Table 3.

Pharmacogenetic Refinement Model in Derivation Cohort (N=969).

Entry into model	variable	Effect on dose* (95% CI)	Cumulative Model R2	P
1	VKORC1-1639 G>A	-20% (-23% to -17%)	23.2%	<0.001
2	ln(INR) †	-12% (-14% to -11%)	35.5%	<0.001
3	Dose-3, per mg‡	+2% (1% to 3%)	44.1%	<0.001
4	Age, per year	-7% (-9% to -6%)	49.9%	<0.001
5	CYP2C9*3	-28% (-32% to -23%)	55.1%	<0.001
6	CYP2C9*2	-15% (-19% to -11%)	57.0%	<0.001
7	BSA║, per 0.25 m2	+7% (4% to 9%)	58.9%	<0.001
8	Target INR	+7% (4% to 10%)	59.7%	<0.001
9	African Origin	-11% (-17% to -5%)	60.4%	0.001
10	Stroke	-23% (-33% to -10%)	61.0%	<0.001
11	Dose-4, per mg	+1% (1% to 2%)	61.5%	<0.001
12	Dose-2, per mg	+2% (1% to 3%)	62.0%	<0.001
13	Diabetes	-9% (-16% to -3%)	62.3%	0.007
14	Amiodarone Use	-12% (-21% to -1%)	62.5%	0.028
15	Fluvastatin Use	-17% (-33% to 4%)	62.6%	0.106

^*Effect on the estimate of the maintenance dose is calculated per variant allele, and per 0.25 unit increase in ln(INR) or target INR.

^†ln is natural logarithm. INR is International normalized ratio.

^‡Dose _−i is dose given i days before INR is measured.

^║BSA is Body Surface Area.