Changes in household purchasing of soft drinks following the UK soft drinks industry levy by household income and composition: controlled interrupted time series analysis, March 2014 to November 2019

Abstract

Background: The WHO recommends taxes on sugar sweetened beverages (SSBs) to improve population health. We examined changes in volume of and amount of sugar in purchases of soft drinks according to household income and composition, 19 months following the implementation of the UK soft drinks industry levy.

Methods: Data were from the Kantar Fast Moving Consumer Goods panel, a market research panel which collects data on weekly household purchases (mean weekly number of households=21 908), March 2014-November 2019. Interrupted time series analysis of volume and sugar purchases was used to estimate absolute and relative differences in the volume and amount of sugar in soft drinks, confectionery and alcohol purchased weekly by household income (<£20 000, £20-50 000 or >£50 000) and composition (presence of children (<16 years) in the household (yes or no)), 19 months after soft drinks industry levy (SDIL) implementation, compared with the counterfactual scenario based on pre-announcement trends and using a control group (toiletries).

Results: By November 2019, purchased weekly sugar in soft drinks fell by 7.46 g (95% CI: 12.05, 2.87) per household but volumes of drinks purchased remained unchanged, compared with the counterfactual. In low-income households, weekly sugar purchased in soft drinks decreased by 14.0% (95% CI: 12.1, 15.9) compared with the counterfactual but in high-income households increased by 3.4% (1.07, 5.75). Among households with children, sugar purchased decreased by 13.7% (12.1, 15.3) but increased in households without children by 5.0% (3.0, 7.0). Low-income households and those with children also reduced their weekly volume of soft drinks purchased by 5.7% (3.7, 7.7) and 8.5% (6.8, 10.2) respectively. There was no evidence of substitution to confectionary or alcohol.

Conclusion: In the second year following implementation of the SDIL, effects on sugar purchased were greatest in those with the highest pre-SDIL purchasing levels (low-income households and those with children). The SDIL may contribute to reducing dietary inequalities.

Trial registration number: ISRCTN18042742. Registered: August 2017.

Keywords: Dietary patterns; Nutrition assessment; Nutritional treatment.

Figure 1. Weight (g) of sugar from soft drink products purchased per household per week in the total population, from March 2014 to November 2019. Observed and modelled amounts of sugar in all soft drinks (drinks liable to the SDIL and non-liable drinks). Light blue points show observed data and light blue lines (with light blue shadows) show modelled data (and 95% CIs) of sugar from purchased soft drinks. The dark blue line indicates the counterfactual line based on pre-announcement trends and had the announcement and implementation not happened. The red line (and shadow) indicates modelled toiletries (control group). The first and second dashed lines indicate the announcement and implementation of SDIL, respectively. SDIL, soft drinks industry levy.

Figure 2. Weight (g) of sugar from soft drink products purchased per household per week, by gross household income levels, from March 2014 to November 2019. Observed and modelled amounts of sugar in all soft drinks (drinks liable to the SDIL and non-liable drinks) by annual gross household income levels of (a) <20 000, (b) £20 000–£50 000 and (c) £50 000 or more. Light blue points show observed data and light blue lines (with light blue shadows) show modelled data (and 95% CIs) of sugar from purchased soft drinks. The dark blue line indicates the counterfactual line based on pre-announcement trends and had the announcement and implementation not happened. The first and second dashed lines indicate the announcement and implementation of SDIL, respectively. The scales on the Y axis vary between panels and modelled toiletries have been removed (see online supplemental figure S1 for inclusion of toiletries) to maximise the resolution of the graphs. SDIL, soft drinks industry levy.

Figure 3. Weight (g) of sugar from soft drink products purchased per household per week, by whether households have children or not, from March 2014 to November 2019. Observed and modelled amounts of sugar in all soft drinks (drinks liable to the SDIL and non-liable drinks) by (a) households with no children and (b) households with children (<16 years). Light blue points show observed data and light blue lines (with light blue shadows) show modelled data (and 95% CIs) of sugar from purchased soft drinks. The dark blue line indicates the counterfactual line based on pre-announcement trends and had the announcement and implementation not happened. The first and second dashed lines indicate the announcement and implementation of SDIL, respectively. The scales on the Y axis vary between panels and modelled toiletries have been removed to maximise the resolution of the graphs. SDIL, soft drinks industry levy.

Figure 4. Volume (mL) of soft drink products purchased per household per week in the total population, from March 2014 to November 2019. Observed and modelled volumes of all soft drinks (drinks liable to the SDIL and non-liable drinks). Light blue points show observed data and light blue lines (with light blue shadows) show modelled data (and 95% CIs) of volumes of purchased soft drinks. The dark blue line indicates the counterfactual line based on pre-announcement trends and had the announcement and implementation not happened. The first and second dashed lines indicate the announcement and implementation of SDIL, respectively. Modelled toiletries have been removed to maximise the resolution of the graphs. SDIL, soft drinks industry levy.

Figure 5. Volume (mL) of soft drink products purchased per household per week, by gross household income levels, from March 2014 to November 2019. Observed and modelled volumes of soft drinks (drinks liable to the SDIL and non-liable drinks) by annual gross household income levels of (a) <20 000, (b) £20 000–£50 000 and (c) £50 000 or more. Light blue points show observed data and light blue lines (with light blue shadows) show modelled data (and 95% CIs) of sugar from purchased soft drinks. The dark blue line indicates the counterfactual line based on pre-announcement trends and had the announcement and implementation not happened. The first and second dashed lines indicate the announcement and implementation of SDIL, respectively. The scales on the Y axis vary between panels and modelled toiletries have been removed to maximise the resolution of the graphs. SDIL, soft drinks industry levy.

Figure 6. Volume (mL) of soft drink products purchased per household per week, by whether households have children or not, from March 2014 to November 2019. Observed and modelled volumes of soft drinks (drinks liable to the SDIL and non-liable drinks) by (a) households with no children and (b) households with children (<16 years). Light blue points show observed data and light blue lines (with light blue shadows) show modelled data (and 95% CIs) of sugar from purchased soft drinks. The dark blue line indicates the counterfactual line based on pre-announcement trends and had the announcement and implementation not happened. The first and second dashed lines indicate the announcement and implementation of SDIL, respectively. The scales on the Y axis vary between panels and modelled toiletries have been removed to maximise the resolution of the graphs. SDIL, soft drinks industry levy.