Statistical modeling of average and variability of time to extubation for meta-analysis comparing desflurane to sevoflurane

Abstract

Background: The recovery profile of an ideal anesthetic or technique would be fast (e.g., mean of 5 min from end of surgery to extubation) with little variability (e.g., always 4-7 min). We used anesthesia information management system (AIMS) data to learn how to model the time from end of surgery to extubation. We applied that knowledge for meta-analyses of trials comparing extubation times after use of desflurane and sevoflurane.

Methods: AIMS data studied were 32,792 cases performed by 95 surgeons that included tracheal intubation and extubation in the operating room (OR) and use of volatile anesthetic(s). Meta-analysis included the 29 randomized controlled trials through 2008 comparing extubation times with desflurane and sevoflurane. Percentage differences in means and standard deviations were studied using random effects meta-analysis and a Bayesian method.

Results: Times to extubation were better fit by (skewed) Weibull distributions than by (symmetric) normal distributions. Drug choice had nearly equally proportional effects on the means and standard deviations of extubation times, as shown by unchanged coefficients of variation (P > 0.10 for 26 of 29 studies) and nonsignificant pooled difference in the coefficient of variation (sevoflurane--desflurane = -1%, 95% confidence interval [CI] -3% to 1%, P = 0.22). Applying these findings, desflurane reduced the mean extubation time by 25% (95% CI 17%-32%, P < 0.0001) and standard deviation by 21% (95% CI 16%-26%). To value the intangible costs (e.g., frustrated waiting surgeons) of prolonged extubation times, we considered the 15% of AIMS cases with times >15 min. These cases averaged 4.9 min longer times from out of the OR to the start of surgery of the surgeon's next case (95% CI 2.7-7.1 min, P < 0.0001). Reduction in the means and standard deviations by 20%-25% would likely reduce incidences of these prolonged extubation times by 71%-82% (95% CI 68%-84%).

Conclusions: Desflurane reduces the average extubation time and the variability of extubation time by 20%-25% relative to sevoflurane. The principal economic value of these end points is their reductions of direct (labor) costs of OR time. However, reductions in intangible costs of prolonged extubation are real, being associated with subsequent delays. Reductions in the average and variance of times to extubation can be interpreted and monitored in terms of corresponding expected 75% reductions in the incidences of prolonged extubation times by using desflurane relative to sevoflurane.