Table 2C.

Integrated discrimination improvement (IDI) and net reclassification improvement (NRI) for 3-year all-cause mortality

	IDI			NRI
	IDI (%)	95% CI (%)	P value	NRI (%)	95% CI (%)	P value
Entire cohort (n=580)
GWTG-HF
+IMRS	2.14	0.99 to 3.30	<0.001	36.2	17.5 to 54.9	<0.001
Subgroup with NT-proBNP (n=341)
GWTG-HF
+IMRS	1.64	0.49 to 2.78	0.005	25.8	3.63 to 49.9	0.02
+NT-proBNP	3.15	1.52 to 4.78	<0.001	36.3	14.4 to 58.3	0.001
+IMRS and NT-proBNP	4.03	2.13 to 5.93	<0.001	39.9	18.0 to 61.8	<0.001
IMRS
+NT-proBNP	3.26	1.62 to 4.90	<0.001	46.2	24.4 to 68.0	<0.001

Variables included in the analysis were IMRS, GWTG-HF risk score, CAD, history of atrial fibrillation, BMI, haemoglobin and log NT-proBNP. HRs normalised for SD. Log likelihood ratio −760.90 for entire cohort; log likelihood ratio −573.90 for subgroup with NT-proBNP available

Variables included in the analysis were age, CAD, diabetes mellitus, atrial fibrillation, heart rate, BMI, WCC, haemoglobin, haematocrit, MCV, MCHC, RDW, platelet, sodium, BUN, creatinine, potassium; HRs normalised for SD. Log likelihood ratio −742.17 for entire cohort; loglikelihood ratio −562.80 for subgroup with NT-proBNP available.

The IDI is the difference between the discrimination slopes of basic models and basic models extended with a predictor variable. The discrimination slope is the difference in predicted probabilities (%) between subjects with and without event. The NRI reflects the improvement in discriminative power by adding a predictor variable to a Cox model already including one of the tested scores (IMRS or GWTG-HF).

GWTG-HF, Get With the Guidelines – Heart Failure;IMRS, Intermountain Risk Score; NT-proBNP, N-terminal pro-B-type natriuretic peptide.