Fast-track cardiac anesthesia, although variably defined, generally aims at tracheal extubation within 8 hours postoperatively.1 As fast-track tracheal extubation after cardiac surgery has become a common practice, methods believed to facilitate this must be validated. This includes the use of intraoperative monitoring modalities, such as candidate depth of anesthesia monitors. In 2 studies on cardiac surgery patients, Bispectral Index (BIS) monitoring did not decrease time to tracheal extubation compared with standard practice.2,3 However, 1 study had a small sample size, limiting its power.3 The second study lacked a specific fast-track anesthesia protocol, which may have resulted in a failure to extubate the trachea at the earliest possible time.2 It therefore remains unresolved whether or not BIS-guided anesthesia affects the time to tracheal extubation after cardiac surgery. To justify the added cost of the implementation of a BIS-based protocol to facilitate fast-track cardiac anesthesia, it is necessary to demonstrate the efficacy of this monitoring modality for achieving earlier tracheal extubation compared with an alternative well-defined protocol based on other standard intraoperative monitors.
Therefore, this study investigated whether BIS monitoring increased the probability of tracheal extubation relative to an end-tidal anesthetic concentration (ETAC)-guided anesthetic protocol in a large sample of fast-track cardiac surgery patients, receiving a primarily volatile-based, low-dose narcotic anesthetic. A secondary aim was to identify significant preoperative and intraoperative factors increasing the instantaneous probability of tracheal extubation.
METHODS
Subjects
After receiving IRB approval and after obtaining written informed consent, 750 patients undergoing elective cardiac surgery involving cardiopulmonary bypass at St. Boniface Hospital, Winnipeg, Manitoba, Canada were enrolled in the “BIS or Anesthesia Gas to Reduce Explicit Recall” (BAG-RECALL) clinical trial (clincaltrials.gov—NCT00682825).4,5 The original study was designed as a 6000 patient, multicenter trial assessing whether a BIS-guided anesthetic protocol reduced the incidence of intraoperative awareness relative to an ETAC-guided anesthetic protocol. The 750 patients formed an a priori determined subsample of BAG-RECALL patients from a single site. This patient subsample was selected due to the site’s routine use of fast-track anesthesia in cardiac surgical patients. Patients were block randomized to receive BIS (n = 374) or ETAC-guided anesthesia (n = 376) using computer-generated assignments. Assignments for each randomization number were placed in opaque, sealed envelopes which were opened just before anesthetic induction.
Twenty-seven of the 750 cardiac surgery patients originally recruited were excluded, leaving 723 patients in the final substudy analysis (BIS-361, ETAC-362). The various reasons for their exclusion from the analysis are depicted in Figure 1.
;)
Figure 1: Modified CONSORT flow document.
Study Protocol
Details of the BAG-RECALL trial can be found elsewhere.4 In summary, in the BIS group, it was recommended that volatile anesthetic be titrated to maintain a BIS value of 40 to 60. In the ETAC group, anesthesiologists were blinded to the BIS reading, and it was recommended that volatile anesthetic be titrated within 0.7 to 1.3 age-adjusted minimum alveolar concentration (MAC). Age-adjusted MAC was calculated using published formulae.6 Practitioners were free to titrate outside of the ranges according to clinical discretion. Practitioners were reminded of these guidelines via visual cues (a small poster on the anesthetist workstation) and by audible alerts triggered by deviations from the recommended BIS or age-adjusted MAC ranges.
The Winnipeg fast-track approach used in this substudy involves a low-dose narcotic, primarily volatile anesthetic technique. Propofol was given for induction of anesthesia unless contraindicated. Anesthesia was maintained using sevoflurane or desflurane. Midazolam and sufentanil were given at the discretion of the attending anesthesiologist, while rocuronium was used for muscle relaxation.
In both groups, the anesthesiologist was permitted to decrease the concentration of volatile anesthetic after chest closure to facilitate emergence. In the BIS group, this decrease was guided by a target BIS <75. Overall, the study volatile anesthetic titration guidelines were intended to increase the vigilance of anesthesiologists, as opposed to acting as mandatory constraints.
