Implicit memory formation in sedated ICU patients after cardiac surgery

J. Clark, L. Voss*, J. Barnard and J. Sleigh

Department of Anaesthetics and Intensive Care, University of Auckland, Waikato Clinical School, Hamilton, New Zealand

*Corresponding author. E-mail: vossl@waikatodhb.govt.nz

Accepted for publication: August 7, 2003


    Abstract
 Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Appendices
 References
 
Background. Recent research into memory formation under sedation has generated conflicting results. We investigated explicit and implicit memory in ICU patients during moderate to deep propofol sedation following cardiac surgery.

Methods. Two different methods of memory testing were used; (1) free-association (F-A) word-pair testing (n=33) to test conceptual implicit memory and (2) process dissociation procedure (PDP) (n=26) to detect perceptual implicit and explicit memory. One hour after surgery, whilst sedated, the F-A group received one of two lists of 10 category-exemplar word-pairs through headphones, while the PDP group was presented with one of two lists of 16 five-letter words. When awake and co-operative, the F-A group was tested using F-A testing, and the PDP group was tested using the PDP.

Results. The F-A group had a mean (SD) correct response rate of 7 (9)% for the target list, and 9 (8)% for the distractor list. The PDP group had a mean (SD) correct response rate of 11 (14) and 10 (13)% for the inclusion and exclusion lists, respectively, with mean correct response rates of 13 (14)% for both the corresponding distractor lists. Neither group showed any significant differences between their responses and a list of distractor words (Wilcoxon tests).

Conclusion. We found no evidence for memory formation in post-cardiac surgery patients under moderate to deep propofol sedation.

Br J Anaesth 2003; 91: 810–14

Keywords: anaesthetics i.v., propofol; sedation, implicit memory


    Introduction
 Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Appendices
 References
 
Memory formation can be implicit or explicit.1 Explicit memories are those memories an individual knows he or she has formed. In contrast, implicit memories are those we form without realizing. Both explicit and implicit memories may result when we knowingly attend to a particular stimulus. However, only implicit memories can be formed from a stimulus if we do not attend.

The concept of implicit memory formation whilst unresponsive has been around for decades. Despite many investigations, evidence for the formation of memories whilst unresponsive as a result of sedation or anaesthesia remains unsubstantiated.2 Some studies have reported that ‘adequate anaesthesia’ is associated with implicit memory formation.3 Conversely, there is evidence that memory formation is abolished at ‘subanaesthetic’ doses of volatile agents.4 5 Some of these conflicting results may be a result of ambiguities in the definition of general anaesthesia and sedation. For example, ‘adequate anaesthesia’ is defined by some according to the bispectral index3 and by others in terms of the response to surgical stimulus.4 5 Clarification of this contentious area of research is important from both a clinical and a psychological perspective. There may be therapeutic benefits related to the implicit processing of positive aural stimuli6 7 while equally, awareness or recollection of unpleasant events may contribute to the trauma of surgery or intensive care.

For the purpose of this paper we have categorized the level of sedation in our subjects according to guidelines published by the American Society of Anesthesiologists8 and the Harvard Medical School.9 Thus, moderate sedation ‘is a drug-induced depression of consciousness during which patients respond purposefully to verbal commands’8 and deep sedation is a ‘state of depressed consciousness or unconsciousness from which the patient is not easily aroused and is unable to respond purposefully to physical stimulation or verbal command’.9 All of the subjects in the current study were either ‘deeply’ or ‘moderately’ sedated according to these guidelines, but probably not ‘anaesthetized’ (see Methods).

We investigated whether moderately to deeply sedated post-cardiac surgery patients can form explicit and/or implicit memories. We chose patients who had had cardiac surgery because they are an ‘at risk’ group for awareness during surgery and are kept sedated for several hours after surgery.

We used two different methods of memory testing; free-association (F-A) testing and the process dissociation procedure (PDP). The former is commonly used to detect implicit memories, while the latter can, in theory, reveal both implicit and explicit memories. To our knowledge, this is the first study of memory formation in sedated patients after cardiac surgery.


