Table 1 Comparisons of multivariable Fine–Gray models, using complete data cohort (N = 14,781)a.

From: Incorporation of clinical and biological factors improves prognostication and reflects contemporary clinical practice

Model

Covariate

C-indexb (95% CI)

p-valuec

Model 0

Stage

0.692

0.681

0.703

–

Model 1

Stage, Age

0.701

0.689

0.712

<0.0001

Model 2

Stage, HR/Her2d

0.745

0.739

0.761

<0.0001

Model 3

Stage, Grade

0.757

0.747

0.767

<0.0001

Model 4

Stage, HR/Her2d, Age

0.752

0.741

0.763

<0.0001

Model 5

Stage, HR/Her2d, Grade

0.740

0.729

0.751

<0.0001

Model 6

Stage, Age, Grade

0.758

0.748

0.768

<0.0001

Model 7

Stage, Age, HR/Her2d, Grade

0.774

0.764

0.783

<0.0001

  1. aPatients with complete data (age, stage, HR, HER2, Grade) included.
  2. bHarrell’s C- Index: The kmi package in R was used to impute censoring times for competing risk data and the rcorr.cens function in the Hmisc package in R was used to estimate the C-index and its confidence interval.
  3. cp-value comparing C-index values between models was computed using the compareC package in R.
  4. dER estrogen receptor, PR progesterone receptor, HER2 HER2-neu receptor, HR hormone receptor; HR+: ER+ or PR+; HR−: ER− and PR−.
Back to article page