Currently, there exists no effective monitor that can predict the probability of a patient being conscious during general anesthesia. The electroencephalogram-derived bispectral index (BIS) is a promising new method to assess anesthetic adequacy. This study used the BIS to predict the probability of recovery of consciousness after a single bolus induction dose of propofol or thiopental.
Twenty unpremedicated surgical patients were anesthetized with 4 mg/kg thiopental and 20 patients with 2 mg/kg propofol. The BIS was monitored throughout the study. After induction, before administration of neuromuscular blocking agent, a tourniquet was applied to one arm and inflated above the systolic blood pressure. This allowed preservation of the ability to move the hand after neuromuscular blocking agent onset. Patients were then prompted to squeeze the investigator's hand every 30 s, until they responded to the request. At the time of response, anesthesia was reinduced and the study terminated.
The BIS at loss of consciousness and recovery of a response was not statistically different between propofol and thiopental. No patient with a BIS less than 58 was conscious. In both groups, a BIS of less than 65 signified a less than 5% probability of return of consciousness within 50 s.
The BIS can be used to predict probability of recovery of consciousness after a single injection of either thiopental or propofol.
Anesthesiologists currently lack a reliable monitor to assess the adequacy of anesthesia in the paralyzed patient. Commonly used hemodynamic measurements or clinical signs such as diaphoresis, lacrimation, and mydriasis are not useful to detect awareness. [1–4]The results of computerized electroencephalogram (EEG) analysis have not yet provided an EEG-derived value that is a consistently reliable measure of the conscious state. This has been due to differing effects of anesthetics on the EEG and large interpatient variability. [5,6]
However, some methods of EEG analysis have had limited predictive value for various clinical measurements other than the conscious state. In one study, the spectral edge successfully predicted the degree of hemodynamic response to intubation, but, in another study, the spectral edge and other EEG derivatives were found to be poor predictors of movement to skin incision, response to verbal command, and development of memory. Ideally, an EEG-derived method of predicting inadequate anesthesia should predict the conscious state in all patients, with any anesthetic agent, under all clinical circumstances. This has not yet been achieved.
This study used the recently developed bispectral index (BIS) as a measure of EEG activity. Computed from EEG data, it is a measure of the hypnotic effect of anesthesia, and its derivation has been described elsewhere. In brief, the bispectral analysis decomposes the EEG signal into its component sine waves using a Fourier transformation. A set of bispectral features are calculated by analyzing the phase relations between the component waves. These bispectral features are combined with other EEG features into a single measurement, BIS, a numerical index ranging between 0 and 100. The BIS has been found to correlate well with sedation and predict a patient's response to stimulus. [10–12]
We examined the ability of the BIS to predict the return of consciousness as measured by the isolated forearm technique (IFT). The technique, first described by Tunstall in 1977, isolates the patient's forearm from the systemic circulation by the use of a tourniquet. After induction of anesthesia, a neuromuscular blocking agent is administered, but the patient still retains the ability to move the unparalyzed arm in response to a command or painful stimulus.
The isolated forearm technique has been used successfully to detect consciousness in anesthetized, paralyzed patients. [14–16]However, before the actual event occurring, the technique does not enable the prediction of return of consciousness and prevention of inadequate anesthesia. The squeezing action can be an equivocal response and may be difficult to interpret. 
We investigated the time interval from administration of induction agent and loss of consciousness to recovery of consciousness after a single bolus induction dose of thiopental or propofol. Throughout this time, the BIS and hemodynamic values were recorded. We then compared the ability of the BIS versus hemodynamic variables to predict recovery of consciousness.
Forty consenting male and female patients were enrolled. They were 18–70 yr old, American Society of Anesthesiologists (ASA) physical status 1–3, less than 200% of ideal body weight, and scheduled for elective surgery during general anesthesia. Exclusion criteria were as follows: renal, hepatic, or neurologic dysfunction; neurosurgery, cardiothoracic, vascular, and upper extremity surgery; use of benzodiazepines, anticonvulsants, alcohol, opioids, or other psychotropic drugs (chronically or within 24 h before the induction of anesthesia); and any patient in whom a rapid sequence induction was indicated.
After skin preparation, 7 Grass E5 gold-cup electrodes were placed according to the international 10/20 system in a frontoparietal montage and referred to Cz. The electrode impedance was kept less than 2 Kohms, and the EEG was recorded and analyzed using an A1000 EEG monitor (Aspect Medical Systems, Natick, MA). The EEG, end-tidal CO2(Capnomac, Datex, Helsinki, Finland), noninvasive blood pressure (Dinamap Critikon, Tampa, FL), and heart rate were recorded via a serial interface and a laptop computer (Toshiba T5200/100). The EEG data were analyzed continuously during successive 2-s data segments, each overlapping by 75%, and a BIS value was generated every 15 s. All data were stored on tape for further off-line analysis.
