The U.S. diabetes belt and factors explaining the excess risk: Multifactorial modeling and machine learning analysis.

Abstract

Research on the epidemiology of diabetes mellitus (DM) has identified a geographically distinct region in the United States (U.S.) known as the diabetes belt (DM Belt), which represents a significant public health concern. This study aimed to examine the factors contributing to the increased risk of DM in the DM Belt compared to the non-DM Belt. Data were analyzed from 398,243 adults aged ≥ 18 years who participated in the 2019 Behavior Risk Factor Surveillance System. DM status was based on participants' self-reported physician-diagnosed DM. The DM Belt was defined at the state level according to the U.S. Center for Disease Control and Prevention's classification. Logistic regression (LR) was used to estimate odds ratios for DM and assess the excess DM risk in the DM Belt versus the non-DM Belt. Random Forest (RF) and stepwise LR were employed to identify and rank key contributors to the excess DM risk. Residents of the DM Belt had a significantly higher prevalence of DM than those in the non-DM Belt (age-sex-adjusted rate: 12.5 % versus 10.5 %, p < 0.001). Low socioeconomic status (SES), physical inactivity, and hypertension were identified as the top three factors explaining the excess DM risk in the DM Belt. These findings underscore the importance of an integrated approach to improving SES, promoting healthy behaviors, managing chronic conditions for reducing DM risk. Addressing these factors can help mitigate health disparities in DM risk across the U.S.