Just before lunch this afternoon I checked out the new FreeStyle Freedom Lite blood glucose monitoring system from Abbott Diabetes Care. The company's public relations firm, Fleishman-Hillard, was nice enough to send me one of the first of these cute little meters and 10 test strips.

After carefully washing my hands, I quickly used one test strip for each of my 10 fingers. Within 20 minutes I had used them up. In that period I also used up my patience with this meter.

In that time I didn't eat anything and didn't exercise. But my blood glucose level apparently varied from a low of 77 mg/dl (4.3 mmol/L) to a high of 96 mg/dl (5.3 mmol/L. In fact, that difference of 19 points came between 1:35 p.m. and 1:37 p.m., just two minutes apart.

My Blood Glucose Level at 1:35 p.m. today

My Level Two Minutes Later

I have no idea which of the 10 tests are right. The arithmetic mean (or average) was 87.6 mg/dl (4.9 mmol/L) and the middle value (or median) was somewhere between 88 mg/dl (4.9 mmol/L) and 91 mg/dl (5.1 mmol/L). I suppose that one of these numbers would be a pretty good guess. But do we really need to check our blood glucose 10 times to come up with an approximation?

We probably do. I also compared the results from the FreeStyle Freedom Lite with those from the AgaMatrix Jazz, which I praised for accuracy and precision here recently. The Jazz meter measured my blood glucose level consistently higher than Abbott's new meter did. But even the Jazz wasn't completely consistent.

Consistency -- which we technically call precision -- comes before accuracy, as I wrote in an article for Diabetes Health magazine in 2004. So why is this new meter not even consistent with itself?

Lots of people with diabetes believe that our blood glucose meters have to be correct within a 15 percent margin of error. Some people think the margin of error is 20 percent.

Guess what? The U.S. government agency that approves the sale of new meters has no accuracy and precision standards. None.

That agency is the U.S. Food and Drug Administration. The then director of the FDA’s Division of Clinical Laboratory Devices, Steve Gutman, M.D., told me years ago that accuracy is not something they consider when they approve the marketing of blood glucose meters. He added that he expects that we will have a standard for meter accuracy after the FDA accepts the proposed International Organization for Standardization (ISO) standard. That standard will allow a 20 percent error. I hear that we are still waiting to act.

The FDA approved the FreeStyle Freedom Lite meter last year. The agency stated, as it usually does when it approves the sale of a new meter, that it was satisfied that the new meter was "acceptable and comparable in terms of safety and effectiveness to the performance of the predicate device." In non-technical language this means that the FreeStyle Freedom Lite is as good as a meter that the FDA has already said could be sold in the United States, in this case an earlier Abbott Diabetes Care meter, the FreeStyle Freedom.