This is a talk on Zoom about my new book, Let’s Ask Marion.
6:30 at the Jewish Community Center. Information and registration (required for Zoom link) here.
I spent a lot of time last week talking to reporters about the widely publicized study in PloS One that correlates sugar and diabetes.
The study is based on an econometric model of data food availability and diabetes prevalence in many countries. Such data are not particularly reliable, but the authors did the best they could with what they had. They are quite forthcoming about the limitations of their model and the data on which it is based [see addition below].
Their principal conclusion: for every 150 kcal/person/day increase in sugar availability (about one can of soda/day), diabetes prevalence increases by about 1%.
Because no other dietary, weight, or behavioral factor shows this kind of effect in their model, it is tempting to interpret the study as demonstrating that sugar is a risk factor for diabetes independent of calorie intake or body weight.
I’m not so sure. Take a look at the summary figures and decide for yourself.
Figure 1. Relationship between obesity and worldwide prevalence of diabetes.
Despite outliers, this figure shows an obvious and strong correlation between obesity and diabetes. Compare this to Figure 2.
Figure 2. Relationship of sugar availability to worldwide diabetes prevalence.
The correlation here is much less obvious. Without statistical tests, you could just as easily draw the line straight across the graph. The statistical significance is much weaker than that in Figure 1.
This means that these data cannot easily distinguish between several possibilities:
(a) Calories –> Obesity –> Diabetes
(b) Sugar –> Diabetes
(c) Sugar –> Calories –> Obesity –> Diabetes
While waiting for science to clarify these distinctions, the bottom line is the same for all of them.
As I explained in yesterday’s post, everyone would be healthier eating less sugar.
Addition: The authors have posted detailed comments about their methods.