    Methods
 Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Appendices
 References
 
Patients
Ethical approval was granted from the Waikato Ethics Committee and informed consent obtained from all patients. An initial group of 33 patients undergoing cardiac surgery were recruited for the ‘free association’ memory test (see below) and a subsequent 26 patients for the ‘process dissociation’ memory test (see below). Two groups were used so that each memory test could be investigated independently. In the F-A group, cardiac surgery was for coronary artery bypass graft (CABG) (n=25), heart valve replacement/repair (n=3), or Bentall’s procedure (n=1). A further four subjects from the F-A group had combined CABG and valve replacement surgery. In the PDP group, surgery was for CABG (n=19), heart valve replacement/repair (n=5), or removal of atrial myxoma (n=1). One subject had combined CABG and valve replacement surgery. Patients were excluded from the study if they had hearing problems, memory problems, or difficulty understanding English.

After undergoing their cardiac operation patients were transferred to the intensive care unit (ICU) to recover. Here, patients were kept sedated (i.v. propofol 100–300 mg h–1) to allow tolerance of the tracheal tube. This was sufficient for 67% of F-A and 92% of PDP subjects to be classified as ‘deeply sedated’ (unresponsive to physical stimulation or voice9) and 18% of F-A subjects as moderately sedated (responsive to voice8). The remainder (15% of F-A and 8% of PDP subjects) were responsive to physical stimulation (moderate pressure to thumb nail bed) but not responsive to voice. General anaesthesia is most commonly defined as the alveolar partial pressure of a gas at which 50% of humans will not respond to a surgical incision.10 11 Thus, without doing a surgical incision we were unable to classify any patient as ‘anaesthetized’. It is unlikely that any of the deeply sedated subjects were anaesthetized, as a propofol dosage of 170 mg h–1 is not normally sufficient to induce a state of general anaesthesia. Accordingly, we refer to the present subject group as being ‘moderately to deeply sedated’.

The initial conditioning phase of the study started 1 h after the patients’ admission to ICU, to allow the effects of isoflurane and neuromuscular blocking agents administered during surgery to wear off. All subjects were intubated and ventilated during the conditioning phase. We recorded each patient’s age, sex, sedation level (using the Observer Assessment of Alertness and Sedation scale12), bispectral index (BIS) score (F-A group only), the dosage of fentanyl and midazolam received during surgery, and the rate of propofol infusion in ICU. The unavailability of a BIS monitor prevented us recording BIS in the PDP group.

Memory testing
Two well-recognized memory tests were investigated, F-A testing (n=33) and PDP (n=26). Both of these tests have been used extensively to investigate memory in sedated and anaesthetized subjects.3 1316

F-A testing
We tested implicit memory by using category-exemplar word-pairs (e.g. animal—lion) given while sedated and subsequently comparing recall of the words heard (targets) with a control list of word-pairs not heard (distractors). Any implicit memory of the conditioning phase in ICU should result in a significantly greater frequency of correct responses to target words relative to distractors.

With most (67%) still deeply sedated, patients in the F-A group were presented with one of two different lists (list A or B, Appendix 1) of 10 category-exemplar word-pairs. All words were presented five times using pre-recorded lists from a computer, through headphones. Patients were randomly assigned to either list A or list B.

The memory testing phase of the study began once patients were awake and lucid, on average 1.5 days after surgery. Patients underwent an F-A test, whereby the category word from every word-pair in both lists A and B (i.e. targets and distractors) was played in random order through headphones to the patient, and the patient was asked to respond with the first word which came to mind. ‘Nothing’ was an acceptable answer. Correct and incorrect responses were manually recorded. The headphones and voice used during the testing phase were identical to those used during the conditioning phase, to give all possible cues to memory.

PDP
The PDP uses word-stem completion tests to measure implicit memory formation, and also to ascertain the existence of any explicit memories that may have been formed.17 The presented words are split into inclusion and exclusion lists, with the patient being given different tasks for each. The inclusion list requires the patient to complete each stem with a five-letter word recalled from ICU, or failing that, the first five-letter word that comes to mind. Both implicit and explicit memory contributes to the patient’s score here. In contrast, the exclusion word list requires the patient to respond with a different word from that heard in ICU, or else in the absence of recall, with the first word that comes to mind. Therefore, any explicit memory present will result in a lower score on the exclusion stems than on the inclusion stems, while implicit memory will raise both scores above the level of random guessing.

