Response to the Comment on “The Performance and Usability of a Factory-Calibrated Flash Glucose Monitoring System” by Bailey et al.

Dear Editor:

Freckmann et al. ¹ raise three important issues with our statement that flash glucose monitoring (FGM) is “intended to be a replacement” for fingerstick capillary blood glucose monitoring (BGM). ²

The first concerns the physiological lag between interstitial fluid (ISF) and blood compartments. However, the mean lag time between the sensor and venous reference is decreasing with newer continuous glucose monitoring (CGM) systems, and for the FGM system it was estimated to be just 4.5–4.8 min. ²

Their second concern is regarding the absolute accuracy required for nonadjunctive CGM use. Data from in silico simulations have suggested that the threshold acceptable mean absolute relative difference (MARD) should be between 7.5% ³ and 10%, ⁴ based on the predicted incorrect recognition or treatment of hypoglycemia and hyperglycemia. Although we agree that greater accuracy is desirable, we believe that using a qualitative impression of flattening in a graph generated even from an informed mathematical model ⁴ to predict diminishing returns from improved accuracy for a MARD lower than 10% may be misleading. A more pragmatic approach might be to establish the minimum MARD where benefit is seen with a given device in a specific clinical scenario. For example, clinical benefit was demonstrated in a closed-loop artificial pancreas system using sensors with a MARD of 14.2%. ⁵

It should be noted that CGM systems that require calibration are only as good as the BGM results used for the calibration. It is well understood that user errors in BGM can be significant. ⁶ The FGM system does not require a fingerstick calibration by the user, ⁷ thereby eliminating the inaccuracy potentially introduced by erroneous or missing fingerstick calibration measurements. Therefore, a factory-calibrated device (e.g., FGM) may have better “real-life” accuracy compared with a CGM device with similar stated accuracy.

The commenters invoke International Organization for Standardization (ISO) 15197:2003. However, both ISO 15197:2003 and ISO 15197:2013 specify requirements for in vitro glucose monitoring systems that measure glucose concentrations in capillary blood (i.e., not ISF) samples.

Finally, the commenters express concern for first-day MARD. It is well known that all sensor-based systems tend to show worse MARD on Day 1 compared with other times. ^{8 –10} The product labeling therefore suggests BGM confirmation when the sensor glucose level is low or results are suspect. To the extent that calibration impacts MARD, factory calibration has the potential to improve Day 1 performance.

Nonadjunctive use of point-of-care (vs. laboratory) glucose monitoring technology has been the subject of controversy since the introduction of capillary BGM in the 1970s. However, by the 1990s most diabetes specialists confidently recommended frequent BGM to adjust insulin doses (despite device accuracy inferior to current meters) based on results of the Diabetes Control and Complications Trial. ¹¹ Today, although this procedure is not approved, many persons with diabetes use CGM nonadjunctively to adjust their insulin doses. ¹² The additional information given regarding rate of glucose changes (i.e., “trend arrows”) may provide additional guidance that is not available with BGM point values. Preliminary results from the nonadjunctive use of a unique FGM device in people with type 2 diabetes on intensive insulin therapy suggest that, despite a lack of alerts, the frequency of hypoglycemia was significantly decreased with similar hemoglobin A1c level reduction compared with a standard BGM comparator group. ¹³

In summary, people with diabetes need better tools to help them avoid the well-known dangers of hypoglycemia and hyperglycemia. Clinical trials with new sensor technologies—alone (such as FGM) and connected to insulin delivery systems—will better inform us regarding their utility than a priori reasoning. Demonstrating equivalent or improved glycemic control using ISF-based CGM/FGM systems for self-management compared with self-monitoring of blood glucose should be the definitive method to establish the appropriate measure of clinical accuracy. ¹³