Pharmacodynamic modeling of muscle relaxants: effect of design issues on results.

Background: Pharmacodynamic studies of muscle relaxants use different dosing regimens (such as administration by bolus vs. infusion and doses that produce complete vs. incomplete paralysis). The authors used published data to evaluate the effect of modeling assumptions on pharmacodynamic estimates.

Methods: The authors used a pharmacokinetic-pharmacodynamic dataset in which patients received cisatracurium, 75 or 300 microg/kg (1.5 or 6 x ED95), to generate plasma concentration (Cp) and twitch depression (effect) curves. They then evaluated the impact of the following: assuming that Cp decreased monotonically versus increasing initially before decreasing monotonically; misrecording effect data by 6 s or less; and doses targeting incomplete versus complete paralysis. Parameters evaluated were the steady state Cp depressing twitch tension 50% (C50) and the rate constant for equilibration between plasma and effect site concentrations (k(e0)).

Results: With the large dose, increasing the time at which Cp peaked from 0.0 to 1.5 min decreased C50 and increased k(e0) markedly; with the small dose, changes in both were small. Misrecording the timing of effect had a larger impact with the large dose compared with the small dose. Doses smaller than ED50 or those producing prolonged, complete twitch depression yielded biased and variable estimates.

Conclusion: The erroneous assumption that Cp decreases monotonically after bolus administration affects accuracy of pharmacodynamic estimates with doses producing rapid, complete twitch depression. Other errors (e.g., misrecording the time of drug administration) impact on pharmacodynamic estimates, particularly with large doses. The authors' findings suggest that investigators performing neuromuscular (and other) pharmacodynamic studies should carefully consider the impact of study design on their parameter estimates.