Consciousness and Anesthesia: An Update for the Clinician

I. INTRODUCTION

It is 7:18am on a Tuesday morning and you are standing in the pre-operative holding area. In front of you is the first patient of the day, a normally self-assured individual looking decidedly nervous. After the history and physical examination is complete, she looks at you intently and asks: “How will you know that I am unconscious during my surgery?”

First reported in 1846 in the original description of etherization by Bigelow¹, the history of intraoperative awareness is as old as the field of anesthesiology itself. In the most extreme cases, a patient may be able to describe in vivid and horrifying detail every event, conversation and sensation that occurred during a painful operation (see Rowan 2002 for a first-person description from a physician²). The unintended experience of intraoperative events can have potentially catastrophic psychological consequences, with a high proportion of patients developing posttraumatic stress disorder³.

This clinical update on consciousness and memory formation during surgery will review the clinical definition, epidemiology, risk factors and prevention of intraoperative awareness with explicit recall, as well as new directions in the monitoring of anesthetic-induced unconsciousness.

II. DEFINITION OF AWARENESS

In the cognitive sciences, “awareness” denotes simply subjective experience, but in clinical anesthesiology it has become shorthand for multiple cognitive processes:

$$\mathbf{\text{Awareness}}\mathbf{=}\mathbf{\text{consciousness}}\mathbf{+}\mathbf{\text{recall}}$$

Recall may be classified as explicit (a form of conscious memory) or implicit (a form of unconscious memory). Studies addressing awareness are concerned with explicit episodic memory. Awareness during surgery may thus be more accurately regarded as the failure to suppress consciousness (comprised of arousal and experience) and the ability to form explicit episodic memories of intraoperative events⁴. To-date we have no objective way for the assessment of subjective experience in a paralyzed or otherwise noncommunicating patient⁵.

Consciousness is possibly the greatest unanswered question in science. Its enigma has dominated the thinking of some of the greatest scientific minds throughout history, from Socrates and Plato in the 5^th century B.C., up to Nobel Prize-winning scientists of the modern era, such as Francis Crick⁶ and Gerald Edelman⁷. The term “consciousness” can be further subdivided as connected consciousness and disconnected consciousness⁸. Connected consciousness refers to consciousness of external or environmental stimuli, whereas disconnected consciousness refers to internally conscious states such as dreaming, in which an individual is unresponsive to environmental input. Both of these forms of consciousness are dissociable from explicit episodic recall, i.e., we may be conscious of external or internal events without remembering them.

An example of this phenomenon is that responsiveness to commands under anesthesia, demonstrated by the isolated forearm technique, is often not associated with memory. In one recent review⁸, 37% of patients assessed with the isolated forearm technique could follow auditory commands and thus has connected consciousness, but only 25% of those who responded under anesthesia had any recall of events. These patients are connected with their environment, but lack the ability to form memories and thus could not be said to suffer from awareness as clinically defined in anesthesiology. It is, however, important to note that it is highly likely that far more patients experience intraoperative consciousness than awareness with explicit recall. Disconnected consciousness, even with associated explicit recall of dream content, is also considered distinct from what we conventionally define as awareness. For example, dreaming under anesthesia with explicit recall is common and may occur in approximately 1 in 4 patients⁹. It is now clear that dreams during anesthesia are not necessarily “near-miss” awareness events.

III. EPIDEMIOLOGY OF AWARENESS

a. Incidence of Awareness

Awareness has been consistently reported across several large studies using the standardized Brice interview¹⁰ as occurring in 1-2/1000 of anesthetics, with some recent studies reporting an incidence of up to 1%. Discrepancies in rate of incidence may relate to the definition of awareness used.

b. Describing the severity of Awareness

Classification tools have been developed to assess the qualitative aspects of the awareness experience. The Michigan Awareness Classification Instrument (MACI) ¹⁵ is one such classification.

