1 Department of Psychology, University of Sheffield, Western Bank, Sheffield, S10 2TP, UK. 2 Northern General Hospital, Sheffield, UK
* Corresponding author: E-mail c.deeprose{at}shef.ac.uk
Accepted for publication August 20, 2004.
![]() |
Abstract |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Methods. Words were played through headphones during surgery to 62 patients receiving propofol and nitrous oxide anaesthesia. Thirty-two patients received fentanyl 1.5 µg kg1 at induction and 30 received no fentanyl. Neuromuscular blocking drugs were not used. Depth of anaesthesia was measured using the bispectral index (BIS). Anaesthetic variables were recorded at 1 min intervals during word presentation. On recovery, implicit and explicit memory were assessed using an auditory word-stem completion test and a yesno word-recognition test, respectively.
Results. BIS, blood pressure, end-tidal carbon dioxide and heart rate during word presentation did not differ between the study groups. The infusion rate of propofol and the patients' ventilatory frequency were significantly higher in the group not receiving fentanyl. No patient had unprompted explicit recall of surgery, although there was above-zero performance in six patients on the yesno recognition task (P<0.05). There was no physiological evidence of awareness during anaesthesia (median mean-BIS=38 in the no-fentanyl group and 42 in the fentanyl group). There was evidence for priming (mean priming score=0.09, P<0.05 in the no-fentanyl study group; mean priming score=0.07, P<0.05 in the fentanyl group) even when patients with momentary light anaesthesia (maximum recorded BIS60) and/or positive recognition scores were excluded from the analysis.
Conclusions. Existing knowledge can be primed by information presented during propofol and nitrous oxide anaesthesia. This priming is evidence of unconscious information processing and not the result of moments of awareness.
Keywords: anaesthesia, depth ; analgesics opioid, fentanyl ; memory, priming
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Lubke and colleagues4 tested trauma patients anaesthetized with isoflurane and found that, although learning is more likely at lighter anaesthetic depths when awareness is more likely, it can also occur at deeper and clinically adequate depths of anaesthesia, defined as bispectral index (BIS) between 40 and 60. We recently demonstrated priming during day surgery with propofol anaesthesia, but not during equally deep anaesthesia prior to surgery.5 The median BIS during word presentation in the during-surgery group was 41.9, supporting the conclusion of Lubke and colleagues4 that priming occurs during deep anaesthesia. However, these findings do not necessarily mean that priming was unconscious. A moment of awareness immediately after word presentation might facilitate priming by that word, even if the word itself was presented during adequate anaesthesia. This problem is particularly pertinent for the study by Lubke and colleagues because their patients were undergoing trauma surgery during which relatively large fluctuations in anaesthetic depth were unavoidable. It is exacerbated by a lag of up to 1 min in BIS recordings.6 We tackle this problem in the present study by assessing priming as a function of the maximum BIS recorded at any point during word presentation. If patients show priming even though BIS never exceeds 60, this would be stronger evidence than we have at present for unconscious memory activation during anaesthesia.
We also tackle another problem of research into learning during anaesthesia. The common failure to replicate significant results, even by the same research groups using the same stimuli, has been noted and casts doubt on the reliability of evidence for priming during anaesthesia.178 It is especially important to determine the reliability of priming occurring in the surgical period because this is when it could potentially do most harm. Thus in the present study we sought to replicate our previous demonstration of learning during surgery,5 using the same anaesthetic regimen, the same patient population and the same experimental stimuli and memory tests.
We extended our previous study by testing the effect on intraoperative priming of fentanyl, commonly used to supplement the hypnotic agent at induction of anaesthesia. We know of no evidence that fentanyl impairs implicit memory formation and it has almost no effect on explicit recall in sedated volunteers.9 We have previously proposed that surgical stress facilitates priming during anaesthesia.5 It is conceivable that fentanyl may reduce priming in patients who are being surgically stimulated because it suppresses the stress response.10 In the present study we tested the effect on intraoperative priming of a dose of fentanyl at induction. All patients were played words during surgery, but only half received fentanyl at induction of anaesthesia. We expected to replicate priming in those patients receiving fentanyl at induction as their conditions were similar to those of the during-surgery group in the previous study, and predicted more priming in the patients without fentanyl because they would have an unsuppressed stress response to surgery.
![]() |
Methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Construction of experimental stimuli and memory tests
The stimuli and memory tests were identical to those used in our previous study.5 We used word-stem completion to test implicit memory and yesno recognition to test explicit memory. Most memory tests can be performed using a combination of implicit and explicit memory. However, our piloting of these tests with an undergraduate sample, as reported in detail in the previous study, indicated that these tests offered reasonably pure and sensitive measures of the type of memory they were designed to test.
