Pharmacokinetic-pharmacodynamic information regarding warfarin is used
to produce a predictive model based on the idea that pharmacodynamic
variability is more important than pharmacokinetic variability in the
overall dose-response variability to warfarin. A modification of the m
aximum effect model is tested on a group of patients initiating oral a
nticoagulation with warfarin. Results indicate that the model can acco
unt for at least half of the total variation in maintenance doses obse
rved (sample coefficient of determination, 0.53) and offer the physici
an a framework for dose requirements at the onset of therapy. The basi
c prediction equation is as follows: Maintenance dose = (11/internatio
nal normalized ratio) - 1, with a coefficient of correlation of 0.73 (
95% confidence limits, 0.46-0.88). Application of this model may impro
ve on the traditional empiric approach to warfarin dose adjustment.