The MACI has good inter-observer reliability, allows for statistical analysis of the severity of awareness and may help the study of interventions to reduce awareness and post awareness sequelae.

IV. RISK FACTORS FOR AWARENESS

Since intraoperative awareness is a rare event, our understanding of the risk factors is imprecise. Observational studies over decades have attempted to characterize risk factors for awareness¹⁶. However, over time there have been substantial changes in clinical practice, so it may be the case that both the incidence and risk factors for awareness have changed over time⁴.

Two broad groups of patients are most at risk for awareness.

1. Those who receive an insufficient dose of anesthetic drug from their anesthesia provider

2. Those who receive what is typically considered a sufficient dose but are resistant to anesthetic effects

a. Underdosed Patients

Anesthesia providers may chose deliberately to give a patient a minimal dose of anesthetic drug due to physiologic instability or to prevent physiologic instability in the fetus. ^16,17,18,19 However, there are also situations in which anesthesia providers unintentionally deliver inadequate anesthesia, such as with vaporizer failure, interrupted intravenous anesthesia, or during difficult intubations in which there is no redosing of sedative-hypnotic.

b. Patient Resistance Factors

Patients may have an acquired or intrinsic resistance to the anesthetic effects. Patients who may have an acquired resistance to anesthetic drugs include:

1. Regular users of benzodiazepines, opiates or other drugs that affect the cytochrome P450 enzyme system such as alcohol.

2. Regular users of other upregulators of the P450 3A enzyme pathway (i.e. St. John’s Wort, efavirenz, nevirapine, barbiturates, carbamazepine, glucocorticoids, phenytoin and rifampicin) ²⁰; conceivably might also increase resistance to anesthesia.

Certain genetic mutations, most notably the melanocortin-1 receptor gene mutation, which confers red hair phenotype, may cause to increased resistance to inhalation anesthesia²¹. Another example of a genetic contribution to anesthetic sensitivity is resistance to the amnesic effects of etomidate²² or isoflurane²³ conferred by the presence of an α-5 subunit mutation of the GABA_A receptor. It is possible that other, as yet unknown, genetic mutations may also be linked with anesthesia resistance or susceptibility.

Patients at high risk of awareness have been grouped together for study in multicenter trials of awareness, aiming to increase the power of these studies. The risk factors used in the largest study of awareness to date, the BAG-RECALL trial²⁴, are listed in table 3.

Table 3:

Risk factors ²⁴

Criteria for High Risk of Awareness
Planned open-heart surgery
Aortic stenosis
Pulmonary hypertension
Use of opiates
Use of benzodiazepines
Use of anticonvulsant drugs
Daily alcohol consumption
ASA status 4
End-stage lung disease
History of intraoperative awareness
History of or anticipated difficult intubation
Cardiac ejection fraction <40%
Marginal exercise tolerance

Note that emergency or obstetric cases were not included in this study, but are also risk factors.

V. PREVENTION OF AWARENESS AND THE DEPTH OF ANESTHESIA

What can be done to prevent awareness? As with all aspects of anesthesia practice, attention to detail is the key; a checklist based approach may help to achieve this.

Muscle relaxants compound the risk of anesthetic underdosing by preventing the patient’s ability to signal distress. In one study, the incidence of awareness was 0.18% when muscle relaxants were used versus 0.10% without¹¹. In another, all patients who reported awareness had received muscle relaxants, reporting the inability to move as the most distressing sensation¹³.

Anesthetic technique used, as well as the type and timing of surgery, has been implicated in increasing the frequency of awareness. Total intravenous anesthesia (TIVA) is associated with a higher frequency of awareness²⁵ versus inhalational agents. This may be because there is currently no standard dosing metric for TIVA that can be monitored intraoperatively, so anesthetic delivery failure might be missed.

