Early reports established the safety and value of a period of intermittent positive-pressure ventilation (IPPV) of the lungs after cardiac surgery [1-4]. Because the aims of treatment were to minimize respiratory and cardiovascular complications, the duration of postoperative IPPV was rarely < 12 h initially and often was > 24 h. Thus, cardiac surgery became unique in adult surgery in requiring postoperative IPPV as a surgery-specific indication, irrespective of the condition of the patient. However, prolonged periods of IPPV are not without complications and the concept that postoperative IPPV, though valuable, should be kept to a minimum was described some time ago .
However, at virtually the same time, high doses of opioids were advocated as being safe and effective in patients undergoing cardiac surgery. Initially, morphine [6,7], but more latterly fentanyl [8,9], sufentanil  and alfentanil , were used. The haemodynamic safety of these agents combined with excellent analgesia and anaesthesia was well established, but the high doses used made a prolonged period of postoperative IPPV of the lungs appear inevitable.
Studies of the effect of different anaesthetic agents on the outcome in coronary artery bypass graft surgery (CABG) [12,13] led to a reappraisal of the value of high-dose opioid techniques, and the use of a balanced technique of opioids and volatile agents enabled the period of postoperative IPPV to be shortened. However, the motivation to shorten the period of postoperative IPPV and extubate the tracheas of patients early varies and often seems to depend particularly on the healthcare system. In many countries, healthcare costs are easily quantified, and postoperative care costs have been shown to be substantially reduced by early tracheal extubation and reducing the time spent in the intensive care unit (ICU) [14,15]. In the UK, the nature of the National Health Service has meant that the relatively fixed provision of facilities has limited the expansion of the caseload. Thus, although the organizational benefits of early tracheal extubation may lie in the reduction of costs as elsewhere, the main benefit lies in more patients being able to use a fixed ICU resource or perhaps avoiding that fixed resource altogether [16,17].
Recently, we took part in a European multicentre audit designed to compare practices across 21 major cardiac centres in 11 European countries. This paper presents a detailed analysis of our contribution to that database. We have audited our current practice in an attempt to gain insights into which factors in our unit correlate with a process of early extubation and ICU discharge, and to examine the relationship between early extubation and in-hospital complications and hospital length of stay.
We prospectively collected data on 311 consecutive unselected patients undergoing cardiac surgery over 15 weeks. The data analysis was restricted to 296 survivors. The primary aim was to measure the time to tracheal extubation, defined as the time from arrival in the ICU, or equivalent, until the first removal of the endotracheal tube. We were also interested in determining which factors predicted early extubation in our centre, and whether early extubation was associated with significant benefits in terms of reduced postoperative complications and early hospital discharge.
We used a standard protocol to determine when patients were ready for extubation, which included the following:
• Effective gas exchange (partial pressure of arterial oxygen, PaO2) > 10 kPa; partial pressure of arterial carbon dioxide (PaCO2) < 7.0 kPa; saturation of arterial oxygen >95% with a fractional inspired oxygen concentration of 0.4 (FiO2) and a period of adequate spontaneous ventilation at an FiO2 < 0.4 with < 5 cm H2O continuous positive airway pressure (CPAP).
• Haemodynamic stability, with little change in drug therapy.
• No requirement for significant volume replacement for active haemorrhage (< 100 mL h−1).
• Central temperature rewarmed to 35.5°C and peripherally rewarming instituted.
• Patients should have regained consciousness and shown themselves to be responsive and grossly intact neurologically.
In addition, urine output was maintained at >0.5 mL kg−1 h−1, crystalloid fluid replacement was given at 0.75-1.25 mL kg−1h−1, and for each patient blood or colloid replacement was prescribed within set limits (transfusion 'trigger' haemoglobin concentration range 8.0-9.5 g dL−1). These variables were set as targets, which were achieved in all patients and did not primarily influence the time to extubation.
Patients undergoing cardiac surgery are initially managed in one of two units. The first, a general ICU, takes all cardiac surgery patients but predominantly the sickest patients or those in whom complications have developed requiring prolonged intensive care. The second is an overnight intensive recovery unit able to provide overnight, or in exceptional circumstances longer care, for cardiac surgery patients . Patients are discharged from these units to a 'step down' high-dependency unit or direct to the ward.
