Table 3.
The Types and Numbers of Correct Predictions Obtained From the Cox Proportional Hazards Model
| Prediction Type | No Covariates | SP Included | SP Plus YKL‐40 Included |
|---|---|---|---|
| Composite outcome | |||
| True favorable predictions, n (%) | 2658 (44.5) | 2910 (48.7) | 2909 (48.7) |
| True unfavorable predictions n, (%) | 1115 (18.7) | 1174 (19.7) | 1179 (19.7) |
| Total number of true predictions, n (%) | 3773 (63.2) | 4084 (68.4) | 4088 (68.5) |
| All‐cause mortality | |||
| True favorable predictions, n (%) | 4768 (79.9) | 4585 (76.8) | 4582 (76.7) |
| True unfavorable predictions, n (%) | 0 (0)a | 392 (6.57) | 398 (6.67) |
| Total number of true predictions, n (%) | 4768 (79.9) | 4977 (83.4) | 4980 (83.4) |
For comparison results without covariates, with standard predictors included, and with YKL‐40 included, are shown. For each patient at time (T) equal to 3, 6, and 9 years and using the patient's individual survival curve the predicted outcome (patient “alive at T” (favorable outcome) compared with patient “not alive at T” [unfavorable outcome]) was read off the survival curve and the results compared with the observed outcome. When no covariates were included in the model, this coincides with the Kaplan‐Meier survival curve which was used to calculate the predictions. To allow for the fact that age violated the proportional hazard assumption we stratified by age categories in addition to center and excluded age from the covariates. Total predictions=5970, n (%)=true favorable predictions/total predictions. SP indicates standard predictors.
As total mortality is <50%, all predictions are automatically favorable (and 79.9% of them pertain to a so far survivor).