Racial and Ethnic Differences in Diabetes Treatment Modality Selection: The Role of Income and Sociodemographic Factors in NHANES 2017–2023, A Cross-Sectional Study

Racial and ethnic disparities in diabetes prevalence and outcomes are well documented; however less is known about whether the selection of treatment modality itself differs across racial groups after adjustment for clinical correlates of disease severity. This cross-sectional analysis used National Health and Nutrition Examination Survey (NHANES) data from the 2017-March 2020 and August 2021-August 2023 cycles. The primary analytical sample comprised 1,688 adults aged 18 years and older with physician-confirmed diabetes, after excluding 46 probable Type 1 cases. Treatment modality was categorized as insulin-only, oral medication-only (base outcome), combination therapy, or no medication, based on self-reported use of insulin and oral antidiabetic medications. Survey-weighted multinomial logistic regression adjusted for age, gender, education, income, birthplace, insurance, HbA1c, body mass index (BMI), and self-reported diabetes duration. Race and ethnicity remained a significant overall predictor of treatment modality after full adjustment (joint Wald p = 0.012). Other/Multi-Racial adults had 57% lower relative risk of insulin-only therapy than Non-Hispanic White (NHW) adults (relative risk ratio [RRR] = 0.43, 95% CI: 0.24-0.76, p = 0.005), and Other Hispanic and Other/Multi-Racial adults had lower relative risk of combination therapy (RRR = 0.56, p = 0.034 and RRR = 0.47, p = 0.043, respectively). Non-Hispanic Black (NHB) adults did not differ from NHWs at the population level. HbA1c, diabetes duration, BMI, and insurance status were the strongest predictors of treatment modality. An exploratory race-by-income interaction model produced a non-significant joint test (p = 0.259) and is reported as hypothesis-generating. Differences in modality use persist after adjustments, suggesting that structural and healthcare-system factors may contribute to treatment variation independently of measured clinical and socioeconomic characteristics.