Systems for self-monitoring of blood glucose (SMBG) have to be sufficiently accurate for effective glycemic control. In accuracy studies, the selected comparison method can influence the apparent level of measurement accuracy. This study aimed at assessing the impact of two different comparison methods on the apparent accuracy of 18 CE-marked current-generation SMBG systems.
Measurement accuracy was assessed in a setting based on ISO 15197:2013 and additional analyses were performed, including minimal deviation within which 95% of results were found, relative bias, and mean absolute relative difference (MARD). Comparison measurements were performed on a glucose oxidase-based (GOD) and a hexokinase-based (HK) method in parallel.
While ≈80% of SMBG systems met the accuracy criteria of ISO 15197 (i.e., ≥95% of results found within ±15 mg/dl or ±15% of comparison method results) when compared to the manufacturers reference method, only ≈67% did so independently from the applied comparison method. The minimal deviation within which 95% of results were found ranged from 7.7 to 19.7 mg/dl or % (GOD-based method) and from 5.3 to 20.1 mg/dl or % (HK-based method). The mean relative bias ranged from -6.6% to +5.7% (GOD-based method) and from -11.1% to +1.3% (HK-based method), whereas MARD results ranged from 3.7% to 9.8% and from 2.3% to 10.5%, respectively.
The results of this study indicate that the selected comparison measurement method may have an appreciable impact on SMBG systems’ apparent level of accuracy. Therefore, further steps towards harmonization of comparison method results are important.