Twenty-eight five-letter relatively common English words were recorded onto a Macintosh Powerbook (1400cs/133, Apple Computer Inc., California) at a sample rate of 44.1 Hz and 16-bit sample size using a microphone and SoundEdit software (16 Version 2, Macromedia, USA). In a copy of these word files, SoundEdit was used to remove the tail of each word, leaving word stems typically three phonemes long. Each word stem was unique in that, although it was possible to complete it by words not used in this study, it could be completed by only one of the 28 words used here as targets and distractors. The mean spontaneous (unstudied) word-stem completion rate was 0.32 in our pilot study with 48 undergraduate subjects.5
The 28 words were assigned to four lists, as shown in the Appendix. Two of these lists were presented to each patient during surgery, with the words from the two lists combined and presented in random order. One of the lists provided targets for the subsequent recognition test. The stems corresponding to the words in the other list were the target items for the word-stem completion test. The two remaining lists provided distractor stimuli, one list per memory test. Each word list appeared equally often as target and distractor stimuli on both the implicit and explicit tests. The words in the study lists and the stimuli on the test lists were presented in a different random order to each patient.
Anaesthetic technique and experimental procedures
The anaesthetic technique differed from that of the previous study only in that one group received no fentanyl at induction and that intraoperative analgesia was given after word presentation if deemed clinically necessary. None of the patients were premedicated. Patients were assigned randomly to a fentanyl or no-fentanyl study group to which the experimenter (CD) was blinded. Patients in the fentanyl group received fentanyl 1.5 µg kg1 followed by a sleep dose of propofol. Patients in the no-fentanyl group received only propofol for induction of anaesthesia. Anaesthesia was maintained in all patients with a target-controlled infusion at a rate between 3 and 9 mg kg1 h1depending on the anaesthetist's clinical judgement. Patients breathed nitrous oxide 66% and oxygen 33% spontaneously through a laryngeal mask.
BIS monitoring commenced in the operating theatre, prior to surgery, using an Aspect-1000 monitor (software version 2.51, Aspect Medical Systems, Framingham, MA, USA) with bifrontal montage (F7, F8, reference Fp2). The anaesthetist was not blinded to BIS, but the anaesthetic was not BIS guided.
For all patients, word presentation began at first surgical incision and ended before the completion of surgery. In each case, 14 words were presented by experimenter CD in random order using a Macintosh PowerBook (1400cs/133, Apple Computer Inc., California) and closed headphones (KOSS TD/80). Each word was repeated consecutively 15 times over a period of 1 min, including a 5 s period of silence at the end of each word series; thus the total presentation time was 14 min. Twenty-eight BIS readings were recorded, one at the beginning and one at the end of each word series. Anaesthetic variables (heart rate, mean blood pressure, end-tidal carbon dioxide concentration, ventilatory frequency and infusion of propofol) were recorded at 1 min intervals during word presentation.
Intraoperative analgesia following word presentation and postoperative analgesia were given as clinically appropriate.
Memory testing
When patients were able to sit up in bed following recovery from anaesthesia and were agreeable to completing the tests, they were asked a series of open-ended questions as used by Russell and Wang11 to probe for any explicit recollection of intraoperative events. Patients were asked: What is the last thing you remember before falling asleep?, What is the first thing you remember about waking up?, Did you dream while you were asleep? and Did you hear any words while you were asleep? Patients then completed the explicit and implicit memory tests in the order determined by counterbalancing. In yesno recognition, patients were presented with seven target words (presented during anaesthesia) and seven distractor words (not presented during anaesthesia) in random order and were asked to state after each whether or not they recalled hearing that word while they were asleep. The word-stem completion test consisted of word stems, in random order, for seven target and seven distractor words. Patients were played each stem and asked to complete it with the first complete word which comes to mind. In each test, the number of distractor hits (yes responses on the recognition test or correct completions on the word-stem completion test) was subtracted from the number of target hits and expressed as a proportion of the total number of items studied in the test (i.e. seven).
