The topography of awareness: a classification of intra-operative cognitive states


  • M. Wang,

  • A. G. Messina,

  • I. F. Russell

The nature of consciousness and memory

In the past it was assumed there were only two cognitive states relevant to anaesthesia: consciousness with memory; and unconsciousness. From this it was assumed that the obvious and straightforward way to determine the intra-operative cognitive state was postoperative reporting by the patient. Thus arose the concept that the postoperative interview was the ‘gold standard’ for determining this binary intraoperative state (albeit retrospectively), and therefore the criterion for an adequate anaesthetic [1]. This conceptual stance has been profoundly influential in the history, ethos and practice of anaesthesia.

Matters became more complicated, however, when studies investigating intra-operative response to command (using the isolated forearm technique (IFT)) began to demonstrate a discrepancy. Many patients making unambiguous, sensible responses to commands intra-operatively showed no evidence of postoperative recall [2–4]. As a consequence of the conceptual stance outlined above, some concluded that the IFT was unreliable [5]. However, the IFT was simply demonstrating that intra-operative consciousness is not necessarily correlated with postoperative recall. This was predictable from cognitive models of human memory function emerging from studies in the 1960s/1970s in both healthy volunteers and those with dense organic amnesic syndromes. Psychologists and neuroscientists were beginning to conceptualise memory in terms of two distinct stores, one with limited capacity and limited storage time (up to 20 s) and one with virtually unlimited storage [6]. Perception, experience and information were thought to pass through the initial ‘short-term’ store and either be moved into the ‘long-term’ store (a process known as encoding) or be completely lost to memory and consciousness. Recent studies of specific anaesthetic agents have demonstrated subtle differences in the mode of memory disruption, but all cause information loss in and around the encoding phase [7–9].

The key point of relevance to anaesthesia here is that the encoding process is peculiarly vulnerable to a variety of potentially disrupting influences that cause amnesia [6], e.g. opioids, hypnotics, benzodiazepines, alcohol and physical brain trauma. Thus it is no surprise that, following a bolus of an intravenous benzodiazepine or propofol, a patient can engage in a pre-operative conversation with the anaesthetist and later have no recall of this whatsoever. It would be quite wrong to imagine the patient had been unconscious at the time of the conversation, despite the lack of recall.

Unconsciousness or amnesia?

The intra-operative goals of general anaesthesia (GA) are often defined as hypnosis (unconsciousness), analgesia (specifically, to mitigate physiological shock) and amnesia. Actually, amnesia is only relevant when hypnosis (and arguably GA) has failed. The fact that most anaesthetic (and narcotic) agents have amnesic effects has obfuscated the primary endpoint of anaesthetic action – intra-operative unconsciousness. This has bolstered the flawed emphasis on postoperative recall as the most important indication of adequate anaesthesia.

This is a crucial ethical issue. If it is possible that a patient is conscious intra-operatively without any means of communication (because of neuromuscular blockade), it is also possible she is in pain or distress. We cannot know about the patient’s intra-operative mental state when no communication with the patient is attempted (as in typical anaesthetic practice). In one study investigating a new midazolam-alfentanil intravenous ‘anaesthetic’ technique, patients questioned intra-operatively using the IFT indicated not only consciousness but also pain [3]. Since all the patients were amnesic postoperatively, only the use of the IFT revealed this occurrence.

It is ethically unacceptable for a patient to be in pain and/or emotional distress intra-operatively without consent, even with assured postoperative amnesia. Moreover, anaesthetists have indicated that if they themselves required a GA, they would want guaranteed intra-operative unconsciousness and not amnesia [10]: there seems to be a dubious double standard here.

Implicit memory

Nevertheless, the extent of the patient’s postoperative conscious recall is still important, since it provides an indication of what is termed explicit memory, i.e. memory of learning or changed behaviour where there is knowledge of the context in which it was acquired. Conversely, implicit memory refers to a change in knowledge or behaviour, or the probability of a particular response to a stimulus, where there is no recall of the circumstances or context in which that learning took place.

During the 1970s and 1980s, cognitive psychologists began to apply implicit memory experimental paradigms, developed in the context of the investigation of memory function in patients with dense amnesic syndromes such as Korsakoff’s psychosis, to patients undergoing GA. Many studies appeared to show evidence of implicit memory in the absence of explicit recall [11]. These studies generally assumed that the patients were in an unconscious state at the time of intra-operative stimulus presentation, even though formal measures of depth of anaesthesia were never employed. However, when the IFT was used to ensure absence of response to command at the time of stimulus presentation, no memory of any type, either implicit or explicit, was found [12]. This led to the thesis that in previous studies successfully demonstrating implicit memory following GA, stimuli had been presented when the patient was conscious, though subsequently amnesic for the intra-operative episode [13].

In summary, these studies indicate that the previous, binary conception of GA will no longer do, and we have to take account of selective impairments of memory function. Specifically, during GA, a patient may be: (i) conscious with subsequent explicit recall; (ii) conscious with subsequent implicit but not explicit recall; (iii) conscious with neither explicit nor implicit recall; or (iv) unconscious with neither implicit nor explicit recall.

Emotional sequelae

Another complicating factor arises out of a consideration of the psychological and emotional consequences of awareness, the most common of which is post-traumatic stress disorder (PTSD). It has been well established, both clinically and experimentally, that PTSD encompasses both explicit and implicit memory processes in its genesis [14].

