Development and validation of prediction model for stage I patients with lower extremity atherosclerotic disease in type 2 diabetes mellitus in China

Lower extremity atherosclerotic disease (LEAD) is the primary cause of ulcers, gangrene, and amputations in patients with type 2 diabetes mellitus (T2DM), stage I is the crucial time for prevention and intervention to improve the prognosis of T2DM-LEAD. The purpose of this study was to develop and validate a personalized predictive model to determine the risk of outcomes in stage I patients with T2DM-LEAD.

There were 1603 stage I patients with T2DM-LEAD at baseline in this retrospective study. Least absolute shrinkage and selection operator regression was applied to filter predictive variables. Cox regression was used to construct a nomogram prediction model. The model’s 3-year and 5-year predictive performance was evaluated in terms of its discrimination, calibration, and clinical utility using the area under the receiver operating characteristic curve, calibration curve, decision curve analysis, respectively.

Patients were randomly divided into a development cohort (n = 1122) and a validation cohort (n = 481). Age, cerebrovascular diseases, diabetic kidney disease, diabetic retinopathy, low-density lipoprotein cholesterol, fibrinogen, D-dimer and anti-platelet drugs were selected as predictive factors. The model presented moderate discrimination in development and validation sets with AUCs of 70.3 (95% CI: 65.2-75.3) and 70.1 (95% CI: 64.5-75.7) for the 3-year prediction. Andthe AUC values for the 5-year prediction in development and validation sets were 72.8 (95% CI: 67.6-78.1) and 75.9 (95% CI: 69.0-82.8), respectively. The calibration curve for the 3-year and 5-year predictions demonstrated good agreement between the predicted and actual probabilities, and decision curve analysis showed a wide range of beneficial clinical utility.

The prediction model can identify the risk of stage I patients with T2DM-LEAD who are likely to develop outcomes events within 3 years and 5 years. It is valuable for clinical decisions and helps healthcare providers and policy makers to develop more personalized clinical treatment strategies, which has significant public health implications.