Skip to main content

Table 2 Joint and Landmark Cox models estimations on the French cohort (n=1067 patients, 427 recurrent events)

From: Validation of death prediction after breast cancer relapses using joint models

  Joint model Cox landmark
  For recurrent events For death For death
Variable HR (95% CI) HR (95% CI) HR (95% CI)
Age
[40−55] vs. >55 1.17 (0.91-1.51) 0.31 (0.16-0.60) 0.56 (0.41-0.76)
≤40 vs. >55 2.41 (1.73-3.37) 1.57 (0.73-3.38) 0.54 (0.31-0.92)
Peritumoural vascular involvement 1.61 (1.26-2.06) 4.74 (2.54-8.85) 1.04 (0.76-1.43)
Tumour size (>20 mm vs. ≤20 mm) 1.95 (1.52-2.50) 6.21 (2.99-12.86) 1.20 (0.88-1.65)
Nodal involvement 1.84 (1.44-2.36) 4.89 (2.47-9.67) 1.95 (1.45-2.60)
Grade
II vs. I 2.18 (1.57-3.01) 7.48 (2.71-20.66) 1.07 (0.75-1.52)
III vs. I 3.09 (2.16-4.41) 44.33 (15.61-125.93) 1.25 (0.83-1.88)
Recurrences before t=5 years
One previous recurrence      7.18 (5.25-9.83)
Two previous recurrences      6.94 (3.05-15.83)
θ=var(ui) 1.07 (se=0.06)     
α 4.45 (se=0.33)     
LCV 1.19     0.93  
  1. HR: Hazard ratio; CI: Confidence interval; LCV: Likelihood cross-validation criterion; Cox Landmark at time t = 5 years.