Multivariate analysis: independent predictors of therapeutic warfarin dose at day 3
Entry into model . | Variable . | % Change in therapeutic warfarin dose (95% CI)* . | R2 after entry, % . | P in final model . |
---|---|---|---|---|
— | Intercept | — | — | < .001 |
1 | INR3 | −46.5 (−33.3 to −57.1) | 34.4 | < .001 |
2 | 1st warfarin dose, per mg | +7.1 (+4.0 to +10.4) | 54.7 | < .001 |
3 | CYP2C9*3, per allele | −38.1(−29.3 to −45.7) | 69.8 | < .001 |
4 | 2nd warfarin dose, per mg | +3.9 (+0.0 to +8.0) | 73.6 | .051 |
5 | CYP2C9*2, per allele | −17.4 (−8.3 to −25.6) | 75.6 | .001 |
6 | EBL × INR3 | +44.9 (+16.6 to +80.2) | 78.1 | .001 |
7 | Smokes | +20.1 (+6.0 to +36.2) | 79.9 | .005 |
8 | VKORC1 haplotype A, per copy | −10.7 (−2.0 to −18.6) | 81.1 | .018 |
9 | Target INR | +14.6 (−7.9 to +42.5) | 81.5 | .218 |
Entry into model . | Variable . | % Change in therapeutic warfarin dose (95% CI)* . | R2 after entry, % . | P in final model . |
---|---|---|---|---|
— | Intercept | — | — | < .001 |
1 | INR3 | −46.5 (−33.3 to −57.1) | 34.4 | < .001 |
2 | 1st warfarin dose, per mg | +7.1 (+4.0 to +10.4) | 54.7 | < .001 |
3 | CYP2C9*3, per allele | −38.1(−29.3 to −45.7) | 69.8 | < .001 |
4 | 2nd warfarin dose, per mg | +3.9 (+0.0 to +8.0) | 73.6 | .051 |
5 | CYP2C9*2, per allele | −17.4 (−8.3 to −25.6) | 75.6 | .001 |
6 | EBL × INR3 | +44.9 (+16.6 to +80.2) | 78.1 | .001 |
7 | Smokes | +20.1 (+6.0 to +36.2) | 79.9 | .005 |
8 | VKORC1 haplotype A, per copy | −10.7 (−2.0 to −18.6) | 81.1 | .018 |
9 | Target INR | +14.6 (−7.9 to +42.5) | 81.5 | .218 |
The optimal pharmacogenetics algorithm that estimated therapeutic dose (mg/day) 3 days after surgery was as follows: Exp[1.0138 − 2.5047 × ln(INR3) + 0.0690 × 1st warfarin dose + 0.0385 × 2nd warfarin dose + 0.2474 x ln(EBL) × ln(INR3) − 0.1912 × CYP2C9*2 − 0.4793 × CYP2C9*3 + 0.1835 × Smokes − 0.1132 × VKORC1+ 0.2724 × Target INR], where ln(EBL) × ln(INR3) is the interaction between logarithm (EBL) and logarithm (INR3), and VKORC1 is the number of A haplotypes.
— indicates not available.
Percent change (95% confidence interval) was calculated for each 0.25-point increase in log(INR3) and 1.5-point increase in the interaction of ln(EBL) and ln(INR3).