This article and accompanying editorial decry the overuse of the A1C test in people with well-controlled diabetes. I find this extremely informative, but not for the same reasons as the authors. Informative in 2 areas; the tendency to confuse guidelines with evidence-based clinical studies, and the unsophisticated use of large commercial databases. Starting with the guidelines for A1C testing, these are clearly a consensus recommendation not based on clinical studies. They have a certain appeal, but most clinical studies associate more frequent visits and more A1C tests with better control. Perhaps checking an A1C every 3 months is helpful to some well-controlled people with diabetes, because it confirms the continuing benefit of their efforts. I’m not sure that the authors’ perjorative term of ‘frequent’ testing for 3-4 yearly tests has any basis. Checking 5 or more times might seem excessive, or it could be due to a problem noted by the authors, that this ‘excessive’ testing was associated with more visits to a greater number of doctors, and could be due to lack of knowledge of previous A1C tests.
However I think that a bigger problem stems from the use of such a large dataset. Out of a large defined population of 31,545 people with A1Cs under 7.0%, 1,830 (5.8%) had 5 or more A1C tests in a year. This 1,830, however, were identified out of a population of “more than 86 million”. Just as clinical trials are highly encouraged to use CONSORT flow diagrams to describe their patient selection, reports such as need to provide us with further information so that we can judge the validity of the conclusions. Since no such explanation was provided, let’s try some ballpark estimates. Eighty six million adults in the database, and a national prevalence of diagnosed diabetes of around 6.7% gives us about 5.7 million people with diabetes. Excluding about 30% who take insulin provides about 4 million people eligible for analysis. If 1,830 had excessive A1C testing, that would be 0.05% of the people with diabetes. That amount of overtesting is clearly not a concern. However, these numbers also call into question the accuracy or completeness of their data, as the 31,545 people with consecutive A1Cs below seven also seems to small, representing just 0.8% of the potential 4 million people with non-insulin treated diabetes.
It’s hard for me to think of a less important concern in the current treatment of diabetes than overuse of the A1C test. Oh, by the way, total laboratory true cost including personnel and overhead is a little over $2.00.