Evaluation of a new lithium dosage-prediction technique.

The accuracy of a lithium dosage prediction technique was studied retrospectively. The dosage-prediction technique evaluated is based on lithium body clearance. The medical records of 71 psychiatric patients (30 men, 41 women) were reviewed. Age, sex, height, weight, serum creatinine, and one-compartment intravenous dosage equations were used. The lithium dose that would produce steady-state concentrations equal to those actually attained clinically was calculated. Forty-five (64%) of the predictions were within one capsule/day of the actual dose. There were 42 episodes (59%) of underprediction and 10(14%) of overprediction by one or more capsules per day. Underprediction occurred significantly more often in women than in men (31/41 versus 11/30). Predictions differed from actual doses by three or more capsules per day in 10 cases, nine of which were underpredictions. These patients did not differ significantly from the rest of the study group in age, sex, weight, serum creatinine, creatinine clearance, concurrent medical problems, or other medications. This lithium dosage prediction technique may be reliable, rapid, and inexpensive, but further refinement and prospective evaluation are necessary.