Awareness in cardiothoracic anaesthetic practice – where now after NAP5?

Cardiac anaesthesia has historically been associated with a higher incidence of unintended awareness compared with other anaesthetic subspecialties 1, 2, but the incidence in modern practice is less certain. The incidence in thoracic anaesthesia is also unclear, mainly because so few studies have addressed this issue at all 3, 4. The recent publication of the 5th National Audit Project (NAP5) report 5 now provides cardiothoracic anaesthetists with a useful point for reflection on current practice. The reported incidence of unintended awareness in cardiac practice ranges from less than 1% to over 20%, depending on the definition of awareness, the size of the study and the method of detection (Table 1). The early studies 6-8 included fewer than 60 patients each, so were subject to sampling error, and were carried out during the era of high-dose opioid anaesthesia. The later studies were prospective, used a balanced anaesthesia technique, and included 600-900 patients, finding an incidence of 0.3-1.14% 11-13. The incidence in the cardiac cohort of a large US multicentre study was similar, at 0.44% 2. Cardiac anaesthesia was thus associated with a two- to tenfold higher risk of unintended awareness than that reported for the general population 2, 16. 44 45 42 Opioid, N2O Opioid, droperidol Halothane 5% 13% 0% 99 204 Various Various 4% 1.5% 605 607 Balanced, routine care Balanced, BIS monitored 0.8% 0.16% 255 270 Balanced volatile, ETAC monitored Balanced volatile, BIS monitored 0.78% 0.74% 1037 1004 Balanced volatile, ETAC monitored Balanced volatile, BIS monitored 0.1% 0.5% The more recent B-Aware 3, B-Unaware 14 and BAG-RECALL 15 studies specifically recruited patients considered to be at high risk of awareness, so a third to a half of these cohorts were cardiac patients. However, they were not specifically cardiac studies, leading to a relatively small cardiac cohort in B-Unaware (525/1941 patients overall). The incidence of unintended awareness in cardiac patients in these studies varied from 0.1% to 0.8% 3, 14, 15, the lowest incidence occurring with protocolised monitoring of end-tidal agent concentrations 14. Why is there an almost tenfold variation in the reported incidence of unintended awareness in cardiac cohorts, and is there still a problem in cardiac anaesthesia, given that the lowest rate so far achieved (0.1% 14) is similar to that in the large general studies 2, 16? Part of the problem may be sampling error, since the largest dedicated cardiac studies thus far include only one tenth the number of patients of Sandin et al.'s landmark study 16. Other issues may be a reluctance both to adopt a more balanced anaesthetic technique for cardiac patients, and to use conventional doses of anaesthetic drugs in cardiovascularly fragile patients 17. In an audit feedback study, Ranta et al. reduced the incidence of unintended awareness during cardiac anaesthesia in their unit from 4% to 1.5%, the main change being more frequent continuous administration of volatile anaesthetic agents 10. A later study from the same group found that lower doses of midazolam were given to those cardiac patients who had unintended awareness 11. Regular audit and feedback, within a protocol-guided anaesthetic dosing schedule, may be all that is required to reduce the incidence of awareness in cardiac patients to that seen in the general population. Causes of unintended awareness during cardiac anaesthesia may include reliance on traditional high-dose opioid techniques 18 in combination with altered pharmacokinetics during cardiopulmonary bypass (including haemodilution, changes in acid-base status, and drug sequestration). Fentanyl in particular is sequestered on the plastics of the bypass circuit 19 and in the lungs 20, leading to the risk of sub-therapeutic analgesia in the context of light anaesthesia. Whilst there is a decrease in total blood concentrations of midazolam and propofol with the onset of bypass, there is a concomitant increase in the free fractions as a result of both haemodilution and a heparin-induced decrease in protein binding. In addition, metabolism and elimination are reduced during the hypothermic phase of bypass 21. For halogenated volatile anaesthetics, the blood:gas partition coefficients decrease with crystalloid haemodilution and increase with hypothermia 22. Wash-in and wash-out of volatile agents administered via the oxygenator of the bypass circuit are inversely proportional to blood:gas solubility, so that relatively insoluble agents are more rapidly taken up or eliminated than more soluble agents 21. The wash-in of isoflurane has an initial rapid phase if there is no isoflurane present in the circulation, with a much slower late phase, and the wash-out half-life is around 13-19 min 22, 23. Blood concentrations of isoflurane are around a half to a third of vaporiser settings 24. Increasing oxygenator sweep gas flow increases the uptake of volatile anaesthetics, whilst changes in pump flow rate have no effect on uptake 21. However, uptake of volatile anaesthetics is virtually non-existent with the newer ‘plasma-tight’ polymethylpentane (PMP) membranes (designed to reduce microbubble formation, blood trauma and plasma leakage), compared with polypropylene 25. When oxygenators with PMP membranes are used, it is important to ensure that blood concentrations of volatile anaesthetics are adequate before going on bypass, since monitoring of exhaust gas concentrations does not reflect blood concentrations 24, 25. In summary, for propofol, midazolam, and isoflurane, the effects of mild hypothermia and haemodilution at the onset of bypass tend to cancel