Table 2.
Logistic regression parameter estimates from models of miscarriage and log10 GCSF as determined by each of the different approaches to regression calibration to convert observed RFUs into concentration in units pg/mL – comparison of data calibrated using all calibration experiment data, or with outliers eliminated from the calibration experiment dataset
| Calibration models | As observed | Outliers removed | |||
|---|---|---|---|---|---|
| OR | 95% CI | OR | 95% CI | ||
| RC | |||||
| Collapsed | linear | 0.338 | (0.127, 0.901) | 0.271 | (0.101, 0.730) |
| Batch specific | linear | 0.73 | (0.457, 1.166) | 0.605 | (0.332, 1.103) |
| Mixed model | linear | 0.666 | (0.389, 1.141) | 0.564 | (0.299, 1.064) |
| Collapsed | curvilinear | 0.214 | (0.054, 0.843) | 0.250 | (0.087, 0.714) |
| Batch specific | curvilinear | 0.812 | (0.569, 1.16) | 0.977 | (0.934, 1.021) |
| Mixed model | curvilinear | ~ a | ~ a | ||
| IRC | |||||
| Collapsed | linear | 0.369 | (0.150, 0.909) | 0.282 | (0.108, 0.737) |
| Batch specific | linear | 0.629 | (0.356, 1.111) | 0.582 | (0.316, 1.073) |
| Mixed model | linear | 0.427 | (0.193, 0.943) | 0.53 | (0.275, 1.022) |
| Collapsed | curvilinear | 0.370 | (0.151, 0.909) | 0.287 | (0.111, 0.740) |
| Batch specific | curvilinear | 0.864 | (0.530, 1.408) | 0.666 | (0.383, 1.157) |
| Mixed model | curvilinear | 0.502 | (0.266, 0.948) | ~ a | |
Abbreviation: GCSF, granulocyte colony–stimulating factor.
a
Inestimable due to a near singular variance matrix for the random effect variance matrix.