Development and validation of a risk screening model for frailty in older patients with diabetic foot

To determine the risk factors of frailty in older diabetic foot (DF) patients and then construct a validated frail nomogram. Data were performed on 552 older DF patients in two hospitals in Tianjin, China. The logistic regression analysis was applied to explore the risk factors of frailty and incorporate these into the construction of a frail nomogram in older DF patients. The model was internally validated using the receiver operating characteristic curve and area under the curve (AUC), calibration curves, Hosmer-Lemeshow test, and decision curve analysis (DCA). Place of residence, regular exercise, duration of diabetic foot, complicated peripheral neuropathy, complicated peripheral vascular disease, multiple toe amputations, abnormal foot skin temperature, fasting blood glucose, sleep quality, and depression were the risk factors of frailty in older DF patients. The nomogram model showed good internal validation results. The AUC values of the modeling and validation groups were 0.923 (95% CI = 0.896 ~ 0.950) and 0.963 (95% CI = 0.926 ~ 0.988), respectively. The Hosmer-Lemeshow test values were P = 0.114 and P = 0.128. Calibration curves showed high agreement between the predicted and actual values. DCA indicated good clinical validity. Our screening model can help clinical staff screen older DF patients at high risk of frailty and provide a theoretical basis for developing early prevention.