There has been much chuntering (for example here and here) recently about sugar and whether or not it should be taxed. This is because there is a rising and worrying incidence of obesity. Excessive consumption of sugar is blamed for this and there is a powerful school of thought that thinks that taxing sugary drinks will reduce the problem. A report published by Public Health England sets out PHE’s case for taxing sugar on sugary drinks like Coca Cola. It does not provide any scientific evidence for sugar being a principle cause of obesity – it is taken for granted that it does; and then sets out evidence for sugar taxes being effective in reducing consumption. Criticism is also thrown at supermarkets for putting these products on special offers and near the checkouts to encourage impulse buying; and so policies are offered to force supermarkets to reduce this kind of temptation.
This blog attempts a number of exploratory statistics to work towards a testing of the PHE hypothesis that dietary sugar is a cause of increased obesity.
Whilst researching datasets for use in another line of inquiry, I discovered a wonderful dataset, going back to 1974, where food consumption by households is presented as consumption per person per week. This shows food consumed under sufficient categories to cover our food consumption as a nation. The data has been gathered consistently since 1974 and so provides a fairly accurate indicator of our national diet and how this has changed over the last 40 years. Given that seriously obese people were something of a rarity 40 years ago, and now they seem to be commonplace, it would suggest that something is going wrong with our eating habits. Another dataset, this time from Public Health England, which is much shorter and dates only from 2000, is graphed below in Figure 1. It should be noted that the vertical axis starts at 21.2% and so the total rise in incidence of people with a BMI (Body Mass Index) greater than or equal to 30 is about 4.4%:
Figure 1: Percentage of people with BMI greater or equal to 30, since year 2000 to 2014. Source data – Public Health England
Even though the percentage increase of 4.4% is small, it still seems an inexorable and alarming rise in only 14 years. So there is some justification for the panic amongst public health officials that more more obese people will lead to more type 2 diabetes, heart disease and so on. Sugar in soft drinks are known to be very high and consumption of these products are increasing, so perhaps PHE has found the culprit and is correct to target them.
However, the Defra family food datasets show up some trends which provoke the curiosity of those of us who are mildly sceptical of claims from a department which has set off a number of wild strategies in the recent past. In fact, the total consumption of many of the broad categories of food has dropped:
Figure 2: Changes in consumption of main categories of food since 1974. Graph from Defra dataset.
A shorter dataset from 1992 to 2014 is available for three other categories not specifically collected from 1974:
Figure 3: Changes in consumption of soft drinks, confectionary and alcohol since 1992. Graph derived from Defra dataset.
It can be seen that several large categories of foods which have been variously branded as ‘unhealthy’ by assorted health officials over the years, have declined dramatically – bread, potatoes, table sugar and preserves, meat and dairy produce and so on. These figures are in grams per person per week (I have approximated 1ml of liquid as equivalent to 1g). If there have been so many reductions, then what has happened to the total food consumed? Figure 4 shows us:
Figure 4: Total consumption per person per week for all food categories since 1974 and 1992. Graph derived from Defra dataset.
This suggests that for all categories excluding soft drinks, confectionary and alcohol, the national consumption of food has declined by 2,142 g/person/week; from 10,629g in 1974 to 8,487g in 2014. The consumption of soft drinks, confectionary and alcohol has risen slightly by 2,56g from 2,095g in 1992 to 2,351g in 2014. In other words, our total consumption of food has dropped by about 2,000g/person/week. On the face of it, this does not seem to indicate a nation which is overeating huge quantities of food. In fact this suggests that we are eating about 18% less food than we were in 1974. As this is not a picture of a nation of gluttons, there is obviously something happening to our national diet which is affecting some of us. Sugar in drinks may be a big part of the answer, and may suggest that PHE is justified in conducting their ‘war against sugar’. We need to take a closer look at the numbers for each main category of food:
Figure 5: Milk and cheese consumption
Figure 6: Fats and Sugar
Figure 7: Bread, cakes, biscuits and cereals.
Figure 8: Meat, meat products and fish.
Figure 9: Potatoes, vegetables and fruit.
Figure 10: Beverages and other food and drink.
Figure 11: Soft drinks, confectionary and alcoholic drinks
It should be noted that the above graphs mostly show only very small variations from one year to the next and that the trends are fairly easy to discern. This suggests that the data is collected consistently from one year to the next and that the errors in sampling are fairly small. However, there are one or two big fluctuations in some of the time series. For example, in Figure 9, there is a large uptick in 2007/2008 in the curve for fresh and processed fruit. This is most likely because of a government drive to get people to eat more fresh fruit and vegetables (the ‘five-a-day’ campaign). Like wise, in Figure 10, there is another sharp fluctuation in ‘other food and drink’ which I cannot explain at the moment. In Figure 9, there was a very pronounced decrease in the amount of potatoes consumed in 1976. This was because of a very bad harvest in 1975 and prices rising by three or four times in a few months. Questions were asked and statements made in the House of Commons and this Hansard entry shows the exchange. The data shows a sharp reduction in consumption and this coincides with my own memory of that time.
Overall, the trend has been steadily downwards for many traditionally British foods such as dairy produce and fresh red meat, potatoes and so on which have halved in many cases. There have been commendable rises in the amount of fruit that we eat and a very slight decrease in the amount of vegetables. Bread, cakes, flour and biscuits have all decreased. Fats and table sugar, jams and honey have plummeted. Sugar, in particular, has dropped to about 25% of its 1974 amount. Conversely there has been a small but discernible increase in the amount of fish consumed. Looking at these items, we should have a lot of very happy dieticians and public health officers because they have, at some stage or another, encouraged us to eat less of the things that have dropped in consumption and more of the ‘healthy’ things like fish and fruit. However, despite the nation dutifully heeding the strictures of these officials, we are getting much, much fatter as a nation.
The clue must therefore rest in the things which have risen in consumption. These are the categories ‘Other cereals and cereal products’, ‘Non carcass meat and meat products’ and ‘Other food and drink’. (Figures 7, 8 and 10 respectively). Analysis of these three headings will proceed in Part 2 of this series of blogs.
The main target of PHE’s ire is that of sugary drinks, i.e. ‘soft drinks’. Data for this category only started to be gathered in 1992, but is shown in Figure 11. Here the curve for soft drinks is a curve which peaks at 2003/4 and then declines to not far above its 1992 level. The 2014 consumption was only 98ml above the 1992 consumption, which corresponds to a 6.7% increase. However, the peak in 2003/4 was 485ml above 1992 figures, corresponding to an increase of 33.5% above the 1992 figure. So there has been a considerable drop in consumption of sugary drinks since 2003/4. On its own, this consumption behaviour should raise question marks about the link between sugar and obesity. When it is added to the Figure 1 curve which shows a steady increase from 2000 to 2014 in obesity, rather than a peak (assuming obesity is also a transient characteristic), then we would expect a flattening out of the obesity curve, at the very least.
The reduction of consumption of ‘sugars and preserves’ and the increase and then decline of consumption of ‘soft drinks’ begins to make the position of PHE on sugar look open to question.
Part 2 of this series will drill down deeper into the data and examine statistical links between the various categories of food consumption. Part 3 will look at the sugar content of each category of food and will balance that against actual consumption.