Potential impact on prevalence of obesity in the UK of a 20% price increase in high sugar snacks: modelling studyBMJ 2019; 366 doi: https://doi.org/10.1136/bmj.l4786 (Published 04 September 2019) Cite this as: BMJ 2019;366:l4786
All rapid responses
Re: Potential impact on prevalence of obesity in the UK of a 20% price increase in high sugar snacks: modelling study
We thank Kevin Hall for this response. The use of static or dynamic weight loss model was raised during the review process for our paper (https://www.bmj.com/content/366/bmj.l4786/peer-review). We agree with the reviewers and Kevin Hall that, over a longer time period, the static model overestimates weight loss for the reasons he describes. Therefore, the timespan we used in our analyses stops at one year.
We do, however, stand by the point we made in response to the paper’s reviewers that, in the shorter run (the first year), the absolute differences generated by the two models are small. The prediction of weight loss by Lin et al (https://www.ncbi.nlm.nih.gov/pubmed/21940223) following a 20% price increase in SSBs shows the two models provide ‘noticeable’ differences within half a year (difference of 0.2 kg [-0.8kg vs -0.6kg]) and at one year the estimated weight loss between models differs by 0.6kg (Lin et al. do not provide confidence intervals). Similar to Lin et al, our models considered multiple sources of uncertainty arising from the use of different datasets and several estimation/modelling steps, which are reflected in the large confidence intervals around our predictions. The lower end of our reported range is similar to the 60% difference estimated by Lin et al between the two models: our predicted mean weight loss of 1.3kg at the one-year point was reported with a confidence interval ranging from 0.51kg and 2.1kg.
If we assume that estimates of weight loss are more likely to be in the lower range of our predictions, our conclusion remains intact: fiscal intervention to increase the price of sweet snacks could be more effective in the UK context than the positive effects observed around the world for taxes on SSBs. This is because price responses are similar between SSBs and sweet snacks, the difference between them being that the latter are a source of far more energy in average UK diets.
The consumption of sugary snacks (or the snack to drink ratio) is relatively high in the UK, especially when compared with other high-income countries. For example, in the UK the market size of sweet snacks is 19.3kg per kg per person in comparison to 161.3L of soft drinks. The respective figures in USA are 14.3kg and 346L (Euromonitor International, 2014). Although not providing us with the exact caloric intake and sugar consumption by food group, these simple figures highlight the need for policies targeting sugary snacks as well as sugary drinks.
Given that reformulation, the mainstay of UK public health policy to improve the nutritional quality of the diets, is proving ineffective in reducing sugar content of cakes, biscuits and confectionery (https://www.gov.uk/government/publications/sugar-reduction-report-on-fir...), other policy options to reduce consumption of sugary snacks are urgently needed.
We neither suggest nor wish to imply that a price rise on sugary snacks is the sole solution to obesity. There is no one solution and, as such, every little step counts towards the bigger goal of a fundamental shift in eating patterns towards a healthier, lower energy diet.
Pauline Scheelbeek, Laura Cornelsen, Theresa Marteau, Susan Jebb, Richard Smith
Competing interests: No competing interests
Inflated expectations of reduced obesity prevalence result from an erroneous static weight loss model
The recent modeling study by Scheelbeek et al. estimated the calorie intake changes in response to 20% price increase in sugary snacks and predicted the corresponding changes in body weight using a so-called "static model for weight loss". The authors acknowledged that the static model "might not fully reflect actual mechanisms of weight change. However, the period considered in this study is one year, and modelled weight changes during that time were expected to be relatively small. Therefore changes in energy expenditure, as described previously, are expected to have marginal impacts on our results "
Unfortunately, the authors' expectations are incorrect. As previously demonstrated in a similar modeling study investigating the potential impact of a sugar-sweetened beverage tax, the static weight loss model substantially overestimates weight losses after one year and the discrepancy exacerbates over time (https://www.ncbi.nlm.nih.gov/pubmed/21940223) when compared to the results of a validated dynamic model of weight change that correctly accounts for physiological changes in energy expenditure and energy partitioning (https://www.ncbi.nlm.nih.gov/pubmed/21872751),
More recently, dynamic models have been developed to also account for changes in appetite with weight loss (https://www.ncbi.nlm.nih.gov/pubmed/27804272) that result in predicted weight changes that are even smaller than previous dynamic models that account for only energy expenditure changes (https://www.ncbi.nlm.nih.gov/pubmed/29896621). The physiological adaptations in response to persistent interventions are expected to result in greatly diminished returns when it comes to weight loss, as has been clearly documented in obesity pharmacotherapy trials (https://www.ncbi.nlm.nih.gov/pubmed/29071809).
Ignoring the physiological changes in appetite, energy expenditure, and energy partitioning results in inflated estimates of the effectiveness of interventions and policies targeted to address obesity. We should not be misled that obesity prevalence can be easily decreased by single interventions and such interventions should not be judged to have failed if they do not live up to exaggerated promises based on erroneous assumptions.
Competing interests: No competing interests