With the majority (92%) still deeply sedated, patients were presented with pre-recorded words from a computer, through headphones. Patients were divided into four sub-groups (groups i–iv), and presented with one of two lists (list Y or Z, Appendix 2) of 16 five-letter words. For each of 640 cycles, the computer randomly chose one of the 16 words to present. Thus, each word was presented approximately 40 times.

Once patients were awake and lucid, usually 3 days after surgery, they were tested for implicit and explicit memory of the words heard in ICU. The average time between the conditioning and testing phases was longer amongst the PDP group, because of the more complex nature of the PDP testing, which required patients to be more fully recovered. Patients were played all word-stems from both lists Y and Z, with the words not heard previously acting as distractors. Half of the words from each list were used in the inclusion phase of testing and the other half in the exclusion phase. Counter-balancing was used to eliminate bias, as described by Lubke.3 Responses were recorded as ‘hits’ if they matched the five-letter word the stem was derived from. Responses in the singular were accepted even if the original word was plural, and patients could choose to pass if they could not think of a word within 5 s.

Data analysis and statistics
The PDP and F-A group responses were analysed using the Wilcoxon test, comparing the number of target hits to the number of distractor hits.14 The equations of Jacoby and colleagues17 were used to quantify any explicit or implicit memory present in the PDP group:

Explicit memory index=inclusion score – exclusion score

Implicit memory index=[exclusion score/(1 – explicit memory)] – distractor score

The Wilcoxon test was used to see if the values generated differed significantly from zero.

Data are presented as mean (SD). In all cases, P<0.05 was considered statistically significant.


    Results
 Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Appendices
 References
 
The clinical features of each group are shown in Table 1. Seven patients were operated on without undergoing cardiopulmonary bypass. There was no difference in memory performance between this group and all other patients ({chi}2-test).


View this table:
[in this window]
[in a new window]
 
Table 1 Patient data. OAAS, Observer Assessment of Alertness and Sedation; OAAS 2=responds to voice, OAAS 3=responds to light pain, OAAS >3=no response to light pain. Recall period=length of time between conditioning and testing phases
 
Of the 33 patients who underwent F-A testing, none could correctly complete more than three out of the 10 word-pairs given previously in the conditioning phase (Fig. 1). The mean (SD) percentage of hits was 6.6 (8.7)% for the target list and 9.4 (8.0)% for the distractor list. There was no evidence of implicit recall amongst this group of subjects (Wilcoxon test). That is, the rate of recall of target words was not significantly different from the rate of hits for distractor words. The subjects who were responsive to voice (n=6) at the conditioning phase exhibited similar target (4 (9)%) and distractor (8 (13)%) hit rates as the non-responsive subjects (Mann–Whitney test). Differences between list A and list B in terms of hit rates for target and distractor lists were also analysed to determine if one list was more likely to elicit hits. Pearson’s {chi}2-test showed no significant difference between lists.



View larger version (16K):
[in this window]
[in a new window]
 
Fig 1 Results of F-A testing for (A) target words and (B) distractor words. Both graphs show a similar distribution of results for both lists, indicating that (A) there was no difference in performance with respect to list heard, and (B) there was no implicit memory to improve performance on the target list.

 
The PDP group returned a mean hit rate of 10.9 (14)% for the inclusion target list and 10.3 (13)% for the exclusion target list (Fig. 2). Both distractor lists had mean hit rates of 12.7 (14)%. Wilcoxon tests showed no significant difference between either target list and the corresponding distractor list. This indicates no implicit memory present amongst the PDP group. Values for both forms of memory were calculated to be 0.9% for explicit memory and –2.2% for implicit memory. These scores were not significantly different from zero. This indicates that there was no evidence of implicit or explicit memory amongst the PDP group. Analysis of lists Y and Z with Pearson’s {chi}2 table showed no difference in hit rate.



View larger version (16K):
[in this window]
[in a new window]
 
Fig 2 Results of process dissociation testing for (A) target words and (B) distractor words. There is no difference in performance between results for target words and distractors, implying that no explicit or implicit memory occurred in this group of patients.

