The standard ISO (International Organization for Standardization) 15197 is widely accepted for the accuracy evaluation of systems for self-monitoring of blood glucose (SMBG)1. However, this is not the same for sensor-based technology. The accuracy of a glucose sensor system is measured by a number of different metrics which may not be that well known and understood. Below, we look at how the FreeStyle Libre system and other sensors are measured for accuracy.
The first of these accuracy metrics is referred to as MARD or the Mean Absolute Relative Difference. This is a measure that shows on average how far away the glucose sensor reading is from a blood glucose reading, irrespective of whether the difference observed is more or less than the blood glucose reading. The MARD value is expressed as a percentage and the lower the percentage value, the more accurate the system. The metric came into prominence with the launch of continuous glucose monitoring (CGM) systems at the turn of the century2.
For comparison, the following MARDs are reported in product labelling when compared to a blood glucose meter. For the FreeStyle Libre system the MARD is reported to be 11.4%3, the MARD of the Abbott FreeStyle Navigator II CGM system was found to be 12.3%4 and the MARD of the Dexcom G5 Mobile is reported to be 12.5%5.
- The FreeStyle Libre system has proven accuracy of 11.4% MARD compared to a blood glucose reference3
- The FreeStyle Libre system is clinically proven to be accurate, stable and consistent over 14 days without the need for finger prick calibration3
A different metric used to demonstrate the clinical relevance of a glucose monitoring system is known as Error Grid Analysis (EGA). Two Error Grid types are commonly used; these are the Consensus Error Grid, sometimes also known as the Parkes Error Grid and the Clarke Error Grid2. Both Error Grids show the clinical acceptability of glucose monitoring system results based on how close they are to a reference glucose result. The Error Grids have zones from zone A to zone E. Results in zones A and B are considered to be clinically acceptable, whereas results outside of zones A and B may have a negative clinical outcome. The higher the percentage of results in zones A and B, the more clinically accurate the glucose system is.
For comparison the following percentages within zones A and B are reported in when compared to a blood glucose meter. For the FreeStyle Libre system the percentage of results within zones A and B is 99.7%,3 the percentage of results in zones A and B of the Abbott FreeStyle Navigator II CGM system is 97.7%,4 and the percentage of results within zones A and B for the Dexcom G4 is reported to be 98.6%3.
• For the FreeStyle Libre system, 99.7% of glucose results fall within Zone A and Zone B of the Consensus Error Grid.3
1. International Organisation for Standardisation. In vitro diagnostic test systems – Requirements for blood-glucose monitoring systems for self-testing in managing diabetes mellitus. ISO 15197:2015
2. Klonoff DC. Continuous Glucose Monitoring: Roadmap for 21st century diabetes therapy. Diabetes Care 2005;28(5):1231-1239
3. Data on File, Abbott Diabetes Care Inc. Clinical Report: Evaluation of the Accuracy of the Abbott Sensor-Based Interstitial Glucose Monitoring System 2014
4. Data on File, Abbott Diabetes Care Inc. Clinical Report: FreeStyle Navigator II Home Use Evaluation
5. Dexcom G5 Mobile User Guide LBL012787 Rev 005 MT22864
6. Parkes LJ, Pardo S, Slatin SL, Ginsberg BH. A New Consensus Error Grid to Evaluate the Clinical Significance of Inaccuracies in the Measurement of Blood Glucose. Diabetes Care 2000;23:1143-1148
7. Clarke WL, Cox D, Gonder-Frederick LA, Carter W, Pohl SL. Evaluating Clinical Accuracy of Systems for Self-Monitoring of Blood Glucose. Diabetes Care 1987;10(5):622-628
8. Pleus S, Schmid C, Link M, Zschornack E, Klotzer H, Haug C, Freckmann G. Performance Evaluation of a Continuous Glucose Monitoring System under Conditions Similar to Daily Life. Journal of Diabetes Science and Technology. 2013;7(4):833-841