We report a comparison of warfarin treatment outcomes in 172 inpatients in two general hospitals randomly assigned to commence warfarin therapy by one of two methods; the first where warfarin dosage was determined using a flexible dose induction protocol, and the other where dosage was prescribed empirically by resident medical staff. The mean INR for each treatment day, the mean time to reach a therapeutic level of INR, the mean maintenance dose and the mean time to reach maintenance dose were not significantly different between the protocol and empirical treatment groups. Although the mean observations of warfarin effect were similar between the two groups, there were more patients with excessive warfarin effect (INR > 4.0) during empirical treatment. In addition, age and identified complicating factors such as interfering drugs, heart failure or alcohol abuse were significant predictors of excessive warfarin effect, regardless of treatment group. The protocol dose administered on day 4 of treatment, which has been used as a determinant of maintenance dose, predicted maintenance dose to within 1mg in 57/83 patients (69%, 95% confidence interval: 59-79%). We conclude that as a method for safely and effectively initiating warfarin therapy, this protocol at least matches the empirical prescribing skills of medical staff in a teaching hospital.
