American Association for Clinical Chemistry
Better health through laboratory medicine
September 2009 Clinical Laboratory News: Higher Standards on the Way for Glucose Meters?

CLN Banner Logo

September 2009: Volume 35, Number 9


Higher Standards on the Way for Glucose Meters?
FDA Seeking ISO Revision to Improve Performance Requirements
By Bill Malone


Critics say blood glucose meters need tighter standards.

The FDA will consider tightening standards for strip-based blood glucose meters, echoing the concerns of many in the clinical lab field who have warned for a decade that the meters are not reliable. In a letter dated June 24, FDA Commissioner Margaret Hamburg, MD, also acknowledged an issue that has been a growing concern among laboratorians—that poor outcomes in recent tight glucose control (TGC) trials might be, in part, due to the use of error-prone meters instead of more accurate point-of-care instruments.

Together, risks to patients who rely on meters for administering insulin or glucose, as well as problems for hospitals which use them for TGC, have raised the stakes in the agency’s choice to stand by the 7-year-old ISO guideline that established the minimum performance criteria for meters. To comply with the ISO 15197 document, 95% of a meter’s values must be within 20% of the reference method for glucose values ≥75 mg/dL and within 15% for lower values. However, critics say this error rate is unacceptable due to the fact that patients depend on their meters to take action and prevent hypoglycemia and other events.

In a 2001 paper, James Boyd, MD and David Bruns, MD, used computer modeling to demonstrate that very large errors in insulin doses could occur when patients use meters with greater than about 15% total error (Clin Chem 2001;47:209–214). “If you want 95 percent of insulin doses to be correct, you need a CV that’s 1 percent or 2 percent, and a bias of 1 or 2,” said Bruns, professor of pathology at University of Virginia in Charlottesville and director of clinical chemistry and toxicology at University of Virginia Health System. “Now, that’s probably not realistic for glucose meters—it’s more the quality of results that come out of a main lab with good equipment. But we think it is realistic to try and avoid very large errors in insulin dose, and that requires meters that do better than plus or minus 20 percent, and more like 15 percent total error.”

Balancing Act

Such criticism of the current standard is not unknown to FDA. Considering recent technological advancements, the agency has recommended that the minimum performance requirement in the ISO document be revised, and if not, “will consider taking action to no longer recognize that part of the standard, and may instead recognize other (higher) performance standards,” wrote Commissioner Hamburg.

However, the agency is also concerned that changing the standard could be burdensome for patients. “It is now clear to us that tighter standards would be challenging, although perhaps possible, for the industry to meet,” said Alberto Gutierrez, MD, director of FDA’s Office of In Vitro Diagnostic Device Evaluation and Safety. “But if the price for tightening the standards is an increase in the cost of the meters and strips—as industry claims—and therefore a likely decrease in use, the result could be a public health disaster.” Gutierrez cited evidence from clinical trials showing that use of glucose meters under current standards greatly increases the ability of diabetics to control their blood sugar levels and prevent long-term damage from the disease. Because healthcare for diabetics is expensive, large socioeconomic disparities exist in treatment. “Any action that reverses the trend of increased control by diabetics would be most unwelcome,” he added.

In fact, FDA worked very hard to tighten the standards before ISO 15197 was completed. In the end, the agency decided that on balance, there was more good in the document than bad, such as using total error to determine accuracy, the type of interferents tested for, and how to present data. When the standard was published, FDA had the option to accept only parts of it, but the agency realized that the meters coming into the market wouldn’t be able to do better, Gutierrez said. “Once the other parts of the standard had been implemented by manufacturers, including the type of data submitted and its presentation, it was clear to the FDA that many meters that were requesting clearance could not meet even the accuracy level the ISO standard was requiring. Therefore, the FDA began using the standard as a way to prevent the worst meters from reaching the market.”

Now, with almost a decade of innovation since the original ISO document was drafted, the agency will work to push for tighter standards for meters, with or without a revised ISO document. The FDA could write its own guidance requiring a tighter standard, but this would be a long process.

Off-Label Use

Beyond individual patients using meters to manage their diabetes, FDA has also recognized the problem of off-label use by hospitals of these basic meters for TGC. This is another issue that Bruns and other concerned clinical chemists have long sought to focus attention on. “It has been frustrating that for a decade now we’ve been telling people in the ICUs not to use these meters because they’re not accurate enough and not precise enough,” said Bruns. “If you’re trying to control glucose in the ICU at some mean concentration plus or minus 7 percent the target range, and you’re using a glucose meter to measure plus or minus 20 percent, it’s hopeless. Even with 15 percent it’s still hopeless. But people keep doing it.”

Bruns also noted that the recent NICE-SUGAR study, which cast doubt on the original landmark TGC trial and caused the medical community to question TGC, is flawed in his view. The study used several different devices—some not so precise and each reading differently—to measure glucose, yet used the same rule to determine how fast insulin should be pumped (NEJM 2009;360:1283–1297). Many hospitals make a similar poor assumption: that meters meant for self monitoring are good enough to use for TGC in hospitalized patients, and therefore will produce the same results as the original landmark study of TGC that used a precise arterial blood gas instrument.

For now, FDA plans a two-pronged approach when it comes to use of meters for TGC, educating hospitals and taking a hard look at how meters are marketed. “We plan to make sure that manufacturers do not market their devices for TGC, or if they want to, that they come to the FDA to get cleared or approved for that use,” said Gutierrez. “We plan to look carefully at how manufacturers are advertising their devices to hospitals, and if the FDA feels a manufacturer is making claims for which they are not cleared, the FDA will take appropriate action. Furthermore, we plan to work hard with professional societies to make sure that the CLSI standard on ‘Point-of-Care Blood Glucose Testing in Acute and Chronic Care Facilities’ clearly represents such use as off-label use and tightens the accuracy standards recommended for such use.”