A computerised sampling strategy for therapeutic drug monitoring of lithium provides precise estimates and significantly reduces dose-finding time

Abstract: The clinical benefit of implementing Bayesian approach for lithium drug monitoring was evaluated. Intervention group (N = 42) and historical control group (N = 55) patients were each divided into two groups: Dosage with immediate‐release lithium carbonate or a sustained‐release formulation, lithium citrate. Bayesian approach was performed in the intervention groups, and estimation of lithium steady‐state trough concentration was obtained from non‐steady‐state blood sample, collected about 12 hr after the first lithium study dose. The estimate was compared with the actually measured steady‐state concentration. In the control group, lithium monitoring was traditionally performed as steady‐state blood sampling. Predicted and measured lithium concentrations were comparable. The desired lithium dose was reached significantly faster in the intervention group compared to control; 2.47 ± 2.22 days versus 9.96 ± 11.24 days (mean ± S.D.) (p = 0.0003). Bayesian approach was an advantage for the clinicians as a fast and safe aid to obtain the optimal lithium treatment dose.