The ability of bispectral index to detect intra-operative wakefulness during total intravenous anaesthesia compared with the isolated forearm technique


Correspondence to: I. F. Russell



It has been suggested that monitoring during total intravenous anaesthesia should include aspects of brain function. The current study used a manually adjusted target-controlled infusion of propofol for anaesthesia, guided to a bispectral index range of 55–60. Intra-operative responsiveness, as assessed by the isolated forearm technique, was compared with whether the bispectral index predicted/identified a patient's appropriate hand movements in responses to commands. Twenty-two women underwent major gynaecological surgery with total intravenous anaesthesia, propofol, remifentanil and atracurium. Sixteen women responded, on 80 occasions, with appropriate hand movements to commands during surgery, of which the bispectral index detected 47 (sensitivity 59%). The bispectral index suggested consciousness 220 times in the absence of movement responses (specificity 85%). The positive predictive value of a bispectral index response was 18%. While two women had vague recall about squeezing fingers, none had recall of surgery. For patients who responded more than once during surgery the bispectral index value associated with a response was not constant. Although there was no difference in the median (IQR [range]) effect site propofol concentration between intra-operative responses (2.0 (1.5–2.3 [1.2–4.0]) μ−1) and eye opening after surgery (2.1 (1.7–2.8 [1.5–3.9]) μ−1), the median (IQR [range]) bispectral index value at eye opening after surgery was significantly higher than that associated with responses during surgery: 75 (70–78 [51–93]) vs 61 (52–67 [37–80]) respectively, (p < 0.001). The manual control of propofol intravenous anaesthesia to target a bispectral index range of 55–60 may result in an unacceptable number of patients who are conscious during surgery (albeit without recall).

Inhalational anaesthesia can be monitored using end-tidal agent levels to help control anaesthetic delivery, but with total intravenous anaesthesia (TIVA), there is no direct method to measure the concentration of drug in the patient's key body compartments. It has been suggested that, compared with inhalational anaesthesia, the incidence of awareness may be higher when TIVA is used [1] and recent large-scale studies support this opinion [2-4]. It therefore seems desirable to use some form of anaesthesia brain monitor (ABM) to assess the effects of anaesthetic drug delivery. Indeed, it has been suggested that using the Bispectral Index Monitor (BIS; Covidien Medical, Boulder, CO, USA) can reduce the incidence of awareness with TIVA from 0.5% to 0.14% [4].

However, one problem in interpreting the results of this and other studies claiming that ABMs can reduce the incidence of awareness [5, 6] is that the conventional definition of ‘awareness during anaesthesia’ requires that patients, in the postoperative period, have recall of intra-operative events. Studies with the isolated forearm technique (IFT) have shown that patients can be conscious during general anaesthesia with the ability to respond to complex conditional commands, yet have no recall of this intra-operative consciousness in the postoperative period [7-10]. This state (conscious during surgery but with no recall) has been called ‘wakefulness’ [11, 12]. ‘Wakefulness’ probably occurs because sub-anaesthetic doses of anaesthetic agents have amnestic effects, which can prevent recall without necessarily preventing intra-operative consciousness or intra-operative pain [8, 13]. As the emphasis in all studies of ABM-monitored anaesthesia has focused on postoperative recall [4-6], the results tells us nothing about the effects of ABM monitoring on the primary and immediate endpoint of anaesthetic action – intra-operative (un)consciousness.

Those few studies of ABMs that have used the IFT to investigate intra-operative consciousness during general anaesthesia in the presence of neuromuscular blockade reveal that, at the individual patient level, ABMs are unable to detect with any reliability when patients attain a level of consciousness at which they can respond to commands [10, 14, 15]. Despite sparse evidence that ABMs can help prevent such intra-operative consciousness, not only has ABM monitoring recently been endorsed in national guidelines [16], but it has been promoted to guide the administration of general anaesthesia drugs and, by so doing, reduce drug consumption and costs [16, 17]. Fully automated closed loop anaesthetic systems based on ABM output are being studied [18].

Of the available ABMs, the most widely studied is the BIS monitor and the current study sought to use a manually adjusted target-controlled infusion (TCI) of propofol for anaesthesia, titrated to a BIS index range of 55–60, to: (a) observe the incidence of intra-operative responsiveness to command as assessed by IFT; (b) evaluate the utility of the BIS index to predict/identify patient responses to commands; and (c) compare the BIS index observed during intra-operative responsiveness with that at eye opening to command at the end of surgery. The hypothesis was that, if BIS detected intra-operative wakefulness appropriately, its values would correspond to responses (if any) obtained using IFT.


