Ultra-Processed Food Intake and Mortality in the PLCO Cohort

Key Takeaways

• Higher ultra-processed food intake was associated with modestly higher all-cause mortality in this cohort.
• Higher risks were also observed for circulatory system, nervous system, and other-cause mortality, while cancer mortality was not associated.
• In joint analysis with diet quality, no consistent association was observed between ultra-processed food intake and mortality within most Healthy Eating Index-2015 quarters, although among participants in the highest quarter of ultra-processed food intake, those with the highest Healthy Eating Index-2015 scores had lower mortality risk than those with the lowest scores. The all-cause association was stronger at higher BMI, and sensitivity analyses were generally consistent with the main result.

In the PLCO cohort, participants in the highest quarter of ultra-processed food intake had higher all-cause mortality over a median 17 years of follow-up, with an adjusted hazard ratio of 1.10 versus the lowest quarter. The analysis compared quarters of intake rather than testing an intervention, placing the association within a large, long-running US cohort. The finding was observational rather than causal.

This population-based cohort analysis was conducted within the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial, which enrolled adults aged 55 to 74 years. Baseline data were collected from 1993 to 2001, and outcomes were ascertained through December 31, 2009. Dietary intake was assessed at baseline with the Baseline Questionnaire and validated Dietary History Questionnaire. Ultra-processed food intake was defined using the NOVA classification and quantified as the percentage of total daily food and beverage weight. All-cause mortality was the primary outcome; cancer, circulatory system disease, nervous system disease, and other-cause mortality were secondary outcomes. Over follow-up, 24,237 deaths were recorded in the cohort.

Adjusted comparisons between the highest and lowest intake quarters showed positive associations for several cause-specific outcomes. Hazard ratios were 1.09 for circulatory system disease mortality (95% CI, 1.02 to 1.17), 1.20 for nervous system disease mortality (95% CI, 1.06 to 1.37), and 1.28 for other-cause mortality (95% CI, 1.18 to 1.39). Cancer mortality was not associated with ultra-processed food intake. Trend tests were significant for the positive associations, and restricted cubic spline analyses did not support nonlinearity, consistent with a broadly linear pattern across the intake range studied.

In joint analyses with the Healthy Eating Index-2015, no consistent association was observed between ultra-processed food intake and mortality within most Healthy Eating Index-2015 quarters. Among participants in the highest quarter of ultra-processed food intake, those with the highest Healthy Eating Index-2015 scores had lower mortality risk than those with the lowest scores. Subgroup analyses suggested that the all-cause association was stronger among participants with higher BMI. Sensitivity analyses generally supported the main finding, and alternative ultra-processed food measurement approaches produced similar results. These secondary findings aligned with the main analysis but remained exploratory.

The authors noted several limitations that could affect interpretation. Dietary intake was measured only at baseline, so later changes during follow-up were not captured. Residual confounding remained possible despite multivariable adjustment, and applicability to other populations or ethnic groups was uncertain. Within this cohort, higher ultra-processed food intake was associated with modestly higher mortality overall, with the pattern differing across causes of death.