Discrepancies between two long-term dietary datasets in the United Kingdom (UK)

Background: Studying dietary trends can help monitor progress towards healthier and more sustainable diets but longitudinal data are often confounded by lack of standardized methods. Two main data sources are used for longitudinal analysis of diets: food balance sheets on food supply (FBS) and household budget surveys on food purchased (HBS). Methods: We used UK longitudinal dietary data on food supply, provided by the Food and Agriculture Organisation (FAO) (FAO-FBS, 1961-2018), and food purchases, provided by the Department for Environment, Food and Rural Affairs (Defra) (Defra-HBS, 1942-2018). We assessed how trends in dietary change per capita compared between FAO-FBS and Defra-HBS for calories, meat and fish, nuts and pulses, and dairy, and how disparities have changed over time. Results: Estimates made by FAO-FBS were significantly higher (p<0.001) than Defra-HBS for calorie intake and all food types, except nuts and pulses which were significantly lower (p<0.001). These differences are partly due to inclusion of retail waste in FAO-FBS data and under-reporting in Defra- HBS data. The disparities between the two datasets increased over time for calories, meat and dairy; did not change for fish; and decreased for nuts and pulses. Between 1961 and 2018, both FAO-FBS and Defra-FBS showed an increase in meat intake (+23.4% and +1.4%, respectively) and a decrease in fish (-7.1% and -3.2%, respectively). Temporal trends did not agree between the two datasets for dairy, calories, and nuts and pulses. Conclusions: Our finding raises questions over the robustness of both data sources for monitoring UK dietary change, especially when used for evidence-based decision making around health, climate change and sustainability.