The threshold fasting reading for a diabetic diagnosis was lowered from 140 to 126 in the late 1990's, which is where the slope of the curve in the above graph undergoes a large change. ...
I also noticed the word "diagnosed" in there, and was wondering that same thing.
I'd be surprised if someone would fail to correct for that ...
I wouldn't be surprised at all. I see it all the time. I would not assume it was corrected until I saw something that said it was corrected. Esp since it is pretty well known that the definition has changed, I might actually expect an asterisk on that graph with the explanation right there in a footnote on the graph. Just takes a couple words, the details can be in the backup. Lacking that, "diagnosed" is "diagnosed", no more, no less.
Another question(s) on these meters and strips (for anyone). I read up a bit, and from what I gather, there is a chemical reaction on the strip when exposed to glucose, and then the meter reads some effect on the strip that correlates to this reaction (the chemical reaction has some measurable electrical response?).
The strips have a manufacturing tolerance, and a code to indicate their offset (IOW, it is hard/expensive to manufacture these strips to tight tolerances, but easy to characterize them by measuring something). So you've got test strip tolerances, and then meter tolerances in reading the effect of the reaction. So my questions:
Can you read a single prepared strip multiple times (with diff meters or the same meter), or does reading it alter the effect, so subsequent readings of that strip are not meaningful?
Can you buy/make a reference solution of glucose, to validate the strip/meter? Maybe this is complicated by all the other stuff in blood, but it isn't that hard to make up dilutions to achieve any ppm you want.
Home brewers often do that with their hydrometers. They are down to single digit concentration levels, but dilutions are easy (768 tsp per gallon). Two dilutions gets you to 1.65ppm.
edit/add:
How would they correct for it? They look at the diabetes statistics, not at what blood sugar level the diagnosis was made. How could they?
Ha
Easy if they had the raw data. Tricky, but not impossible to be reasonably accurate through other means (i.e, if they knew the distribution of people between 126 and 140 prior to the change, it seems they could detect a deviation from this).
-ERD50