Table 4.

Proportional hazard ratios (HR), 95% confidence intervals (CI) and chi-square P obtained by fitting survival data to 2- and 3-variable Cox proportional hazard model. The models yield qualitatively similar results when age as a continuous variable or second-order terms (cross-products of the independent variables) are added to the models, though none of these variables approach statistical significance. However, when cross-products are included in the 3-variable model fit to first failure as distant metastases, both IHC (HR = 4.99, CI 1.14–61.4, P = 0.030) and lymph node status (HR = 6.76, CI 2.12–25.2, P = .0015) are significant.

3-Variable Model	Proportional Hazard Ratio (Confidence Interval)
	p
	Any Failure	Local Failure	Distant Metastases
IHC:	1.07 (0.57–2.12)	0.81 (0.40–1.70)	2.78 (0.91–12.1)
<30 vs ≥30	p = 0.83	p = 0.58	p = 0.075
Lymph Nodes:	3.24 (1.66–6.16)	2.34 (1.00–5.03)	4.74 (1.83–12.4)
Positive vs Negative	p = 0.0008	p = 0.049	p = 0.0016
Histology:	2.56 (1.22–6.05)	5.37 (1.87–22.7)	1.13 (0.44–3.16)
Squamous vs Other	p = 0.012	p = 0.0008	p = 0.79
2-Variable Model	Proportional Hazard Ratio (Confidence Interval)
	p
	Any Failure	Local Failure	Distant Metastases
IHC:			2.82 (0.57–2.12)
<30 vs ≥30			p = 0.069
Lymph Node Status:	3.28 (1.69–6.18)	2.27 (0.98–4.83)	4.58 (1.83–11.64)
Positive vs Negative	p = 0.0007	p = 0.056	p = 0.0015
Histology:	2.57 (1.23–6.08)	5.27 (1.83–22.2)
Squamous vs Other	p = 0.0108	p = 0.0009