Following ethical approval from the Local Research Ethics Committee, women about to undergo major gynaecological surgery were approached, and their written consent was obtained to undergo surgery using BIS monitoring (BIS A 2000, Revision 3.01: Covidien Medical) to guide the administration of propofol, in association with clinical signs and the IFT. Women aged > 60 or < 18 years of age, those with hearing difficulties, or those of ASA physical status > 2 were not studied.

During the pre-operative interview, it was ascertained what name the patient usually used, and whether she was right- or left-handed. While the patient was informed that I would speak to her during surgery and that she would be able to move her hand to indicate that she was awake, there was no mention of specific commands on the minidisc player or the content of the recorded message (see below).

A Sony Walkman minidisc recorder (Sony MZ R900, Tokyo, Japan) was used to create and play continuously to the patient, through padded headphones (MDR-V700; Sony, Tokyo, Japan), a 1-min sequence. The sequence consisted of three tracks. Track 1 was a command (15 s), track 2 was radio static (15 s) and track 3 was information for the patient to remember (15 s). The minidisc player, in its programme repeat mode, played these three tracks in the order 1, 2, 3, 2, until it was switched off. The command was specific for each patient, “Name, name, this is Dr Russell speaking. If you can hear me, open and close the fingers of your right hand, open and close the fingers of your right hand.” If the patient's dominant hand was the left hand, then ‘left hand’ was substituted for ‘right hand’ in the command. The information the patient was asked to remember was also patient specific, “Name, name, this is Dr Russell speaking. Here are some special words I want you to remember: green pear, sharp lemon, sour gooseberry.” The minidisc player was switched on (‘tape on’) at the time of the skin incision, and switched off (‘tape off’) at the start of skin closure before the TCI infusion pumps were switched off. From the time when the TCI pumps were turned off until a response was obtained, the patient was asked at one-minute intervals “name, name open your eyes”. The BIS at eye opening was noted.

Velband (Smith and Nephew, London, UK) padding was wrapped around the dominant forearm and a tourniquet was applied over this. Small ECG electrodes were placed over the ulnar and median nerves at the elbow and these were connected to a nerve stimulator (Innervator; Fisher & Pykel Healthcare Ltd, Maidenhead, Berkshire, UK) set at 60 mA delivered current. The arm was placed on an armboard close to 90° from the table, so that the forearm and hand could be observed at all times. The hand was restrained with a strap around the palm and the armboard. After loss of consciousness, the tourniquet was inflated to 200 mmHg, and the hand response to nerve stimulation was observed. Atracurium (0.4−1) was then administered. Between 20 and 30 min after intubation, the tourniquet was deflated. Neuromuscular integrity of the forearm was assessed at regular intervals with both train of four and a short tetanic stimulus to ensure adequate muscle power. If further neuromuscular blockade was required during surgery, the cuff was re-inflated and an intravenous bolus of 0.2–0.3−1 atracurium was administered. Again, after 20–30 min, the cuff was deflated. If a hand response was noticed during surgery, consciousness was verified by speaking directly to the patient and asking her, “Name name, squeeze my fingers once.” If there was a response, then the command to squeeze my fingers twice was given (or vice versa). Before speaking directly to the patient, the minidisc player was stopped (see below) and one earpiece was eased from the ear.

Routine monitoring was applied: ECG; non-invasive blood pressure (cuff around the dominant upper arm); and pulse oximetry (probe on a finger of the non-dominant hand). Following placement of a 16-G intravenous cannula in the non-dominant forearm, if a patient had given consent, a low thoracic epidural was inserted, before induction of anaesthesia, at what was estimated to be the T10 level. After a negative aspiration test, 3 ml levobupivacaine 0.5% was administered as a test dose, followed by a further 7 ml. The epidural was topped up with further 10-ml increments of levobupivacaine 0.5% at 90–120 min intervals throughout surgery.

Before anaesthesia was induced, BIS electrodes were applied to the patient's forehead and BIS monitoring commenced. The BIS monitor was connected via its RS232 interface to a laptop computer and the BIS data were saved to the laptop for later analysis. The BIS monitor was set to a smoothing rate of 15 s.

