Table 3.
Glucose sensor error/accuracy metrics | |||
---|---|---|---|
Relative difference* | Absolute relative difference† | ||
Rising | Mean | −8.1 | 11.0 |
95% CI | −5.8 to −10.3 | 9.3–12.6 | |
Median | −6.6 | 9.7 | |
n (data pairs) | 36 | 36 | |
Falling | Mean | 2.3 | 8.5 |
95% CI | 0.02–4.6 | 7.0–9.9 | |
Median | 1.2 | 7.4 | |
n (data pairs) | 36 | 36 | |
P rising vs. falling | < 0.001 | < 0.001 |
Relative (per cent) difference is defined as [(sensor glucose − reference glucose)/reference glucose] ×100. The magnitude of relative difference was greater (and more negative) during rising vs. falling glucose, indicating that the sensor error was substantially greater when glucose was rising, at a time when lag was greater.
Absolute relative difference refers to absolute value of the relative difference and was lower when glucose was falling, indicating greater accuracy in that condition.