Assessment of different techniques for subcutaneous glucose monitoring in Type 1 diabetic patients during 'real-life' glucose excursions.
Publikation aus Health
Mader J., Weinhandl H., Plank J., Bock G., Korsatko S., Ratzer M., Ikeoka D., Köhler G., Univ.-Prof. Dr. Thomas Pieber, Ellmerer M.
Diabetic Medicine 27 (3): 332-338., 2010
To compare the accuracy of two marketed subcutaneous glucose monitoring devices (Guardian RT, GRT; GlucoDay S, GDS) and standard microdialysis (CMA60; MD) in Type 1 diabetic patients.
Seven male Type diabetic patients were investigated over a period of 26 h simulating real-life meal glucose excursions. Catheters of the three systems were inserted into subcutaneous adipose tissue of the abdominal region. For MD, interstitial fluid was sampled at 30- to 60-min intervals for offline glucose determination. Reference samples were taken at 15- to 60-min intervals. All three systems were prospectively calibrated to reference. Median differences, median absolute relative differences (MARD), median absolute differences (MAD), Bland-Altman plot and Clark Error Grid were used to determine accuracy.
Bland-Altman analysis indicated a mean glucose difference (2 standard deviations) between reference and interstitial glucose of -10.5 (41.8) % for GRT, 20.2 (55.9) % for GDS and 6.5 (35.2) % for MD, respectively. Overall MAD (interquartile range) was 1.07 (0.39; 2.04) mmol/l for GRT, 1.59 (0.54; 3.08) mmol/l for GDS and 0.76 (0.26; 1.58) mmol/l for MD. Overall MARD was 15.0 (5.6; 23.4) % (GRT), 19.7 (6.1; 37.6) % (GDS) and 8.7 (4.1; 18.3) % (MD), respectively. Total sensor failure occurred in two subjects using GRT and one subject using GDS.
The three investigated technologies had comparable performance. Whereas GRT underestimated actual blood glucose, GDS and MD overestimated blood glucose. Considerable deviations during daily life meal glucose excursions from reference glucose were observed for all three investigated technologies. Present technologies may require further improvement until individual data can lead to direct and automated generation of therapeutic advice in diabetes management.