Predictive techniques applied to imipramine therapeutic drug monitoring

The aim of this study was to establish the performance of pharmacokinetic methods employing little data on serum drug concentrations obtained in routine therapeutic drug monitoring of imipramine. Forty-three and 123 serum levels were obtained in 8 adult depressive patients (aged 57-80 y) and 34 enuretic children (aged 5-13 y), respectively. Forecasting of the serum concentrations was performed based on mean population pharmacokinetic parameters (method A), with knowledge of one steady-state serum concentration (method B), and from two or more steady-state serum concentrations (method C). The accuracy and precision of each method were evaluated from the mean prediction error (ME) and from the root mean squared prediction error (RMSE), respectively. The values of ME and RMSE of methods B and C proved to be significantly lower than those found using method A. Method C was the most precise and accurate in both populations. Method A underestimates the serum concentrations observed in adults (ME greater than 0) but overestimates them in children (ME less than 0), although to a lesser extent. The study shows that it is possible to obtain a good estimation of individual dosage needs from one or more serum concentrations obtained at steady state. Clinical application of these methods (B and C) yields an increase in the efficiency and safety of the treatment, particularly in special populations such as geriatric and pediatric patients.