Standardized institutional criteria were used to determine eligibility for tracheal extubation (Table 1). This protocol included the achievement of necessary benchmarks related to respiration, hemodynamics, as well as the ability to respond to commands and perform simple motor tasks from a supine position. All anesthesiologists involved with the study were experienced consultant practitioners with at least 5 years of experience. Although anesthesiologists were aware that their patients had been recruited into the BAG-RECALL study, they were unaware that time to tracheal extubation was a study end point. This blinding was instituted to minimize potential practitioner bias on emergence practices.
Table 1: Institutional Protocol-Driven Tracheal Extubation Criteria
All patients were monitored using a BIS Quatro™ Sensor and version XP™ of the BIS software (Covidien, Mansfield, MA); however, anesthesiologists in the ETAC group were blinded to the BIS readings. Patient characteristics and surgical variables such as intraoperative blood loss, anesthesia duration (time from induction to skin closure), and surgery type were documented by research staff. Intraoperative signals, including BIS and ETAC values, were recorded continuously at 1 Hz using TrendFace Solo software (ixellence GmbH, Wildau, Germany).
Statistical Analysis
The maintenance phase of anesthesia, defined as 10 minutes after induction until 20 minutes before the surgery end time, was isolated to allow calculation of the median age-adjusted MAC values during this maintenance phase. Data reduction was performed using Matlab engineering software (v. 7.5; MathWorks, Natick, MA). Data were analyzed using SPSS statistical software (IBM, v. 19.0; SPSS Inc., Armonk, NY), R statistical software (R Development Core Team, v. 2.14.0, Vienna, Austria), and G-power software (v. 3.1.2, Universität Düsseldorf, Düsseldorf, Germany).
To ensure the adequacy of our convenience sample size, an a priori power analysis was performed for a univariate Mann–Whitney U test comparing time to tracheal extubation in the BIS and ETAC groups. A consensus was reached among the clinicians acting as authors on this study that a 30-minute difference in time to tracheal extubation would be the smallest clinically significant intergroup difference. Assuming a difference of 30 minutes between the mean times to tracheal extubation (Cohen d = 0.75), an α of 0.05, and a β of 0.05 (power = 0.95), 50 patients per group would be needed to show a significant difference between the BIS and ETAC groups. Our convenience sample of >360 patients per group exceeded this number by >7 times. A post hoc sensitivity analysis conducted showed that the minimal detectable effect size in this study was 0.27, given our α, β, and our sample size. This indicates adequate power to detect a difference of much smaller than 30 minutes.
Univariate comparisons tested for differences between the 2 study groups. Kolmogorov–Smirnov goodness-of-fit tests with Lilliefors correction indicated that the continuous predictors and the outcome variable were not normally distributed (all P < 0.001). Therefore, nonparametric and semiparametric tests were used in subsequent comparisons. Continuous variables were compared using the Mann–Whitney U test, and frequencies were compared using χ2 or Fisher exact test, as appropriate. Wilcoxon–Mann–Whitney odds ratio was used as an effect size measure for the univariate time to tracheal extubation comparisons.7
Time to tracheal extubation comparisons were repeated after stratification based on the extubation outcome (successfully fast-tracked or failed fast track). A Kaplan–Meier plot stratified by group assignment was constructed to compare the time to tracheal extubation functions of the BIS and ETAC groups in patients who were successfully fast-tracked. Differences between the curves were evaluated using log-rank testing.
