regression variables screened finally 7 variables inc-luding age,marital status,education level,cardiovascular and cerebrovascular dise ases,frequency of intradialytic hypotension,physical exercises,and resilience.2.The results of univariate factor logistic screening variables were included in t-he multivariate logistic regression analysis.The final result showed that age(OR=2.056 95%CI:1.005～4.204 P=0.048),the frequency of intradialytic hypotension(OR=2.738 95%CI:1.681～4.460 P<0.001),cardio-cerebrovascular disease(OR=2.559 95%CI:1.211～5.405 P=0.014)and the number of complications(OR=1.705 95%CI:1.184～2.456 P=0.004)were prediction factors for PDF in MHD pati-ents.After the variables screened by LASSO regression were included in the m-ultiv ariate logistic regression,age(OR=2.056 95%CI:1.027～4.118 P=0.042),the freq uency of intradialytic hypotension(OR=2.909 95%CI:1.810～4.676 P<0.001),ca rdiovascular and cerebrovascular diseases(OR=3.546 95%CI:1.756～7.163 P<0.001)and resilience(OR=0.184 95%CI:0.091～0.370 P<0.001)were fou-nd.3.Five prediction factors obtained by the univariate and multiple factor logistic r egression was used to build a nomograph model.In the model group,the AUC of the model was 0.850(95%CI:0.781～0.919),and the discrimination is me-d ium.Combined with the Hosmer-Lemeshow goodness of fit test(χ~2=11.874,P=0.220>0.05)and calibration curve,it suggested that the model had good consist-e ncy.DCA curve showed that the model had higher net benefit when the thresh-o ld value of the model was 0.10～0.86.In the validation group,the AUC of them odel was 0.832(95%CI:0.814～0.950)and the discrimination was medium.C-om bined with the Hosmer-Lemeshow goodness of fit test(χ~2=9.987,P=0.352>0.05)and calibration curve,the model showed good consistency.The DCA curve indic ated that when the threshold value of the model was 0.02～0.70,the modelshowed a higher net benefit.4.Four predictors(age,the frequency of intradialytic hypotension,cardiovascular a nd cerebrovascular diseases,resilience)of LASSO regression and multivariate log istic regression was used to establish a nomograph model.In the model group,t he AUC of the model was 0.831(95%CI:0.766～0.896).Combined with the Hos mer-Lemeshow goodness of fit test(χ~2=10.623,P=0.302>0.05)and calibration curv e,it suggested that the model had good discrimination and calibration.The DCA curve showed that when the threshold was 0.08～0.72,the model gained the pos itive net benefit.In the validation group,the AUC of the model was 0.840(0.726～0.955).Combined with the the Hosmer-Lemeshow goodness of fit test(χ~2=10.129,P=0.340>0.05)and the calibration curve,it suggested that the model had go od discrimination and calibration.The DCA curve showed that when the thres-ho ld value of themodel was 0.03～0.60,the model revealed the positive net bene-fit.5.Model based on the univariate factor logistic analysis and screening variables was BIC=193.83;Model based on LASSO regression screening variables was BI C=189.19.DiscussionThis study find that age,the frequency of intradialytic hypotension,cardiovascul-ar and cerebrovascular diseases,the number of complications,and psychological r esilience are prediction factors for PDF in MHD Patients.The nomogram model finally constructed by the univariate factor logistic analysis and LASSO regressi-on both have good discrimination,calibration and clinical practicability,but after comprehensive comparison,the nomograpm model constructed by LASSO regressi on has better predictive value. |
PDF Full Text Request