Table 6.
The final simple and multivariate Cox proportional hazard regression model of prognostic factors associated with the survival of NPC patients in Hospital USM (n = 134)
| Variables | Simple Cox regression | Multiple Cox regression | ||||
|---|---|---|---|---|---|---|
| b | Crude HR (95% CI) | p value | b | Adjusted HR (95% CI) | p value | |
| Age (year) | 0.02 | 1.02 (1.01, 1.04) | 0.005 | 0.03 | 1.03 (1.01, 1.04) | 0.001 |
| Metastases | ||||||
| No | 0.00 | 1.00 | – | 0.00 | 1.00 | – |
| Yes | 1.11 | 3.02 (1.29, 7.09) | 0.011 | 0.92 | 2.52 (1.01, 6.28) | 0.048 |
| AJCC stage | ||||||
| I–II | 0.00 | 1.00 | – | 0.00 | 1.00 | – |
| III | 0.45 | 1.81 (0.75, 4.38) | 0.190 | 0.88 | 2.42 (0.98, 5.99) | 0.056 |
| IV | 0.44 | 3.41 (1.44, 8.08) | 0.005 | 1.40 | 4.05 (1.66, 9.88) | 0.002 |
A backward stepwise Cox proportional hazards regression model was applied. Two-way interaction and multicollinearity were assessed and not found. A log-minus-log plot, a hazard function plot and Schoenfeld residuals were applied to test the model assumptions. Martingale residuals, Cox-Snell residuals, deviance residuals and influence analysis were used to assess the fit of the model and to identify influential cases
b regression coefficient