Sugar tax for obesity- how strong is the evidence?
A report published in February 2016 by the UK Health Forum and Cancer Research UK concluded that a 20% tax on sugary drinks would prevent 3.7 million people in the UK from becoming obese over the next decade. This would stabilise obesity rates at the current level of 29% when otherwise it is projected to rise to 34% of the population. This report projected £10million of annual savings for the NHS health and social care budget i.e. less than 0.01% of current healthcare expenditure from the public purse. Celebrities like TV chef Jamie Oliver have been vocal in their support of this tax and, according to Cancer Research UK, 55% of the population support such a tax. Is there really any credible evidence that such a tax would really have any noticeable effect upon obesity rates? As a nutritionist, I am no fan of sugar, have no food industry connections and almost never consume sugary drinks but I believe that there is no credible evidence that this tax would work.
In the six weeks leading up to the 2015 general election, not one of the almost 100 opinion poll results predicted the 7% lead that the Conservatives would achieve. More than half predicted a Labour lead and most predicted a very narrow margin of 0-3% between the two largest parties. In essence what opinion pollsters do is select a representative sample of people and ask them how they intend to vote and then use a computer model to predict the outcome of the election in terms of share of the vote and in some cases the share of seats for each party at Westminster. Many of these organisations are well resourced, have long and wide experience of conducting polls and are able to adjust their computer models by comparing their predictions with actual outcomes of real elections; their impartiality should be ensured because it damages their credibility if they get incorrect results.
Why is this discussion of opinion polls relevant to a discussion of the evidence for predicting the effect of a sugar tax on obesity rates? Because the methods used by opinion pollsters look positively robust and straightforward compared to those used to predict the likely outcome of a tax on sugary drinks. There are two main investigative approaches in science:
• Experimental – identical control and experimental groups are set up but only the experimental group is exposed to the intervention under test. E.g. one has identical groups of animals, in identical environments and diets but one is exposed to an experimental “intervention” and any difference between the two groups can be confidently attributed to the intervention.
• Observational – one is only able to collect data and then draw inferences and predictions from this data e.g. whether two measured variables like a dietary/ lifestyle factor and a disease risk are associated. Observational methods have many limitations; there are many variables which one cannot control and one is limited to trying to find inevitably imprecise ways of allowing for other variables when looking at any particular association. It has been argued with substantial evidence from precedent that most conclusions drawn from observational studies are wrong and they are mainly useful in generating hypotheses which can then be tested in experiments. They have had many spectacular successes but in most of these cases the effect of the variable under test is very large e.g. identifying the link between smoking and lung cancer (smoking increases the risk by 10-40 times), identifying the link between front sleeping and cot death (risk increased by 3-8 times), identifying the link between occupational asbestos exposure and mesothelioma (risk increased by more than 10 times).
In order to estimate the effects of a sugar tax the authors of the 2016 report had to:
1) Estimate the effect the tax would have on sugary drink consumption – they estimated a 16% reduction.
2) Estimate what the effect this would have on the consumption of other foods; at least some of the lost sugary drink calories would be replaced by calories from other foods
3) Estimate the overall change in per capita calorie intake; they estimated a 15 calories/day reduction i.e. just over half a per cent of average daily intakes.
4) Estimate what effect a daily reduction of 15 calories would have on obesity prevalence in the population.
Every stage in this process involves making several assumptions and uses methods that are inevitably subject to major errors and uncertainties e.g. it is accepted that it is impossible to measure with accuracy and certainty the food and calorie intake of free-living individuals let alone that of national populations. Given the experience of opinion pollsters and the general tendency of observational methods to get the wrong answer in scientific studies, how can one have any confidence in these estimates especially given the highly speculative nature of the methods and the small effects predicted?
