Intervention Study for Validation of Application of a Prediction Model in Maintenance Hemodialysis Patients

Chronic kidney disease (CKD) is a major public health concern. Progression of CKD leads to end-stage renal disease (ESRD), a total and permanent failure of kidney function requiring kidney transplant or maintenance hemodialysis. Among the numerous risk factors of mortality, severe malnutrition, now defined as protein energy wasting (PEW), which is manifested by low serum levels of albumin or pre-albumin, sarcopenia and weight loss, is one of the strongest predictors of mortality, morbidity and infection-related hospitalizations in hemodialysis (HD) patients. When PEW is present or the patient is at high risk, oral intradialytic or daily supplements, aiming at increasing energy and protein intake, while in selected cases intra-dialytic parenteral nutrition may be used for the clinician. Previous reports showed that providing patients with intradialytic food, liquid supplements, or parenteral nutrition has been shown to compensate for this loss. In patients with ESRD, it is important to prevent PEW through a regular nutritional status evaluation and nutritional counseling by a renal dietitian. Intradialytic administration of oral or parenteral supplementation of nutrients is safe and should be encouraged in selected patients. Although the problem of malnutrition became more and more serious due to the increasingly aging and Diabetes Mellitus (DM) proportion in HD population, there are still no universally accepted methods to early detect or predict this problem. Thus, the intention for early intervention to suitable patients faces difficulty. In Taiwan Prediction Model Project Plan (TPMPP), the investigators try to develop a model using common available clinical data with the utilization of current machine learning methods to early detect the patients with high risks for malnutrition and the investigators wish through the routine application of this model, the investigators can do nutrition intervention in much earlier stage and improve the patients' outcome. In the first stage of TPMPP, the investigators have developed some prediction models through an abundant retrospective database collected in Fresenius Database (NephroWeb) from 2011-2018. Before real implementation of these models in clinical practice, the investigators need to validate these models with prospective design with proper intervention, this is the main purpose of this study. The Body Composition Monitor with the principle of Bioimpedance Spectroscopy (BCMBIS, Fresenius Medical Care) is a simple and accurate tool for the assessment of body composition. Recently, the lean and fat mass indexes (LTI, FTI), derived from BCM, have become good predictors of mortality in HD individuals, which have been more sensitive to malnutrition, and altogether anthropometric indicators may result in an early detection of mortality risk in this population. Thus, the investigators also try to apply these parameters in this study as important nutrition outcomes for follow-up.