MEASURING THE DEPTH OF ANESTHESIA

Depth of anesthesia was initially described as a classification of clinical observations in response to a single inhaled agent, Diethyl Ether. First described by John Snow in 1847²⁶ and formalized in 1937 by Guedel²⁷, these “planes” are now rarely seen in the modern era of rapid induction with intravenous drugs.

Modern measures of anesthesia depth rely upon either one or both of two modalities:

1. end-tidal drug monitoring

2. electroencephalographic (EEG) based techniques

a. Minimum Alveolar Concentration (MAC)-BASED TECHNIQUES

Calculation of the (MAC), which reflects a common expression of anesthetic dosing, dates from 1965²⁸. Pharmacologically, MAC is the ED₅₀, i.e., the effective dose at which 50% of the population will not move in response to a standard noxious stimulus. The concept of MAC has since been expanded to include MAC-Awake, the MAC at which 50% of the population will regain consciousness, which is usually noted at between 0.3 to 0.5 MAC. The very nature of MAC being an ED₅₀ and thus a measure of the mean implies that some patients with an end-tidal MAC of 1.0 will be outliers and thus might be aware.

MAC alarms have been demonstrated to reduce the risk of awareness. An awareness rate of 2/2852 (0.07%) of patients in a surgical population deemed at high risk of awareness (in which the expected incidence is up to 1%) was achieved using alarms set to 0.7MAC²⁴. MAC is a reliable metric, but is limited to inhaled anesthetics, motivating the use of end-organ monitors such as the EEG.

b. EEG BASED TECHNIQUES

EEG based anesthesia monitoring technologies fall into 2 general categories: spontaneous electroencephalography (raw or processed) and evoked potentials. There is an extensive range of monitors available, reviewed elsewhere²⁹. Clinical anesthesiologists can be trained to interpret a raw EEG at a basic level³⁰ and even to predict Bispectral index (BIS) values with accuracy similar to a second BIS monitor used in tandem³¹. However, processed EEG can potentially aid interpretation for the untrained and can facilitate numerical thresholds for the purpose of subanesthetic alerts.

There has been an expansion of technologies to process and interpret the raw EEG, putatively producing a measure of the depth of anesthesia. At a basic level, these devices apply Fourier transformation to the EEG signal before analyzing each frequency domain of the EEG or spectrum and its contribution to the total signal³². Some devices use a proprietary EEG algorithm, calibrated against observed responses to anesthesia, while others use publically available EEG interpretation algorithms of the entropy within the EEG signal³³. Many commercially available monitors are broadly comparable in reflecting anesthetic dosing³⁴.

The most widely used device with the largest record of use in randomized controlled trials is the Bispectral Index ™ monitor (BIS ™ monitor; Covidien Medical, Boulder, CO). BIS uses a proprietary EEG algorithm, which generates values from 0 (EEG silence) to 100 (fully awake), with “surgical anesthesia” defined as a range of 40-60. Following the first studies in the late 1990s validating the device^{35, 36}, BIS has been investigated as an awareness monitor in a series of large trials^{17, 24, 37}.

Studies of the BIS and Awareness

The B-Aware trial, the first randomized, multicenter trial comparing BIS monitoring versus routine care in a population of 2,463 patients deemed at high risk for awareness, demonstrated a 80% relative reduction in awareness versus standard care (2 versus 11 cases)¹⁷. The B-Unaware trial, a further study of 1,941 patients similarly deemed to be at high risk of awareness, failed to demonstrate a difference in outcome between a BIS-guided and end-tidal anesthetic concentration (ETAC)-guided protocol, with each group reporting 2 cases of awareness³⁷.