Statistical methods
Two BIS scores were analysed for each patient: the mean of the 28 BIS scores recorded during word presentation (mean-BIS), and the highest of those 28 scores, representing the lightest anaesthetic depth reached during word presentation (max-BIS). These scores were compared across groups using the MannWhitney test for independent samples. The other anaesthetic variables were compared across groups using two-tailed t-tests for independent samples. One-sample t-tests were used to test whether the memory scores in the fentanyl and no-fentanyl study groups exceeded zero, and to test whether memory scores exceeded zero in patients with max-BIS<60. These tests were one-tailed because we were attempting to replicate previous evidence for priming during surgery5 and predicted that the priming effect would be at least as large when fentanyl was omitted from the anaesthetic regimen. We report two-tailed confidence intervals for further comparison. Pearson's correlations were used to test the relationship between memory scores and anaesthetic variables. Spearman's rank correlations were used to test the relationship between BIS (mean-BIS and max-BIS) and memory scores on recovery. For these analyses, we report values corrected for ties. Statistical significance was assessed with
=0.05 unless otherwise stated. All analyses were performed using StatView 5.0 (SAS Institute Inc., Cary, NC).
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
Preliminary analyses showed no effect of test order (word-stem completion followed by yesno recognition, or vice versa) on task performance. No patient revealed spontaneous or prompted recall on the structured interview for intraoperative events. Performance on the yesno recognition task (P<0.05) was above chance for the sample as a whole, as shown in Table 3. When split by group, this was clearly non-significant in the fentanyl study group (P=0.18) and just beyond significance in the no-fentanyl study group (P=0.051). Examination of the unprocessed data revealed six patients overall with non-zero scores, five of whom also made two or more false alarms (i.e. yes responses to distractor words).
|
|
|
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
We suggest that this implicit memory is evidence for unconscious priming. The mean BIS scores recorded during word presentation indicated relatively deep anaesthesia (mean mean-BIS of 42 compared with 44,5 44,13 and 4812 in previous studies of intraoperative priming during BIS-monitored propofol anaesthesia). The maximum BIS values for each patient during word presentation ranged from 22 to 88. In the study by Kerssens and colleagues13 data were excluded from 19 of 102 patients because their BIS readings were <40 or >60 during word presentation. Struys and colleagues12 did not report max-BIS values but did report the mean proportion of BIS readings >60 during maintenance of propofol anaesthesia. Even when the anaesthetist aimed to keep depth within the BIS range 4060, 20% of BIS readings were >60. When depth was not BIS-guided 26% were >60. In our study, only 12% of BIS readings were >60 even though BIS was not deliberately used to guide maintenance of a stable anaesthetic depth. Therefore we suggest that the anaesthetic used in our study was deep and relatively well controlled. Priming remained above chance even when patients whose max-BIS was 60 were excluded from the analysis. There was no suggestion that priming was occurring at lighter anaesthetic depths when awareness was more likely; thus there was no correlation between implicit memory scores and BIS. Our patients did not receive neuromuscular blocking drugs and therefore were free to move if they became conscious. None did so.
No patient had spontaneous or convincing prompted recollection of intraoperative events evidenced by the ability to report specific details. However, one patient, who also performed above chance on the yesno recognition task, did report a dreamlike memory of hearing a voice. Performance on the yesno recognition task slightly, but not significantly, exceeded chance in the no-fentanyl group. Inspection of the data showed that just six patients responded yes to one or more target words. Patients were not told how many of the words had been presented during anaesthesia, but were asked to say yes to any words they remembered hearing during anaesthesia. Five of these patients also responded yes to two or more distractor words (of a possible seven). No other patient made false alarms. This pattern of responding suggests a general bias towards making a yes response in the five patients. This bias may have facilitated implicit memory influences, such as familiarity in the absence of explicit recollection, on yesno recognition performance. Three of these patients also had above-zero scores on the implicit memory test. Thus the sensitivity of the yesno recognition task may compromise its purity, allowing implicit memory to contaminate performance. Given the fact that our patients did not receive neuromuscular blocking drugs and that BIS recordings indicated deep anaesthesia, we suggest that this performance on the yesno recognition task is spurious rather than an indication of awareness during surgery or explicit recall. This is consistent with the lack of correlation between our measure of awareness (BIS) and yesno recognition scores.
The mean implicit memory score was slightly higher for the group who received no fentanyl than for the group who did receive fentanyl (mean 0.09 compared with 0.07), but the difference between groups was not statistically significant. We had predicted that fentanyl would reduce priming during anaesthesia because it suppresses the stress response to surgery that we hypothesized facilitates intraoperative priming. This study provided no evidence for an effect of fentanyl, although our dose was relatively small and the memory scores were in the predicted direction. From the present findings and those of our previous study,5 it would appear that any lingering effect on the surgical stress response of a dose of fentanyl at induction is insufficient to prevent priming during surgery.
