Table 3.
External Validation, Predictive Performance a
| Time | Outcome | Measure | Point of Consultation | 2–4 Wk After Consultation |
|---|---|---|---|---|
| 2 mo | Pain score | Calibration slope | 0.650 (0.549 to 0.750) | 0.848 (0.767 to 0.928) |
| CITL | 0.649 (0.506 to 0.792) | 0.378 (0.249 to 0.507) | ||
| R 2 median (IQR) | 11.1% (10.0% to 12.1%) | 25.3% (24.9% to 26.0%) | ||
| High pain | Calibration slope | 0.436 (0.338 to 0.535) | 0.657 (0.556 to 0.758) | |
| CITL | 1.081 (0.951 to 1.21) | 0.798 (0.665 to 0.931) | ||
| O/E | 1.599 (1.552 to 1.647) | 1.369 (1.329 to 1.41) | ||
| C statistic | 0.649 (0.618 to 0.679) | 0.726 (0.696 to 0.753) | ||
| Pseudo R2 median (IQR) | 12.1% (10.5% to 13.3%) | 27.1% (26.1% to 27.9%) | ||
| 6 mo | Pain score | Calibration slope | 0.593 (0.499 to 0.688) | 0.735 (0.656 to 0.815) |
| CITL | −0.93 (−1.088 to −0.773) | −1.262 (−1.408 to −1.116) | ||
| R 2 median (IQR) | 10.4% (9.4% to 11.4%) | 20.8% (20.2% to 21.3%) | ||
| High pain | Calibration slope | 0.526 (0.417 to 0.635) | 0.71 (0.598 to 0.823) | |
| CITL | 0.028 (−0.101 to 0.157) | −0.307 (−0.438 to −0.176) | ||
| O/E | 1.015 (0.984 to 1.046) | 0.88 (0.854 to 0.908) | ||
| C statistic | 0.663 (0.632 to 0.693) | 0.721 (0.692 to 0.749) | ||
| Pseudo R2 median (IQR) | 8.7% (7.6% to 10.2%) | 22.1% (21.1% to 23.1%) | ||
a Predictive performance of models for pain on external validation in the STarT MSK Main Trial data. CITL = calibration in the large; IQR = interquartile range; O/E = observed/expected ratio.