Changes in soft drinks purchased by British households associated with the UK soft drinks industry levy: a controlled interrupted time series analysis

Objective To determine changes in household purchases of drinks 1 year after implementation of the UK soft drinks industry levy (SDIL). Design Controlled interrupted time series. Participants Households reporting their purchasing to a market research company (average weekly n=22 091), March 2014 to March 2019. Intervention A two-tiered tax levied on soft drinks manufacturers, announced in March 2016 and implemented in April 2018. Drinks with ≥8 g sugar/100 mL (high tier) are taxed at £0.24/L, drinks with ≥5 to <8 g sugar/100 mL (low tier) are taxed at £0.18/L. Main outcome measures Absolute and relative differences in the volume of, and amount of sugar in, soft drinks categories, all soft drinks combined, alcohol and confectionery purchased per household per week 1 year after implementation. Results In March 2019, compared with the counterfactual, purchased volume of high tier drinks decreased by 140.8 mL (95% CI 104.3 to 177.3 mL) per household per week, equivalent to 37.8% (28.0% to 47.6%), and sugar purchased in these drinks decreased by 16.2 g (13.5 to 18.8 g), or 42.6% (35.6% to 49.6%). Purchases of low tier drinks decreased by 170.5 mL (154.5 to 186.5 mL) or 85.8% (77.8% to 93.9%), with an 11.5 g (9.1 to 13.9 g) reduction in sugar in these drinks, equivalent to 87.8% (69.2% to 106.4%). When all soft drinks were combined irrespective of levy tier or eligibility, the volume of drinks purchased increased by 188.8 mL (30.7 to 346.9 mL) per household per week, or 2.6% (0.4% to 4.7%), but sugar decreased by 8.0 g (2.4 to 13.6 g), or 2.7% (0.8% to 4.5%). Purchases of confectionery and alcoholic drinks did not increase. Conclusions Compared with trends before the SDIL was announced, 1 year after implementation, volume of all soft drinks purchased combined increased by 189 mL, or 2.6% per household per week. The amount of sugar in those drinks was 8 g, or 2.7%, lower per household per week. Further studies should determine whether and how apparently small effect sizes translate into health outcomes. Trial registration number ISRCTN18042742.


Supplementary material B
The model specification is given by: the fall in purchases in the weeks immediately after Christmas; and the increase in confectionery purchases seen at Easter, for toiletries these were set to 0. To adjust for temperature-related trends in drink consumption the average UK monthly temperature was included in the intervention group with the average study period temperature used for toiletries. 5Quadratic functions of announcement trends were included where they improved model fit -assessed using likelihood ratio tests.Stationary was examined using augmented Dickey-Fuller tests. 41Autocorrelation between preceding time points was examined using autocorrelation and partial-autocorrelation plots.An appropriate autocorrelation structure was determined and then compared to alternative models using likelihood ratio tests.Visual inspection of the data suggested no additional benefit would be gained from including polynomial terms.
BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s)  The Y-axis varies in scale between panels to maximise the resolution of figures; modelled purchases include averaged effects for seasonality and the impact of December and January (Christmas period).The control category of toiletries is shown in Figure 3.

Sugar
High tier (≥8g sugar per 100ml) All Manufacturers 8.9 (6.8, 10.9) -0.05 (-0.08, -0.03) -11.5 (-15.3,-7.7) -0.1 (-0.2, -0.02) -2.6 (-6.9, 1.6) -0.2 (-0.Bold indicates a significant difference at the 95% confidence interval level. BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s) Notes.Points are observed data, black lines (with shadows) are modelled data (and 95% confidence intervals); red lines indicate counterfactuals had the announcement (red solid line) and implementation (red dashed line) not happened; the first dashed vertical line indicates the point of announcement; the second dashed vertical line indicates the point of implementation; The Y-axis varies in scale between panels to maximise the resolution of figures; modelled purchases include averaged effects for seasonality and the impact of December and January (Christmas period).The control category of toiletries is shown in Figure3.BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s) .Observed and modelled amount of sugar in no levy drinks containing sugar (low sugar) and drinks exempt from the Soft Drinks Industry Levy purchased per household per week, March 2014-March 2019 (weighted) Notes.Points are observed data, black lines (with shadows) are modelled data (and 95% confidence intervals); red lines indicate counterfactuals had the announcement (red solid line) and implementation (red dashed line) not happened; the first dashed vertical line indicates the point of announcement; the second dashed vertical line indicates the point of implementation; pre-post announcement & implementation (March 2014 -March 2019) Absolute change (ml) Relative change (%) Absolute change (ml) Relative change (%) Absolute change (ml) Relative change (%) Volume High tier (≥8g sugar per 100ml) An Explanation and Elaboration article discusses each checklist item and gives methodological background and published examples of transparent reporting.The STROBE checklist is best used in conjunction with this article (freely available on the Web sites of PLoS Medicine at http://www.plosmedicine.org/,Annals of Internal Medicine at http://www.annals.org/,and Epidemiology at http://www.epidem.com/).Information on the STROBE Initiative is available at www.strobe-statement.org. Note:

Table 2a :
Modelled level and trend changes in volume of drinks (ml) and confectionery (g) purchased per household per week (95% CI) in relation to the UK SDIL,March 2014 -March 2019 (weighted)Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s) Bold indicates a significant difference at the 95% confidence interval level.*Milkcomprises drinks in the following categories: semi-skimmed; specific low fat % milk (e.g.1% fat milk); whole milk; buttermilk; modified milk; other milk; other non-cows milk; rice drink; soya milk.Skimmed milk is excluded from all analysis in this paper due to missing data.BMJ

Table 2b :
Modelled level and trend changes in sugar in drinks and confectionery (g) purchased per household per week (95% CI) in relation to the UK SDIL,March 2014 -March 2019 (weighted) Bold indicates a significant difference at the 95% confidence interval level.*Milk comprises drinks in the following categories: semi-skimmed; specific low fat % milk (e.g.1% fat milk); whole milk; buttermilk; modified milk; other milk; other non-cows milk; rice drink; soya milk.Skimmed milk is excluded from all analysis in this paper due to missing data.

Table 3a :
Level and trend changes in volume of, and sugar in, all soft drinks (ml) and confectionery (g) purchased per household per week (95% CI) in relation to the UK SDIL,March 2014 -March 2019 (weighted)

Table 3b :
Absolute and relative change in volume of, and sugar in, all soft drinks and confectionery (g) purchased per household (95% CI) per week in relation to the UK SDIL,March 2014-March 2019 (weighted) The levy liable drinks category is a combination of high tier, low tier and no levy drinks.*Milk comprises drinks in the following categories: semi-skimmed; specific low fat % milk (e.g.1% fat milk); whole milk; buttermilk; modified milk; other milk; other non-cows milk; rice drink; soya milk.Skimmed milk is excluded from all analysis in this paper due to missing data.

Table 4b :
Modelled absolute and relative change in volume of, and sugar in, all drinks (ml) purchased per household (95% CI) per week in relation to the UK SDIL,March 2014-March 2019; excluding small manufacturers (weighted)