Risk assessment modeling for diabetes-related lower extremity amputation using socioeconomic and behavioral health indicators

Abstract

Lower extremity amputation (LEA) is a costly complication of diabetes, with disproportionate impacts among socially and medically vulnerable populations. While clinical predictors of LEA are well established, the role of behavioral health comorbidities, particularly alongside socioeconomic disparities, in amputation risk is unexplored. To evaluate sociodemographic and behavioral health predictors of LEA, a retrospective analysis of adult patients with diabetes was performed utilizing the Nationwide Inpatient Sample from 2018 to 2022. Sociodemographic and behavioral health characteristics were compared between patients that did and did not undergo LEA. Multivariable logistic regression was used to identify independent predictors of LEA using 2018–2021 data and was then tested on the 2022 validation cohort. Among 7.5 million adult patients with diabetes with inpatient events, patients undergoing LEA were significantly younger, more likely to be male, from lower income quartiles, and disproportionately Black, Hispanic, and Native American (p < 0.001). Substance use disorder was significantly more common among LEA patients (16.0 % vs 11.6 %). In adjusted analysis, several behavioral health conditions were independently associated with increased LEA risk. The model demonstrated modest predictive performance in the testing cohort (AUC = 0.658). While not yet suitable for standalone clinical use, this model highlights that behavioral risk factors, especially in the context of sociodemographic characteristics, warrants increased attention in limb preservation strategies.