Clinical utility of six methods of predicting phenytoin doses and plasma concentrations

Abstract

The usefulness of six reported methods of predicting phenytoin doses or plasma concentrations in epileptic patients is compared. Forty adult epileptic outpatients who were taking phenytoin sodium orally were followed prospectively for nine months; 16 met requirements for inclusion in the study. Six methods of predicting phenytoin dosage or plasma concentrations—one equation, two nomograms, and three graphs—were compared to actual phenytoin doses and corresponding steady-state serum levels. Criteria establishing good, high, and low predictions were specified for each method. No method produced consistently accurate results. Good dosage predictions occurred in 42-60% of the predictions with the six methods. Low predictions occurred in 0-50% of the estimates; 8-53% of the predictions were high. It is concluded that certain methods are clinically useful under some conditions if recommended cautions are observed.