Association of ultra-processed food consumption with all cause and cause specific mortality: population based cohort study.
Level 3 - non-randomized controlled study
Prospective cohort study
PubMed 38719536 · doi:10.1136/bmj-2023-078476
What was done
Prospective cohort study following 74,563 women in the Nurses' Health Study (1984–2018) and 39,501 men in the Health Professionals Follow-up Study (1986–2018) free of baseline cancer, cardiovascular disease, and diabetes. Intake of ultra-processed foods was measured every four years via semiquantitative food frequency questionnaires. Multivariable Cox proportional hazard models assessed associations with all-cause, cancer, cardiovascular, and other-cause mortality.
What was found
Over a median follow-up of 34 years in women (30,188 deaths) and 31 years in men (18,005 deaths), participants in the highest quarter of ultra-processed food consumption had a 4% higher all-cause mortality compared with the lowest quarter (HR 1.04, 95% CI 1.01 to 1.07; 1536 vs 1472 deaths per 100,000 person years) and a 9% higher mortality from non-cancer, non-cardiovascular causes (HR 1.09, 95% CI 1.05 to 1.13). No associations were observed for cancer or cardiovascular mortality. Subgroup analysis showed stronger mortality associations for meat/poultry/seafood ready-to-eat products (HRs 1.06 to 1.43), sugar-sweetened and artificially sweetened beverages (HR 1.09, 95% CI 1.07 to 1.12), dairy-based desserts (HR 1.07, 95% CI 1.04 to 1.10), and ultra-processed breakfast foods (HR 1.04, 95% CI 1.02 to 1.07). No consistent associations between ultra-processed foods and mortality were seen within dietary quality quarters assessed by the Alternative Healthy Eating Index-2010 score.
Why it matters
This study shows that overall mortality risk associated with ultra-processed foods is modest and largely driven by specific subgroups like processed meats and sweetened drinks, highlighting that overall dietary quality remains a critical factor.
Limits
The observational design cannot establish causality and is subject to residual confounding. Dietary intake was self-reported through food frequency questionnaires, introducing potential misclassification. The study population comprised US health professionals, which may limit generalizability to more socioeconomically and demographically diverse groups.
Cited by
- context The average American consumes 7.5 servings of ultra-processed foods per day.