Population pharmacodynamics of midazolam administered by target controlled infusion in SICU patients after CABG surgery.

BACKGROUND

Midazolam is used commonly for sedation in the surgical intensive care unit. A suboptimal dosing regimen may lead to relative overdosing, which could result in delayed extubation and increased cost. This multicenter trial characterized midazolam pharmacodynamics in patients recovering from coronary artery bypass grafting.

METHODS

Three centers enrolled 90 patients undergoing coronary artery bypass grafting. All patients received sufentanil and midazolam via target-controlled infusion. After surgery, midazolam was titrated to a Ramsay sedation score of 5 for 2 h and then decreased to maintain a sedation score of 3 or 4 for at least another 4 h. Pharmacodynamic parameters were derived using NONMEM. The model was cross-validated to test performance.

RESULTS

The probability of a given level of sedation was related to the midazolam concentration by this equation: P(Sedation > or = ss) = Cn/(Cn + C(50,ss)n), where ss is the sedation score, C is the sum of the midazolam concentration and a term reflecting the dissipating effect of anesthesia: C = [midazolam] + theta x e(-Kt), where theta = 256 ng/ml and K = 0.19 h(-1). C(50,ss) values for Ramsay scores of 2 to 6 were 5.7, 71, 171, 260, and 659 ng/ml, respectively. The model predicted 57% of the data points correctly and 88% within one sedation score.

CONCLUSIONS

Despite previous reports of high interindividual variability in midazolam pharmacodynamics in patients in the surgical intensive care unit, these cross-validation results suggest that, when midazolam is administered using a target-controlled infusion device, the level of sedation can be predicted within 1 sedation score in 88% of patients based on the target midazolam concentration and the time since the conclusion of the anesthetic.