A time-updated scoring system derived from a nomogram to predict 3-month mortality in maintenance hemodialysis patients

TL;DR

This study created a simple scoring system to predict 3-month mortality in patients undergoing hemodialysis, helping doctors provide better care.

Contribution

A novel, time-updated scoring system was developed and validated for 3-month mortality prediction in hemodialysis patients.

Findings

The scoring system includes five predictors with assigned point values and achieved a C-index of 0.72 and 0.73 in training and validation sets.

The model effectively stratified patients into low, intermediate, and high-risk groups (P < 0.0001).

The system is described as easy-to-use and accurate for predicting mortality in hemodialysis patients.

Abstract

This study aimed to develop and validate a novel risk score for death within the next 3 months in patients with maintenance hemodialysis (MHD).

All the data were derived in the Wuhan Hemodialysis Quality Control Center and were divided into the training set (2019–2021, n = 19,735) and the validation set (2022–2023, n = 15,265). The primary outcome was the all-cause mortality within 3 months after regular monthly laboratory tests. The predictive model was displayed as a nomogram and modified into a novel scoring system based on the coefficients of multivariable logistic regression.

There were 1684 (8.5%) patients and 1670 (10.9%) patients who died in the development and validation set, respectively. The final novel score system was calculated based on the five predictors: age ≥60 years (2 points), dialysis duration <1 year (2 points), catheter usage (1 point), hemoglobin <110 g/L (1 point), and albumin <35 g/L (3 points). This model with a C-index of 0.72 and 0.73 on the two sets and exhibited a significant ability in stratification of patients into low-risk, intermediate-risk, and high-risk groups (P < 0.0001).

This easy-to-use applicable scoring system accurately predicts 3-month mortality in HD patients, facilitating risk stratification and personalized care.

Figures