Ron Brazg, United States of America

Rainier Clinical Research Center Endocrinology

Presenter of 1 Presentation

ORAL PRESENTATION SESSION

PERFORMANCE OF THE GUARDIAN™ SENSOR 3 CONTINUOUS GLUCOSE MONITORING SYSTEM WITH NO CALIBRATION

Abstract

Background and Aims

Continuous glucose monitoring (CGM) is becoming an important tool for glycemic management and is an integral part of automated insulin delivery (AID) systems. Medtronic developed a new sensor algorithm for use with the Guardian™ Sensor 3 (GS3) glucose sensor that requires no calibration, and the present study assessed its performance.

Methods

There were 160 subjects (aged 18-80 years, ~64% with T1D) who wore GS3 sensors in the arm and abdomen. Data were compared with Yellow Springs Instrument (YSI) values and processed using the Zeus algorithm and incorporating different calibration schemes. The primary endpoint was determination of overall sensor glucose (SG) values within 20% of YSI reference, or ±20 mg/dL when SG was <80 mg/dL (20%/20). Other endpoints included overall mean absolute relative difference (MARD) between system and reference values, ±15 mg/dL at SG <70 mg/dL, and Consensus Error Grid (CEG) analyses.

Results

For no calibrations and the arm location, 20%/20 and ±15 mg/dL agreement rates were 88.0% (N=20,612 paired data points) and 89.6% (N=2456 paired data points), respectively. For the abdomen, the rates were 88.0% (N=18,423) and 92.1% (N=1963), respectively. MARD was 10.64% for the arm and 10.78% for the abdomen. MARD for the <70 mg/dL SG range was 12.61% for the arm and 14.78% for the abdomen. The CEG analyses determined 99.9% of points within the A+B range for both the arm (N=20,590) and abdomen (N=18,409) locations.

Conclusions

These data on the performance of the no-calibration Guardian™ Sensor 3 system may support non-adjunctive insulin dosing in standalone CGM and AID systems.

Hide