Table 3. Impact of adding different adipocytokine, hepatic or inflammatory markers as quartiles in the predictive capacity of a clinical + biological (C+B) risk score for type 2 diabetes.

	HL-test (p-value)	AIC	BIC	AROC § (95% CI)	AROC §§ (95% CI)
Kahn's C+B score	0.84	1123.6	1136.1	0.901 (0.883–0.919)	0.681 (0.648–0.715)
Kahn's C+B score + IL-1β	0.97	1126.3	1157.6	0.902 (0.884–0.920)	0.692 (0.658–0.726)
Kahn's C+B score + IL-6	0.96	1127.1	1158.4	0.901 (0.883–0.919)	0.684 (0.650–0.717)
Kahn's C+B score + TNF-α	0.95	1124.9	1156.2	0.902 (0.883–0.920)	0.686 (0.652–0.720)
Kahn's C+B score + hs-CRP	0.47	1126.9	1158.2	0.901 (0.883–0.919)	0.686 (0.653–0.720)
Kahn's C+B score + leptin	0.99	1122.9	1154.2	0.901 (0.883–0.919)	0.682 (0.648–0.715)
Kahn's C+B score + adiponectin	0.56	1125.4	1156.7	0.902 (0.884–0.920)	0.688 (0.654–0.722)
Kahn's C+B score + γGT	0.78	1126.5	1157.8	0.902 (0.884–0.920)	0.688 (0.654–0.722)
Kahn's C+B score + all variables	1.00	1137.9	1281.8	0.908 (0.891–0.925) **	0.699 (0.665–0.733)

Statistical analysis by logistic regression. Each line shows the results of the original model (first line with HL, Hosmer-Lemeshow goodness-of-fit test (only p-values are reported); AIC, Akaike's information criterion; BIC, Bayesian information criterion; AROC, area under the ROC curve; IL-1β, interleukin 1 beta; IL-6, interleukin 6; TNF-α, tumour necrosis factor alpha; hs-CRP, high sensitive C reactive protein; γGT, gamma glutamyl transpeptidase. § using the type 2 diabetes risk predicted by the model as a continuous variable; §§ splitting the type 2 diabetes risk into two categories (not at risk and at risk). Data from 208 participants who developed type 2 diabetes mellitus and 3634 controls. ** significantly different (p<0.01) from the baseline model (Kahn's C+B score).