BMC Geriatrics · Nov 16, 2026
Identification and prediction of frailty in elderly inpatients with diabetes based on Bayesian network analysis: a multicenter cross-sectional study
By Peng S, Xu J, Guo R et al.
Background Frailty, the third major complication of diabetes, is dynamic and reversible. Early identification and targeted intervention can effectively prevent or delay its progression. However, current research insufficiently explores diabetes-specific indicators and struggles to describe the multi-level interrelations among multiple factors, limiting the exploration of potential intervention targets. Aims To describe frailty prevalence in elderly inpatients with diabetes, and to identify factors influencing frailty as well as explore latent relationships among these factors using Bayesian network analysis. Methods From April to December 2024, this study recruited elderly inpatients with diabetes (age ≥ 60 years) from four public general hospitals in Hunan, China. Data were collected via structured questionnaires. The Least Absolute Shrinkage and Selection Operator (LASSO) and multivariate logistic regression analysis were used to identify factors influencing frailty. Bayesian network analysis using the max-min hill-climbing (MMHC) algorithm was conducted to explore potential relationships among factors. Results Among 427 elderly inpatients with diabetes, 146 (34.19%) exhibited frailty. Age ≥ 75 years, number of comorbidities ≥ 3, fasting blood glucose (FBG) > 10.0 mmol/L, and malnutrition/at risk were directly related to frailty. Hypoglycemia occurrence and depression were associated with frailty indirectly through malnutrition/at risk. Personal monthly income was associated with frailty indirectly through the number of comorbidities, hypoglycemia occurrence, and malnutrition/at risk. Conclusion The high prevalence of frailty among elderly inpatients with diabetes warrants clinical attention. Incorporating frailty screening into the routine clinical assessment system for diabetes may be necessary. Future studies could further explore multidimensional management strategies targeting modifiable factors such as nutrition, comorbidities, and glycemic control to prevent or ameliorate frailty. Clinical trial number: Not applicable.