Accuracy of 11 methods for predicting theophylline dose.

Eleven methods for pharmacokinetic determination of theophylline dose were compared, based on the ability of each method to predict theophylline serum concentration and clearance for adults with asthma. Predictions by each method were compared with actual serum theophylline concentrations before and after an aminophylline loading dose was administered to treat bronchospasm in 22 patients. Follow-up serum theophylline concentrations were obtained after maintenance therapy with i.v. aminophylline in 6 patients and an oral sustained-release theophylline product (Theo-Dur, Key) in 16 patients; maintenance doses administered to the patients were calculated by the method of Chiou et al. Two variations of the method of Chiou et al., seven Bayesian methods using one or more measured serum theophylline concentrations, FDA's standardized clearance estimate, and a population-based clearance-estimation method were compared. A one-compartment, open model with first-order elimination was assumed for all calculations. All 11 methods predicted steady-state concentration with minimal bias and good precision. The Chiou and Bayesian methods performed similarly, with the highest precision found for the Bayesian method that incorporated four measured concentrations. The population-based method had the highest correlation coefficient and the lowest mean error for predicting steady-state concentration. Decreasing the number of measured concentrations had a minimal effect on the various Bayesian methods. All methods evaluated are sufficiently accurate for clinical application in patients with stable, uncomplicated asthma.