Effect-Site Target-Controlled Infusion in the Obese: Model Derivation and Performance Assessment.

BACKGROUND: The aim of this study is to derive a propofol pharmacokinetic (PK) pharmacodynamic (PD) model to perform effect-site target-controlled infusion (TCI) in obese patients, and to analyze its performance along with that of other available PK models. METHODS: In the first step of the study, a 3-compartment PK model linked to a sigmoidal inhibitory Emax PD model by a first-order rate constant (keo) was used to fit propofol concentration-bispectral index (BIS) data. Population modeling analysis was performed by nonlinear mixed effects regression in NONMEM (ICON, Dublin, Ireland). PK data from 3 previous studies in obese adult patients (n = 47), including PD (BIS) data from 1 of these studies (n = 20), were pooled and simultaneously analyzed. A decrease in NONMEM objective function (ΔOBJ) of 3.84 points, for an added parameter, was considered significant at the 0.05 level. In the second step of the study, we analyzed the predictive performance (median predictive errors [MDPE] and median absolute predictive errors [MDAPE]) of the current model and of other available models using an independent data set (n = 14). RESULTS: Step 1: The selected PKPD model produced an adequate fit of the data. Total body weight resulted in the best size scalar for volumes and clearances (ΔOBJ, -18.173). Empirical allometric total body weight relationships did not improve model fit (ΔOBJ, 0.309). A lag time parameter for BIS response improved the fit (ΔOBJ, 89.593). No effect of age or gender was observed. Step 2: Current model MDPE and MDAPE were 11.5% (3.7-25.0) and 26.8% (20.7-32.6) in the PK part and 0.4% (-10.39 to 3.85) and 11.9% (20.7-32.6) in the PD part. The PK model developed by Eleveld et al resulted in the lowest PK predictive errors (MDPE = <10% and MDAPE = <25%). CONCLUSIONS: We derived and validated a propofol PKPD model to perform effect-site TCI in obese patients. This model, derived exclusively from obese patient's data, is not recommended for TCI in lean patients because it carries the risk of underdosing.

Tight computerized versus conventional glucose control in the ICU: a randomized controlled trial.

PURPOSE: The blood glucose target range and optimal method to reach this range remain a matter of debate in the intensive care unit (ICU). A computer decision support system (CDSS) might improve the outcome of ICU patients through facilitation of a tighter blood glucose control. METHODS: We conducted a multi-center randomized trial in 34 French ICU. Adult patients expected to require treatment in the ICU for at least 3 days were randomly assigned without blinding to undergo tight computerized glucose control with the CDSS (TGC) or conventional glucose control (CGC), with blood glucose targets of 4.4-6.1 and <10.0 mmol/L, respectively. The primary outcome was all-cause death within 90 days after ICU admission. RESULTS: Of the 2,684 patients who underwent randomization to the TGC and CGC treatment groups, primary outcome was available for 1,335 and 1,311 patients, respectively. The baseline characteristics of these treatment groups were similar in terms of age (61 ± 16 years), SAPS II (51 ± 19), percentage of surgical admissions (40.0%) and proportion of diabetic patients (20.3%). A total of 431 (32.3%) patients in the TGC group and 447 (34.1%) in the CGC group had died by day 90 (odds ratio for death in the TGC 0.92; 95% confidence interval 0.78-1.78; p = 0.32). Severe hypoglycemia (<2.2 mmol/L) occurred in 174 of 1,317 patients (13.2%) in the TGC group and 79 of 1,284 patients (6.2%) in the CGC group (p < 0.001). CONCLUSIONS: Tight computerized glucose control with the CDSS did not significantly change 90-day mortality and was associated with more frequent severe hypoglycemia episodes in comparison with conventional glucose control.