Different Levels of Ultraprocessed Food and Beverage Consumption and Associations with Environmental Sustainability and All-cause Mortality in EPIC-NL
Reina E. Vellinga, Iris van den Boomgaard, Jolanda M.A. Boer, Yvonne T. van der Schouw, Marjolein C. Harbers, W. M. Monique Verschuren, P. van ’t Veer, Elisabeth H. M. Temme, Sander Biesbroek
Abstract
The adverse health effects of high ultra-processed food and drink consumption (UPFD) are well documented. However, its environmental impact remains unclear and the separate effects of ultra-processed foods (UPF) and drinks (UPD) on all-cause mortality are not previously studied. To assess the association between levels of UPFD, UPF and UPD consumption with diet-related environmental impacts and all-cause mortality in Dutch adults. Habitual diets were assessed by FFQ in 1993-1997 among 38,261 participants of the EPIC-NL cohort. The mean follow-up time was 18.2 years (SD 4.1), 4,697 deaths occurred. FFQ-items were categorized according to the NOVA classification. Associations with quartiles of UPFD, UPF, and UPD consumption and environmental impact indicators were analyzed using general linear models and with all-cause mortality by Cox proportional hazard models. The lowest UPFD, UPF, UPD consumption quartiles were used as comparator. The average UPFD consumption was 181 (SD 88) g per 1000 kcal. High UPF consumption was statistically significantly inversely associated with all environmental impact indicators (Q4vsQ1: -13.6% to -3.0%) whereas high UPD consumption was, except for land use, statistically significant positively associated with all environmental impact indicators (Q4vsQ1: 5.9% to 1.2%). High UPFD consumption was heterogeneously associated with environmental impacts (Q4vsQ1: 2.6% to -4.0% ). After multivariable adjustment, the highest quartiles of UPFD and UPD consumption were significantly associated with all-cause mortality (HRQ4vsQ1 1.20, 95%CI 1.10,1.30 and HRQ4vsQ1 1.19, 95%CI 1.09,1.29, respectively). UPF consumption of Q2 and Q3 were associated with a borderline significant lower risk of all-cause mortality (HRQ2vsQ1 0.93, 95% CI 0.85,1.00; HRQ3vsQ1 0.91, 95% CI 0.84,0.99), while Q4 was not statistically significant (HRQ4vsQ1 1.05, 95% CI 0.96,1.15). Reducing UPD consumption could lower environmental impact and all-cause mortality risk, however this was not shown for UPF. When categorizing foods consumption by their degree of processing trade-offs are observed for human and planetary health aspects.