Table 2.

Comparisons of biomarkers with clinical variables of the PCE and other models with differences in AUC, NRI, and IDI for risk prediction of global CVD

	C-statistics Primary Model (95% CI)	C-statistics Extended Comparison Model (95% CI)	ΔAUC (95% CI)	NRI (95% CI)	Continuous NRI (95% CI)	IDI (95% CI)
Basic model vs PCE	0.597 (0.57, 0.63)	0.647 (0.62, 0.68)	0.050 (0.028, 0.084)	0.057 (0.005, 0.184)	0.274 (0.174, 0.434)	0.012 (0.006, 0.023)
PCE vs hs-CRP model	0.647 (0.623, 0.679)	0.665 (0.642, 0.6991)	0.018 (0.006, 0.034)	0.035 (−0.022, 0.104)	0.167 (0.026, 0.299)	0.007 (0.002, 0.015)
PCE vs hs-cTnT model	0.647 (0.623, 0.679)	0.686 (0.663, 0.720)	0.039 (0.022, 0.057)	0.137 (0.051, 0.248)	0.361 (0.226, 0.496)	0.019 (0.011, 0.030)
PCE vs NT-proBNP model	0.647 (0.623, 0.679)	0.739 (0.709, 0.771)	0.092 (0.063, 0.117)	0.304 (0.190, 0.401)	0.407 (0.270, 0.538)	0.066 (0.040, 0.099)
PCE vs Complete model	0.647 (0.623, 0.679)	0.750 (0.724, 0.781)	0.103 (0.074, 0.129)	0.370 (0.228, 0.455)	0.484 (0.359, 0.634)	0.075 (0.051, 0.112)
PCE model vs Lab model	0.647 (0.623, 0.679)	0.738 (0.710, 0.7683)	0.091 (0.055, 0.121)	0.302 (0.145, 0.412)	0.355 (0.177, 0.550)	0.068 (0.041, 0.104)

NRI was calculated using cutpoints of 2.4% and 8% over 4 years to categorize participants as low, intermediate, or high risk.

Basic model: age, gender, and race

PCE model: age, gender, race, total cholesterol, HDL-C, SBP, antihypertensive medication use, current smoking, and diabetes status

hs-CRP model: PCE variables + hs-CRP

hs-cTnT model: PCE variables + hs-cTnT

NT-proBNP model: PCE variables + NT-proBNP

Complete model: = PCE variables + hs-cTnT, NT-pro-BNP, and hs-CRP

Lab model: age, gender, race + hs-cTnT, NT-pro-BNP, and hs-CRP

Abbreviations: AUC, area under the receiver operative characteristic curve (C-statistics); IDI, integrated discrimination index; NRI, net reclassification index.