Anaesthesia was induced and maintained using an effect site TCI of propofol [19] with an initial effect site target concentration set at 4 μ−1. An effect site infusion of remifentanil [20] (initial effect site target 2−1) was also used. Both drugs were infused with Alaris PK TCI infusion pumps (Cardinal Health, 1180, Rolle, Switzerland). If consciousness was not lost at these target concentrations (i.e. there was a response to the command, “Name, open your eyes”), then the target concentrations of both drugs were increased. Following intubation, the TCI pumps were manually adjusted to maintain the BIS in the range of 55–60. This range was chosen for two reasons: the manufacturer's guidance states that a range of 45–60 is appropriate for balanced anaesthesia with opioids and that within this range, the patient will be ‘unresponsive to verbal stimuli’ [16, 21]; and it is desirable to minimise propofol use [16]. If the BIS did rise > 60 with no IFT response and clinical signs remained stable, then anaesthesia was not deepened. Regardless of the BIS value, if a patient responded to command (see above), the TCI propofol target was increased until responding stopped. If responding to command stopped spontaneously, the TCI propofol target was not increased.

At the end of surgery, following eye opening to command, the tracheal tube was removed and the patient transferred to the recovery room. Before the patients were returned to the ward, they were interviewed by the author using a structured format to investigate both implicit and explicit memory using the information given via the headphones [9]. Because of split hospital sites, it was not possible to perform follow-up interviews on subsequent days.

For data analysis, the definition of consciousness was a verified response to command. The definition of a BIS index predicting/identifying consciousness in association with a patient response to command was as a BIS index > 60 continuously for at least 60 s within the time period extending from 2 min before to 2 min after the patient response. If the BIS index rose > 60 continuously in association with a hand movement response, and the BIS index was still > 60 at the time of a subsequent movement response, then it was assumed that BIS had also identified consciousness associated with this subsequent response. In the absence of any patient response to command, a BIS index > 60 continuously for at least 60 s was taken to indicate the mistaken prediction/identification of consciousness. A single BIS value < 60 in the 1-min time intervals was ignored.

Parametric and non-parametric statistical tests were used as appropriate with the statistical program SPSS (v 19) (IBM Corporation, Armonk, New York, USA).


A total of 22 women were recruited (Table 1). Overall, 1573 commands were played to the 22 patients. Excluding the eye opening response at the end of surgery, 16 (73%) women responded to commands during surgery. In total, these 16 women responded to command 80 times and of these, the BIS monitor detected 47 (Table 2). Patients responded to command over a wide range of BIS values with the result that the positive predictive value (PPV) of a BIS response was only 18% (Table 1, Fig. 1). Fifty-two per cent of the responses occurred in association with a BIS < 60 (Fig. 1).

Table 1. Patients' characteristics, anaesthetic information and duration of anaesthesia and taped commands. Responders are those patients who exhibited at least one hand movement to command with IFT. Values are mean (SD), median (IQR [range]) or number
 All patients (n = 22)Responders (n = 16)*Non-responders (n = 6)*
  1. *No statistically signicant differences for any of the variables.

Age; years45.5 (8.6)44.6 (8.7)47.7 (8.8)
Height; cm160.1 (6.7)160.2 (7.0)159.9 (6.4)
Weight; kg76.5 (23.4)79.2 (25.5)69.3 (16.4)
BMI; kg.m−229.9 (9.3)30.8 (9.5)27.6 (9.0)
Duration of anaesthesia; min80 (64–121 [42–132])81 (65–121 [42–131)76 (53–115 [42–132])
Duration of tape; min66 (47–99 [30–120])66 (50–96 [30–119])63 (38–102 [30–120])
Premedication; n101
Epidural; n18135
No epidural; n431
Table 2. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of a BIS response (i.e. BIS > 60 for 60 s)
 IFT responseNo IFT Response 
BIS response47220PPV 18%
No BIS response331273NPV 98%
 Sensitivity 59%Specificity 85% 
Figure 1.

Number of responses within a particular BIS index range. Because the individual instantaneous BIS values (shaded columns) may change significantly from reading to reading, the BIS values averaged over the 30 s after the IFT response (clear columns) are also shown.

There was a statistically significant difference between the BIS associated with eye opening to command after surgery and that associated with a response to intra-operative commands, but there was no such difference in propofol concentrations. At eye opening, median (IQR [range]) BIS was 75 (70–78 [51–93]), while the BIS at intra-operative response to command was 61 (52–67 [37–80]) (p < 0.001). At eye opening, median (IQR [range]) effect site concentration of propofol was 2.1 (1.7–2.8 [1.5–3.9]) μ−1and at intra-operative response to command, it was 2.0 (1.5–2.3 [1.2–4.0]) μ−1.