 

    Discussion
 Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Appendices
 References
 
We found that patients moderately to deeply sedated after cardiac surgery showed no evidence of implicit memory using F-A testing or PDP testing; nor any evidence of explicit memory using the PDP. These findings suggest that patients at this level of sedation may not benefit from positive (or be detrimentally affected by negative) aural stimuli. This requires further investigation. Our findings support other studies of deeply sedated patients.4 5 Conversely, a study by Cork and colleagues18 showed implicit memory under propofol sedation. The discrepancy in the findings between these studies is probably attributable to differences in the level of sedation, as the patients in Cork’s study were only moderately sedated (e.g. they could tap their fingers to command). Also, their patients underwent ambulatory procedures under sedation rather than general anaesthesia for cardiac surgery.

There is some evidence for the formation of implicit memories during general anaesthesia. How then can we explain the ability of the brain to form implicit memories whilst fully anaesthetized, but not when sedated with lower levels of propofol? Some researchers suggested that the stress of surgery might play an important role in the generation of implicit memories under general anaesthesia.14 The stress of surgery causes the release of adrenaline and cortisol, both of which may aid the formation of memory.19 Conflicting results may also be explained in some instances by differences in experimental design. For example, isolated forearm testing (IFT) is recommended in anaesthetized, paralysed subjects to confirm that they are unresponsive to stimulation.16 A number of studies that have demonstrated memory formation during anaesthesia, coupled with neuromuscular block, did not use the IFT technique.15 20 21 The importance of repetition in implicit memory formation has also been emphasized by some22 and investigations exploiting higher repetitions than in the present study have demonstrated implicit memory during general anaesthesia.15 22 Another possible explanation for the failure to detect implicit memory under sedation is that memory function is not proportional to propofol dosage.

A major problem we had carrying out this study was that of patient exhaustion. It is difficult to separate memory function from other aspects of cognitive function. The nature of this particular study involved the recruitment of patients undergoing cardiac surgery—an operation notorious for impairing cognitive function for some time post operation.23 While other studies used subjects whose operations were sufficiently minor as to allow memory testing to proceed within hours of the conditioning phase, no patients in this study were adequately recovered from their surgery to complete the testing phase before 24 h had passed, and many took longer than this. However, a previous study has shown that the length of interval between the conditioning and testing phases does not significantly affect memory performance3 and is therefore unlikely to have confounded the results.

An initial pilot survey was done of six non-patient (non-sedated) volunteers who completed (without prior conditioning) the 32 word-stems used in the PDP. The volunteers’ guess rate was higher than the patient group, with an average hit rate of about 20%, compared with 13% for distractor lists amongst the patients (because of a higher rate of passing and inappropriate responses). This suggests that cognitive impairment following cardiopulmonary bypass surgery, perhaps as a result of sleep fragmentation, or pain, may have detrimentally affected the patients’ ability to generate suitable words.

In conclusion, we found that explicit and implicit memory formation is abolished in patients after cardiac surgery who are under moderate to deep propofol sedation.


    Acknowledgements
 
The authors wish to thank the University of Auckland, and Waikato Hospital for their research grants, the staff of Ward 14 and ICU at Waikato hospital for their support, and all the patients who took part in this study.


    Appendices
 Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Appendices
 References
 


View this table:
[in this window]
[in a new window]
 
Appendix 1A F-A group
 

View this table:
[in this window]
[in a new window]
 
Appendix 2A PDP group
 

    References
 Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Appendices
 References
 
1 Graf P, Schacter DL. Implicit and explicit memory for new associations in normal and amnesic subjects. J Exp Psychol Learn Mem Cog 1985; 11: 501–18[CrossRef][ISI][Medline]

2 Andrade J. Investigations of hypesthesia: using anesthetics to explore relationships between consciousness, learning, and memory. Conscious Cogn 1996; 5: 562–80[CrossRef][ISI][Medline]

3 Lubke GH, Kerssens C, Phaf H, Sebel PS. Dependence of explicit and implicit memory on hypnotic state in trauma patients Anesthesiology 1999; 90: 670–80[CrossRef][ISI][Medline]

4 Gonsowski CT, Chortkoff BS, Eger EI, 2nd, Bennett HL, Weiskopf RB. Subanesthetic concentrations of desflurane and isoflurane suppress explicit and implicit learning. Anesth Analg 1995; 80: 568–72[Abstract]