BAG-RECALL, the most recent and largest multicenter trial to date, assessed BIS-guided and MAC-guided protocols in 5,713 patients at high risk for awareness. The BAG-RECALL trial demonstrated conclusively that BIS protocol was not superior to an ETAC protocol in preventing awareness²⁴. 7 of 2,861 patients (0.24%) monitored with BIS experienced definite awareness versus 2 of 2,852 (0.07%) of those in whom ETAC alerts were used. It has been suggested by Cochrane meta-analysis³⁸ that the BIS may prevent awareness compared to routine care, but in patients receiving potent volatile anesthetics an ETAC-guided protocol is a cost-effective alternative. Furthermore, a multicenter trial recently demonstrated that the BIS reduced awareness incidence in patients receiving TIVA compared to standard care (without target-controlled infusion pumps) ²⁵.

Limitations of the BIS

The BIS monitor and similar technologies that also rely upon processed EEG have some practical limitations. Anesthetic drugs such as ketamine³⁹ and nitrous oxide⁴⁰ do not act by potentiating the effects of gamma-aminobutyric acid; they increase EEG frequency, which can cause a paradoxical increase in the BIS value. Nitrous oxide and Ketamine may produce unconsciousness via an excitatory pathway, which is not reflected in the BIS and other processed EEG indices. Xenon also produces atypical changes but to a lesser extent, with initially stable processed EEG indices rising after approximately an hour of anesthesia time⁴¹.

It is also unclear how well BIS reflects anesthetic dosing over a clinically relevant range of MAC. Analysis of over 3 million data points from 1,100 patients of end-tidal anesthetic concentration (ETAC) versus BIS value demonstrated a poor correlation between them and a substantial lag in changes in BIS value versus changes in the ETAC⁴². Furthermore, patients may lose consciousness at one BIS value and regain it at another. Similar hysteresis of drug concentration to processed EEG values appears to occur with TIVA⁴³. However, this is not due to processing lags of processed EEG⁴⁴, but likely to the dissociable neurobiology of induction and emergence⁴⁵. Emergence from anesthesia is dependent on the re-activation of the circuits of arousal, which may not be involved in the mechanism of anesthetic induction⁴⁶. This distinct neurobiology may reflect neural inertia⁴⁷, the resistance of the brain to transitions between the conscious and unconscious state and vice versa. Neural inertia explains why the current generation of anesthesia monitors does not demonstrate the same values at the point of loss of consciousness and emergence and implies that there should be a small buffer to awareness if the MAC transiently decreases. Certain patients may have reduced neural inertia, leaving them without this “buffer zone” and allowing for more rapid state transitions from unconsciousness to consciousness.

For an anesthetic monitor to produce a higher fidelity measure of awareness, new techniques rooted in the neurobiology of consciousness are required. Identifying a common neural correlate of both depressive and excitatory anesthetic-induced unconsciousness is the key to developing a superior monitor of awareness⁴⁸.

VI. OBJECTIVE MEASURES OF CONNECTED CONSCIOUSNESS AND MEMORY

Can we monitor memory?

From observations of patients with selective neurological injuries, it is known that the medial temporal lobe plays a critical role in explicit recall; lesions of the hippocampus may prevent the formation of episodic memory⁴⁹. In particular, an oscillatory pattern in the 4-12 Hz, or θ range, has been linked to memory processes in both primates and non-primates⁵⁰. Certain anesthetics have been shown to slow these θ oscillations⁵¹; the question of whether this slowing is the cause of memory blockade or merely an epiphenomenon remains unanswered. Because of the location of explicit recall centers in the medial temporal lobe (and thus deep to the cortex), memory monitoring at present appears to be a difficult candidate for an awareness monitor.

How do we measure consciousness?

There is no discreet consciousness center in the brain. There are a number of wake- and sleep-promoting nuclei, located within the pons, midbrain, hypothalamus and basal forebrain. These centers regulate sleep and are hypothesized to co-ordinate in a “flip-flop” manner to regulate sleep-wake cycles. However, brain arousal and sleep-wake cycles are not sufficient for the subjective dimension of consciousness, as patients in vegetative states may have such cycles in the absence of experience⁵². Furthermore, like the structures of the medial temporal lobe mediating explicit recall, subcortical structures mediating arousal are not amenable to real-time monitoring in the operating room. Cortical and thalamocortical activities are thought to be more closely related to experiential processing and are accessible to study via multiple neuroimaging modalities.

One currently prevailing consciousness hypothesis is that consciousness depends on the integration of information from dynamic interactions across large-scale neural networks in the brain^{53, 54, 55}. Accordingly, general anesthesia might suppress consciousness by either disrupting⁵⁶ or unbinding⁵⁷ this integrative process. With this in mind, the key to measuring connected consciousness may lie not in measuring the activity of one specific center or region of the brain, but instead in tracking the connectivity across neural areas.

a. NEUROIMAGING ANESTHETIC MODULATION OF THE CONSCIOUS STATE

Neuroimaging efforts to assess the effect of anesthetics can be divided into three basic categories:

1. assessment of the effects on anesthesia on baseline activity, reflected by cerebral metabolism and cerebral blood flow

2. responsiveness of neuronal networks to a sensory input or task

3. functional, directional or effective connectivity of networks in the brain.

a. Suppression of metabolic activity

Positron emission tomography (PET) can be used to assess cerebral metabolic rate (CMR). General anesthetics globally reduce CMR and blood flow (CBF), by 30-70% at the point of loss of consciousness from propofol infusion⁵⁸, with similar results for isoflurane⁵⁹ and midazolam⁶⁰. It is unclear if the reduction in CMR / CBF is directly responsible for loss of consciousness or a consequence of altered network interactions. Consciousness can co-exist with a low CMR state in certain neurological conditions⁶¹. Ketamine increases CMR⁶²; thus, the excitatory pathway to unconsciousness does not conform to the patterns of primarily inhibitory anesthetics.

b. Suppression of the Thalamus

PET scanning suggests that thalamic suppression is the common site of action for anesthetics⁶³, a potential mechanism of unconsciousness shared with non-REM sleep and persistent vegetative state⁶⁴. However, preliminary electrophysiologic evidence that thalamic suppression follows cortical suppression in time favors an indirect role in anesthetic-induced unconsciousness rather than a causal one⁶⁵. Furthermore, a recent functional magnetic resonance imaging study suggested a more profound functional disconnection with the putamen at the point of propofol-induced unconsciousness, with thalamic connectivity relatively well preserved⁶⁶.

c. Metabolic and functional suppression of frontoparietal networks

The frontoparietal network consists of the prefrontal and superior/posterior parietal cortex and is linked to attention, perception, working memory and consciousness. The frontoparietal network has been shown to be suppressed during anesthesia, non-REM sleep, seizures and coma⁶⁷; breakdown in frontoparietal connectivity is associated with loss of consciousness⁶⁸. Conversely, return of function there is associated with emergence from vegetative state⁵².

Magnetic resonance imaging assessment of functional connectivity during propofol exposure shows suppression of frontoparietal networks with preserved activity in the sensory cortices, suggesting that anesthetic-induced unconsciousness is a higher-order process⁶⁹. Neuroimaging studies also confirm that propofol disrupts functional interactions between sensory and high-order processing systems preventing the conscious perception of incoming stimuli⁶⁹. This breakdown in communication between the higher and lower cortical processing networks appears to be key in losing consciousness. In support of this, the induction of anesthesia is associated with a loss of effective connectivity (implying causal interactions) across the cortex, as measured by EEG⁷⁰.

b. EEG-based Connectivity in Frontoparietal Networks

Current neuroradiological techniques have advanced our understanding of anesthetic-induced unconsciousness, but have poor temporal resolution and are impractical for the operating room. One potential method to assess consciousness is to measure cortical effective connectivity and, in particular, feedback connectivity from the anterior to posterior areas of the brain⁷¹. For example, evoking activity within the primary visual cortex and its associated feedforward pathway to higher processing areas is not necessarily associated with conscious experience; a feedback pathway is required⁷². In non-REM sleep and propofol anesthesia the anterior-to-posterior propagation of slow waves is similar^{73, 74} and may lead to a disruption of directional connectivity that is common to the two states.

