Table 2:
Adherence with analytical performance specifications for comparators (C-APS).
| Sample | Bias | Imprecision | C-APS met2 | ||||
|---|---|---|---|---|---|---|---|
| Device | n | original | Corrected1 | n | CV | original | final |
| Venous | |||||||
| INT | 184 | +3.0% | +0.0% | 2059 | 0.3% | No | Yes |
| YSI 1 | 64 | -2.3% | +0.1% | 706 | 0.2% | No | Yes |
| YSI 2 | 55 | -1.5% | +0.0% | 630 | 0.3% | Yes | Yes |
| YSI 3 | 65 | -1.7% | +0.0% | 724 | 0.6% | Yes | Yes |
| Capillary | |||||||
| CNX (vs. INT) | 142 | +2.8% | +0.1% | 2061 | 1.6% | No | Yes |
| CNX (vs. YSI) | 142 | +3.8% | +0.2% | 2061 | 1.6% | No | Yes |
As bias was corrected retrospectively, compliance with C-APS is given before and after retrospective correction.
CGM: continuous glucose monitoring system, FDA: Food and Drug Administration, BGMS: blood glucose monitoring system
CV: coefficient of variation, INT: Cobas Integra 400 plus, YSI: YSI 2300 STAT Plus, CNX: Contour Next
1 For venous samples on INT and the three YSIs, bias was retrospectively corrected based on a split-sample approach with a higher-order method. For capillary CNX samples, bias correction was implemented using a split-sample approach using corrected capillary INT or YSI results (as indicated in the table).
2 “Original” indicates application of C-APS to original bias and imprecision, whereas “final” indicates application of C-APS to retrospectively corrected bias and imprecision.