Implicit emotional memory following general anaesthesia

There are now many intriguing examples in the anaesthetic and medical journals of postoperative development of PTSD-like psychological problems in patients who have no explicit recall of their operation whatsoever [15]. These examples suggest the possibility of implicit emotional memory, and this is mediated by the amygdala. This type of traumatic memory is distinguished from standard implicit memory paradigms because of the emotional and affect-laden component of the implicit memory which is absent from typical implicit memory priming studies referred to above [11]. Whilst general implicit memory processes may be of academic interest to cognitive psychologists, implicit emotional memory is directly relevant to the genesis of more worrying, clinically significant postoperative psychological sequelae such as PTSD. Implicit emotional memory can be said to occur when an association is created between an intra-operative emotional state (e.g. anxiety or distress) and physical aspects of the operating environment (e.g. the sounds of monitors, the patient’s supine body position). Postoperatively, an encounter with a similar, previously innocuous, physical cue may trigger the emotion – but without any explicit recall of being conscious intra-operatively (because of the amnesic effects of anaesthetic drugs) despite the persisting emotional effect.

Sandin and colleagues [16] demonstrated that when patients were followed up prospectively, 65% of those reporting awareness did not do so until the second or third interview, partly because of delayed memory effects. There are also anecdotal reports of breakthrough memory occurring decades after the index traumatic event [17]. Moreover, in accordance with the Diagnostic and Statistical Manual IV [18], the formal diagnosis of PTSD requires a minimum of one month’s symptoms following the traumatic event.

Thus on the basis of these considerations, the classification of intra-operative states of consciousness must account for at least three time-points: intra-operative; immediate postoperative; and at least one month postoperative.

Intra-operative measures of cognitive state or depth of anaesthesia

Anaesthetic culture and training have promulgated the erroneous belief that so-called ‘clinical signs of anaesthesia’ (tachycardia, hypertension, lacrimation and sweating) reliably indicate intra-operative wakefulness, despite empirical evidence to the contrary [19]. It has long been known from psychophysiological studies of acute episodic stress and pain in animals and humans that a bradycardic and hypotensive reaction is just as common as the reverse [20]. Nevertheless, whilst the absence of these ‘clinical signs’ cannot be taken as a reliable indication that the patient is oblivious, their presence can be taken as a significant indication of light anaesthesia [19].

Generally, although cases of traumatic anaesthetic awareness in the absence of intra-operative neuromuscular blocking drugs (NBDs) have been documented, anaesthetic awareness is associated with neuromuscular blockade. In the absence of NBDs, body movement on the operating table can be taken as an indication of increasing probability of consciousness or ‘light’ anaesthesia. While some may consider the IFT controversial, there is general agreement that when NBDs are not used, sensible intra-operative motor responses to command unquestionably indicates consciousness and therefore, inadequate anaesthesia. The IFT should be considered as a special example of response to command.

Brain monitors based on passive or evoked EEG parameters are now more common (e.g. Bispectral Index (BIS), Narcotrend, Entropy). While their output has some correlation with brain depression, their ability to detect states of consciousness in the individual case continues to be disputed [21–23]. Moreover, their practical availability in the UK, whilst increasing, remains patchy. Further, monitors of depth of anaesthesia such as BIS unhelpfully conflate amnesia and intra-operative unconsciousness since typically the 40–60 ‘adequate anaesthesia’ window is validated against postoperative recall and not intra-operative unconsciousness.

Finally, all theatres are now equipped with end-tidal gas monitoring technology, which provide an indirect correlate of anaesthetic depth, although this too conflates amnesia with unconsciousness.

A classification of intra-operative states

The framework we present in Table 1 allows the identification of six grades of intra-operative state. Unlike other classifications [24], it focuses on intra-operative mental state as key to understanding postoperative outcomes. Intra-operative movement and response to command are seen as essential to determining intra-operative mental state. The classification is arranged hierarchically in order of increasing indications of intra-operative consciousness and memory on the basis of categorical criteria for each of the three time points. The table includes key studies or reviews that exemplify each of these categorical states. This classification provides a convenient and pragmatic framework for the evaluation and comparison of past and present studies of anaesthetic awareness and memory. It is currently being used to classify cases of awareness by the National Audit Project (NAP) 5 Review Panel (see and forms the basis of a forthcoming Cochrane Review concerning the prevention of unplanned anaesthetic awareness. It also provides a practical guide for clinical anaesthetists in evaluating and conceptualising the nature of anaesthetic states of consciousness and the assessment of anaesthetic depth.

Table 1.   Classification of intra-operative cognitive states (first presented at the 7th International Symposium on Memory and Awareness in Anaesthesia, Munich, Germany, March 2008).
 Intra-operative statePostoperative stateDescriptorExemplar study/ review
ImmediateLate (>1 month)
  1. PTSD, post-traumatic stress disorder.

0UnconsciousNo signs; no response to commandNo recallNo recallAdequate anaesthesia[12]
1ConsciousSigns/response to commandNo recallNo recall or emotional sequelaeIntra-operative wakefulness with obliterated explicit and implicit memory[4]
2Conscious; word stimuli presentedSigns/response to commandNo explicit recall, implicit memory for word stimuliNo explicit recall; implicit memory for word stimuli but no emotional sequelaeIntra-operative wakefulness with subsequent implicit memory[11]
3ConsciousSigns/response to commandNo recallPTSD/nightmares but no explicit recallIntra-operative wakefulness with implicit emotional memory[15]
4ConsciousSigns/response to commandExplicit recall with or without painExplicit recall but no emotional sequelae Awareness but resilient patient[16]
5ConsciousSigns/response to commandExplicit recall with distress and/or painPTSD/nightmares with explicit recall Awareness with emotional sequelae[25]

Competing interests

No external funding declared. MW is a member of the National Institute for Health and Clinical Excellence’s Diagnostics Advisory Committee on depth of anaesthesia monitoring, and also sits on the National Audit Project (NAP) 5 Steering Panel. AGM has been commissioned by the Cochrane Collaboration to lead the Cochrane review on anaesthetic awareness.