Skip to main content
. 2018 Mar 3;9(2):683–697. doi: 10.1007/s13300-018-0392-6

Table 3.

Accuracy results based on ISO 15197:2013 criteria, results within consensus error grid (CEG) zones A and B, results within surveillance error grid zones with “no risk” or “slight lower risk”, relative bias according to Bland and Altman, and calculated insulin dosing error

System User Results within Relative bias Calculated insulin dosing error
±15 mg/dl/± 15% CEG zones A + B SEG zones
No risk Slight, lower risk Median (50th percentile) 99% ranges between the 0.5th and 99.5th percentile
Risk score
0–0.5 > 0.5–1.0
% % n % Units
A Lay-users 96 100 100 0 − 7.6 − 0.9 − 2.7 to − 0.1
Study personnel 99 100 197 3 − 6.8 − 0.8 − 2.4 to 0.0
B Lay-users 100 100 100 0 − 2.2 − 0.3 − 1.0 to + 0.3
Study personnel 100 100 200 0 − 2.2 − 0.2 − 1.0 to + 0.2
C Lay-users 98 100 97 3 + 2.8 + 0.3 − 0.6 to + 1.4
Study personnel 99.5 100 198 2 + 1.6 + 0.2 − 0.7 to +1.4
D Lay-users 92 100 97 3 − 5.6 − 0.6 − 2.4 to +0.6
Study personnel 96 100 193 7 − 5.3 − 0.7 − 2.5 to + 0.3

Results are shown for measurements performed by lay-users (n = 100) and by study personnel (N = 200) when evaluated by using the manufacturer’s comparison method (hexokinase for system A, glucose oxidase for systems B, C, and D)