DeCleene · JAMA 2026 · Bayesian spatiotemporal modeling and forecasting study · n=11315421

US State-Level Prevalence of Adult Obesity by Race and Ethnicity From 1990 to 2022 and Forecasted to 2035.

Level 3 - non-randomized controlled study

Epidemiological modeling and forecasting using repeated cross-sectional population survey data

PubMed 41604179 · doi:10.1001/jama.2025.26817 · record verified 2026-08-26

What was done

Researchers modeled US adult (≥20 years) obesity prevalence (BMI ≥30) from 1990 to 2022 and forecasted trends through 2035 across states, sex, age groups, and three racial and ethnic groups (Hispanic, non-Hispanic Black, non-Hispanic White). The analysis integrated measured BMI from the National Health and Nutrition Examination Survey (NHANES) and bias-corrected self-reported BMI from the Behavioral Risk Factor Surveillance System (BRFSS) and Gallup Daily Survey (total n = 11,315,421). Statistical methods included spatiotemporal Gaussian process regression, annualized rate of change models, and meta-regression Bayesian spline models.

What was found

US adult obesity prevalence increased from 19.3% (95% UI, 17.3%-21.3%; 34.7 million individuals) in 1990 to 42.5% (95% UI, 40.2%-45.0%; 107 million individuals) in 2022, and is projected to reach 46.9% (95% UI, 43.9%-49.9%; 126 million individuals) by 2035. In 2022, age-standardized prevalence ranged from 40.1% (95% UI, 37.8%-42.5%) in non-Hispanic White males to 56.9% (95% UI, 54.1%-59.9%) in non-Hispanic Black females. Obesity was highest in Midwestern and Southern states and among middle-aged adults, with large increases seen among the youngest adult age groups, particularly females.

Why it matters

This study provides comprehensive, state-level historical estimates and future projections demonstrating that US adult obesity has more than doubled since 1990 and is expected to continue rising across all major demographic groups without substantial intervention.

Limits

The dataset relies heavily on self-reported height and weight that required mathematical bias-correction algorithms. Results are reported for only three broad racial and ethnic groups, leaving out other populations. Forecasts through 2035 are model-based projections that may not account for recent or future shifts in medical interventions, such as anti-obesity pharmacotherapies.