Figure 2, a section of a BIS/electromyography (EMG) record, demonstrates that, for an individual patient, the BIS associated with multiple responses to command may vary between responses: the BIS indices associated with responses 1–5 in Fig. 2 were 51, 49, 60, 46 and 50, respectively: at eye opening, BIS was 68. As can be seen in Fig. 2, the highest BIS values in this patient were associated with significant EMG activity (Pearson correlation coefficient = 0.5, p < 0.0005).

Figure 2.

A section of BIS (black line) and EMG (green line) output along with effect site propofol concentration (blue dotted line; 10 times the real μ−1 value) for a patient with five IFT commands and/or responses (downward pointing arrows 1–5). The effect site propofol concentrations are a combination of written data collected during the anaesthetic with missing values calculated using TIVATrainer© ( Upward pointing arrows depict specific events: A = atracurium 20 mg; Ep = epidural top-up of 10 ml levobupivacaine 0.5%; TCI ↑ = target-controlled infusion of propofol increased; TCI ↓ = target-controlled infusion of propofol decreased. Response 1 was verified when speaking directly to the patient, but then the patient stopped responding so the propofol target concentration was not increased. Response 2 was not verified when speaking directly to the patient so the propofol target concentration was not increased. Response 3 was verified and the propofol target concentration was increased. Response 4 had not been verified before response 5 commenced. Response 5 was verified and the propofol target concentration was increased.

Figure 3 is a section of the BIS/EMG record from a different patient. This shows two large rises in the BIS index, again associated with EMG activity (Pearson correlation coefficient = 0.8, p < 0.0005). There was no response to command associated with the second rise in BIS value (at 29 min), so a bolus of atracurium was given. The BIS then declined to < 40 despite the propofol concentration declining during this time.

Figure 3.

A section of BIS (black line) and EMG (green line) output with effect site propofol concentration (blue dotted line; 10 times the real μ−1 value) for a patient with two IFT responses (downward pointing arrows 1 and 2). The effect site propofol concentrations are a combination of written data collected during the anaesthetic with missing values calculated using TIVATrainer© ( Upward pointing arrows depict specific events: Ind = induction; Int = intubation; TCI ↓ = target-controlled propofol concentration decreased, TCI ↑ = target-controlled concentration of propofol increased; A = atracurium 20 mg. Response 1 was verified and the propofol target was increased. Response 2 was unverified, as responding then ceased, probably due to increasing propofol concentration. There was no response to command associated with the second rise in BIS value (at 29 min), but this rise was preceded by tightening of the abdominal muscles, creating access problems for the surgeons. A bolus of atracurium was given and the BIS declined to < 40.

The BIS was > 60 for 60 s or more for a median (IQR [range]) of 17 (8–32 [0–52])% of the time between tape-on and tape-off. There was no difference in this proportion between responders (i.e. those patients who responded with movement to command) and non-responders, 17 (7–24 [0–50])% and 23 (11–41 [7–52])%, respectively.

The median (IQR [range]) of time from when the TCI pumps were switched off until the patients opened their eyes to command was 2.7 (2.1–4.1 [1.1–6.0]) min, with no difference between responders and non-responders, 2.4 (2.0–5.2 [1.2–6.0]) min and 3.5 (2.8–3.9 [2.3–4.5]) min, respectively.

On direct questioning, two of the 16 patients who responded during surgery had a vague memory about ‘squeezing fingers’ and one patient in the non-responding group remembered extubation. Apart from the above, there was no other evidence of either explicit or implicit memory in any patient and, in particular, no patient had any recall of surgery. Three other women (one responder, two non-responders) recalled dreaming, but neither of the non-responding women could remember her dream content, apart from it being a good dream. The dream of the woman who responded was also described as ‘good’ and involved pop music.

At no time was the neuromuscular integrity of the isolated hand affected. The train of four was always 4 and a tetanic stimulus was well maintained.


