Social and Behavioral Factors Associated With Diabetes in Southern California vs the US.

Importance: Diabetes disproportionately impacts communities across the US and in Southern California, where there is a large Hispanic population. Diabetes prevalence continues to increase, and more studies are needed to understand behaviors and socioecological factors contributing to diabetes risk regionally and nationally. Objective: To identify social and behavioral correlates (SBC) of diabetes in Southern California and to contrast these findings with national data to help ensure that public health strategies are appropriately targeted and regionally responsive. Design, setting, and participants: This ecological and cross-sectional study used modeled estimates from census tracts in Southern California and nationwide, drawing from the Centers for Disease Control and Prevention's 2024 Population Level Analysis and Community Estimates (PLACES) dataset. Exposures: Based on PLACES data availability and prior evidence, 24 SBC were selected and were grouped into health outcomes and conditions, prevention, health risk behaviors, health-related social needs, and broader social determinants of health. Main Outcomes and Measures: Diagnosed diabetes prevalence, also from the PLACES dataset, represented the percentage of adults ever diagnosed with diabetes by a health care professional, excluding gestational diabetes. A total of 24 indicators were tested for their association with diabetes prevalence using a gradient-boosted regression tree method. Results: In Southern California (5420 census tracts composed of approximately 18.5 million adults), diagnosed diabetes had a mean (SE) prevalence of 11.29% (0.06), ranging from 1.4% (0.05) to 33.6% (1.37), compared with 11.52% (0.02) nationally (62 480 tracts composed of approximately 253 million adults). In Southern California, leveraging an explainable machine learning approach, 5 key correlates were identified accounting for 67% of the estimate in the model: physical inactivity (31%), routine check-ups (14%), binge drinking (11%), lack of insurance (6%), and food insecurity (5%). While some key correlates overlapped regionally and nationally, their estimated contributions differed. Additionally, obesity, receipt of food stamps, being aged 65 years or older, and persons of ethnic or racial minority background (all except White, non-Hispanic) were key correlates nationally but not in Southern California. Conclusion: In this cross-sectional study, key correlates of diagnosed diabetes in Southern California included physical inactivity, access to health care, and food insecurity, while nationally, obesity, food stamp participation, older age (≥65 years), and racial or ethnic minority status were also key. These findings point to possible regional differences in the factors linked to diabetes prevalence and highlight the need for further research to determine their significance and potential for guiding targeted interventions.