Predicting frailty in diabetic polyneuropathy: psychosocial and lifestyle factors outweigh traditional clinical predictors

Aims: To develop and validate a predictive model for frailty in older adults with diabetic polyneuropathy (DPN), integrating psychosocial and lifestyle-related factors beyond traditional clinical characteristics.

Methods: This cross-sectional study included 400 older individuals with clinically diagnosed DPN from a tertiary hospital in China. Frailty status was assessed using the Tilburg Frailty Indicator. A broad range of variables were collected, including demographic data, diabetes-related clinical characteristics (disease duration, complications, comorbidities), levels of physical activity, sleep quality, nutritional status, and psychosocial measures. Multivariable logistic regression and ROC-curves with decision curve analysis were used to identify the strongest predictors of frailty and to construct an optimized prediction model.

Results: Frailty prevalence was 28 % among individuals with DPN. Importantly, the strongest predictors of frailty were not traditional diabetes-related clinical factors such as diabetes duration, number of complications, or comorbidity burden. Instead, psychosocial factors, lower physical activity levels, poor sleep quality, and impaired nutritional status contributed most to the predictive model. The final multifactorial model demonstrated good discriminatory ability for identifying frail individuals (AUC 0.89), while traditional clinical variables did not contribute significantly and were not retained in the final model.

Conclusions: Frailty in individuals with DPN was best predicted by psychosocial and lifestyle-related factors rather than conventional markers of diabetes severity. These findings suggest that frailty risk assessment in DPN should extend beyond routine clinical parameters and incorporate multidimensional lifestyle and psychosocial screening.

Comments: This work provides clinically relevant insight into the multifactorial nature of frailty in DPN, an area that has received relatively limited attention compared with frailty research in other chronic disease populations. By demonstrating the added predictive value of behavioural and psychosocial determinants of health beyond classical diabetes-related variables, the study reinforces the need for holistic and interdisciplinary management strategies. This holistic approach to frailty prediction may be particularly suitable for elderly individuals with DPN, in whom chronic pain and sensorimotor deficits are frequently accompanied by depression, impaired sleep, fear of falling, increased sedentary behaviour, and social isolation. As such, this study provides valuable knowledge on frailty risk factors in DPN, an area currently lacking robust models, as existing approaches often focus predominantly on physical aspects and diabetes severity related factors while overlooking the broader multifactorial consequences of DPN. The study’s main limitation is its cross-sectional design, which precludes causal inference. Additionally, the findings require validation in other populations. Longitudinal studies are needed to determine whether modifying these factors can reduce frailty progression. In this context, there remains a clear lack of multifactorial intervention studies targeting both falls and frailty in individuals with DPN.

Anders Stouge

Reference. Xie X, Huang Y, Wang Y, Chen W, Liang X, Xiong C, Zou X. Frailty Prediction Model for Elderly Diabetic Peripheral Neuropathy Patients. Diabetes Metab Syndr Obes. 2025 Dec 22;18:4683-4697. doi: 10.2147/DMSO.S570083. PMID: 41459327; PMCID: PMC12742580.

🔗 https://www.dovepress.com/frailty-prediction-model-for-elderly-diabetic-peripheral-neuropathy-pa-peer-reviewed-fulltext-article-DMSO