Results of the multivariate analysis of the prognostic value of eosinophilia using Cox regression methods
| Subtype(s) . | Endpoint . | Factors . | N . | P-value . | Hazard Ratio . | CI (HR) . |
|---|---|---|---|---|---|---|
| All | FFTF | Selection | 1200 | <.0001 | 1.60 | 1.28-1.99 |
| All | FFTF | All | 861 | <.0001 | 1.75 | 1.34-2.27 |
| All | SV | Selection | 1199 | <.0001 | 1.99 | 1.44-2.74 |
| All | SV | All | 861 | .0001 | 2.21 | 1.50-3.25 |
| NS | FFTF | Selection | 770 | .0004 | 1.67 | 1.26-2.22 |
| NS | FFTF | All | 589 | .0005 | 1.77 | 1.48-2.43 |
| NS | SV | Selection | 768 | <.0001 | 2.89 | 1.88-4.44 |
| NS | SV | All | 574 | <.0001 | 2.94 | 1.78-4.84 |
| MC | FFTF | Selection | 230 | .019 | 1.84 | 1.11-3.07 |
| MC | FFTF | All | 182 | .032 | 1.92 | 1.06-3.51 |
| MC | SV | Selection | 229 | .56 | 1.26 | 0.59-2.68 |
| MC | SV | All | 180 | .51 | 1.36 | 0.54-3.43 |
| Subtype(s) . | Endpoint . | Factors . | N . | P-value . | Hazard Ratio . | CI (HR) . |
|---|---|---|---|---|---|---|
| All | FFTF | Selection | 1200 | <.0001 | 1.60 | 1.28-1.99 |
| All | FFTF | All | 861 | <.0001 | 1.75 | 1.34-2.27 |
| All | SV | Selection | 1199 | <.0001 | 1.99 | 1.44-2.74 |
| All | SV | All | 861 | .0001 | 2.21 | 1.50-3.25 |
| NS | FFTF | Selection | 770 | .0004 | 1.67 | 1.26-2.22 |
| NS | FFTF | All | 589 | .0005 | 1.77 | 1.48-2.43 |
| NS | SV | Selection | 768 | <.0001 | 2.89 | 1.88-4.44 |
| NS | SV | All | 574 | <.0001 | 2.94 | 1.78-4.84 |
| MC | FFTF | Selection | 230 | .019 | 1.84 | 1.11-3.07 |
| MC | FFTF | All | 182 | .032 | 1.92 | 1.06-3.51 |
| MC | SV | Selection | 229 | .56 | 1.26 | 0.59-2.68 |
| MC | SV | All | 180 | .51 | 1.36 | 0.54-3.43 |
Analyses were performed for all cases, for nodular sclerosis (NS) cases only and for mixed cellularity cases (MC) only, for each of the 2 endpoints survival (SV) and freedom from treatment failure (FFTF). Two methods were used, including either all factors (“all”) or the selection of multivariate significant factors (“selection”), as described in Materials and Methods. “N” is the number of cases included, which varied as a result of missing values. The results of each analysis are given as follows: P-value for the significance of eosinophilia, allowing for the other included factors, estimated hazard ratio for eosinophilia (with eosinophilia versus without), and 95% confidence interval for this hazard ratio.