Table 1.

Performance of genetic and combined (clinical and genetic) risk models for AF using rare and common variants.

OR	Frequency	Variants (n)	AUC			IDI	NRI(>0.01)		NRI categorical
			Genetic	Combined	Δ			Total	Cases	Controls
Rare variants
2	0.0001	1	0.50	0.76	0	0	0	0	0	0
	0.001	1	0.50	0.76	0	0	0	0	0	0
	0.005	1	0.50	0.76	0	0	-0.01	0	0	0
	0.01	1	0.51	0.76	0	0	-0.03	0	0	0
5	0.0001	1	0.50	0.76	0	0	0	0	0	0
	0.001	1	0.50	0.76	0	0	-0.01	0	0	0
	0.005	1	0.51	0.76	0	0	-0.05	0.01	0.01	0
	0.01	1	0.52	0.77	0.01	0.01	-0.10	0.02	-0.01	0.03
10	0.0001	1	0.50	0.76	0	0	0	0	0	0
	0.001	1	0.50	0.76	0	0	-0.03	0.00	0	0.01
	0.005	1	0.52	0.77	0.01	0.02	-0.11	0.02	-0.01	0.03
	0.01	1	0.54	0.78	0.02	0.03	-0.13	0.04	-0.02	0.06
Common variants
1.14-1.45a	0.03-0.84*	10	0.59	0.77	0.01	0.01	0.20	0.04	0.01	0.04

^aUsing parameters from the top 10 (that is, in terms of P value) uncorrelated SNPs in the CHARGE AF meta-analysis; in the table are listed the range of OR and allele frequency [22]. Variables included in the clinical risk score were: age, weight, height, systolic blood pressure (SBP), diastolic blood pressure (DBP), diabetes, medication for hypertension, history of congestive heart failure, history of myocardial infarction, smoking status, and race. Disease risk is 4% and population size is 200,000 for rare variants scenarios and 20,000 for common variants scenarios. Results are median values from 200 simulations.

AUC, area under the receiver operating characteristic curve; ΔAUC, change in AUC between the model with and without genetic variants; IDI, integrated discrimination improvement; NRI, net reclassification improvement (cutoffs 2.5% and 5%); OR, odds ratio.