Significant multivariable predictors of the instantaneous probability of tracheal extubation were determined using time-dependent Cox regression analysis.8 The Cox model estimates the instantaneous probability that tracheal extubation will occur, based on a set of included variables. Variable selection was theory-driven, based on previous models of tracheal extubation times,9,10 as well as pathophysiological rationale. Patients who died in the intensive care unit before tracheal extubation were recorded as censored cases, that is, patients who did not reach the tracheal extubation end point (Fig. 1). Variables were entered into the model using forced entry. This entered all candidate predictors in the final model in a single step and kept them in the model regardless of their statistical significance. This was done to avoid problems associated with stepwise variable entry, such as inflated type 1 error rates.11 The randomness of the missing values within the dataset was also assessed, using Little Missing Completely at Random (MCAR) test.12 Eleven (1.5%) patients had missing blood loss values, representing 0.2% of all data. Since the MCAR test demonstrated a degree of nonrandomness (P < 0.001), the missing values were therefore estimated using 10 imputations of Markov Chain Monte Carlo simulation. However, the imputation model results were not found to differ from those derived from the original dataset. Thus, the missing values were addressed through listwise deletion.
Systematic testing of the Cox model assumptions was performed on the dataset to identify (1) predictors that needed to be transformed before being entered into the model and (2) time-related interactions that needed to be added to the model. The proportional hazards assumption, which states that the model-independent variables are not time dependent, was tested through graphical and correlation analysis of Schoenfeld residuals.13 Schoenfeld residuals are the difference between a given subjects’ observed and expected covariate values, resulting in multiple covariate specific residuals for each subject. Variables that did not meet the proportional hazards assumption were accounted for through a time interaction factor that was added to the Cox model, allowing the hazard ratio (HR) to change over time.
Lack of predictor multicollinearity was established through an examination of the tolerance statistic and variance inflation factor. The assumption of additive covariate effects was examined by incorporating interaction terms between key variables. In light of the study purpose, and to maximize model parsimony, interactions involving group assignment were given preference.
The linearity of the continuous predictors on the log-risk scale was assessed by comparing locally weighted scatterplot smoothing (lowess) fitted Martingale residual plots14 for both the original continuous variables and various transformations of these variables. An examination of the monotonicity and linearity of the HRs of the continuous variable quartiles was also done to identify thresholds in the relationship between the variable and the log-odds. If either method revealed a threshold in this relationship, the specific continuous predictor was replaced with its categorical equivalent for entry into the multivariate model. Otherwise, the most appropriate functional form was selected.
RESULTS
Baseline Comparisons
Patient baseline characteristics and nonanesthetic intraoperative variables are presented in Table 2. The 2 groups were very similar with the exception of intraoperative blood loss between the 2 groups (P = 0.036, but the shift median difference was small [−50 mL], and its 95% confidence interval extended from −50 to −1.23 × 10−05 mL, indicating that any difference present would be small. Furthermore, this variable was accounted for as a covariate in the subsequent Cox model).
Table 2: Patient Demographics and Surgical Details
Anesthetic Dosing and Protocol Compliance
There were no differences in the anesthetic drugs doses between the 2 groups (Table 3). There were also no differences between groups in the frequency with which the different anesthetic drugs were used (Table 3). A small number of patients (n = 19) received induction drugs other than propofol; however, there were no significant differences between groups in the frequency with which this occurred (Table 3).
Table 3: Anesthetic Variables
On average, patients in the BIS group were found to have a BIS value >60 in 2.4% of the operative time (interquartile range [IQR], 0.18–9.94], and <40 in 29.5% of the operative time (IQR, 10.6–64.2). On average, patients in the ETAC group were found to have an ETAC of <0.7 in 13.7% of the operative time (IQR, 8.41–20.5), and >1.3 in 2.6% of the operative time (IQR, 0–12.3).
Univariate Time to Tracheal Extubation Comparisons
Time to tracheal extubation comparisons are shown in Table 4. The majority of patients (75% in the BIS group; 78% in the ETAC group, (P = 0.426) were successfully extubated within 8 hours after chest closure. There were no significant differences between groups in time to extubation, even when stratified based on extubation outcome. Figure 2 shows the Kaplan–Meier curves for extubation in the BIS and ETAC groups, up to the fast-track cutoff (8 hours postoperatively; P = 0.111). Furthermore, the measures of effect sizes for the time to extubation comparisons were small (Wilcoxon–Mann–Whitney odds ratios between 1.04 and 1.14).
