TOPLINE:

Electronic health records (EHRs) and retail kiosk data from millions of participants revealed a national prevalence of obesity of more than 40%, with substantial geographic variation across the US. These findings demonstrated that alternative data sources could support near‑real‑time population health monitoring, offering value given declining response rates to traditional surveys.

METHODOLOGY:

• Monitoring of overweight and obesity has relied on national surveys such as the National Health and Nutrition Examination Survey and the Behavioral Risk Factor Surveillance System; however, these surveys have limited geographic resolution, declining response rates, and 2- to 3-year lags; therefore, alternative data sources should be evaluated for population-level monitoring.
• Researchers analyzed cross-sectional data from 85,186,060 nonpregnant adult patients in EHRs and 1,180,273 retail kiosk users between January 2024 and July 2025 to estimate the prevalence of overweight (BMI, 25 to < 30) or obesity (BMI > 30).
• BMI was calculated from measured height and weight in EHRs and from self-reported height with measured weight at retail kiosks.
• Information on individual demographics (age groups, 18-19 years, 20-44 years, 45-64 years, and ≥ 65 years; sex; and race/ethnicity), area-level variables (median household income, percent unemployed, percent without high school diploma, and percent uninsured), and urban/rural classification was obtained.
• Prevalence was estimated directly from EHRs and using multilevel regression and poststratification for kiosk data, and these estimates were compared with national survey and modeled estimates.

TAKEAWAY:

• The national prevalence of obesity was 43.7% according to EHRs and 43.3% according to kiosks, whereas the national prevalence of overweight was 39.1% and 33.6%, respectively. In the National Health and Nutrition Examination Survey, the prevalence of obesity and overweight was 39.5% and 31.7%, respectively.
• Higher obesity rates were observed among women, non-Hispanic Black adults, individuals aged 45-64 years, and residents of rural areas.
• Geographic analysis revealed a higher burden of obesity across Southern and Midwestern states than states in the West and Northeast; county-level estimates of the prevalence of overweight or obesity correlated with modeled data (Spearman rho, 0.64 for EHRs and 0.34 for kiosks).

IN PRACTICE:

“Findings from this study suggest that administrative and convenience samples can provide more real-time and local estimates of high BMI and complement national surveys, which are limited by time lags and poor spatial resolution and have declining response rates,” the authors wrote.

SOURCE:

The study was led by Jithin Sam Varghese, Emory Global Diabetes Research Center of Woodruff Health Sciences Center and Emory University in Atlanta. It was published online in Obesity.

LIMITATIONS:

Systematic selection bias may have affected the EHR sample. Additionally, information bias may have arisen from variations in reporting between healthcare systems and misreporting of self-reported height at kiosks. The final direction and magnitude of biases remain unknown.

DISCLOSURES:

The authors report no funding for this study. Two authors declared being employees of Pursuant Health, which provided free screenings of blood pressure and weight in Walmart and select CVS stores through over 4600 voluntary self-service kiosks.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication.