Table 3

Discrimination and Reclassification

Derivation CohortValidation Cohort
AUC (95% CI)p ValueAUC (95% CI)p Value
Harrell’s C-Statistic
 NT-proBNP0.69 (0.64–0.72)0.65 (0.61–0.68)
 NT-proBNP plus suPAR0.72 (0.68–0.76)for comparison: 0.040.68 (0.64–0.71)For comparison: 0.17
 Multivariate model0.72 (0.68–0.75)0.68 (0.64–0.71)
 Multivariate model plus suPAR0.74 (0.70–0.78)for comparison: 0.020.70 (0.66–0.74)For comparison: 0.32
Improvement (95% CI)Improvement (95% CI)
Net reclassification index
 suPAR in addition to NT-proBNP32.9% (21.3%–44.5%)0.00654.9% (43.9%–65.8%)<0.001
 suPAR in addition to multivariate model24.9% (13.3%–36.5%)0.03235.8% (24.8%–46.7%)0.001
Integrated discrimination improvement
 suPAR in addition to NT-proBNP0.022 (0.014–0.028)0.0080.060 (0.047–0.073)<0.001
 suPAR in addition to multivariate model0.011 (0.006–0.017)0.050.022 (0.013–0.030)0.01

Harrell’s C-statistic was calculated to compare the discriminatory power to predict mortality of soluble urokinase-type plasminogen activator receptor in addition to N-terminal pro B-type natriuretic peptide and a multivariate model comprising all variables independently associated with mortality in multivariate Cox regression (age, male sex, New York Heart association functional classification and N-terminal pro B-type natriuretic peptide). An improvement in individual risk stratification was assessed using the category free net reclassification index. The p values in bold indicate a value <0.05.

AUC = area under the curve; other abbreviations as in Tables 1 and 2.