Table 4: Primary Outcome: Univariate Tracheal Extubation Analysis
Figure 2: Kaplan–Meier plot of time to extubation in fast-track patients (those extubated within 8 h of chest closure): Log-rank testing showed no difference between the 2 curves (P = 0.111). BIS = bispectral Index (n = 274); ETAC = end-tidal anesthetic gas concentration (n = 284).
Cox Regression Model of Tracheal Extubation
Table 5 outlines the adjusted HRs, indicating the probability of a patient being extubated at a given point after the surgery. Factors associated with a change in the instantaneous probability of tracheal extubation included body mass index (BMI), surgery type, logistic EuroSCORE,15 and the time of day when the surgery finished. Variable-time interactions were not found to be significant in the model presented. Additional details regarding how to interpret HRs for continuous and categorical variables can be found in Appendix 1.
Table 5: Adjusted Hazard Ratios Indicating the Probability of a Patient Being Extubated at a Given Point After the Surgery
DISCUSSION
These results show that the use of BIS-guided anesthesia administration in patients undergoing cardiac surgery was not associated with a higher instantaneous probability of tracheal extubation compared with patients having anesthetic drugs titrated based on an ETAC protocol. Instead, the patient’s BMI, logistic EuroSCORE, type of surgery, and the time of day the surgery ended affected the instantaneous probability of tracheal extubation.
Furthermore, it was found that group assignment did not have significant interactions with time from chest closure, age, blood loss, surgery type, logistic EuroSCORE, or BMI. An interaction was found between group assignment and anesthesia duration. However, the HR for this interaction was close to 1, and its 95% confidence interval extended to 1. Thus, group assignment to BIS or ETAC is unlikely to have a clinically significant impact on the instantaneous probability of tracheal extubation.
Our findings are consistent with previous studies of patients undergoing cardiac surgery, which found that BIS monitoring did not affect time to tracheal extubation.2,3 However, our study overcomes the limitations of these prior studies, namely the lack of an appropriate control group and well-defined fast-track cardiac anesthetic protocol.
A retrospective study on 198 patients found that 15% of patients with a BMI ≥30 failed to achieve tracheal extubation within 6 hours of cardiac surgery compared with only 2% of patients with a BMI <30.16 This may have been due to an increased risk of postoperative complications such as pulmonary atelectasis and the increased risk of difficult reintubation in these patients.17 Patients with a high logistic EuroSCORE are at a higher risk of short-term mortality.15 Consequently, they are likely at a higher risk of postoperative complications, which could necessitate longer intubation times. Complex surgical procedures, such as aortic aneurysm repair, may be associated with a greater risk of postoperative complications than less complex procedures. A retrospective analysis of 5798 patients found that operations on the aorta, as well as presence of a high logistic EuroSCORE were associated with a higher incidence of respiratory failure.18 This is consistent with our findings that non–coronary artery bypass graft surgery patients are more likely to have their trachea extubated later than patients undergoing isolated coronary artery bypass graft surgery.
Our study found that patients whose operations were completed in the evenings (4:00–7:00 PM) had a decreased instantaneous probability of tracheal extubation relative to patients whose operations ended in the afternoon (12:00–4:00 PM). This could be related to several factors. First, those patients with operations ending in the evening and night could have been more likely to have experienced intraoperative complications. However, our lack of patients undergoing emergency surgery, and the fact that 2 other potential indicators of intraoperative complications (blood loss and anesthesia duration) did not strongly influence the instantaneous probability of tracheal extubation, makes this unlikely. A reduced number of experienced intensive care unit staff during the evening and night might have influenced the decision to extubate the trachea. Alternatively, it is possible that patients with later finishing surgeries may simply be sedated with a plan for a morning tracheal extubation for convenience.
There are several ways BIS-guided titration could have reduced time to tracheal extubation. First, it could have reduced time to tracheal extubation by identifying the minimal amount of anesthetic needed to maintain adequate anesthesia in a patient. However, there were no significant differences in drug dosages between the 2 groups. Second, BIS monitoring could have directly influenced the anesthesiologist’s decision by helping them identify patients unsuitable for fast-track extubation. In this regard, data from the B-Unaware trial found that those who had BIS values <45 for long periods of time were at an increased risk of 1-year mortality.19 Thus, in this study, low BIS values in some patients could have theoretically influenced clinicians to prolong time to tracheal extubation. Despite this theoretical plausibility, our inability to find a significant difference in tracheal extubation times between the BIS and ETAC protocols indicates that BIS and ETAC monitoring had similar influences on the anesthesiologist’s decision to extubate the trachea.
We found that there was a larger proportion of time spent within the suggested ETAC range (0.7–1.3 age-adjusted MAC) in the ETAC group compared with the proportion of time spent within the suggested BIS range (40–60) in the BIS group. Prior research by our group has shown that overall, BIS is generally not sensitive to changes in age-adjusted MAC between 0.42 and 1.51 during the maintenance phase of anesthesia in surgical patients, and BIS values <50 may be seen at age-adjusted MAC levels <0.7.20 Therefore in some patients, it would be difficult to titrate anesthesia to achieve a BIS value >40 without putting them at an increased risk of intraoperative awareness. The small percentage of deviation >60 (BIS group) and <0.7 age-adjusted MAC (ETAC group) is not surprising, since anesthesiologists were encouraged to maintain a BIS of up to 75 (BIS group) or decrease anesthetic levels (ETAC group) around the time of chest closure.
There were several limitations to this study. As an effectiveness study, anesthesiologists could diverge from the suggested target BIS or ETAC range if they felt that such action was in the best interest of the patient. While this may have decreased time spent in the recommended range of BIS or ETAC values, making the protocol more restrictive would have prevented anesthesiologists from giving equal weight to other important factors in their clinical judgment. As mentioned above, the anesthesiologists were also allowed to decrease anesthetic levels at the time of sternal reapproximation to maintain a BIS <75, when their patients were in the BIS group. In contrast, anesthesiologists were simply encouraged to decrease anesthetic levels at the time of sternal approximation, without an equivalent target, when their patients were in the ETAC group. Third, anesthesiologists were permitted to use ketamine, which can cause slight increases in BIS during sevoflurane anesthesia.21 However, ketamine was normally used in low doses as a coanalgesic (Table 3). Furthermore, our univariate comparisons demonstrate that there was no difference in the frequency in the use of ketamine, and the relative doses of ketamine, between the 2 groups (Table 3). The administration of additional sedatives and analgesic drugs was also permitted. Again, however, the frequency of use of these medications, and their overall doses, were similar between groups (Table 3). Finally, the decision to extubate the trachea is also complex. Criteria used to determine readiness for tracheal extubation, such as the patient’s postoperative hemodynamic status, would be predictors of time to tracheal extubation. However, we wanted to identify demographics and surgical variables that would influence the probability of tracheal extubation at a given point, presumably by decreasing the time to tracheal extubation readiness.
This study has several strengths compared with previous trials that examined the influence of BIS on time to tracheal extubation. The appropriate sample size ensured that the role of BIS monitoring in early tracheal extubation would be unlikely to be confounded by a possible type 2 error. Furthermore, the inclusion of a control group receiving ETAC-guided anesthesia titration provides an appropriate comparator to test the effectiveness of BIS monitoring in promoting the instantaneous probability of tracheal extubation. This controls for the possibility that BIS monitoring could simply decrease tracheal extubation time relative to clinical practice due to the lack of a titration protocol in the latter. Finally, by using a fast-track anesthetic technique, patients could be extubated very shortly after surgery. The sooner the anesthesiologist is contemplating tracheal extubation, the more likely the intraoperative BIS and ETAC measures would be remembered, and taken into consideration in deciding when to extubate the patient. Therefore, this study involved circumstances where group assignment could have influenced the instantaneous probability of tracheal extubation.
On balance, the use of intraoperative BIS-guided anesthesia titration did not strongly increase the instantaneous probability of tracheal extubation compared with ETAC anesthesia titration for fast-track cardiac surgery. Instead, the decision to extubate the trachea was more influenced by patient characteristics and perioperative course.
