A comparison of pharmacokinetic versus empirical lithium dosing techniques.

Three methods for estimating maintenance dosage requirements of lithium carbonate were retrospectively evaluated in 20 inpatients who met criteria of the Diagnostic and Statistical Manual, Third Edition, for "bipolar disorder, manic phase." Dosing methods evaluated included a pharmacokinetic method, the single-point method of Perry et al.; a population-based nomogram approach, the Zetin et al. method; and a physician-based empirical dosing procedure. The ability of each dosing procedure to produce dosing recommendations that resulted in a targeted steady-state serum lithium concentration was evaluated. The empirical dosing procedure demonstrated a significant tendency (bias) to underestimate the dose necessary to produce a desired steady-state serum lithium concentration. Comparison of the predictive accuracy of the various dosing methods failed to demonstrate any statistically significant differences among the dosing procedures. There was a strong trend, however, for the Perry method to produce predictions of steady-state lithium levels that were more frequently within 0.2 mEq/L of actual levels.