EEG studies have demonstrated that frontal-to-parietal feedback connectivity is preferentially inhibited by propofol in healthy volunteers⁷⁵ and both propofol and sevoflurane in surgical patients⁷⁶. Feedback connectivity has also been shown to increase during recovery from general anesthesia (Figure 1)^75,76,77; the importance of frontoparietal connectivity in recovery from anesthesia has been confirmed by a recent PET study⁷⁸. Selective disruption of feedback connectivity disruption occurs not only in anesthesia but also in vegetative states, while in minimally conscious states directional connectivity patterns are similar to conscious controls⁷⁹.

Figure 1:

Preferential inhibition of feedback connectivity (in striped bars) in the frontoparietal network. Note that at surgical anesthesia (in green) with either propofol (n=9) or sevoflurane (n=9), both directions of frontal-parietal connectivity are suppressed. The measure of directional connectivity in this study was symbolic transfer entropy. ⁷⁵

Directional connectivity, non-invasively measured using multichannel EEG is a possible novel metric for assessing consciousness that is connected to the environment. It is disrupted by different classes of anesthetics and in pathologic conditions such as the vegetative state.

VII. CONCLUSIONS

Intraoperative awareness with explicit recall occurs in 1-2 cases per thousand and is linked to the neuroscientific problem of consciousness. Current brain monitoring techniques may be useful in preventing awareness during routine care or TIVA, but are not superior to protocols based on anesthetic concentration. Real-time measure of functional and directional connectivity across different brain regions known to be important for connected consciousness may be the next wave of intraoperative brain monitors.

Table 1:

Major studies of awareness incidence

Study / Year	Population / Location	Incidence	Percent Aware
Sandin 2000	Adult – Sweden	19 / 11,785	0.16%
Sebel 2004	Adult – USA	25 / 19.575	0.13%
Errando 2008	Adult – Spain	39 / 3,921	1.0%
Davidson 2011	Pediatric – Multiple	33 / 4,486	0.73%

11–14

Table 2:

Michigan Awareness Classification Instrument

Class 0:	No awareness
Class 1:	Isolated auditory perceptions
Class 2:	Tactile perceptions (e.g. surgical manipulation of endotracheal tube)
Class 3:	Pain
Class 4:	Paralysis (e.g. feeling one cannot move, speak, or breathe)
Class 5:	Paralysis and pain

An additional designation of D for distress is included for patient reports of fear, anxiety, suffocation, sense of doom, sense of impending death, etc. If multiple experiences are reported, the highest classification category is used.

TABLE 4.

Awareness Prevention Checklist modified from Mashour Anesthesiology 2011⁴

• Equipment and drug check including dosages; are all drugs labeled and infusions running into veins?

• Consider an amnesic premedication or redosing in patients at high risk such as those with a history of awareness.

• Avoid muscle relaxants, or minimize the dose given with nerve stimulator monitoring.

• Use a potent inhalational agent rather than total intravenous anesthesia, if possible.

• Set a minimum alveolar concentration alarm to at least 0.5 to 0.7 MAC.

• If on cardiac bypass, monitor anesthetic agent concentration within the circuit.

• Treat hypotension by another means than reducing the anesthetic concentration.

• If concerned about hemodynamic compromise, use drugs with the widest therapeutic index; administer benzodiazepines or scopolamine.

• Supplement hypnotic agents with systemic opioid or local analgesia to minimize experience of pain if awareness occurs.

• Consider the use of a brain monitor; i.e. raw or processed EEG.

• Routinely monitor the brain if using total intravenous anesthesia.

• Evaluate known risk factors for awareness; consider if a higher dose of anesthetic may be required.

• Re-dose the intravenous anesthetic is intubation is difficult or prolonged, or during rigid bronchoscopy.