Juul · The American journal of clinical nutrition 2022 · serial cross-sectional study · n=40,937

Ultra-processed food consumption among US adults from 2001 to 2018.

Cited 384 times in the scientific literature.

Level 4 - case-series / case-control

Serial cross-sectional study of nationally representative survey data (by design analogy, Oxford CEBM Level 4).

PubMed 34647997 · doi:10.1093/ajcn/nqab305 · record verified 2026-08-26

What was done

Dietary intake data collected via 24-hour recalls from 40,937 US adults (aged >19 years) across 9 cross-sectional waves of the NHANES (2001–2002 to 2017–2018) were analyzed. Food intake was categorized using the NOVA framework into minimally processed foods, processed culinary ingredients, processed foods, and ultra-processed foods, measured as the relative contribution to daily energy intake (%kcal). Linear regression models adjusted for age, sex, race/ethnicity, education, and income evaluated linear secular trends overall and stratified by sociodemographic characteristics.

What was found

Adjusting for population characteristics, ultra-processed food consumption among all US adults increased from 53.5% to 57.0% of total calories from 2001–2002 to 2017–2018 (P-trend < 0.001), showing consistent increases across all sociodemographic subgroups except Hispanics. Conversely, minimally processed food consumption decreased from 32.7% to 27.4% of total calories (P-trend < 0.001) across all sociodemographic strata. Processed culinary ingredients increased from 3.9% to 5.4% of total calories (P-trend < 0.001), while processed food intake remained stable at approximately 10% of total calories (P-trend = 0.052).

Why it matters

This study provides national surveillance demonstrating that ultra-processed foods constitute the majority of caloric intake among US adults and continue to rise, displacing minimally processed foods across nearly all demographics.

Limits

The findings rely on self-reported 24-hour dietary recalls, which are subject to recall bias and underreporting. The repeated cross-sectional design cannot assess longitudinal dietary trajectories within individuals, and the abstract does not assess direct clinical disease outcomes.

Cited by