Accuracy of the Pharmacogenetic Dosing Table in the Warfarin Label in Predicting Initial Therapeutic Warfarin Doses in a Large, Racially Diverse Cohort

Abstract

To compare the accuracy of the pharmacogenetic dosing table included in the warfarin label with two empiric dosing strategies in predicting initial therapeutic warfarin doses, and to identify factors that influence the accuracy of the table. Retrospective cohort study. International Warfarin Pharmacogenetic Consortium database. A total of 3727 racially diverse patients receiving stable doses of warfarin who had international normalized ratios (INRs) between 2.0 and 3.0. Mean absolute error (MAE) and mean percentage of patients whose predicted doses were within 20% of their actual therapeutic doses (percentage within 20%) were compared between the pharmacogenetic table and two empiric dosing strategies. In the first strategy, warfarin 5 mg/day was used, and in the second strategy, dosing varied depending on the patient's race. The mean percentage of patients whose actual doses were within the ranges of the table (percentage within range) was also calculated. Warfarin dosing using the table resulted in a lower MAE (10.9 mg/wk) and a higher percentage within 20% (41.5%) than both empiric dosing strategies (MAE 12.3-12.6 mg/wk, percentage within 20% of 31.8-32.7%). In addition, using the table showed similar MAE, mean percentage within 20%, and mean percentage within range across the different racial groups. Dosing according to the table had higher mean percentage within 20% (56.4% vs 15.4%) and higher mean percentage within range (53.3% vs 19.2%) in the intermediate-dose group (> 21 but ≤ 49 mg/wk) than in the low-dose group (≤ 21 mg/wk). Being female and taking amiodarone were identified as factors that significantly increased the likelihood that the patient's predicted dose would be outside of 20% of their actual therapeutic dose or outside the range of the pharmacogenetic table. Using the pharmacogenetic dosing table included in the warfarin label resulted in higher accuracy of dosing prediction than two empiric dosing strategies. The table had similar accuracy across racial groups and better accuracy in patients receiving an intermediate dose of warfarin. When the table is used for warfarin dosing, women and patients receiving amiodarone may require more intensive monitoring of their warfarin therapy.