MARDS

The MARD values of the simulated glucose sensors, expressed in ranges, give a more practical interpretation of the above results.

Accordingly, it is advisable to adopt a stricter maximum MARD for a blood glucose meter to deal with use in clinical practice.

Third, although the MARD is a quick method to assess glucose sensor accuracy, there is no consensus on the gold standard glucose metric.

Our data suggest that the MARD should preferably be smaller than 7.1% for intermittent glucose sensors and 11.0% for CGM devices and never be higher than 15.7% for intermittent glucose sensors and 17.8% for CGM devices.

The optimal MARD scores of <10 or <11 suggested, respectively, in Wilinska and Hovorka (8) and Van Herpe et al.

[2] Nonstandard abbreviations: ICU, intensive care unit; CGM, continuous glucose monitoring; MARD, mean absolute relative difference; TE, total absolute error.