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

Dear Editor:

Bailey et al. ¹ recently presented results obtained with a factory-calibrated flash glucose monitoring (FGM) system. In their article, the authors provide results of a study in which 72 subjects wore two FGM systems in parallel for 14 days.

We would like to initiate an open discussion about one statement in the introduction saying that the FGM system is “intended to be a replacement for capillary [blood glucose] measurement.” The statement, which can also be found in the system's instructions for use, is possibly challenged by the results presented.

One concern is the fact that glucose is measured in two different compartments: blood versus interstitial fluid (ISF). In periods with stable glucose levels, the concentrations measured will be more or less the same; however, in periods with rapid changes (“glucose swings”), measurement results can differ because of physiological processes. ²

In a publication regarding non-adjunct use of continuous glucose monitoring (CGM) data, Kovatchev et al. ³ concluded that CGM systems should achieve mean absolute relative difference (MARD) results of 10% or better (i.e., lower) compared with blood glucose (BG) results to enable reliable insulin dosing based on CGM results. This conclusion is derived from an in silico estimate for CGM data, not FGM data. However, considering that the same measurement principles (i.e., glucose measurement in ISF) are used with CGM and FGM, with differences concerning automated data transfer and need for calibration with CGM, this should be applicable to FGM as well. Figure 2 in the article of Bailey et al., ¹ however, shows that approximately 50% of measurement results had a MARD of 10% and more (up to 21–23% MARD).

From a clinical point of view, at least some of the FGM results might have led to clinically relevant insulin dosing errors. In practice, both patients and clinicians do not know if the measurement results shown happen to be “good” or “bad” sensor data when making the dosing decision. In addition, results are also reported for percentage of sensor readings within ±15 mg/dL and ±20% of capillary and venous BG readings, which corresponds to the system accuracy limits of the International Organization for Standardization protocol ISO 15197:2003. Systems for self-monitoring of BG should show at least 95% of results within these limits, whereas the FGM system showed 86.2% and 82.8% against capillary and venous BG, respectively. ¹ Another issue is the run-in effect observed with all sensors measuring glucose in ISF. For the FGM system, the MARD results on the first day can be calculated as approximately 15% and 17% against capillary and venous BG, respectively. This is well above the MARD results observed during Days 2–14, and such MARD results also challenge the replacement claim. At least on this day, additional BG measurements are advisable to ensure patient safety.

We would like to emphasize that the results presented ¹ about the performance of the FGM system are impressive and encouraging; however, we see the need for a discussion in the scientific community regarding the feasibility of the replacement of BG measurements with FGM and CGM systems.