# Spatial bayesian modeling of diabetes mellitus (DM) risk in the United States

## TL;DR

This study maps diabetes risk across the US, finding high risk in the South and identifying Florida and Maine as needing attention.

## Contribution

The study identifies new high-risk regions for diabetes mellitus using spatial Bayesian modeling.

## Key findings

- Diabetes risk is non-randomly distributed, with elevated risk concentrated in the US South.
- Smoking and obesity significantly increase diabetes risk, while temperature and pollution factors do not.
- Florida and Maine are identified as high-risk areas requiring targeted interventions.

## Abstract

Spatial targeting of disease burden has long been instrumental to achieving optimal health resource allocation in the United States (US). Building on a previous study, the current research examined the spatial variation and hotspots of Diabetes Mellitus (DM) risk across the US.

The outcome variable was the number of DM cases per county acquired from the CDC-Behavioral Risk Factor Surveillance System (BRFSS). Air pollution data (PM2.5 & NO2) were acquired from the Socioeconomic Data and Application Centre (SEDAC), obesity cases from the BRFSS, and maximum temperature data from WorldClim. The percentage of alcoholics and smokers was obtained from the County Health Rankings & Roadmaps Association. We model the spatial pattern of DM risks by coupling the integrated nested Laplace approximation (INLA) and the Local Indicator of Spatial Autocorrelation (LISA).

The spatial distribution of diabetes mellitus (DM) risk was non-random (I = 0.428; p < 0.001). The elevated risk was concentrated in the US South, where multiple counties were classified within the highest standard deviation band, at or above 2.5 standard deviations above the national expected risk (RR = 1). After adjusting for spatial confounding, smoking increased the risk of DM by 8.3% (RR = 1.083, 95% CI: 1.07, 1.095), and obesity also increased DM risk by 0.7% (RR = 1.007; 95% CI: 1.000, 1.014). However, maximum temperature (RR = 0.998; 95% CI: 0.982, 1.013), NO2 concentration (RR = 0.996; 95% CI: 0.984, 1.008), Social Vulnerability Index (RR = 0.99; 95% CI: 0.98, 1.001), and PM2.5 concentration (RR = 0.993; 95% CI: 0.97, 1.017) were not statistically significantly related to DM risk.

The US South was found to be the region most at risk for the DM burden. Unlike previous research, our study also identified Florida and Maine as areas requiring equal attention. Although cross-sectional, the findings from this study provide essential information for targeted public health interventions.

Not applicable.

The online version contains supplementary material available at 10.1186/s12889-025-25622-8.