I posted this on Nutrition and Metabolism Facebook in response to a similar request. It is, however, specifically critical of the paper as in my own comments on this page.

Figure 2 from the paper (shown at https://dl.dropbox.com/u/... with highlighted parts) shows how changes in sugar availability in a country are associated with changes in the prevalence of diabetes. Strictly speaking, such associations are not necessarily predictive but they are always interpreted that way -- certainly in this case which got so much press coverage on the risk for sugar.

The line drawn in Figure 2 shows you the average drift of the data points but I make the point that there is too much variation for this to be meaningful...

... imagine that your country has not increased sugar availability and has not changed diabetes level (center of the graph). Now what happens if your government lowers availability of sugar. According to the graph, you are almost as likely to see an increase in diabetes (red points) as a decrease in diabetes (original blue points) and since we don't really know what causes diabetes, the risk may be real, that is, people who get diabetes may avoid sugar. For example, it is popularly thought, correctly or incorrectly, that sugar causes diabetes and people with a family history might specifically avoid sugar. Whatever the reasons, the figure says that you are taking a risk in having your country lower sugar availability -- in fact, the 5 countries with the highest change in diabetes had lowered their sugar availability.