Changes in food and drink purchasing behaviour in England during the first 3 months of the COVID-19 pandemic: an interrupted time series analysis

Abstract

Objective: This study examined changes food and drink purchasing during the first 3 months of the COVID-19 pandemic in England, and if changes varied by population subgroups.

Design: We investigated changes in take-home food and drink purchasing and frequency of out-of-home (OOH) purchasing using an interrupted time series analysis design. The start of pandemic restrictions (the intervention) was defined as 16 March 2020, when first announced in the UK.

Setting: London and the North of England.

Participants: 1245 households reporting take-home and 226 individuals reporting OOH purchases between January 2019 and mid-June 2020 from the GB Kantar Fast Moving Consumer Goods Panel.

Results: The marginal mean estimate of total take-home energy purchased was 17·4 % (95 % CI 14·9, 19·9) higher during the pandemic restriction period compared with the counterfactual. Increases of 35·2 % (95 % CI 23·4, 47·0) in take-home volume of alcoholic beverages and 1·2 % (95 % CI 0·1, 2·4) in foods and drinks high in fat, salt and sugar were observed. Reductions in purchased energy from fruit and vegetables (-7·3 %, 95 % CI -10·9, -3·6), ultra-processed foods (-4·0 %, 95 % CI -5·2, -2·8) and in OOH purchasing frequency (-44·0 %, 95 % CI -58·3, -29·6) were observed. Changes in chocolate and confectionery, soft drink and savoury snack purchases levelled off over time. Changes in all studied outcomes varied by sociodemographic characteristics and usual purchasing.

Conclusions: Pandemic restrictions were associated with positive and negative changes in food and drink purchasing, which differed by individual characteristics. Future research should ascertain if changes persist and translate into changes in health.

Keywords: COVID-19; Food and drink purchasing; Interrupted time series; Ultra-processed foods.

Figure 1.

Adjusted weekly mean estimates of food and drink purchasing before and during pandemic restrictions, and the counterfactual. Vertical line = 16 March 2020, start of pandemic restrictions. The counterfactual was estimated by extrapolating the pre-pandemic trend. Marginal means were estimated from interrupted time series two-part models: part 1 (logit) and part 2 (generalised linear model) with negative binomial distribution. Models were adjusted for season, region, festivals, age, sex, and occupational social grade of the main shopper, number of adults, and presence of children. Cluster-robust se were used. Data period: 1 January 2019 to 14 June 2020. Y axes limits were set manually to best display changes; therefore, some do not originate in 0. HFSS, high in fat, salt and sugar; OOH, out-of-home; UPF, ultra-processed foods.