Table 3.
Assessments of Model Calibration in New Cohorts from the US CFFPR, 1993–2016*.
| Parameter | 1993–1997 Validation Cohort, n = 5,810 | New 1993–1997 Cohort, n = 9,941 | New 1993–1998, n = 9,757 | New 1999–2004, n = 13,073 | New 2005–2010, n = 15,043 | New 2011–2016, n = 17,635 |
|---|---|---|---|---|---|---|
| Using Eq. (2) | ||||||
| α0 | 0.046 (0.062) | 0.33 (0.055) | 0.62 (0.058) | 0.74 (0.060) | 0.59 (0.066) | 0.64 (0.062) |
| α1 | 1.00 (0.034) | 1.05 (0.027) | 1.06 (0.027) | 1.06 (0.025) | 1.03 (0.025) | 1.04 (0.024) |
| Using Eq. (2) with α1 fixed to be 1 | ||||||
| α0 | 0.029 (0.047) | 0.26 (0.037) | 0.52 (0.036) | 0.61 (0.035) | 0.53 (0.036) | 0.54 (0.033) |
*Results from using two strategies to assess model calibration are shown. See Statistical Analysis in Methods following Eq. (2). Results are shown as “Estimate (Standard Error).” Prognostic risk score is equivalent to log-odds of death within 5 years.