I delved into the technical summary as I have an interest in these extrapolations. The claim is that a 20% excise tax levied on sweetened soft drinks could avoid 3.7 million people (about 5%) becoming obese (>=30 BMI) by 2025 and that without the tax obesity will rise from 29% in 2015 to 34% in 2025.
I've had issues with the "inevitable upward trend" claims about obesity before, as the Foresight report of the Blair years drew an exponential growth curve (rising to 55% obesity by 2050) that hasn't really materialised :
|UK Actual obesity data (3y average, Public Health England)|
10 years ago the chart above shows obesity at about 22.5%, lets' call it 22% which gives us a 10 year growth of 3% to 2014. This seems inconsistent with the claimed 5% in 10 years projection even if we gloss over the apparent problem with the baseline figure.
Does the analysis overstate the baseline prevalence of obesity as well as projected growth in order to deliver a greater forecast benefit from the proposed measure ?
Moving on from how fat we all are, who drinks this stuff anyway ? Our annual household consumption of sugar sweetened beverages (SSB) is probably less than 5 litres, if that. The technical summary reveals some concerns about the limitations due to lack of data of SSB stratified by BMI however there is published data from (for example) the EPIC-Norfolk study which shows us that the consumers of SSBs had a mean BMI 1 unit lower than that of the consumers of diet equivalents - in other words the heavier people were perhaps inclined to select the diet option. This shouldn't be a surprise, we've all witnessed the "Big Mac and fries with a diet coke" mindset in action.
The EPIC-Norfolk data is also interesting in that of the 24,700 participants only 12,800 were consumers of SSBs with 5,600 consumers of diet equivalents. That's right, only about HALF of the population drink any SSBs at all in that dataset.
I do recall another dataset where the median consumption of SSBs in adults was zero - ie less than half of the population surveyed drank any "sugary fizzy drinks". From memory that was in Scotland.
As the UKHF modelling used data from the UK's National Diet and Nutrition Survey: (Headline Results from Years 1, 2 and 3 (combined) of the Rolling Programme 2008/09 – 2010/11) I had a dig into their SSB consumption data. Men were reported as consuming a mean average of 170 grams/day of SSBs with a standard deviation of 270 g/day whereas women consumped 109 g/day with an sd of 193 g/day. These large sds clearly point to a very non-normal (skewed) distribution of consumption.
Digging back to earlier work based on the NDNS data we can see that the 2008/09 median consumption of "Soft drinks, not low calorie" was 63 g/day in men with an interquartile range (IQR) of 0-248 and 22 g/day in women with an IQR of 0-149. From this we can deduce that 50% of women reported consuming 22g/day or less of SSBs while at least 25% of them consumed none.
In men 50% consumed 63 g/day or less and at least 25% consumed none. 25% of men, the upper quartile, were consuming at least 248 g/day ie at least one can or serving sized bottle.
It looks to me as if the target population of a sugar tax on sugar sweetened drinks is quite a small part of the whole, and as such it is difficult to see how the BMI of large swathes of the population will be reduced as claimed by the charities today.