SAN DIEGO – Slightly more than half of veterans targeted for screening have unrecognized diabetes or prediabetes, results from a recent analysis showed.

However, screening such patients by measuring hemoglobin A_1c "would result in major misclassification – missing disease when it is present and, to a lesser extent, mislabeling normals as having disease," Sandra L. Jackson, M.P.H., said at the annual scientific sessions of the American Diabetes Association.

The findings are based on a study of 789 individuals from the Atlanta VA Medical Center that assessed the use of targeted screening to detect prediabetes and diabetes, and to compare HbA_1c testing with the oral glucose tolerance test (OGTT), said Ms. Jackson, a graduate student in the nutrition and health sciences doctoral program at Emory University, Atlanta.

Although screening to detect unrecognized diabetes and prediabetes is recommended, the best strategy for screening in patients in primary care settings is unknown. The upside of the OGTT, Ms. Jackson said, is that it’s established in clinical use, it can detect all patients with prediabetes, and the glucose measurement itself is accurate. However, "on the downside, it requires [fasting] and morning testing. It’s burdensome for patients and health care systems, and it has poor day-to-day reproducibility."

The upside of HbA_1c, she continued, is that it does not require a fast, "and it’s only a single blood draw, so it’s much more convenient, there’s less day-to-day variation, and there’s greater preanalytic stability. On the downside, measurement may be problematic as platforms vary, point-of-care testing can be unreliable, there’s a lack of agreement on cutoffs, and there may be racial and age disparities such that blacks and older persons may have higher HbA_1c independent of glucose."

The researchers defined hyperglycemia according to American Diabetes Association (ADA) criteria: prediabetes as a fasting OGTT of 100-125 mg/dL or a 2-hour OGTT of 149-199 mg/dL, and diabetes as a fasting OGTT of 126 mg/dL or greater or a 2-hour OGTT of 200 mg/dL or greater.

They categorized HbA_1c results according to three sets of diagnostic criteria: the International Expert Committee (IEC) (prediabetes 6.0%-6.4%, diabetes 6.5% or greater), ADA (prediabetes 5.7%-6.4%, diabetes 6.5% or greater), and Department of Veterans Affairs/Department of Defense (VA/DoD) (prediabetes 5.7%-6.9%, diabetes 7.0% or greater).

The mean age of the 789 study participants was 58 years, 95% were men, 74% were black, and their mean BMI was 30.5 kg/m².

Screening was offered to patients meeting National Institutes of Health guidelines for screening: without known diabetes, and with age greater than 45 years or a BMI of more than 25 with another risk factor.

Fully 10% of patients met criteria for diabetes based on the OGTT, which was a higher rate compared with the HbA_1c guidelines (6.7% by the IEC, 6.7% by the ADA, and 1.5% by the VA/DoD guidelines, respectively). "This would indicate that these cutoffs are insensitive compared with the OGTT for detecting diabetes," she said.

According to the OGTT, 42% had prediabetes: 27% had isolated impaired fasting glucose, 6% had isolated impaired glucose tolerance, and 9% had both.

In patients with diabetes by OGTT criteria, HbA_1c classification by IEC criteria labeled 32% correctly, 38% incorrectly as having prediabetes, and 29% incorrectly as being normal; ADA criteria labeled 32% correctly, 50% incorrectly as having prediabetes, and 18% incorrectly as being normal; and VA/DoD criteria labeled 12% correctly, 71% incorrectly as having prediabetes, and 18% incorrectly as being normal.

In patients with prediabetes by OGTT criteria, HbA_1c classification by IEC criteria labeled 36% correctly, 6% incorrectly as having diabetes, and 59% incorrectly as being normal; ADA criteria labeled 61% correctly, 6% incorrectly as having diabetes, and 33% incorrectly as being normal; and VA/DoD criteria labeled 66% correctly, 1% incorrectly as having diabetes, and 33% incorrectly as being normal.

The prevalence of diabetes increased in a stepwise fashion with increasing BMI, from 1.5% among those with a normal BMI (18.5-24.9) to 15% among those who met criteria for class III obesity (BMI more than 40). "For every 1 unit increase in BMI, we observed a 10% increase in the odds of having diabetes," she said.

Ms. Jackson also reported that with the IEC, ADA, and VA/DoD cutoffs for diabetes, screening with HbA_1c was specific but insensitive, with a false negative rate of 68% at the 6.5% cutoff and a false negative rate of 89% at the 7.0% cutoff.

"Many veterans – and probably many Americans – targeted by national guidelines have unrecognized diabetes and prediabetes, yet screening with HbA_1c would miss many," she said. "Given these findings, we should consider or develop alternative diabetes screening strategies that can be used opportunistically during outpatient visits, without fasting, but are more accurate."

The study was supported by a grant from the VA’s Health Services Research and Development Service. Ms. Jackson said that she had no relevant financial conflicts of interest.

HbA1c Misclassifies Patients in Diabetes Screening Program

HbA_1c Misclassifies Patients in Diabetes Screening Program