OP20 Estimating small area differences in consumption of ultra-processed foods in England and Wales

Background

The association between higher consumption of ultra-processed foods (UPF) and adverse health outcomes, including obesity, is increasingly evident. England has one of the highest UPF intakes worldwide. UPF consumption may be influenced by exposures to poor local nutrition environments. However, diet data are rarely available at local or regional levels. Local data would support approaches to improve these environments more equitably. We therefore aimed to create small area estimates of UPF consumption for England and Wales using local and individual-level data.

Methods

Using spatial microsimulation (SMS), we combined individual-level survey data (National Dietary and Nutritional Survey 2008–2018, n=15,655) and spatial data from the 2021 Census to create a unique estimate of all-population UFP consumption for all 7264 middle-layer output areas (MSOA) in England and Wales. SMS requires constraint variables present in both data sets. We used sex and age cohorts as constraining variables to distribute individuals from the NDNS data into the estimates for each geographical area, by using an iterative proportional fitting procedure. The main model outcomes quantified the proportion of daily energy (kcal) UPF contributed to individuals’ diets as ‘low’ (< 53%), ‘medium’ (53–67%), or ‘high’ (> 67%). All simulation modelling was executed in R.

Results

Overall UPF consumption is very high in England and Wales; 43% of the population derives over two-thirds of their daily energy from UPF, and 58% derives more than half of their daily energy from UPF. We found substantial geographical variation: high consumers made up more than a third (36%) of the population in some MSOAs, double the level of 16% seen in other MSOAs. Conversely, low consumers constituted as little as 32%, or as much as 50% of the population of other MSOAs. Initial internal validation was promising and additional exploration for external validation via local diet surveys is ongoing.

Conclusion

Detailed diet data are not generally available at a local or regional level in the UK. However, spatial microsimulation estimates could perhaps provide more meaningful geographical data to local authorities than simple deprivation indicators. These estimates of where UPF consumption are highest might better reflect the impact of UPF marketing environments and fast-food outlet licensing, thus potentially informing public health policy and action.