Data were collected from standard sources including medical and nursing records. These included baseline demographic details, cardiovascular risk factors and perioperative variables. All data were collected and entered on file at the time of activity; no data were entered retrospectively. However, data were downloaded onto the main study file in Vienna University, Austria, and at that time our data were screened for inconsistencies, obvious errors and blank fields by the Vienna study group. This, in effect, represented an independent check of our own audit data.
Statistical handling of the data required that the data be considered as a whole cohort and divided into groups. We chose four separate groups based on a prospective categorization:
- Group 1: extubated within 6 h of surgery.
- Group 2: extubated between >6 and 24 h.
- Group 3: extubated between >24 and 48 h.
- Group 4: extubated >48 h.
The data was collected using Microsoft® Access97® and analysed using a commercial statistics package, Stata 6.0®. The prevalence of the pre-, peri- and immediate postoperative variables was stratified by the four extubation groups. Univariate analysis was performed on all variables to examine the relationship between each variable and the main outcome: time to tracheal extubation. The time to extubation data were not normally distributed and, therefore, logistic transformation was carried out . A risk ratio estimating the effect of individual factors on the time to extubation was calculated. The analysis was carried out on the geometric mean of the time to extubation of the trachea. For categorical variables, the reference group used was either the lowest risk subgroup (e.g. no inotrope usage) or alternatively the largest group (e.g. primary CABG). Continuous variables (e.g. the time of the extracorporeal circulation) were tested for linearity and if this was appropriate the risk ratio was presented for each unit increase in that variable.
A predictive model of the time to tracheal extubation was produced using multiple regression. Variables that were significant in the univariate analysis (P < 0.05) were selected for inclusion in the multivariate model. Variables were added to the model in a step-wise fashion and retained if they remained independent predictors for delayed extubation after adjusting for the other variables in the model. Confidence intervals and significance tests were corrected for slight non-normality by using robust standard errors . The non-parametric U-test compared two postoperative variables not included in the above model.
Table 1 shows the number of patients undergoing each type of operation who were extubated at the times shown. Patient characteristic and other pre-, peri- and postoperative details are presented in Tables 2 and 3, stratified according to their time to extubation. Of patients, 39% were extubated within the first 6 h, and a further 50% were extubated between >6 and 24h. Thus, nearly 90% of the cohort was extubated within the first 24 h; the incidence rate of failed extubations was 1%. The cumulative frequency of the time to extubation for the whole cohort is shown in (Fig. 1). Interestingly, although the median time to extubation was nearly three times longer in Group 2 compared with Group 1 patients, the duration of stay in the ICU was similar in the two groups (Table 3).
Table 4 shows the effect of individual parameters on the time to extubation in the cohort as a whole. Gender, body mass index, a previous pattern of angina or myocardial infarction, diabetes mellitus, preoperative atrial fibrillation, and preoperative cerebrovascular accident with residual deficit demonstrated no significant association with prolonged intubation in this series. There was also no statistically significant correlation with age in our series.
In contrast, low ejection fraction (<30%), evidence of renal disease (creatinine > 166 μmol L−1) and chronic obstructive pulmonary disease requiring bronchodilator therapy were individual factors predicting prolonged intubation time. In addition, we found that the Euroscore - a composite score designed to predict perioperative mortality  - could also predict prolonged time to extubation.
Taking the operation of primary coronary revascularization on cardiopulmonary bypass (CABG) as the standard, we found that patients undergoing coronary revascularization without cardiopulmonary bypass (OPCAB) were extubated significantly sooner, and patients undergoing re-operation for CABG surgery and complex surgery were intubated significantly longer. In this series, mitral and aortic valve replacements did not significantly prolong the time to extubation. Taking elective surgery as standard, patients undergoing urgent and emergency procedures were intubated for a significantly longer time.
Our analysis of intraoperative variables showed that a longer duration of aortic cross-clamp and extracorporeal circulation predicted a significantly prolonged intubation time. Regarding the univariate analysis of postoperative variables, minimum arterial pH, maximum arterial base deficit, the use of inotropic drugs in progressive dosage and the use of an intra-aortic balloon pump were associated with a prolonged intubation time (Table 5).
