Survival analysis and risk predictive model of maintenance hemodialysis patients

Objective  To explore the survival rate of maintenance hemodialysis （MHD） patients and construct an survival prediction.

Methods  From December 2018 to May 2019, 370 cases of MHD patients were recruited. Clinical data and outcomes were recorded. Cox regression was employed for examining the influencing factors of survival prognosis. And survminer, survival, rms and survival receiver operating characteristic （ROC） data packets of R were utilized for constructing and validating the survival predictive model. DynNom data packet and DNbuilder function were employed for constructing online dynamic nomogram.

Results  A total of 370 cases of MHD patients were included. Multivariate Cox regression analysis indicated that the risk factors for survival prognosis included age, diabetes mellitus （DM）, vascular access, ratio of neutrophil-to-lymphocyte and serum albumin. According to the results of Cox regression analysis, a survival predictive nomogram was constructed with a C-index of 0.703（95%CI: 0.6589-0.7471）. And the areas under ROC of 1/2/3-year survival were 0.742, 0.734 and 0.737 respectively.

Conclusion  Based upon age, DM, vascular access, ratio of neutrophil-to-lymphocyte and serum albumin, online dynamic nomogram helps to predict 1/2/3-year survival rates of MHD patients.