Divergent Socioeconomic Pathways to Biologically Uncontrolled Diabetes by Gender: A Bayesian Analysis of NHANES 2021–2023

Abstract

Standard health evaluations often treat gender as a mere covariate, masking distinct socioeconomic pathways to disease. We applied a gender-stratified Bayesian framework to evaluate how socioeconomic indicators relate to biologically uncontrolled diabetes risk in the volatile post-pandemic context. Using NHANES 2021–2023 data (N = 5,995), we conducted a Bayesian multilevel logistic regression utilizing targeted prior regularization and partial pooling to stabilize estimates for underrepresented subgroups. This approach provides a rigorous assessment of structural factors including income, education, employment, and health insurance. Our analysis identified a sharp gender divide. While educational attainment serves as a universal buffer, structural drivers differ fundamentally by gender. For men, labor market status acts as a high-fidelity signal of functional reserve; active job seekers show significantly lower odds of uncontrolled diabetes (OR = 0.26, 95% CrI: 0.07–0.70), reflecting health selection at re-entry. Conversely, uninsured women show a lower diagnostic likelihood (OR = 0.47, 95% CrI: 0.24–0.87), reflecting structural invisibility rather than metabolic protection. Gender-blind models miss critical structural vulnerabilities. Translating these findings into actionable guidance requires a paradigm shift in screening design.