AUTHOR=Yin Jie , Xu Yiyong , Cai Mian , Fang Xiwei 

TITLE=Risk prediction models for sarcopenia in elderly people: a systematic review and meta-analysis

JOURNAL=Frontiers in Medicine

VOLUME=Volume 12 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/medicine/articles/10.3389/fmed.2025.1589583

DOI=10.3389/fmed.2025.1589583

ISSN=2296-858X

ABSTRACT=ObjectivesThis study aims to systematically review and evaluate risk prediction models for sarcopenia in older adults. The goal is to offer a reference for clinicians in selecting or developing suitable sarcopenia risk prediction models for the elderly.MethodsA systematic search was performed across CNKI, Wanfang Database, VIP Database, SinoMed, Embase, PubMed, Web of Science, and Cochrane Library for studies on risk prediction models of sarcopenia in older adults. The time frame for the search was from the creation of these databases to 13 August 2024. The literature was independently vetted by two researchers, who also gathered data and assessed the included studies’ applicability and bias risk.ResultsA total of 29 studies with 70 sarcopenia prediction models were included, with a total sample size of 140,386 and 13,472 sarcopenia events. Frequently reported independent predictors in multivariate models included BMI, age, and gender. Meta-analysis showed a combined AUC of 0.9125 [95% CI (0.9254–0.8996)], indicating good overall model predictive performance. Issues in modeling included inappropriate predictive factor screening methods, insufficient sample sizes, and lack of external validation, resulting in high study bias risk and limited model generalizability.ConclusionCurrent elderly sarcopenia risk prediction models have considerable room for improvement in overall quality and applicability. Future modeling should follow PROBAST guidelines to reduce bias risk, incorporate predictive factors with theoretical foundation and clinical significance, and strengthen external validation.Systematic review registrationhttps://www.crd.york.ac.uk/PROSPERO/Diew/CRD42025636116, identifier CRD42025636116.