Risk prediction models for short-term mortality within 30 days after stroke: a systematic review
10.3760/cma.j.cn115682-20240329-01688
- VernacularTitle:脑卒中后30 d内短期死亡风险预测模型的系统评价
- Author:
Qian ZHANG
1
;
Chun CHEN
;
Juan DING
;
Ren LIU
;
Tingting CHEN
;
Jinlong ZHENG
;
Jiaqian KUANG
Author Information
1. 长江大学医学部，荆州 434020
- Keywords:
Stroke;
Short-term mortality;
Prediction model;
Systematic review
- From:
Chinese Journal of Modern Nursing
2024;30(28):3893-3900
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To systematically evaluate the bias risk and applicability of short-term mortality risk prediction models within 30 days after stroke, providing a basis for selecting or developing standardized risk prediction models.Methods:Research on short-term mortality risk prediction models within 30 days after stroke was electronically retrieved from China National Knowledge Infrastructure, WanFang Data, VIP, and China Biomedical Database, PubMed, Web of Science, Embase, Cochrane Library and CINAHL. The search period was from database establishment to December 5, 2023. Two researchers independently conducted literature screening and quality evaluation.Results:Twelve studies were included, and a total of 31 models were internally validated, with 7 models undergoing external validation based on internal validation. 26 models reported discriminative power, and 18 models reported calibration methods. The most frequent predictors of modeling were age, hypertension, atrial fibrillation, diabetes and admission Glasgow Coma Scale score. Due to methodological problems such as insufficient sample size, improper handling of missing variables, and inadequate reporting of modeling information, all included studies were rated as high risk of bias.Conclusions:The research on short-term mortality risk prediction models for stroke patients is still in the development stage. Although it has good applicability, the risk of bias is relatively high. Future research should be designed and reported based on prediction model risk of bias assessment tool (PROBAST) and transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD) to avoid common problems summarized in this study and reduce the risk of bias.
