AUTHOR=Xu Chaoran , Xiang Lijun , Luo Yansi , He Li , Tai Liwen , Liu Yaman , He Kaixin , Du Min , Zhang Xiaomei 

TITLE=Risk prediction models for discharge disposition in patients with stroke: a systematic review and meta-analysis

JOURNAL=Frontiers in Neurology

VOLUME=Volume 16 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/neurology/articles/10.3389/fneur.2025.1637606

DOI=10.3389/fneur.2025.1637606

ISSN=1664-2295

ABSTRACT=AimsMultivariate prediction models can be used to estimate the risk of discharged stroke patients needing a higher level of care. To determine the model’s performance, a systematic evaluation and meta-analysis were performed.MethodsChina National Knowledge Infrastructure (CNKI), Wanfang Database, China Science and Technology Journal Database (VIP), SinoMed, PubMed, Web of Science, CINAHL, and Embase were searched from inception to September 30, 2024. Multiple reviewers independently conducted screening and data extraction. The Prediction Model Risk of Bias Assessment Tool (PROBAST) checklist was used to assess the risk of bias and applicability. All statistical analyses were conducted in Stata 17.0.ResultsA total of 4,059 studies were retrieved, and after the selection process, 14 studies included 22 models were included in this review. The incidence of non-home discharge in stroke patients ranged from 15 to 84.9%. The most frequently used predictors were age, the National Institutes of Health Stroke Scale (NIHSS) score at admission, the Functional Independence Measure (FIM) cognitive function score, and the FIM motor function score. The reported area under the curve (AUC) ranged from 0.75 to 0.95. Quality appraisal was performed. All studies were found to have a high risk of bias, mainly attributable to unsuitable data sources and inadequate reporting of the analytical domain. All statistical analyses were conducted in Stata 17.0. In the meta-analysis, the area under the curve (AUC) value for the five validation models was 0.80 [95%CI (0.75–0.86)].ConclusionResearch on risk prediction models for stroke patient discharge disposition is still in its initial stages, with a high overall risk of bias and a lack of clinical application, but the model has good predictive performance. Future research should focus on developing highly interpretive, high-performance, easy-to-use machine learning models, enhancing external validation, and driving clinical applications.Systematic review registrationhttps://www.crd.york.ac.uk/PROSPERO/, CRD42024576996.