Table 2. Results for Passing–Bablok regression (slope, intercept) and Bland–Altman plot analysis (percentage bias, limits of agreement (LoA)).
| PB regression | BA plot analysis | |||||
|---|---|---|---|---|---|---|
| Index test | Number of samples | Slope (two-sided 95%CI) | Intercept (two-sided 95%CI) | Number of samples | Mean percentage difference (two-sided 90%CI) | LoA (k-factor approach) |
| QuikRead go* | 34 | 0.963 (0.930, 0.988) | -0.483 (-0.998, -0.039) | 34 | ==7.2% (-9.0%, -5.3%) | -24.8% to 10.5% |
| Spinit* | 38 | 0.921 (0.875, 0.936) | -0.543 (-0.807, 0.369) | 38 | -11.1% (-12.5%, -9.7%) | -25.7% to 3.5% |
| INCLIX | 37 | 0.896 (0.809, 0.975) | 3.042 (0.214, 6.635) | 37 | -1.2% (-4.7%, 2.3%) | -41.0% to 38.6% |
| GS1200 | 38 | 0.843 (0.772, 0.904) | -1.246 (-3.576, -0.516) | 38 | -26.7% (-31.7%, -21.6%) | -74.3% to 20.9% |
| LS 4000 | 39 | 0.822 (0.735, 0.939) | 0.495 (-1.401, -1.329) | 39 | -12.5% (-16.7%, -8.2%) | -50.3% to 25.4% |
| IFP-3000 | 35 | 1.345 (1.27 7,1.411) | -1.251 (-3.947, 1.409) | 31 | 23.4% (17.7%, 29.0%) | -28.8% to 75.5% |
| EPITHOD 616 | 36 | 1.390 (1,390, 1.303) | -0.937 (-3.679, 0.765) | 36 | 31.3% (25.0%, 37.5%) | -29.1% to 91.6% |
| Standard F 200 | 22 | 1.571 (1,379, 1,948) | -4.278 (-10.358, -2.468) | 22 | 28.0% (14.3%, 41.6%) | -79.2% to 135% |
Tests were ranked based on the best slope (i.e., the nearest to 1).
The number of samples, each measured in duplicate, refers to the number included in the specific analysis out of 40 samples. Where this number differs from 40, it means that some samples were excluded from the analysis because the values were (for PB and BA) outside the AR or (for BA) violated the assumption of homogeneity of variances (mean <3 mg/L, IFP 3000).
*Spinit and QuikRead go showed the best slope and intercept, near to 1 and 0, respectively.
NB: One outlier found with both replicates was excluded for LS4000. For EPITHOD 616 and GS1200, strong deviating replicates were found for one sample in both cases, so a third measurement was then performed and the pair with similar values was chosen for analysis.