All of US BMI and Obesity Trends Across Pandemic Periods
Key Takeaways
- US adults with repeated BMI records in the All of Us program had a modest peak rise in mean BMI followed by a later decline.
- Obesity prevalence followed the same pattern, falling below its earlier level afterward.
- Adjusted BMI and obesity odds were higher among female, Black/African American, and Hispanic/Latino participants than among male, White, and non-Hispanic/Latino participants, respectively, and were lower with higher household income.
- Age and employment most consistently modified changes over time for both outcomes.
Investigators retrospectively examined US adults aged at least 18 in All of Us Controlled Tier electronic health record (EHR) data for their analysis of BMI trends. The primary cohort included 38,632 adults with at least one BMI measurement in each study period.
The pre-pandemic period ran from January 1, 2018, through March 12, 2020; the pandemic peak from March 13, 2020, through December 31, 2021; and the later period from January 1, 2022, through October 31, 2023.
Annual BMI was averaged from EHR-recorded height and weight; obesity was defined by an average annual BMI of at least 30 kg/m² rather than diagnosis codes. Repeated-measures linear mixed-effects BMI models and generalized estimating equations (GEE) logit obesity models adjusted for age, sex, race, ethnicity, household income, and employment and tested interactions with time. The cohort was predominantly female, White, and non-Hispanic/Latino. An expanded sensitivity cohort of 83,698 adults required BMI measurements in at least two periods.
Mean BMI (standard deviation) was 30.05 (6.67) before the pandemic, 30.14 (6.76) at the pandemic peak, and 29.96 (6.66) in the later period. These descriptive period means are not individual weight-change estimates.
Obesity prevalence was 43.1% before the pandemic, 43.4% at the peak, and 42.8% afterward. The later prevalence was below its earlier level.
For both outcomes, the small raw rise at the peak was followed by a larger decline, leaving later values below pre-pandemic levels. Adjusted continuous-time terms were positive and quadratic terms negative for both outcomes; these modeled patterns are not raw period-to-period differences. Across periods, female versus male, Black/African American versus White, and Hispanic/Latino versus non-Hispanic/Latino participants had higher adjusted BMI and obesity odds, whereas both outcomes were lower with higher household income. Adults aged 50 or older versus younger adults, and unemployed versus employed adults, had smaller modeled changes over time for both outcomes. Additional BMI time interactions involved sex, Black/African American race, and ethnicity. Findings in the expanded cohort were broadly similar, although some individual associations and interactions changed statistical significance.
All of Us participants are not nationally representative, and requiring measurements in every period may favor adults with continuing healthcare contact. EHR measurement and encounter frequency can vary; BMI does not capture body composition, and income and employment reflected baseline information. Physical activity, stress, infection, obesity medication use, and bariatric surgery were not reliably available; nationwide period boundaries also mask state-level variation.
These observational associations cannot show that pandemic restrictions caused weight changes or that sociodemographic differences reflect biological mechanisms. The authors concluded that changes were modest, while adjusted differences by sex, race, ethnicity, and income remained evident in this US cohort.
Clinician Questions
How were 2020 BMI measurements divided between pre-pandemic and pandemic-peak periods in All of Us?
For adult BMI measurements in All of Us EHR data, investigators assigned observations before March 13, 2020, to the pre-pandemic interval and those on or after that date to the pandemic-peak interval; they did not combine measurements across that boundary into a single annual average.
Did obesity prevalence follow different trajectories for All of Us adults younger than 50 and those 50 or older?
In descriptive All of Us adult EHR data, obesity prevalence rose at the pandemic peak and remained slightly higher afterward among adults younger than 50, whereas it declined steadily across the periods among adults 50 or older. These patterns do not establish that age or the pandemic caused the changes.
Were class I, II, and III obesity trajectories evaluated separately across the All of Us pandemic periods?
Investigators classified obesity among All of Us adults as one binary outcome based on average annual BMI. They identified separate trajectories for class I, II, and III obesity as a question for future research rather than reporting class-specific longitudinal findings.
Recommended Reading
- For population BMI patterns linked to health: Food-Label Reading Linked to Higher Cardiovascular Health Scores
This content is intended for healthcare professionals for educational purposes. The article and its image were developed with AI assistance and reviewed by members of the ReachMD team prior to publication. Please refer to the original source linked for the complete and most up-to-date information.