Accuracy of noncompartmental pharmacokinetic parameters estimated from bolus injection and steady-state infusion data.

A Monte Carlo simulation study was carried out to examine the accuracy of parameters derived from curve moments. Impulse response (IR) and washout (WO) concentration-time curves, based on a triexponential model, were analyzed by numerical integration and regression analysis. Both designs were tested according to their robustness to measurement error and model misspecification. Performance of the methods was judged using the median error (ME) and the median absolute error (MAE) of 1000 simulations. The WO design provided better estimates of mean disposition residence time and worse estimates of the normalized variance of disposition residence times (CVD2) than its rival. At 20% measurement noise, the MAE of CVD2 was less than 13%. The WO design was much more robust to model misspecification. Numerical integration performed as good as, or better than, regression analysis. Both methods are very sensitive to tail-area error, meaning that special attention needs to be paid to this aspect of experimental design. This study demonstrates that it is possible to obtain good estimates of higher moment parameters in a well-designed experiment.