Table 6 shows a multiple regression analysis with ventilator hours as the dependent variable. By combining these significant univariate pre- and perioperative predictors of prolonged intubation, we produced a predictive model that accounted for 32% (r2 adjusted) of the variation in ventilation times. Incorporating the most significant postoperative variables improved the predictive model to 46%.
Figure 2 shows a breakdown of the time to tracheal extubation in patients undergoing first-time CABG surgery according to the identity of the anaesthesiologist. The median times to extubation ranged from 5 to 9 h, but there were no significant differences between the individual anaesthesiologists, and each anaesthesiologist showed a considerable range in time to extubation. Three patients extubated after 48 h were excluded from this analysis since other factors were responsible for the prolonged ventilation.
Table 7 shows the impact of the time of the surgery and early postoperative blood loss on extubation time in first-time CABG patients. We found that patients undergoing surgery in the morning were extubated significantly earlier than those admitted to the postoperative care facility after 1.30 p.m. However, blood loss in Group 1 patients over the first 6 h was virtually identical to those extubated after 6 h.
Table 8 shows the proportion of patients within each group whose hospital discharge was delayed for >7 days. There was no significant difference between Groups 1 and 2 in the proportion of patients whose hospital discharge was delayed and the duration of hospital stay was similar in Group 1 and 2 patients, but significantly longer in Groups 3 and 4. The proportion of patients whose discharge was delayed due to medical complications as opposed to social factors was also similar between Groups 1 and 2. Approximately 85% of patients in Groups 1 and 2 were discharged home, and a further 10% went to convalescent homes. The nature of the complications that delayed discharge is shown in Figure 3. Arhythmias, respiratory complications - including chest infection - and leg wound problems were predominant. The cumulative frequency graph of the days to hospital discharge is shown in Figure 4.
Previously, cardiac surgery was an indication for prolonged postoperative ventilation of the lungs. Gradually, this period of IPPV has been shortened, linked to a perception that IPPV should only be continued until patients have recovered from surgery and not beyond. Some institutions have virtually eliminated this period of IPPV in many patients ; others have initiated protocols that have markedly reduced artificial ventilation times [23,24].
In the UK, the restricted number and difficulty in gaining access to intensive care beds has led to the development of other areas where postoperative cardiac surgery patients could be managed . In 1986, Aps and colleagues reported the postoperative management of cardiac surgery patients in a general recovery room, the patients' tracheas being extubated at the end of surgery . However, the further development of this concept into an overnight intensive recovery unit has enabled patients' lungs to be ventilated for a while and it is currently responsible for managing over 75% of our cardiac surgery patients, as well as other major surgery patients .
We divided our patients into four groups based on their extubation times. Those in Group 1 were patients in whom early and full recovery had taken place and who theoretically could be discharged to a 'step-down' high-dependency unit on the same day, thereby allowing the reuse of the intensive care or overnight recovery unit bed. Patients in Group 2 represented those who failed to be extubated early but who could still be easily managed in the overnight recovery unit. The increasing age and poor physical status of cardiac surgery patients has meant that we recognize a third group - patients who require up to 48 h IPPV and are contained in Group 3. Group 4 patients represent those patients who are in particularly poor condition preoperatively or who are undergoing non-elective, often complex surgery, or in whom the perioperative period was not straightforward, and sometimes a combination of all of the above.
One aspect worthy of comment is the length of stay of patients in Groups 1 and 2 in the intensive care or overnight recovery units. These numbers were nearly identical, despite much earlier extubation times in Group 1. The reasons for this are unclear, but we believe they are most probably organizational rather than clinical. Patients extubated later in the day will stay overnight in these units unless there is good cause to move them. This problem appears widespread and may well affect the results of surveys of ICU length of stay following cardiac surgery. For example, in a comparison of new treatments or postoperative analgesia and sedation regimens, clinicians are now well advised to record the time to fulfil the criteria for ICU discharge rather than the actual time of discharge itself. The similar length of stay of the two groups in the intensive care or overnight recovery units may reflect good hospital management, i.e. not moving patients around the hospital late at night. Alternatively, it may illustrate a failure to maximize the use of postoperative facilities that needs addressing.