The first 20 subjects received thiopental, and the second group of 20 received propofol. Intravenous access was in the nondominant hand. No premedication was given. Before induction, the patients practiced squeezing the investigator's fingers twice to command. After baseline EEG and hemodynamic data collection, an intravenous injection of 4 mg/kg thiopental or 2 mg/kg propofol was administered by the investigator for 20 s. After loss of consciousness (failure to squeeze fingers in response to command), a tourniquet was inflated over the arm of the dominant hand to a pressure of 150 mmHg higher than systolic blood pressure. Vecuronium (0.1 mg/kg) was administered, and mask-assisted ventilation with 100% oxygen was used to achieve normocapnia. The patient was then verbally prompted by the investigator with the words “Squeeze my fingers” every 30 s. After responding to the command, he/she was then immediately prompted with “Squeeze my fingers twice.” Only after a positive response of two squeezes was a patient considered to have recovered consciousness. At that time, the study was concluded and anesthesia reinduced.
All patients were interviewed for recall on the first postoperative day using a standardized set of postoperative questions. This included inquiring for the recall of any unpleasant dreams during intubation or surgery as well as explicit memory recall for operative events.
Data are mean (+/- SD), and P < 0.05 was considered statistically significant. The gender and age of the patients in the two groups were compared using chi-square analysis and Student's t test, respectively. The BIS value at induction, loss of consciousness, minimum BIS (represented as the lowest BIS value after induction for each patient), and recovery of consciousness were compared within each induction agent group using analysis of variance. The Student-Newman-Keuls procedure was used for multiple comparisons. Event times and BIS at each event also were compared between groups using analysis of variance.
The probability of a patient being conscious at any time after achieving a certain BIS value was calculated for both the propofol and thiopental group. Each patient's BIS trend exhibited a rapid decrease from induction to the minimum BIS and then began a gradual increase, which culminated in recovery of consciousness. To calculate the probability of awareness after any BIS value, the time that each patient reaches the BIS for which a probability is being calculated is considered ty= 0. A subject was classified as conscious at any BIS value equal to or greater than that at which he responded. The probability of consciousness at any time tyafter achieving a BIS value of BISywas calculated by dividing the number of conscious patients by the total number of patients (conscious or not). Equation 1
Probabilities of response to command were calculated for combinations of BIS values between 5 and 95 (increasing in increments of 5) and time intervals from 0 to 650 s after achieving that given BIS value (increasing in 5-s increments). 
For thiopental, the heart rate at return of consciousness was significantly greater than that at 1 min before induction and at loss of consciousness (analysis of variance, Student-Newman-Keuls multiple comparison procedure). There were no other statistical differences in heart rate or blood pressure values in either group (Table 2).
(Figure 1and Figure 2) present the BIS trends plotted against time for each individual patient in both groups; the dots represent time of recovery of consciousness. The mean time of relevant events are presented in Table 3. After induction with propofol, patients lost consciousness 46 +/- 15 s after induction compared with 43 +/- 15 s for the thiopental group (not statistically significantly different).
The maximal induction agent effect (minimum BIS) was achieved 62 +/- 29 s after loss of consciousness with the propofol group and 43 +/- 16 s in the thiopental group. Patients in the propofol group recovered consciousness 529 +/- 176 s after induction and 421 +/- 175 s after minimum BIS value. In the thiopental group, recovery of consciousness occurred earlier, at 330 +/- 153 seconds after induction and 244 +/- 151 s after minimum BIS. These event times were statistically significantly different between groups.
Mean BIS values for each relevant time point are presented in Table 4. The baseline mean BIS before induction was 92 +/- 2 for both groups of patients, with no difference between groups. At loss of consciousness, mean BIS was not different between groups (89 +/- 9 for propofol and 90 +/- 13 for thiopental).
After induction, minimum BIS was significantly lower in the propofol group (27 +/- 9) compared with the thiopental group (34 +/- 10). At the time of return of response to command, the BIS in the propofol group (80 +/- 7) was not different compared with thiopental (81 +/- 5).
To characterize the relation between responsiveness, BIS, and time during recovery, the calculated probabilities of 5%, 50%, and 95% of patients responding to command after they have reached each BIS value or “threshold” are plotted against time. This is shown in Figure 3. For example, in the propofol group, when the BIS reaches 65, 50% of all patients will respond to command within 240 s of achieving a BIS of 65, and 95% within 580 s. From these two figures, it can be seen that, as BIS increases, the probability of a patient becoming aware occurs more rapidly than at lower BIS values. The last subject responded at 1,052 s after propofol. No recall of events occurred in any of the patients studied in either group, despite all subjects responding to verbal command.