APPENDIX 1—INTERPRETING HAZARD RATIOS
Table 5 presents adjusted hazard ratios (HRs) from the final Cox model, indicating the probability of a patient being extubated at a given instant after the surgery (assuming that the patient has not yet been extubated). With categorical variables, the HR gives a probability of extubation relative to a reference group. For instance, isolated cardiac valve repair, combined coronary artery bypass graft and valve surgery, and “other” cardiac operations had HRs <1, indicating that patients undergoing any of these surgeries had a lower instantaneous probability of having their trachea extubated relative to patients undergoing isolated coronary artery bypass graft procedures. If the variable is continuous, the HR reflects the change in the hazard that would occur with a unit increase in the value of the variable. For example, as logistic EuroSCORE15 increased, the probability of tracheal extubation occurring at any given time decreased. This means that a patient with a logistic EuroSCORE of 2% would have a HR of (0.926)2 = 0.857 (the baseline HR to the power of the units of the variable), whereas a patient with a logistic EuroSCORE of 15% would have a HR of (0.926)15 = 0.316. Thus, the patient with a logistic EuroSCORE of 2% would be 0.857/0.316 = 2.71 times as likely to be extubated as the morbidly obese patient immediately after chest closure.
DISCLOSURES
Name: Alexander Villafranca, BESS, MSc.
Contribution: This author helped design and conduct the study, helped design the substudy and collect the study data, analyze the data, wrote the manuscript, is the primary author of the manuscript, and is the author responsible for archiving the study files.
Attestation: Alexander Villafranca has seen the original study data and approved the final manuscript.
Name: Ian A. Thomson, BAS.
Contribution: This author helped design and conduct the study, analyze the data, and write the manuscript.
Attestation: Ian A. Thomson has seen the original study data and approved the final manuscript.
Name: Hilary P. Grocott, MD, FRCPC, FASE.
Contribution: This author helped design the substudy, analyze the data, and write the manuscript and approved the final manuscript.
Attestation: Hilary P. Grocott reviewed the analysis of the data and approved the final manuscript.
Name: Michael S. Avidan, MB, BCh, FCASA.
Contribution: This author helped design and conduct the study, analyze the data, and write the manuscript.
Attestation: Michael S. Avidan has seen the original study data and approved the final manuscript.
Name: Sadia Kahn, BSc, MD.
Contribution: This author helped design and conduct the study and write the manuscript.
Attestation: Sadia Kahn has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
Name: Eric Jacobsohn, MBChB, MHPE, FRCPC.
Contribution: This author helped design and conduct the study, analyze the data, write the manuscript and was the senior and corresponding author.
Attestation: Eric Jacobsohn has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
This manuscript was handled by: Charles W. Hogue, Jr., MD.
REFERENCES
1. Cheng DC, Karski J, Peniston C, Raveendran G, Asokumar B, Carroll J, David T, Sandler A. Early tracheal extubation after coronary artery bypass graft surgery reduces costs and improves resource use. A prospective, randomized, controlled trial. Anesthesiology. 1996;85:1300–10
2. Leslie K, Myles PS, Forbes A, Chan MT, Short TG, Swallow SK. Recovery from bispectral index-guided anaesthesia in a large randomized controlled trial of patients at high risk of awareness. Anaesth Intensive Care. 2005;33:443–51
3. Puri GD, Murthy SS. Bispectral index monitoring in patients undergoing cardiac surgery under cardiopulmonary bypass. Eur J Anaesthesiol. 2003;20:451–6