Regarding the impact of individual factors on extubation time, much of our data agree with previous studies, although there are some differences. We did not find age an independent predictor, although this may be the result of a Type II error (Table 4). Our observations about early extubation of patients not subjected to cardiopulmonary bypass should also be interpreted with care. The numbers are small, and at the time under study most of our non-cardiopulmonary bypass practice was restricted to relatively low-risk patients who might be expected to be extubated sooner whatever the procedure. We have not risk-adjusted the non-cardiopulmonary bypass and CABG patients for comparison in this analysis. Nevertheless, it would be expected that all clinical factors taken together are capable of identifying cohorts of patients who are expected to extubate early. Perhaps, more importantly, we may be able to identify patients in whom early extubation is unlikely, thereby allowing improved use of resources. Multiple regression analysis shows that taking all factors together, our ability to predict which patients will or will not be extubated early is restricted to just under 50%. In some cases, this failure of predictive ability will be an improved clinical outcome for the patient, since a complex case who is extubated early may well be predicted to extubate late, and thus the patient becomes both a predictive failure and a clinical success.
Institutional factors may also affect the duration of intubation and examination of the data concerning the time of admission to the intensive care or overnight recovery units show that this is clearly the case. There is no apparent clinical reason why first-time CABG patients scheduled for morning surgery are extubated so much faster than those undergoing surgery in the afternoon. The reason for this observation is almost certainly related to staffing and organization. Again, there may be a difference between fulfilling the criteria for extubation and actually being extubated.
Another influencing factor might be the type of anaesthesia. It is well known that early extubation following high-dose opioid analgesia is difficult , although there are reports showing it to be possible . None of the 10 cardiac anaesthesiologists at our institution routinely use high-dose opioid anaesthesia, but the doses of drugs used vary and it is probably easier to consider the impact of anaesthesia by identifying each anaesthesiologist rather than by a detailed analysis of the interactions between the type and dose of drugs. We restricted this comparison to the most prevalent and standardized surgical procedure, first-time CABG surgery, in an attempt to try and ease comparison. The most striking aspect of the resulting data is the range in extubation times for all 10 anaesthesiologists, which implies that a number of factors play a large and confounding role in time to extubation. These ranges may be reduced by better adherence to postoperative protocols.
We found that haemorrhage, a factor often implicated in delayed extubation, was no different at 6 h between Groups 1 and 2 and therefore could not account for differences in times to extubation. There was a significant trend to a lower minimum pH in patients extubated later and greater use of inotropic drugs was associated with increasing duration of intubation.
Does early extubation confer any longer term benefit or disadvantage to the patients? At our institution, Group 1 patients had a marginally shorter duration of hospital stay, but a similar proportion of patients were delayed beyond 7 days compared with those patients in Group 2. The delay was mostly due to clinical complications, and although there were many complications documented, the main causes (new atrial fibrillation, respiratory problems including infection and leg wounds) are amenable to improvement and were common to both groups. Importantly, there were no disadvantages to early tracheal extubation. For the patient, the avoidance of an unnecessary period of IPPV of the lungs with the attending drug usage and risk of complications, and for the institution the ability to staff the intensive care or overnight recovery units at a lower level than otherwise might be necessary are the major gains. However, one might suggest that the clinical and organizational advantages are not as great as might be expected.
In conclusion, our current practice allows early extubation of the trachea and accelerated recovery in 39% of an unselected sequential 15 week cohort of cardiac surgery patients. All but 10% of patients were extubated within 24 h. Modelling shows extubation time can be predicted by clinical factors in 46% of patients. Length of stay in the intensive care or overnight recovery units and hospital stay is not significantly shorter in those patients who extubated < 6 h compared with those who extubated between > 6 and 24 h. There appeared to be no clinical reason for the comparable stay in the intensive care or overnight recovery units. Future efforts should be directed towards improving the size of the Group 1 cohort, reducing the time spent in the intensive care or overnight recovery units in this group for nonclinical reasons and minimizing in-hospital complications that lead to a delay in early discharge.
