Are Meters Accurate Enough?
More accurate blood glucose meters would help all of us. The manufacturer of my meter says that if the meter gives a reading that is within 20 percent, it is working fine. This means that if the true reading is 100, I could get results ranging from 80 to 120 and the meter is working fine. Are there reasons for this inaccuracy? Has anyone tried to make them better? Karl D. Smith, Broadview Heights, Ohio
David E. Bruns, MD, University of Virginia School of Medicine, responds:
These are concerns with all medical testing: How accurate are the methods that are available? How accurate must the methods be to meet clinical needs? Both questions are unavoidable because all methods of measurement have error. Both questions are difficult to answer especially in the case of blood glucose meters.
Many studies have attempted to evaluate the accuracy of blood glucose meters, but many, if not most, of the studies suffer from deficiencies. For example, many papers report the performance of a meter only in the hands of a single highly skilled medical technologist, working under ideal conditions, often paid by the manufacturer (directly or indirectly), sampling a few strips and meters. Such a study does not tell us how the meters perform in the hands of people who are not highly trained medical technologists and who will be doing the testing day or night, under less-than-ideal conditions, potentially with distractions, vision problems, and all. Some of the better studies do address this question, and find that meters do not perform as well under real-life conditions as in laboratories.
Having said that, however, it is also clear that meters have improved over time. Many of the innovations in meters can be expected to make the quality of results less dependent on the environment and on the operator's skill. Such improvements include "no-wipe" strips, automatic timing, automatic detection of sample volume, and small sample size.
It is difficult to define how well meters perform and how accurate they need to be. We can probably all agree that it would not matter much if a meter reported a result of 99 mg/dl as 100 mg/dl. In contrast, we surely would agree that it would be terrible if a meter reported a glucose of 25 mg/dl as 100 mg/dl. Somewhere between such extremes we must find where an error makes a difference.
The current Guidelines on Laboratory Testing in Diabetes propose a goal of less than 5 percent total error for glucose meters. The U.S. Food and Drug Administration does not require such a high performance, and no meters have been shown to meet that goal in real-world use. That is not to say, however, that the current performance of meters is such as to make them unacceptable. It simply says that the technology, despite all the improvements, may not yet be at a level to produce a meter with that characteristic. Moreover, we do not really know if a person's glucose control will be improved by using a meter with a total error less than 5 percent rather than a meter with a total error of 10 percent.
Clearly, use of glucose meters, even the earlier kinds used in the Diabetes Complications and Control Trial, has had a profound impact on control of blood glucose.