Sugar is currently in fashion as the culprit to blame for our spiralling obesity rates but for many years fat was regarded as the most important dietary factor. More than thirty years ago the National Advisory Committee on Nutrition Education recommended a reduction in total fat intake partly to help reduce obesity. They suggested that fat should be reduced initially from around 40% of food calories to around 35% which they believed could be achieved by 1990. It has actually taken thirty years to just about reach this short term target and obesity rates have trebled during that time. There is even a school of thought which suggests that if sugar consumption is reduced (e.g. by greater use of calorie-free sweeteners) that this might lead to an overall increase in calorie intake because some of the sugar would inevitably be replaced by fat which would be less well detected by our bodies’ appetite control mechanisms.
There is data from official sources which suggests that average per capita calorie intakes have dropped by around a quarter (i.e. by several hundreds of calories per day) since around 1960 yet obesity rates have at least quadrupled over that period. As we have become less active and expended less energy so our food intake has also dropped but the fall in energy use has been slightly faster that the fall in consumption leading a net surplus of calories and increased storage of energy as body fat. Obesity rates have increased despite falling energy intakes. Even in the unlikely event that the tiny 15 calorie/ day reduction predicted by the Health Forum is correct will it actually stem the rise in obesity?
If a sugar tax were introduced it would be almost impossible to check some of the predictions made in this report. One could test with some degree of confidence whether the predicted large reduction in sugary drink consumption actually occurred; factors that might reduce the effect would be if consumers partly offset the price rise by switching to cheaper brands or if producers or retailers absorbed some of the expected price rises. Would consumers or suppliers find other ways of partly circumventing the effects of this tax? It is not possible to measure average calorie intake accurately enough to see whether calorie intake really falls by 15 calories per day; year-to-year fluctuations in recorded intakes are greater than this and it also seems likely that average intakes have fallen by much more than this amount in recent years in several sectors of the population. It would be some years before one could detect with any confidence whether rises in obesity rates were really being checked as predicted and given the uncertainties in predicting future changes in obesity it would be all but impossible to show with any certainty what the specific effect of the sugar tax had been. Overall we have some very tenuous predictions based upon necessarily flawed methods and no way of testing whether most of these predictions have actually been fulfilled. So unlike with opinion polls we would not even be able to test the accuracy of the predictions.
I agree with the view that obesity is a major public health problem that needs concerted urgent action. I agree that a reduction in sugar consumption in Britain would be desirable, if only for its positive benefits for dental health and probably increased vitamin and mineral intakes. I do not think, however, that a sugary drink tax would make a major contribution to the obesity problem. It seems like an easy cosmetic option that would give the appearance of taking decisive action because really effective measures are likely to be complex, costly and difficult to effect.
I believe that the major driving force behind the massive and continuing increase in obesity rates and especially childhood obesity are our declining activity levels and physical fitness. Activity levels have declined because of major reductions in the energy we have to expend in everyday life, for example:
• Increasing mechanisation means that there are few jobs that are now truly manual
• Energy-saving devices have reduced the energy necessary to complete everyday household and garden chores
• Children now walk and cycle much less than they used to
• Children are given much less freedom to independently roam and engage in unsupervised outdoor play
• The condensing of the school day has reduced the time for outdoor activity during breaks
• The focus on league tables and academic targets has tended to lessen the opportunities for children to take part in physical activities during the school week
• Increased availability of screen-based pursuits encourages children in particular to be sedentary for more of their leisure time.
Policies that seek to increase activity levels in the face of these forces that are driving the trend to inactivity is inevitably going to be difficult, slow and perhaps costly. There is no simple, easy-to-implement national measure that will give a substantial large boost to the activity level of the population as a whole; no sedentary tax that would even be a net beneficiary to the public purse! How are we going to make it safer for children to walk or cycle to and from school? How can we create safe places where children can roam and play without direct adult supervision? How can we get children who are not athletically gifted to engage more in active leisure-time pursuits?
I do not have easy answers to any of these and other questions but I am concerned that introducing a sugar tax as a populist solution to the problem of obesity means that we may be deflected from even trying to seriously address them for a few more years.