5 Chortkoff BS, Gonsowski CT, Bennett HL, et al. Subanesthetic concentrations of desflurane and propofol suppress recall of emotionally charged information. Anesth Analg 1995; 81: 728–36[Abstract]

6 Maroof M, Ahmed SM, Khan RM, Bano S, Haque AW. Positive suggestion during surgery reduces post hysterectomy emesis. Can J Anaesth 1997; 44: 227[ISI][Medline]

7 Oddby-Muhrbeck E, Jakobsson J, Enquist B. Implicit processing and therapeutic suggestion during balanced anaesthesia. Acta Anaesthesiol Scand 1995; 39: 333–7[ISI][Medline]

8 American Society of Anesthesiologists Task Force. Practice guidelines for sedation and analgesia by non-anesthesiologists. Anesthesiology 2002; 96: 1004–17[ISI][Medline]

9 Holzman RS, Cullen DJ, Eichhorn JH, Philip JH. Guidelines for sedation by nonanesthesiologists during diagnostic and therapeutic procedures. The Risk Management Committee of the Department of Anaesthesia of Harvard Medical School. J Clin Anesth 1994; 6: 265[ISI][Medline]

10 Evers AS: Cellular and molecular mechanisms of anesthesia. In: Barash PG, Cullen BF, Stoelting RK, eds. Clinical Anesthesia, 3rd Edn. Philadelphia: Lippincott-Raven Publishers, 1996

11 Eger EI, 2nd, Saidman LJ, Brandstater B. Minimum alveolar anesthetic concentration: a standard of anesthetic potency. Anesthesiology 1965; 26: 756–63[ISI][Medline]

12 Glass PS, Bloom M, Kearse L, Rosow C, Sebel P, Manberg P. Bispectral analysis measures sedation and memory effects of propofol, midazolam, isoflurane, and alfentanil in healthy volunteers. Anesthesiology 1997; 86: 836–47[ISI][Medline]

13 Cork RC, Kihlstrom JF, Schacter DL. Absence of explicit or implicit memory in patients anesthetized with sufentanil/nitrous oxide. Anesthesiology 1992; 76: 892–8[ISI][Medline]

14 Stapleton CL, Andrade J. An investigation of learning during propofol sedation and anesthesia using the process dissociation procedure. Anesthesiology 2000; 93: 1418–25[CrossRef][ISI][Medline]

15 Adams DC, Hilton HJ, Madigan JD, Szerlip NJ, et al. Evidence for unconscious memory processing during elective cardiac surgery. Circulation 1998; 98: 289–93

16 Kerssens C, Lubke GH, Klein J, van der Woerd A, Bonke B. Memory function during propofol and alfentanil anesthesia: predictive value of individual differences. Anesthesiology 2002; 97: 382–9[CrossRef][ISI][Medline]

17 Jacoby LL, Toth JP, Yonelinas AP. Separating conscious and unconscious influences of memory: measuring recollection. J Exp Psychol Gen 1993; 122: 139–54[CrossRef][ISI]

18 Cork RC, Heaton JF, Campbell CE, Kihlstrom JF. Is there implicit memory after propofol sedation? Br J Anaesth 1996; 76: 492–8[Abstract/Free Full Text]

19 Mishkin M, Appenzeller T. The anatomy of memory. Sci Am 1987; 256: 80–9[ISI][Medline]

20 Munte S, Schmidt M, Meyer M, et al. Implicit memory for words played during isoflurane- or propofol-based anesthesia: the lexical decision task. Anesthesiology 2002; 96: 588–94[ISI][Medline]

21 Jelicic M, De Roode A, Bovill JG, Bonke B. Unconscious learning during anaesthesia. Anaesthesia 1992; 47: 835–7[ISI][Medline]

22 Block RI, Ghoneim MM, Sum Ping ST, Ali MA. Human learning during general anaesthesia and surgery. Br J Anaesth 1991; 66: 170–8[Abstract]

23 Bruggemans EF, Van de Vijver FJ, Huysmans HA. Assessment of cognitive deterioration in individual patients following cardiac surgery: correcting for measurement error and practice effects. J Clin Exp Neuropsychol 1997; 19: 543–59[ISI][Medline]