Schnabel · JAMA internal medicine 2019 · prospective cohort study · n=44551

Association Between Ultraprocessed Food Consumption and Risk of Mortality Among Middle-aged Adults in France.

Cited 394 times in the scientific literature.

Level 3 - non-randomized controlled study

Prospective cohort study

PubMed 30742202 · doi:10.1001/jamainternmed.2018.7289 · record verified 2026-08-29

What was done

This prospective observational cohort study evaluated 44,551 French adults aged 45 years or older from the NutriNet-Santé study followed for a median of 7.1 years (2009–2017). Baseline dietary intake was assessed using at least one set of three web-based 24-hour dietary records over the first two years of follow-up. Foods were categorized using the NOVA classification system to determine the proportion (by weight) of ultraprocessed foods in the diet. All-cause mortality was tracked via the French national mortality registry (CépiDC), and multivariable Cox proportional hazards regression models adjusted for sociodemographic, anthropometric, and lifestyle factors.

What was found

Ultraprocessed foods accounted for a mean (SD) of 14.4% (7.6%) of food weight and 29.1% (10.9%) of total energy intake. Higher consumption was significantly associated with younger age (45–64 years: 14.50% vs older; P < .001), lower income (<€1200/month: 15.58%; P < .001), lower educational level (no diploma/primary: 15.50%; P < .001), living alone (15.02%; P < .001), BMI ≥30 (15.98%; P < .001), and lower physical activity (15.56%; P < .001). Over follow-up, 602 deaths (1.4%) occurred. After multivariable adjustment, each 10% increment in the proportion of ultraprocessed food consumption was associated with a 14% increase in all-cause mortality risk (HR 1.14; 95% CI, 1.04–1.27; P = .008).

Why it matters

This study provides early prospective epidemiological evidence linking higher consumption of ultraprocessed foods directly to increased all-cause mortality risk in middle-aged adults, supporting dietary guidelines that recommend limiting industrial food processing.

Limits

Dietary intake relied on self-reported web-based records, introducing potential measurement error. The cohort predominantly comprised women (73.1%) and health-conscious volunteers, limiting generalizability. The total number of deaths was relatively small (602 events, 1.4% mortality rate), and residual confounding from unmeasured socioeconomic or lifestyle variables cannot be ruled out.

Cited by