Risk prediction models for hospitalization-associated disability among older adults: A systematic review and meta-analysis

BACKGROUND: Hospitalization-associated disability is a significant adverse outcome in older adults, leading to prolonged hospital stays and increased healthcare burden. Although several risk prediction models have been developed to identify hospitalization-associated disability, their quality, risk of bias and clinical applicability remain unclear. OBJECTIVES: To systematic review existing evidence on risk prediction models for hospitalization-associated disability among older adults. METHODS: Eight databases were searched from inception to October 15, 2025. Data from the included studies were extracted based on the CHARMS. The PROBAST+AI was used to assess quality, risk of bias, and applicability of the included models. RESULTS: A total of 18 prediction models from 16 studies ultimately included in this review. All models were developed using logistic regression, with age, cognitive function, instrumental activities of daily living, and nutritional status being the most frequent predictors. Discrimination, assessed by the pooled AUC/C-statistic/C-index, 0.76 (95 % CI: 0.76-0.77) for model development, 0.77 (95 % CI: 0.74-0.80) for internal validation, 0.71 (95 % CI: 0.68-0.74) for external validation. Calibration was reported in 81.8 % of development models, 20 % of internal validations, and 38.5 % of external validations, with all results demonstrating favorable calibration except for 1 model. All studies had low quality or high risk of bias. CONCLUSIONS: Although most of the included models demonstrated moderate discrimination, significant methodological limitations and inadequate validation were identified. Future research should focus on enhancing the quality in model development, mitigating risk of bias in model validation, and improving clinical applicability.