This letter is in response to the study by Jungheim et al. (1) in the September issue of Diabetes Care. They studied a subcutaneous glucose monitoring system in 23 subjects with diabetes over a 3-day period. They report a satisfactory mean absolute difference (MAD) between sensor values and capillary blood glucose of 14.8%.
I agree with the authors that their sensor, which uses microdialysis, appears to function in a stable fashion. It is impressive that they found it necessary to calibrate the device only once during this period, at a mean time of 4.7 h after insertion. As the authors point out, this stability (if verified) probably constitutes an advantage over intracorporeally placed sensors, which tend to undergo a decline in sensitivity over time and therefore require periodic recalibration.
However, I believe that another part of this report is misleading. Their first sentence states that devices for continuous glucose monitoring should display the data in real time rather than retrospectively. I think all workers in this field and all potential users of these devices would agree. It is for precisely this reason that I think that it is invalid for the authors to subtract 31 min from the time value for each sensed glucose value. I understand their rationale: the microdialysis system creates an intrinsic lag (time for fluid transportation through the dialysis system). Nonetheless, when a patient with diabetes uses this system in real time, this lag cannot be conveniently removed by subtraction. In fact, this could be a problem in the clinical situation of rapidly declining glycemia when the patient needs to know the glucose level immediately. I believe that in the displayed figure, the time values for the sensor readings were retrospectively corrected for this 31-min delay; thus, it would appear much different without such a correction. It would be more clinically relevant to show the actual time values. Furthermore, it appears that the MAD of 14.8% was also based on time-corrected values, and I suspect that the MAD would be substantially greater without this correction. Studies performed by my colleagues and I using intracorporeal sensors have not used corrections for sensor delay (2–6).
I think readers may have taken away a misleading message from this report, i. e., that dialysis-based sensors are likely to be more stable over time than intracorporeal systems. I think a more balanced set of conclusions would be that dialysis-based sensors are likely to be more stable over time than intracorporeal systems; on the other hand, dialysis-based sensors are more likely than intracorporeal sensors to exhibit error caused by delays intrinsic to the device, and judgments of superiority of one system over the other await prospective studies where sensor accuracy is assessed in real time.
Address correspondence to W. Kenneth Ward, MD, Associate Scientist, Legacy Clinical Research and Technology Center, 1225 NE 2nd Ave., Portland, OR 97232. E-mail: email@example.com.
W.K.W. is an employee of iSense.