Table 4 Results of the flexible spline-based model (N=269)

From: Flexible modeling improves assessment of prognostic value of C-reactive protein in advanced non-small cell lung cancer

Variables

HR (95% CI) a

P -value for test of no association

Stage: (IIIB+pleural effusion/4 vs IIIA/IIIB)

1.859 (1.284, 2.691)

<0.001

ECOGb performance status: (2 vs 0–1)

1.336 (0.923, 1.935)

0.116

Smoking status: (ever vs never)

2.248 (1.419, 3.561)

<0.001

Chemotherapy type: (single vs double)

1.462 (0.990, 2.160)

0.041

Log2 CRP: (per doubling of CRP values)

*

0.003 (overall P-value)#

Albumin: (per ↓c of 1 g l–1)

**

0.001 (overall P-value)#

Log2 LDH: (per doubling of LDH values)

2.281 (1.661, 3.142)

<0.001

Alkaline phosphatase: (per ↑of 10 U l–1)

1.012 (0.980, 1.041)

0.366

Neutrophil counts: (per ↑d of 1 × 109 l–1)

1.072 (1.025, 1.122)

0.001

Lymphocytes: (per ↓ of 1 × 109 l–1)

1.313 (1.035, 1.666)

0.012

Deviancee

1873.3

 

AIC

1909.3

 
  1. Abbreviations: AIC=Akaike's information criterion; CRP=C-reactive protein; LDH=lactate dehydrogenase.
  2. aAdjusted hazard ratio (HR) and 95% confidence interval (95% CI).
  3. bEastern cooperative oncology group.
  4. c↓: decrease.
  5. d↑: increase.
  6. eDeviance=−2*log-likelihood.
  7. *Both the time-dependent (P=0.033) and non-linear (P=0.015) effects were significant. The estimated non-linear effects, at selected follow-up times, are shown in Figure 2.
  8. **Both the time-dependent (P=0.0001) and non-linear (P=0.038) effects were significant. The estimated non-linear effects, at selected follow-up times, are shown in Figure 3.
  9. #P-value for a likelihood ratio test, with 5 degrees of freedom, of the null hypothesis of no association, obtained by comparing the deviances of (i) a flexible model where both time-dependent and non-linear effects of a given variable are modeled by splines, vs (ii) a simpler ‘reduced’ model, which does not include the variable being tested (see the section on “Statistical analyses” for details of the test).