Table 3. Predictive performance of prediction models A–D.
| Predictive performance | Model A | Model B | Model C | Model D | ||||
|---|---|---|---|---|---|---|---|---|
| Harrell’s C | 0.73 | (0.71–0.76) | 0.72 | (0.70–0.75) | 0.70 | (0.68–0.73) | 0.69 | (0.67–0.72) |
| Harrell’s C corrected for optimism by bootstrap | 0.73 | 0.72 | 0.70 | 0.70 | ||||
| Harrell’s C after 10-fold cross validation | 0.72 | (0.69–0.74) | 0.71 | (0.69–0.73) | 0.69 | (0.67–0.72) | 0.69 | (0.66–0.71) |
| Harrell’s C after censoring at 2 years of follow-up | 0.73 | (0.70–0.77) | 0.72 | (0.69–0.75) | 0.70 | (0.67–0.73) | 0.69 | (0.66–0.72) |
| Harrell’s C using complete cases only | 0.74 | (0.70–0.77) | 0.73 | (0.69–0.76) | 0.71 | (0.68–0.73) | 0.69 | (0.67–0.71) |
| Range of 2-year predicted risks* | 0%–54% | 0%–46% | 0%–32% | 0%–24% | ||||
| Harrell’s C after external validation | NA | NA | 0.64 | (0.62–0.66) | 0.65 | (0.63–0.66) | ||
| Baseline 2-year recurrence-free probability* | .9998462 | .9996196 | .9235595 | .9019939 | ||||
Model A denotes maximum model (i.e., all candidate predictor variables); model B, clinical variables and some laboratory markers easy to assess in the clinic; model C, clinical and genetic variables only; model D, clinical variables only.
*The baseline recurrence-free probability S0 can be used to calculate the absolute 2-year risk of recurrence for a patient using the following equation: risk of recurrence = 1 − S0** exp(prognostic score). Here, the prognostic score is equal to beta1*x1 + beta2*x2 + beta3*x3 + …, where the x1, x2 x3, etc., represent the variables in the prediction model, and beta1, beta2, beta3, etc., represent the corresponding regression coefficients.
Abbreviation: NA, not applicable