Ultra-processed food intake and trajectories of global and domain-specific cognition in the very old: The Newcastle 85+ Study

Dietary intake is increasingly recognised as a modifiable determinant of cognitive function, with emerging evidence linking higher intake of ultra-processed foods (UPF) to poorer cognitive outcome, especially in older adults aged 60 and above(1-3). However, evidence on specifically the very old (aged 85 and above) are scarce. Since they are a rapidly growing population in the United Kingdom (UK) where the number of adults aged 85 and over has increased by 186% between 1981 and 2021 (4) . Due to this demographic shift, it is essential to study specifically this populations UPF intake in relation to cognition. The Newcastle 85+ Study (n = 790) provides a unique opportunity to investigate nutritional determinants of cognition in the very old (≥ 85 y); the mean (standard deviation) age at baseline was 85.5 (0.4) y. Baseline diet (2006/07) was assessed with two non-consecutive 24-h recalls (5) and foods were classified according to the NOVA food classification system (6) . Global cognition (Mini-Mental State Examination, MMSE) were assessed at baseline and at 1.5, 3 and 5 years, while the Clinical Dementia Rating (CDR), composite attention scores [Power of Attention (PoA), Continuity of Attention (CoA), Reaction-Time Variability (RTA)] and single-task reaction times [simple reaction time (SRT), choice reaction time (CRT), and vigilance reaction time (VIGRT)] times were assessed at baseline, and at 1.5 and 3 years. UPF intake was categorised into tertiles using the Nova system as a proportion of total energy (%TE). UPF intake supplied 56 % of total energy intake at baseline. Cross-sectionally, higher UPF intake was associated with poorer baseline performance on selected attention and processing-speed measures, but not with global cognition. In fully adjusted regression models, each standard-deviation increase in UPF intake was associated with worse baseline PoA (β = 0.020, SE = 0.010, P = 0.045) and slower baseline SRT (β = 0.049, SE = 0.023, P = 0.032). No statistically significant associations were observed for MMSE, CoA, RTV, CRT, VIGRT, or CDR after full adjustment. Additionally, longitudinal mixed-effects models showed no statistically significant associations between baseline UPF intake and subsequent change in MMSE, attention domains, reaction-time measures, or CDR over up to five years of follow-up. Findings indicate that a higher intake of %TE of UPF was cross-sectionally associated with poorer attention and processing-speed performance, but not global cognition, while longitudinal mixed-effects models showed no association with cognitive decline over three and five years. These findings suggest that associations between UPF intake and cognition may vary within older age groups, highlighting the need to avoid treating adults aged ≥60 years as a homogeneous population and to adopt age-stratified approaches in research on UPF. Further research is needed to confirm whether these observed links reflect a causal influence of high UPF intake on cognition or whether the links are driven by overall diet quality, is essential. Future work should also test how the United Kingdom food-policy initiatives could improve both dietary quality and cognitive health in the very old.