1. Spencer FC, Benson DW, Liu WC, Bahnson HT. Use of mechanical respirator in the management of respiratory insufficiency following trauma or operation for cardiac or pulmonary disease. J Thorac Cardiovasc Surg
2. Ditzeitlin GL. Artificial respiration after cardiac surgery. Anaesthesia
3. Cooperman L, Mann PE. Postoperative respiratory care: a review of 65 consecutive of open heart surgery on the mitral valve. J Thorac Cardiovasc Surg
4. Lefemine AA, Harken DE. Postoperative care following open heart operations: routine use of controlled ventilation. J Thorac Cardiovasc Surg
5. Sykes MK, Adams AP, McCormick PW, Bird B, Greenburgh S. The effect of mechanical ventilation after open-heart surgery. Anaesthesia
6. Lowenstein E, Hallowell P, Daggett WM, Austen WG, Laver MB. Cardiovascular response to large doses of intravenous morphine in man. N Engl J Med
7. Arens JF, Benbow BP, Ochsner JL, Therd R. Morphine anesthesia for aortocoronary bypass procedures. Anesth Analg
8. Sanford TJ Jr, Smith NT, Dec Silver H, Harrison WK. A comparison of morphine, fentanyl and sufentanil anaesthesia for cardiothoraic emergency and extubation. Anaesth Analg
9. De Lange S, Boscoe MJ, Stanley TH, Pace N. Comparison of sufentanil-O2
for coronary artery surgery. Anesthesiology
10. Truman KJ, McCarty RJ, el Ganzouri AR, Spiess BD, Ivankovich AB. Sufentanil-midazolam anesthesia for coronary artery surgery. J Cardiothorac Anesth
11. De Lange S, Stanley TH, Boscoe MJ. Alfentanil-oxygen anaesthesia for coronary artery surgery. Br J Anaesth
12. Tuman KJ, McCarthy RJ, Spiess BD, DaValle M, Dabir R, Ivankovich AD. Does choice of anaesthetic agent significantly affect outcome after coronary artery bypass surgery. Anesthesiology
13. Slogoff S, Keats AS. Randomised trial of primary anesthetic agents on outcome of coronary artery surgery. Anesthesiology
14. Cheng DC. Fast-track cardiac surgery, economic implications in postoperative care. J Cardiothorac Vasc Anesth
15. Chong JL, Pillai R, Fisher A, Grebenik C, Sinclair M, Westaby S. Cardiac surgery moving away from intensive care. Br Heart J
16. Westaby S, Pillai R, Parry A, Giannopoulos N, Grebenik C, Sinclair M. Does modern cardiac surgery require conventional intensive care. Eur J Cardiothorac Surg
17. Hutter JA, Aps C, Hemsi D, Williams BT. The management of cardiac surgical patients in a general surgical recovery. J Cardiovasc Surg
18. Aps C. Fast-tracking in cardiac surgery. Br J Hosp Med
19. Altman DG. Preparing to analyse data. In: Bland J, Altman DG, eds. Practical Statistics for Medical Research.
London, UK: Chapman & Hall, 1999.
20. Huber PJ. The behaviour of maximum likelihood estimates under non-standard conditions. In: Proceedings of the 5th Berkeley Symposium in Mathematical Statistics and Probability.
Berkeley, USA: University of California Press, 1967: 221-233.
21. Roques F, Nashef SA, Michel P, et al.
Risk factors and outcome in European cardiac surgery: analysis of the EuroSCORE multinational database of 19030 patients. Eur J Cardiothorac Surg
816-822; discussion: 822-823.
22. Lee TWR, Jacobsohn E. Pro: tracheal extubation should occur routinely in the operating room after cardiac surgery. J Cardiothorac Vasc Anesth
23. London MJ, Shroyer AL, Coll JR. Early extubation following cardiac surgery in a veterans population. Anesthesiology
24. Cheng DC. Anesthetic techniques and early extubation: does it matter? J Cardiothorac Vasc Anesth
25. Aps C, Hutter JA, Williams BT. Anaesthetic management and postoperative care of cardiac surgical patients in a general recovery ward. Anaesthesia
26. Moldenhauser CC. New narcotics. In: Kaplan JA, ed. Cardiac Anesthesia,
Vol. 2. New York, USA: Grune & Stratton, 1983: 31-79.