Previous studies of intraoperative priming during BIS-monitored propofol anaesthesia have provided mixed results. As already discussed, our own previous study5 using a word-stem completion test found evidence for intraoperative priming when mean BIS was 44. Struys and colleagues12 played patients the Robinson Crusoe story during surgery and found that three of 58 patients associated Robinson Crusoe with the cues Friday or desert island on recovery. Although they do not report a baseline measure for this association, Schwender and colleagues14 reported no spontaneous associations of Robinson Crusoe with Friday in a control group of 15 patients who had not been played the story, so it is conceivable that the three positive responses represent implicit memory for the story. Kerssens and colleagues13 presented repetitions of the phrase yellow banana green pear during surgery with BIS between 40 and 60, and measured implicit memory with a category generation task that required patients to name the first three exemplars of fruits and colours that came to mind. They found no difference in the hit rate (banana, pear, yellow or green responses) between the experimental group and a control group that heard bird sounds during surgery. We suggest that previous failures to find priming during propofol anaesthesia may lie in the type of memory test used. Anaesthesia leaves low-level auditory processing intact, and perceptual implicit memory tests such as the word-stem completion task, which rely upon the same neural networks that subserve initial processing of stimuli,15 are more likely to detect priming during anaesthesia than conceptual tests. Conceptual tests such as category generation and word association tasks demand higher-level semantic processing of stimuli and priming of links between stimuli, probably in the association areas of the frontal and temporal lobes to which the flow of information is disrupted by anaesthesia.16
To conclude, we replicated our finding of auditory priming in patients undergoing surgery with propofol and nitrous oxide anaesthesia. Priming exceeded chance even when patients who experienced moments of light anaesthesia (BIS60) were excluded from the analysis. Thus memories can be primed even when patients are unconscious.
![]() |
Appendix |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
![]() |
Acknowledgments |
---|
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
2 Andrade J. Learning during anaesthesia: a review. Br J Psychol 1995; 86: 479506[ISI][Medline]
3 Wang M. The psychological consequences of explicit and implicit memories of events during surgery. In: Ghoneim MM, ed. Awareness During Anesthesia. Oxford: Butterworth Heinman, 2001
4 Lubke GH, Kerrsens C, Phaf H, et al. Dependence on explicit and implicit memory on hypnotic state in trauma patients. Anesthesiology 1999; 90: 67080[CrossRef][ISI][Medline]
5 Deeprose C, Andrade J, Varma S, et al. Unconscious learning during surgery with propofol anaesthesia. Br J Anaesth 2004; 92: 1717
6 Hoymork SC, Raeder J, Grimsmo B, et al. Bispectral index, predicted and measured drug levels of target-controlled infusions of remifentanil and propofol during laparoscopic cholecystectomy and emergence. Acta Anaesthesiol Scand 2000; 44: 113844[CrossRef][ISI][Medline]
7 Shanks DR, St John MF. Characteristics of dissociable human learning systems. Behav Brain Sci 1994; 17: 36795[ISI]
8 Veselis RA. Editoral II: Gone but not forgottenor was it? Br J Anaesth 2004; 92: 1613
9 Veselis RA, Reinsal RA, Feshchenko VA, et al. The comparative amnestic effects of midazolam, propofol, thiopental and fentanyl at equisedative concentrations. Anesthesiology 1997; 87: 74964[ISI][Medline]
10 Schriker T, Carli F, Schreiber M, et al. Propofol/sufentanil anesthesia suppresses the metabolic and endocrine response during, not after, lower abdominal surgery. Anesth Analg 2000; 90: 4505
11 Russell IF, Wang M. Absence of memory for intra-operative information during surgery under adequate general anaesthesia. Br J Anaesth 1997; 78: 39
12 Struys ML, Versichelen L, Byttebier G, et al. Clinical usefulness of the bispectral index for titrating propofol target effect-site concentration. Anaesthesia 1998; 53: 412[ISI][Medline]
13 Kerssens CJ, Klein J, van der Woerd A, et al. Auditory information processing during adequate propofol anesthesia monitored by electroencephalogram bispectral index. Anesth Analg 2001; 92: 121014
14 Schwender D, Madler C, Klasing S, et al. Anesthetic control of 40-Hz brain activity and implicit memory. Conscious Cogn 1994; 3: 12947[CrossRef][ISI]
15 Gabrieli JDE. Cognitive neuroscience of human memory. Ann Rev Psychol 1998; 49: 87115[CrossRef][ISI][Medline]
16 Angel A. Central neuronal pathways and the process of anaesthesia. Br J Anaesth 1993; 71: 14863[ISI][Medline]