Objective:To systematically evaluate the risk prediction models for refeeding syndrome in critically ill patients in China and abroad, with the aim of providing feasible risk assessment tools for clinical healthcare professionals.Methods:A computerized search of PubMed, Embase, Web of Science, Cochrane Library, CINAHL, Proquest, Scopous, China Biomedical Literature Database, China National Knowledge Infrastructure (CNKI), Wanfang, and VIP was conducted for relevant literature from database inception to July 1, 2024. Two researchers independently reviewed the literature and extracted the relevant information. Quality assessment was performed using a risk of bias assessment tool specifically designed for predictive modeling.Results:A total of 8 papers were included, reporting 8 risk prediction models for refeeding syndrome in critically ill patients, with sample sizes ranging from 109 to 806 cases, outcome event rates from 22.5% to 69.8%, and area under the curve values from 0.74 to 0.95. The most frequently reported predictors included albumin (ALB), pre-albumin (PAB), blood potassium, use of diuretics, and Acute Physiology and Chronic Health Evaluation Ⅱ (APACHE Ⅱ) scores.Conclusions:The risk prediction models for refeeding syndrome in critically ill patients have good predictive performance but are still in the development stage. A high-quality, low-bias clinically applicable model should be established through multicenter, large-sample, and standardized studies.

Objective:To systematically evaluate the risk prediction models for refeeding syndrome in critically ill patients in China and abroad, with the aim of providing feasible risk assessment tools for clinical healthcare professionals.Methods:A computerized search of PubMed, Embase, Web of Science, Cochrane Library, CINAHL, Proquest, Scopous, China Biomedical Literature Database, China National Knowledge Infrastructure (CNKI), Wanfang, and VIP was conducted for relevant literature from database inception to July 1, 2024. Two researchers independently reviewed the literature and extracted the relevant information. Quality assessment was performed using a risk of bias assessment tool specifically designed for predictive modeling.Results:A total of 8 papers were included, reporting 8 risk prediction models for refeeding syndrome in critically ill patients, with sample sizes ranging from 109 to 806 cases, outcome event rates from 22.5% to 69.8%, and area under the curve values from 0.74 to 0.95. The most frequently reported predictors included albumin (ALB), pre-albumin (PAB), blood potassium, use of diuretics, and Acute Physiology and Chronic Health Evaluation Ⅱ (APACHE Ⅱ) scores.Conclusions:The risk prediction models for refeeding syndrome in critically ill patients have good predictive performance but are still in the development stage. A high-quality, low-bias clinically applicable model should be established through multicenter, large-sample, and standardized studies.