Prediction of Long-Term Survival after Coronary Artery Bypass with Bilateral Internal Thoracic Artery Grafting: External Validation of A Contemporary Nomogram

Figures

• 

  Figure 1. External calibration of the original score in the isolated CABG-BITA dataset at 5, 10, 15 and 20 years with smooth estimates with pointwise 95% confidence intervals plotted against predicted mortality probabilities. The solid line is the line of identity, denoting perfect calibration. Some miscalibration in the large is evident at 5 and 10 years with underprediction of mortality probabilities.

• 

  Figure 2. Three-dimensional surface plots, contour maps and representative slices for fixed values of the score and fixed time points showing the time-varying effect of the score. (A) Three-dimensional surface plot showing the relation between the score and follow-up time on the y- and x-axes, respectively. The z-axis shows the contribution of each combination on the log hazard ratio for all-cause mortality. (B) Contour map that uses a color gradient to visualize the effect of the combination on the log hazard ratio, where the darker shades of red denote an increase in the log hazard ratio. (C) Representative slices at fixed time points of 5, 10 and 15 years of the follow-up. The score and the log hazard ratio are shown in x- and y-axes, respectively. (D) Representative slices at fixed values of the score at 25, 75 and 125 points. The follow-up in years and the log hazard ratio are shown in x- and y-axes, respectively.

• 

  Figure 3. Prediction survival curves adjusted for the time-varying coefficient of the score in 8 subgroups according to the points total in the isolated CABG-BITA dataset (n = 2846). There were used cut off points at round numbers 0–10 (n = 233), 11–30 (n = 389), 31–50 (n = 414), 51–70 (n = 621), 71–90 (n = 502), 91–110 (n = 289), 111–130 (n = 259) and >130 (n = 139).

• 

  Figure 4. Spline interaction analysis. (A) Relationship of the hazard ratio in males relative to females with CABG-BITA across the score. The solid line is the hazard ratio, and the dotted lines are the 95% confidence intervals around the hazard ratio. A hazard ratio of more than 1 suggests increased all-cause mortality in males relative to females, as shown in this figure at higher than 85 points of the score. (B) Cross-sectional spline interaction analysis between the score (continuous variable) and the groups of BMI (categorical variable). The solid colored lines represent the log hazard ratios of the BMI categories, and the shadows are shown the corresponding 95% confidence intervals.