This is the first investigation that directly investigates intra-operative responsiveness with the IFT, for the duration of major surgery, during BIS-guided anaesthesia in the presence of neuromuscular blocking drugs. With 16/22 women in the current study responsive to commands (i.e. a state normally regarded as being conscious) during surgery, the results do not support the notion that manual adjustment of TCI propofol in an attempt to keep the BIS in the range 55–60 is an appropriate anaesthetic technique. The overall incidence of consciousness with recall in this patient group was high (~10%). These are surprising results because the levels of BIS to which propofol was titrated conform to recommended practice [17, 21] approved by national regulatory bodies [16]. Nonetheless, the incidence of consciousness could have been higher had the IFT not been in use and, without the IFT to identify responsive patients so allowing anaesthesia to be deepened, patients might plausibly have had more detailed and/or traumatic recall.

It is difficult to keep the BIS in such a narrow range when manually titrating the infusion, and for 26% of the time the ‘instantaneous’ BIS value was > 60 (for half of this time, it was between 60 and 65). The other consideration is the accuracy of the BIS in identifying consciousness under these conditions. The sensitivity of the BIS response as defined in the methods (i.e. > 60 for 60 s) was only 59%. If the more ‘traditional’ real-time (instantaneous) BIS is used, then the sensitivity is even worse, at 50%. Thus, as I found in this study, there is a real possibility that titrating propofol to this level of BIS could result in a high proportion of patients who are conscious during surgery. Even with the IFT serving as a back-up, two women (~10%) had recall of the commands, but there was no recall of the surgery. It is unknown what the level of recall might have been without the IFT.

Despite the widespread use of ABMs and encouragement to reduce anaesthetic drug use by using ABM-guided anaesthesia [17], there have been few studies directly investigating ABM output and consciousness in the presence of neuromuscular blocking drugs. In the only other investigation of consciousness during surgery using ABM-guided anaesthesia in conjunction with the IFT and neuromuscular blockers [10], all 12 patients were conscious at some stage during surgery. Two other studies have investigated ABMs in conjunction with the IFT in paralysed patients before surgery started [14, 15]. In both these studies, the Narcotrend, the BIS index and Patient State Index were unable to detect consciousness reliably. The data from the current study, together with those of these three studies [10, 14, 15], suggest that in the presence of neuromuscular blocking drugs, ABMs are not able to identify the return of consciousness with any reliability.

A major problem in this type of study relates to the definition of a BIS response indicative of consciousness. BIS is a dimensionless number, representing a continuum from fully conscious to an iso-electric EEG, and the BIS index can only be associated with a probability of a patient being conscious. Despite this, a BIS index maintained between 45 and 60 is widely promoted as an ‘acceptable range’ [16] in which patients will be ‘unresponsive to verbal stimuli’ [21]. (i.e. unconscious]. There is, however, no agreement as to what might constitute a ‘BIS response’ to a particular event. Possibilities include: a single BIS value at a specified time interval from the stimulus [15]; the highest single BIS value within a specified time interval after the stimulus [22]; a 15% rise in the BIS index within 2 min of the stimulus [23]; the average BIS over a period of 15 s after the event [24]; the average BIS over a period of 1 min after the event [23]; the average BIS over a period of 2 min after the event [25]; or the BIS area under the curve over a period of 2 min after the event [26]. Some investigators do not specify what BIS values were used in their analysis [27, 28]. In the current study, a BIS index continuously > 60 for 1 min was chosen as a ‘BIS response’ indicative of consciousness, although in practice a single BIS value < 60 was ignored if all the other values during the minute were > 60. Re-analysing the results of this current study using an instantaneous BIS value at the time of the response, the BIS value averaged over 30 s after the response, a 15% rise in BIS over 2 min, or a BIS > 60 for 60 s, made no difference to the overall results. Whatever the definition, frequently, the BIS did not indicate consciousness when patients responded to commands during surgery, and even more frequently, the BIS suggested patients were conscious in the absence of any response to commands (Table 2, Fig. 1).

A problem for BIS in the presence of neuromuscular blocking drugs is the varying degrees of paralysis encountered during anaesthesia. The EEG and EMG frequencies overlap in the 35–47 Hz range and this EMG activity is interpreted as high-frequency, low-amplitude waves, falsely elevating the BIS [29, 30]. Figures 2 and 3 demonstrate how such ‘contamination’ from the EMG signal can result in high BIS values and when the EMG signal disappears, either spontaneously or after atracurium, the high BIS values fall to low values. The increased EMG activity may be because the neuromuscular blocking drug is wearing off, or the patient is waking up. Without using the IFT, it would not be clear how to respond to such high BIS values – by deepening anaesthesia or by administering further neuromuscular blockade. It is possible that these EMG-driven BIS responses could be prevented by maintaining near complete muscle paralysis, but whether this would result in a more or less reliable BIS indication of intra-operative consciousness is not known. If anaesthetising a patient with a previous awareness experience, or a patient at high risk of awareness, this EMG effect on the BIS index creates a dilemma for the anaesthetist. Current advice is, to an extent, contradictory. On the one hand, it is suggested that the use of an ABM monitor may be justified on a case-by-case basis [1, 31, 32]; however, it is also suggested that if neuromuscular blocking agents are to be used, they should be used at doses that do not cause complete paralysis [31, 33]. If both suggestions are followed, then the minimal use of neuromuscular blockers could create conditions where EMG activity reduces the utility of the BIS index.

