The complete clinical data of 198 patients with stable Warfarin dose were collected. The VKORC1-1173C>T, CYP2C9*2 and CYP2C9*3 genotypes were detected by polymerase chain reaction-high resolution melting technique after whole blood genomic DNA extraction with peripheral blood 1.8 mL of sodium citrate anticoagulation 1:9. The 4 prediction algorithms, including IWPC for mixed population, WEN for Chinese Taiwan population, HUANG for Chinese mainland population and OHNO for Japanese, were evaluated and compared by mean absolute error (MAE) and prediction accuracy.

The MAE of HUANG, OHNO and WEN in 4 prediction algorithms and fixed dose model was <±1.0 mg/d with ideal prediction percentage >40%, with the minimum MAE in HUANG. The ideal prediction algorithm with highest percentage was OHNO.

The Warfarin stable dose algorithms by pharmacogenetics for Asian population combined with basic clinical information can predict the dose of Warfarin among Shanghai with clinical application significance.