09-10-2012 Reply to Dr. Schaafsma's comments on the DRINK study

[A clarification for readers unfamiliar with the Sugar Foundation (Suikerstichting): the Sugar Foundation has hired Dr Schaafsma to ask critical questions about our study.[i] The Sugar Foundation is funded and governed by Suiker Unie, a major manufacturer of beet sugar. The Sugar Foundation handles public relations with health professionals and the public on the topic of sugar and health.[ii]]

We mentioned this figure in the paper that described the design of our study.[iii] It was derived using the model of Wang et al.[iv] We needed such an estimate in order to perform a power calculation. However, the accuracy of this number is doubtful because there are almost no experimental data on the relation between energy imbalance and weight loss or gain in children. Thus, such models cannot be validated properly.
Recently, Kevin Hall of NIDDK and coworkers have produced a valid model to estimate weight changes in adults: the Body Weight Simulator. We calculated what this model predicts for our study. Intake in the sweetener group would have been 104 kcal/d less than in the sugar group if adherence had been 100%. The estimated adherence of about 85% reduced this difference to 88 kcal/d. Actual adherence could have been smaller because 10% of children did not return their cans, and not all empty cans were necessarily consumed.
The model predicts that 88 kcal/d produces a weight difference in 541 days of about 2.1 kg in an active fully-grown 18-year-old adult (PAL 1.75). This model is, however, not valid for growing children.

2.      ‘It is difficult to imagine….’
Yes, many people have a problem imagining this. Eating 26 g sugar less for 541 days corresponds with a total of 56 000 kcal. It is tempting – but erroneous – to divide this by 3500 kcal/lbto yield a weight difference of 16 lbs (7 kg). However, such thinking ignores that bodies are dynamic systems, not static buckets. See Hall, Lancet2011 or Katan & Ludwig, JAMA 2010 for detailed explanations.

3.     ‘To what extent is energy compensation involved here?’
That is impossible to tell. A review commissioned by aspartame-producer Ajinomoto estimated that when sugar is replaced by sweeteners such as aspartame, 16-32% of the calories of the sugar that was removed is compensated for by increased intake of other foods.[v]The weight difference in our study was about 1 kg, so the uncompensated weight difference would then have been 1.15 to 1.30 kg -- not impossible in view of the uncertainties described above. The other extreme would be the 2.1 kg estimated above for adults, which corresponds with 52% compensation. The true value probably lies somewhere in between.

4.     ‘it is remarkable that the largest part of the result was achieved already in the first 6 months of the study’
That is not remarkable, it is what the experimental data and the physiological models derived from them predict. Weight loss is initially rapid and then slows down as the amount of weight-dissipating tissue decreases. See the publications of Hall and of Katan, above, and the references in these papers.

5.     ‘The weight difference of 1 kg consisted for about 500 g of fat. The rest was LBM.’
About 0.3 kg of this lean body mass consists of the cell material, blood vessels and nerves in the fat tissue, plus the muscles and bones that the body creates to carry the extra fat. Gains or losses of such associated tissues always accompany gains or losses of body lipids (triglycerides). The triglycerides and the associated cell material together are called ‘obesity tissue’. About two thirds of obesity tissue is lipids, and the rest is non-lipidic.[vi]

6.     ‘I suppose that correction for the height difference may not result in a change of the BMI’.
No, the correction for height refers to lean mass only. The change in fat (lipids) was measured directly and is represented completely by the 0.5 kg.

7.     ‘the most simple explanation for the observed results is that considerable energy compensation has taken place.’
As discussed above, compensation was probably somewhere between 16% and 48%. The actual number is not very relevant for policy decisions. What is relevant is that one small can of sugary beverage produced a difference in obesity tissue of about 0.8 kg.
The results of our trial and of the accompanying trial of Ebbeling et al[vii] confirm the results of a large number of observational studies and short-term experiments, and they strongly suggest that sugar-containing beverages are fattening.
Such beverages are by far not the only cause of the obesity epidemic, but the evidence that their removal will reduce overweight is now compelling. Replacing juices and soft drinks by water – preferably tap water -- or diet drinks will help to reduce obesity and associated diseases, and should therefore be an urgent priority. 

Cordially,

Martijn B. Katan,
also on behalf of Janne C. De Ruyter, Margreet R. Olthof and Jacob C. Seidell