4. Avidan MS, Palanca BJ, Glick D, Jacobsohn E, Villafranca A, O’Connor M, Mashour GABAG-RECALL Study Group. . Protocol for the BAG-RECALL clinical trial: a prospective, multi-center, randomized, controlled trial to determine whether a bispectral index-guided protocol is superior to an anesthesia gas-guided protocol in reducing intraoperative awareness with explicit recall in high risk surgical patients. BMC Anesthesiol. 2009;9:8
5. Avidan MS, Jacobsohn E, Glick D, Burnside BA, Zhang L, Villafranca A, Karl L, Kamal S, Torres B, O’Connor M, Evers AS, Gradwohl S, Lin N, Palanca BJ, Mashour GABAG-RECALL Research Group. . Prevention of intraoperative awareness in a high-risk surgical population. N Engl J Med. 2011;365:591–600
6. Nickalls RW, Mapleson WW. Age-related iso-MAC charts for isoflurane, sevoflurane and desflurane in man. Br J Anaesth. 2003;91:170–4
7. O’Brien RG, Castelloe J. Exploiting the link between the Wilcoxon-Mann-Whitney test and a simple odds statistic. In: Proceedings of the Thirty-first Annual SAS Users Group International Conference.. 2006 San Francisco, CA SAS Institute Inc.:Paper 209–31
8. Kleinbaum DG, Klein MKleinbaum DG, Klein M. Extension of Cox proportional hazards model for time dependent covariates. In: Survival Analysis: A Self-Learning Text. 2005 New York Springer Science and Business Media Inc.:242–87
9. Cislaghi F, Condemi AM, Corona A. Predictors of prolonged mechanical ventilation in a cohort of 5123 cardiac surgical patients. Eur J Anaesthesiol. 2009;26:396–403
10. Wong DT, Cheng DC, Kustra R, Tibshirani R, Karski J, Carroll-Munro J, Sandler A. Risk factors of delayed extubation, prolonged length of stay in the intensive care unit, and mortality in patients undergoing coronary artery bypass graft with fast-track cardiac anesthesia: a new cardiac risk score. Anesthesiology. 1999;91:936–44
11. Harrell FE Regression Modeling Strategies: With Applications to Linear Models, Logistic Regression, and Survival Analysis. 2001 New York, NY Springer-Verlag
12. Little RJA. A test of missing completely at random for multivariate data with missing values. J Am Stat Assoc. 1988;83:1198–202
13. Schoenfeld D. Partial residuals for the proportional hazards regression model. Biometrika. 1982;69:239–41
14. Therneau T, Grambsh P, Fleming T. Martingale-based residuals for survival analysis. Biometrika. 1990;77:147–60
15. Roques F, Michel P, Goldstone AR, Nashef SA. The logistic EuroSCORE. Eur Heart J. 2003;24:881–2
16. Parlow JL, Ahn R, Milne B. Obesity is a risk factor for failure of “fast track” extubation following coronary artery bypass surgery. Can J Anaesth. 2006;53:288–94
17. Eichenberger A, Proietti S, Wicky S, Frascarolo P, Suter M, Spahn DR, Magnusson L. Morbid obesity and postoperative pulmonary atelectasis: an underestimated problem. Anesth Analg. 2002;95:1788–92 table of contents
18. Filsoufi F, Rahmanian PB, Castillo JG, Chikwe J, Adams DH. Predictors and early and late outcomes of respiratory failure in contemporary cardiac surgery. Chest. 2008;133:713–21
19. Kertai MD, Pal N, Palanca BJ, Lin N, Searleman SA, Zhang L, Burnside BA, Finkel KJ, Avidan MSB-Unaware Study Group. . Association of perioperative risk factors and cumulative duration of low bispectral index with intermediate-term mortality after cardiac surgery in the B-Unaware Trial. Anesthesiology. 2010;112:1116–27
20. Whitlock EL, Villafranca AJ, Lin N, Palanca BJ, Jacobsohn E, Finkel KJ, Zhang L, Burnside BA, Kaiser HA, Evers AS, Avidan MS. Relationship between bispectral index values and volatile anesthetic concentrations during the maintenance phase of anesthesia in the B-Unaware trial. Anesthesiology. 2011;115:1209–18
21. Hans P, Dewandre PY, Brichant JF, Bonhomme V. Comparative effects of ketamine on Bispectral Index and spectral entropy of the electroencephalogram under sevoflurane anaesthesia. Br J Anaesth. 2005;94:336–40