To determine whether a patient is aware while undergoing general anesthesia presents a problem. A monitor that measures physiologic changes associated with the conscious state would be an improvement on current methods, which are dependent on responses that only indirectly reflect consciousness. The limitations of current clinical methods to assess anesthetic adequacy have been well documented. [1,4]
Hemodynamic variables are inadequate. [4,20]Also, movement responses to skin incision (often used for research purposes to determine minimum alveolar concentration or minimum inhibitory concentration) are not directly related to consciousness. It has been shown that the minimum alveolar concentration of isoflurane required to prevent recall of information is far less than that required to prevent movement to skin incision. It appears that movement responses are mediated by anesthetic action on both the spinal cord and higher centers. Antognini and Schwartz demonstrated an increase in isoflurane minimum alveolar concentration required to prevent movement to skin incision in goats when circulatory bypass allowed isolated brain anesthesia (minimum alveolar concentration increased from 1.2% to 2.9%). Therefore, mechanisms in the higher central nervous system that control memory function and consciousness may be anesthetized adequately, whereas spinal cord mechanisms that suppress movement to surgical stimulus may not. A more direct method of evaluating consciousness is needed, rather than the current practice of observing hemodynamic changes or movement responses.
The BIS provides an EEG-derived measurement of anesthetic effect on the brain. Studies with volunteers showed that the BIS is effective for monitoring sedation and memory function and predicting consciousness with a variety of anesthetic agents. [23,24]It analyzes data during a short enough time interval to give a real-time change in anesthetic agent effect, allowing appropriate drug titration. By contrast, the isolated forearm technique detects return of consciousness as an “all or none response” once anesthesia becomes inadequate. This use of a technique to define recovery of response as response to verbal stimuli combined with measurement of the BIS enabled study of the association between return of consciousness and a quantitative, real-time measurement of anesthetic adequacy. It should be noted that return of consciousness was defined in the absence of any noxious or surgical stimulus.
In this study, the IFT was modified to give an unequivocal endpoint by requesting a double squeeze after the single squeeze. A disadvantage of this method is that the command to squeeze was given every 30 s, and the delay until the patient responded with a single, then a double squeeze could mean that return of consciousness may have occurred up to a minute earlier than defined by the chosen end-point. This study design feature may also explain, in part, the observation that some individuals had BIS levels greater than 60 for some time before responding.
The findings in this study are consistent with data from other workers. In volunteers who received midazolam, isoflurane, or propofol, Kearse et al. reported that BIS values between 60 and 80 were associated with an increased probability of consciousness but no recall. The presence of consciousness with an absence of recall after the event is demonstrated, necessitating a clear distinction between consciousness with and without recall. Patients may be fully aware during surgery but may have no recall of any intraoperative event. Russel used the isolated forearm technique to detect awareness in 13 of 55 patients undergoing gynecologic procedures, but only one patient had any clear factual recall of an intraoperative event. Patients may have no recall of any intraoperative event but still have an implicit memory and suffer problems such as sleep disturbances. Another finding supported by this current study is that hemodynamic signs do not correspond with a state of awareness. Patients may be fully aware but still not show any clinical or hemodynamic signs to suggest inadequate anesthesia. However, it should be emphasized that the patients in this study were not undergoing surgery, and therefore recovered consciousness in the absence of noxious surgical stimulation. In the thiopental group, the statistical difference between heart rate 1 min before induction and at loss of consciousness when compared with recovery of consciousness is of little clinical significance, because all mean values would be acceptable to the clinician.
In this study, the wide range of recorded times until recovery of consciousness illustrates the potential for awareness during intubation and skin incision. One subject who received thiopental responded to command as early as 114 s after initial loss of consciousness, demonstrating that some patients can recover consciousness very rapidly and possibly before intubation. Also, if induction agent effect-site concentration decreases rapidly and maintenance agent concentrations are still inadequate, a patient may be conscious at skin incision. The data support the possibility that current conventional clinical monitoring may result in an undetected return of consciousness in a paralyzed patient, even if appropriate drug doses are given, because of the large variation in duration of drug action.
According to our data, patients recovering in the propofol bolus group have less than a 5% chance of being aware within 50 s if the BIS is 65. There is an even smaller probability of awareness in the thiopental-induced group. These probability values may vary in relation to the size of the bolus dose. If the actual BIS value at which patients recovered consciousness is examined, it could be inferred from this study that a patient maintained at a BIS of less than 55 would appear to have only an extremely small probability of awareness (only 2 patients recovered consciousness at a BIS below 60, and no patient was conscious below a BIS of 58;Figure 2).
With this study, we showed that the probability of recovery of consciousness after a single injection of propofol or thiopental could be predicted using the BIS. There was no absolute single BIS value at which consciousness returned, but the lower the BIS value, the greater the probability of a patient being unconscious. Whether the measured end-point has a similar mean BIS value for other agents is unknown. Nor can we be certain that similar probability values for return of consciousness also occur for a surgical stimulus as opposed to the less noxious verbal stimulus used in this study.
The authors thank Dr. H. Bennett for suggesting the double squeeze/isolated forearm paradigm.