Improving diets to halve obesity across Britain

Key finding 1

Daily calories purchased per person at the population level are much higher than in previous national surveys, and we are still likely underestimating true consumption

Our latest analysis shows that, in Britain, we purchase on average ~2,730 calories per day (not including alcohol purchased for consumption outside of the home and not accounting for food waste).

This figure is: 

• 2% lower than a population-weighted average of the Doubly-Labelled Water (DLW) adjusted consumption in the National Diets and Nutritional Survey (NDNS) in 2020 (~2,800 kcal per person, per day), the most robust method available for estimating calorie consumption at a population level
• 54% higher than the total consumption estimated in the National Diet and Nutrition Survey (NDNS) in 2020 (1,775 calories per person per day), which records self-reported diet activity (unadjusted compared to the Doubly Labelled Water Study)
• 28% higher than the estimated calories purchased in the Family foods dataset (sourced from LCFS) in 2019, based on purchasing expenditure (2,140 calories purchased per person per day). 
• 22% higher than the average recommended daily intake of 2,245 calories per day (weighted by GB 2021 population estimates for males and females).

We note that these are not perfect comparisons – we are comparing estimates of consumption (NDNS) with purchase (Kantar, LCFS), and all of the data sources cited (including the Kantar Worldpanel) in our opinion have inherent uncertainties and limitations (the key limitations are outlined in the technical appendix). Our methods for estimating this figure also necessarily introduce uncertainty, also outlined in the technical appendix. 

However, despite being significantly higher than NDNS and LCFS raw estimates of consumption and purchase, we believe our estimate to be, if anything, an underestimate of average calories purchased in GB. The key reasons for this are:

• Our estimate remains under the DLW-adjusted NDNS figure for consumption. Whilst this is a relatively small-sample study, it is the best available data for correcting for under-reporting in the NDNS (see the technical appendix for an explanation of this method). Because our purchase figures do not account for food waste, we would expect them to be higher than the DLW NDNS figure (which does).
• The Kantar Worldpanel dataset we used does not include alcohol consumed out of home, so the purchase of alcohol out of home is excluded from this estimate.

Key finding 2

We get the majority of our calories from supermarkets, making them critical partners in shifting our food environment

Our analysis estimates that on average, we get ~2,390 (88%) of our daily calories from food and drink prepared and consumed in the home (in-home). The remaining ~340 (12%) of our daily calories come from food cooked or prepared out-of-home (OOH) (this is an updated figure since our How eating out contributes to our diets report was released; see technical appendix for details). This is considerably lower than the 25% figure often cited in the literature, although in line with recent estimates from NDNS and LCFS. Of the ~340 kcal consumed out of home, approximately one quarter (or 3% of our total diets) is ‘food to go’ purchased from supermarkets or convenience stores. 

However, our How eating out contributes to our diets report showed that the number of daily calories purchased out-of-home is distributed very unevenly across the population, and for some, it’s likely to be a much more significant proportion of their diet. 

Nesta’s previous analysis shows that the 11 largest retailers make up over 95% of the grocery market. We have estimated that through retailer targets, we could reduce calorie intake among populations living with excess weight by approximately 80 kcal per person per day.

Key finding 3

Relatively minor changes are required to reduce the calories in a weekly shopping basket by 8.5%

Based on our analysis of food and drink purchasing data in Great Britain, we’ve created examples of what the weekly shop of a ‘typical’ person looks like when eating ~2,730 calories per day (or 19,110 calories per week). We then show the same basket but with certain policies applied to reduce the overall calorie content by around 8.5%, which is the level of reduction needed in the diets of those living with excess weight to halve obesity prevalence.

The example basket is broken down by the item purchased, the average calories per week contribution and the cost per item. The design of the baskets is predominantly driven by the proportion of total calories purchased from different food and drink categories, both in and out of the home. The basket's total cost per week is £43.65 (in line with the Family Food FYE 2023 value of £43.33 per person per week spent in a household). See designing the GB and low-income basket section of our technical appendix for a detailed breakdown of the process.

Note: calories and prices reflect overall quantities of each product purchased per week, eg, the cost of a loaf used in modelling is £1, while the price of bread used is £1.20, which reflects the quantity purchased.

This work focuses on calorie reduction rather than replacing a basket with a fully balanced and optimally healthy diet. However, we have sought to ensure that changes to the baskets do not lead to a decrease in nutritional quality. 

Less than a third of the products (10 out of the 28) have changed to marginally reduce calories. These changes could be achieved either through product replacement (facilitated through interventions such as price promotions, advertising, product placement and availability changes), reformulation, or portion size reduction. Many of the changes are relatively small – either unlikely to be noticed at all by a consumer, or unlikely to represent a significant change in their weekly shop. For example, a 5% reformulation reduction in calories is applied to the chocolate bar in the basket and the full-fat Greek yoghurt is substituted for low-fat. 

We believe there are numerous policies that could incentivise in-home and out-of-home businesses to make these changes. For example, in our previous research, we proposed that we need health targets for food retailers that set an industry standard and incentivise companies to prioritise healthy sales.

Key finding 4

Similar changes can be applied to the weekly shop of people in lower income groups, with an 8.5% reduction in calories still being reached

The second basket is designed based on data from a segment of the population in the lowest socioeconomic class (SEC) level (social class E). For this report, we refer to the considered groups as low-income, though more nuanced factors are taken into consideration in the SEC grouping.

Unlike the overall GB analysis above, we are only able to compare percentages of calories purchased at social grades (rather than absolute calories) using the data we have available. The total number of calories is, therefore, the same in the low-income basket as the average calculated for the GB population. We discuss this in more detail in our data approach section.

Note: calories and prices reflect overall quantities of each product purchased per week, eg, the cost of a loaf used in modelling is £1, while the price of bread used is £1.20, which reflects the quantity purchased.

Conclusion

With two-thirds of adults in Great Britain living with excess weight, we must act to improve our diets. This analysis has shown that it would only take minor changes to the food we buy to achieve the 8.5% daily calorie reduction needed to halve the prevalence of obesity. Importantly, this work supports the case that changes could be achieved without increasing the cost of a typical weekly shop or reducing the joy we take from food. What’s more, they could be achieved by relatively imperceptible changes in our food environment. With these small, easy-to-achieve changes, we could make significant improvements to our health across Great Britain.

Data and approach

For full details of the dataset used for this analysis, limitations and our methodology, see our technical appendix document.