A double-blind, randomized comparison of i.v. lorazepam versus midazolam for sedation of ICU patients via a pharmacologic model.

BACKGROUND Benzodiazepines, such as lorazepam and midazolam, are frequently administered to surgical intensive care unit (ICU) patients for postoperative sedation. To date, the pharmacology of lorazepam in critically ill patients has not been described. The aim of the current study was to characterize and compare the pharmacokinetics and pharmacodynamics of lorazepam and midazolam administered as continuous intravenous infusions for postoperative sedation of surgical ICU patients. METHODS With Institutional Review Board approval, 24 consenting adult surgical patients were given either lorazepam or midazolam in a double-blind fashion (together with either intravenous fentanyl or epidural morphine for analgesia) through target-controlled intravenous infusions titrated to maintain a moderate level of sedation for 12-72 h postoperatively. Moderate sedation was defined as a Ramsay Sedation Scale score of 3 or 4. Sedation scores were measured, together with benzodiazepine plasma concentrations. Population pharmacokinetic and pharmacodynamic parameters were estimated using nonlinear mixed-effects modeling. RESULTS A two-compartment model best described the pharmacokinetics of both lorazepam and midazolam. The pharmacodynamic model predicted depth of sedation for both midazolam and lorazepam with 76% accuracy. The estimated sedative potency of lorazepam was twice that of midazolam. The predicted C50,ss (plasma benzodiazepine concentrations where P(Sedation > or = ss) = 50%) values for midazolam (sedation score [SS] > or = n, where n = a Ramsay Sedation Score of 2, 3, ... 6) were 68, 101, 208, 304, and 375 ng/ml. The corresponding predicted C50,ss values for lorazepam were 34, 51, 104, 152, and 188 ng/ml, respectively. Age, fentanyl administration, and the resolving effects of surgery and anesthesia were significant covariates of benzodiazepine sedation. The relative amnestic potency of lorazepam to midazolam was 4 (observed). The predicted emergence times from sedation after a 72-h benzodiazepine infusion for light (SS = 3) and deep (SS = 5) sedation in a typical patient were 3.6 and 14.9 h for midazolam infusions and 11.9 and 31.1 h for lorazepam infusions, respectively. CONCLUSIONS The pharmacology of intravenous infusions of lorazepam differs significantly from that of midazolam in critically ill patients. This results in significant delays in emergence from sedation with lorazepam as compared with midazolam when administered for ICU sedation.