each other out. With the increasing trend towards ‘fast-track’ management, and balanced anaesthetic techniques with moderate doses of fentanyl (or other opioid) for cardiac surgery, the use of high-dose opioid-only anaesthesia has waned. So, where does that leave us? The estimated incidence of spontaneous reports of awareness in the UK in the NAP5 study was 1:15 000 in the baseline survey 26 and 1:19 000 in the main study 5, remarkably less than in the formal studies with a Brice interview 2, 16. The possible reasons for this have been addressed elsewhere 27, 28, and it is important to stress that NAP5 does not provide a definitive incidence of unintended awareness during cardiothoracic anaesthesia. The NAP5 project received five or four spontaneous reports of unintended awareness during cardiac anaesthesia, depending on whether one reads Chapter 11 or Chapter 14, and four reports during thoracic anaesthesia (Table 2). However, only two of the four cardiac reports originated during the primary surgical procedure, with one of these having occurred quite a few years previously. One case occurred during re-operation for bleeding, and one case in a child in the catheter laboratory; this latter case could arguably be classified as paediatric or cardiological, rather than cardiac surgical. The situations causing unintended awareness during cardio-thoracic anaesthesia were similar to those in the general population, with the majority occurring at induction. The thoracic cases were qualitatively different to the cardiac ones (Table 2), in that the cardiac cases were mainly due to (deliberately?) light anaesthesia, while the thoracic cases were mainly cognitive or technical failures of anaesthetic delivery. The overall incidence of reports of 1:8600 for cardiothoracic practice is similar to the 1:8200 incidence when neuromuscular blocking drugs are used, perhaps indicating a similar root cause. Co-morbidity is high in both cardiac and thoracic practice, leading to reluctance to use conventional doses of the primary anaesthetic agent, although there were no episodes of awareness specifically linked to the cardiothoracic procedure (e.g. during cardiopulmonary bypass). The National Institute for Health and Care Excellence (NICE) has suggested the use of depth of anaesthesia monitoring, particularly bispectral index (BIS), where there is a high risk of unintended awareness 29. Though the definition of ‘high risk’ remains unclear, the increased incidence of awareness in cardiothoracic anaesthetic practice suggests that this is a risk category. A potential problem in cardiac anaesthesia is that the electrodes used for depth of anaesthesia monitoring will often compete for space with the near-infrared spectroscopy devices that are also increasingly used, suggesting a potential market for an integrated monitor. Using a depth of anaesthesia monitor may be useful during rigid bronchoscopy 30, and may be the only feedback of adequacy of anaesthesia during cardiopulmonary bypass. However, the impact of BIS monitoring in cardiac practice is variable, with a reduction in unintended awareness in cardiac patients in B-Aware 3, no effect in B-Unaware 14, and a higher incidence in the monitored group in BAG-RECALL 15. Part of the problem could be that BIS varies widely during fentanyl-midazolam anaesthesia, which is commonly used for cardiac patients, and may be unreliable with this drug combination 31. On the other hand, cardiothoracic may be the most appropriate specialty for evaluating depth of anaesthesia monitors. Patients with multiple co-morbidities may be compromised by the side-effects of excessive administration of anaesthetic agents. Conversely, abnormal cardiovascular physiology provides an unreliable surrogate for assessing adequate depth of anaesthesia. Anaesthetic requirements also differ before, during and after bypass. A lower concentration of isoflurane is required to maintain a BIS of 40-50 during mild hypothermic bypass with fentanyl than in the pre-bypass period 32, 33. In addition, BIS is higher in normothermic patients than during mild hypothermia at 32°C 34. These studies suggest that standard cardiac anaesthetic recipes may err on the light side for the current trend of ‘tepid’ bypass, particularly during the re-warming phase at the end of bypass. On the grounds of the risk of both underdosing and overdosing anaesthetic agents, perhaps there should be more habitual use of depth of anaesthesia monitoring in cardiothoracic cases. Cardiothoracic patients remain at risk of light anaesthesia, compounded by the potential for distraction in a complex surgical setting. The risk in cardiac anaesthesia is mainly in the pre-bypass period, which might be minimised by normalisation of practice to balanced anaesthesia techniques. While the findings of NAP5 are instructive, there are too few reports of unintended awareness during either cardiac or thoracic anaesthesia for us to be confident that we have a robust incidence of awareness during the primary procedure in either subspecialty, a point made in the cardiothoracic chapter of the NAP5 report. This is important to cardiothoracic patients and their anaesthetists, because of the persisting historical perception that cardiac anaesthesia in particular constitutes a high risk for unintended awareness. We suggest that a prospective subspecialty project be undertaken to evaluate the incidence of both spontaneously reported and Brice-elicited awareness in cardiothoracic practice. DS was NAP5 Moderator (independent of the NAP5 Project Team and Review Panel), and a Specialist Committee member for the NICE diagnostic guidance document on depth of anaesthesia monitoring. No external funding declared.