Association of 37 markers of ultra-processing with all-cause mortality: a prospective cohort study in the UK Biobank.

Background: In contrast to ultra-processed foods (UPFs), the association between specific markers of ultra-processing (MUPs) and all-cause mortality remains unexplored. As MUPs such as flavour, emulsifier, and sweetener may contribute differently to health outcomes, evaluating their individual associations with mortality is crucial. Methods: This population-based, prospective, exploratory cohort study included 186,744 UK Biobank participants aged 40 to 75, recruited between 2006 and 2010, each of whom completed at least one Oxford WebQ. To assess intake of UPFs and specific MUPs, up to ten matching commercial products were researched for each Oxford WebQ item and analysed concerning 57 previously described MUPs grouped into nine MUP categories. Of these, 37 specific MUPs were identified in one or more products. A food item was regarded as UPF if it contained at least one MUP. Cox proportional hazard regression models for all-cause mortality were applied, incorporating gram intake from UPFs, as well as from foods containing each of the nine MUP categories and 37 specific MUPs, in relation to gram total food intake (expressed as percentage of total food intake (%TFI)), using penalised cubic splines. The hazard ratio (HR)-nadir was defined as the lowest HR on the %TFI axis and rescaled to 1. Findings: Over a mean follow-up of 11.0 years, 10,203 deaths occurred. Five MUP categories were significantly (linear P < 0.05) related to all-cause mortality and the following HRs (confidence interval (CI)) were observed relative to the HR-nadir: flavour (40% versus 10 %TFI (HR-nadir); HR (CI) 1.20 (1.08 to 1.33); P < 0.0011), flavour enhancer (2% versus 0 %TFI; 1.07 (1.00 to 1.16); P = 0.025), colouring agent (20% versus 3 %TFI; 1.24 (1.10 to 1.39); P < 0.0001), sweetener (20% versus 0 %TFI; 1.14 (1.06 to 1.23); P < 0.0043), and varieties of sugar (10% versus 4 %TFI; 1.10 (1.03 to 1.16); P = 0.025). Furthermore, 13 specific MUPs were significantly associated with all-cause mortality (linear and/or non-linear P < 0.05), i.e., glutamate, ribonucleotide, acesulfame, saccharin, sucralose, caking agent, firming agent, gelling agent, thickener, fructose, inverted sugar, lactose, and maltodextrin. Among these, gelling agent was inversely associated with mortality risk. UPF intake was significantly (P < 0.05) associated with all-cause mortality. Major findings remained robust in sensitivity analyses. Interpretation: Total MUP-based definition UPF, five MUP categories, and several specific MUPs within these categories were significantly associated with all-cause mortality. Thus, interventions targeting UPF may be most effective when focused on these candidates, which should also be prioritised for future mechanistic studies. Funding: Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) (Project: SFB 1052/2 C6 to MF).