Computer-assisted individualized lidocaine dosage: clinical evaluation and comparison with physician performance.

The performance of a computerized dosing aid in achieving a target serum concentration of lidocaine in the middle of the recommended therapeutic range (3.5 mg/L) was evaluated in 63 patients treated for acute ventricular arrhythmias. In all patients a serum concentration measurement was obtained shortly after starting lidocaine infusion. In 22 patients a microcomputer program based on a Bayesian forecasting technique was used for dosing recommendations, whereas in 41 the serum concentration was interpreted and the dose was adjusted by the unaided physician. Both groups were similar with respect to the average concentration achieved (control: 3.8 +/- 1.13 [SD] mg/L, computer-aided: 3.5 +/- 0.59 mg/L). However, the interindividual variability was significantly larger in the control group (95% confidence interval: 1.5 to 6.1 mg/L vs 2.3 to 4.7 mg/L [p less than 0.01]). Nine of the 41 patients in the control group had a lidocaine concentration outside the recommended therapeutic range of 2 to 5 mg/L compared to only 1 of 22 in the computer-aided group. Lidocaine concentrations greater than 2 mg/L were associated with significantly more effective suppression of ventricular arrhythmias (p less than 0.05). The results show that Bayesian forecasting outperforms the physician in early adjustment of lidocaine dosage based on serum concentration measurements.