Here is the actual study (3 links deep): http://www.unitedhealthgroup.com/hrm/UNH_WorkingPaper5.pdf http://www.unitedhealthgroup.com/hrm/UNH_WorkingPaper5.pdf .

Here is the actual study (3 links deep): http://www.unitedhealthgroup.com/hrm/UNH_WorkingPaper5.pdf http://www.unitedhealthgroup.com/hrm/UNH_WorkingPaper5.pdf . Take a look at page 8, the one that contains this prediction. It was made by taking the rates of diabetes and pre-diabetes in 2007 and in 2010 (37.0% and 39.9% respectively), and drawing a line through them to get 51.9. That includes a 26.3% (2007) and 28.4% (2010) rate of "pre-diabetes", which really just means a HbA1c test that was higher than average. But between 2007 and 2010, the diagnostic test for pre-diabetes, the HbA1c, got much cheaper and became part of routine screenings where it wasn't before, so the slight uptick observed may actually be an artifact of increased testing, and it is definitely not part of a trend that can be linearly extrapolated ten years out.