Principal Findings:

Indication for warfarin was preoperative orthopedic surgery in 60% of patients, deep vein thrombosis and/or pulmonary embolism in 24%, and atrial fibrillation in 14%. Mean number of INRs checked was 7.6 per patient. The average warfarin dose per week decreased as the number of variants increased, with the highest does in those with wild type variant (44.7 mg for wild type variant, 37.4 mg for 1 variant, 26.3 mg for 2 variants, 17.6 for 3 variants, and 8 mg for 4 variants; p < 0.001 for trend).

The initial to final stable dose change in patients with a wild type variant (n = 56) was significantly lower in the genotype-guided warfarin regimen compared with the standard dose regimen (-0.5 mg/wk vs. +13.6 mg/wk, p = 0.001), showed no difference in patients with 1 variant (n = 75) (1.9 mg/wk vs. 2.4 mg/wk, p = 0.85) and was significantly higher in patients with >1 variant (n = 44) (-0.6 mg/wk vs. -10 mg/wk, p < 0.001). The primary endpoint of percent out-of-range INRs did not differ between the randomized groups (30.7% for the genotype-guided warfarin regimen vs. 33.1% for the standard dose regimen, p = 0.47). In a subgroup analysis, the percent out-of-range INRs were less frequent in the genotype-guided warfarin group than the standard group among patients with either wild type or multiple variants (29.3% vs. 39.1%, p = 0.03) but the results went the opposite direction in patients with a single variant (33.6%vs. 27%, p = 0.14). There were fewer dose changes per patient in the genotype-guided warfarin group than the standard group (mean 3.0 vs. 3.6, p = 0.035). There was no significant difference in the frequency of INR ≥4 between groups (34 vs. 42). INRs ≥4 were more common in multiple variant allele carriers (46% vs. 29%, p = 0.029 in the pooled population).