An important observation was the fact that, in any single patient who responded to command on multiple occasions, the BIS associated with these responses varied. Not only were the absolute values different, but the BIS trend may be stable, going up or going down at the time. Another disconcerting observation was the fact that the BIS was significantly higher at eye opening than it was in association with movement responses during surgery. It is unclear why movement to command after surgery (eye opening) occurred at a higher BIS value than movement to command (hand movement) during surgery. This may be because neuromuscular blockers may limit the EMG contamination of the BIS signal during surgery such that BIS is artificially lower during surgery than at the end of surgery when the effects of these drugs have been fully or near-fully reversed, or because the end of surgery is a dynamic phase where propofol concentrations are in rapid decline such that there is a correspondingly rapid rise in BIS index, further accentuated by the additional EMG activity. The fact that there was no difference between the effect site propofol concentrations associated with intra-operative responses and eye opening at the end of surgery would support (1) above.

Several studies claim that using the BIS monitor can reduce the incidence of awareness during general anaesthesia [4-6]. However, in these studies, the definition of ‘awareness’ is postoperative recall of intra-operative events. None of the studies provides any data to reassure us that the patients were unconscious during surgery. Some believe that lack of recall rather than unconsciousness during general anaesthesia is what matters [34]; however, the ethical aspects of operating on patients who are conscious during ‘general anaesthesia’, even if there is no subsequent recall, is debatable: these patients may be in pain at the time [8]. Not only do patients expect to be unconscious during general anaesthesia, but a survey of anaesthetists indicates that they too wish to be unconscious during general anaesthesia [35].

The current study has its limitations. Defining consciousness is difficult. However, with regard to general anaesthesia, it is accepted that when patients are not paralysed, then a patient's sensible intra-operative motor response to a command (e.g. “Name name, open your eyes”, or “name, name squeeze my fingers”) would indicate that the patient is conscious and therefore anaesthesia is inadequate [13, 36]. The IFT is simply a means to allow a paralysed patient to respond to command [13] and thus, if a patient responds to command in the presence of neuromuscular blocking drugs, then that patient should logically be regarded as conscious [13, 21]. There is some debate as to why, if patients are able to move, they do so only in response to command and not in response to the surgery. It seems that the IFT may initially identify a very early stage of ‘connected consciousness’ where patients perform ‘goal-directed movements’ to command, but do not respond to surgery [37]. As anaesthesia lightens, ‘connectedness’ increases, and with it the chances of recall. In this study, the lack of full ‘connectedness’ is exemplified by the fact that most patients had no recall and the two patient with vague memories of squeezing fingers had no memory of the surgery.

It could be argued that the BIS index target range of 55–60 is too near the ‘margins of consciousness’ [38]. Yet, current BIS guidance indicates that 45–60 is acceptable [16, 17, 21] and there is therefore no justification for such criticism. Furthermore, there are recommendations in the Cochrane database that one should keep the BIS index between 60 and 80 near the end of surgery [17]. However, this latter recommendation cannot be supported by the present results: 19 patients opened their eyes to command at BIS levels < 80.

While it is possible that additional interviews later in the postoperative period may have uncovered further evidence of recall [39], this seems unlikely. In previous IFT studies, additional interviews did not reveal additional recall, suggesting that, as used, the IFT probably identifies initially a level of consciousness where no encoding of events into long-term memory occurs.

In conclusion, when using the IFT, if appropriate hand movement response to command is an indicator of consciousness during general anaesthesia, then the BIS correlates very poorly with this, exhibiting poor positive predictive values and sensitivity.

Competing interests

The BIS monitor and electrodes were purchased from funds raised by the 6th International Symposium on Memory and Awareness, Hull, 2004. No other funding or competing interests declared.