Postoperative cognitive dysfunction (POCD) has been reported after 28%–100% of cardiac operations (1–3) and after 7%–26% of noncardiac surgical procedures (4). The incidence of POCD is substantially less (2%–10%) when assessed 1–6 mo postoperatively and 5% at 6 mo (1,4,5). POCD differs from delirium, in that delirium is considered an acute, potentially reversible confusional state (6,7) and is a common feature of physical illness or drug intoxication. POCD is associated with a decline in performance of activities of daily living of elderly medical and surgical patients (8–10) and can be a substantial burden on family and social support systems (10).
Although no definitive causes for POCD have been identified, possible contributing factors are cerebral effects of anesthesia, patient age, and education (11,12). Use of diazepam premedication nearly doubled the incidence of 7-day POCD when compared to the general incidence of cognitive dysfunction after noncardiac surgery (13). Recently, Moller et al. (4) found age to be the only risk factor for POCD at 3 mo after noncardiac surgery but did not assess the effect of the depth of anesthesia on delayed POCD.
The aim of the current study was to investigate whether depth of anesthesia as determined by the bispectral index (BIS) affects 4–6 wk postoperative neurocognitive function in patients older than 50 yr.
After written informed consent and IRB approval, patients were randomized to higher BIS (HIBIS) (i.e., lighter anesthetic) or lower BIS (LOBIS) (i.e., deeper anesthetic) groups. To ensure comparability, groups were block randomized by spine, abdominal, and pelvic surgery. The HIBIS group was maintained at a BIS target of 50–60 during surgery and 55–70 during wound closure. The LOBIS group was maintained at a BIS target of 30–40 during surgery and 50–60 during closure.
Patients older than 50 yr with Mini-Mental State Examination (MMSE) (14) scores ≥23 were included in the study. They were scheduled to undergo elective surgical procedures expected to take 2–3 h, such as laminectomy, hysterectomy, or other abdominal surgery.
We excluded outpatients, patients who were difficult to follow-up after hospital discharge, those with another operative procedure planned within 1 wk, patients with known central nervous system or major psychiatric disease, prior brain surgery, significant cardiovascular disease, and those who could not understand English instructions or were otherwise unable to cooperate. We also excluded patients with alcoholism, other drug dependence, or known sensitivity or allergy to any component of the study anesthetic regimen.
The anesthetic protocol was the same in both groups. Patients were premedicated with midazolam 1–2 mg and fentanyl 50–100 μg. The anesthetic was induced with propofol 1.5–2.5 mg/kg and fentanyl 2–4 μg/kg; it was maintained with nitrous oxide 60%–70%, isoflurane, fentanyl 1–2 μg · kg−1 · h−1 and neuromuscular block.
The anesthetic depth was titrated using isoflurane to achieve the desired BIS target ranges in each group. Fentanyl was used only to provide background analgesia, not to adjust depth of anesthesia. There were no restrictions on the use of muscle relaxants other than their mandatory use. Ondansetron, rather than droperidol or phenothiazines, was administered for nausea. Ketorolac 15–30 mg IV was administered during closure or postoperatively, except in spine surgery. Arterial blood pressure (BP) was controlled to within 25% of the most recently obtained preoperative (before the day of surgery) systolic value. If the BIS was outside the target range, BP was adjusted by anesthetics (isoflurane). If BP required further treatment despite BIS within the targeted range, it was controlled as follows: Hypotension (systolic <75% of baseline) was treated with phenylephrine 50–100 μg IV if heart rate (HR) was >70 bpm, or with ephedrine 5–10 mg IV if HR was < 70 bpm. If transient hypovolemia was suspected, these drugs were administered as a temporizing measure while volume was corrected. Hypertension was treated with nitroglycerin 80–160 μg IV if HR was <70 bpm, and with labetalol 10–20 mg IV if HR was >70. Phenylephrine and nitroglycerin were administered by infusion, as necessary. HR was controlled to 50–90 bpm using glycopyrrolate, propranolol or esmolol, as appropriate.
Data were recorded intraoperatively by downloading the BIS continuously to a laptop computer; isoflurane MAC hours and total doses of propofol and fentanyl were documented at the end of each case. BP and HR were recorded every 5 min. Immediate postanesthetic observations included time to tracheal extubation, time to following commands, time to orientation, time to Aldrete Score ≥9, and a modified emergence score (15). A battery of cognitive tests was administered preoperatively, at 1 wk and at 4–6 wk postoperatively.
A 30- to 40-min cognitive test battery was administered to all patients in a standardized manner by trained research personnel at baseline (BL, before surgery), immediate postoperative follow-up (IPO, after surgery and before discharge), and at long-term postoperative follow-up (LTPO, 4–6 wk after surgery).1 The primary cognitive outcome measures consisted of the Processing Speed Index (PSI), Working Memory Index (WMI), and a Verbal Memory Index (VMI), which were derived from selected subtests of the Wechsler Adult Intelligence Scale–III (WAIS–III) (16) and the Wechsler Memory Scale–III (WMS–III) (17). The PSI is composed of the Digit Symbol and Symbol Search subtests from the WAIS–III and is a measure of graphomotor and mental processing speed. Raw scores from each subtest are converted to age-corrected scaled scores, which are then summed to yield a PSI score with a mean of 100 and a standard deviation of 15. The PSI was normed on a United States census-matched sample of 2450 subjects (16). The WMI and VMI scores are composed of subtests from the WMS–III, which was co-normed with the WAIS–III on a United States census-matched subsample of 1250 subjects (17). WMI is composed of the Letter-Number Sequencing and Spatial Span subtests of the WMS–III, verbal and nonverbal measures of complex attention and concentration. Raw scores on each subtest are converted to age-corrected scaled scores, which are then summed to yield a WMI score with a mean of 100 and standard deviation of 15. A VMI measure of verbal learning and recall was constructed from the immediate and delayed trials of the Verbal Paired Associates and Word Lists subtests from the WMS–III. The age-corrected scaled scores for the immediate and delayed trials of each of these subtests were summed, and the composite was converted to a VMI score with a mean of 100 and a standard deviation of 15.
The study was designed to detect differences of 0.5 sd or more in test-retest change scores between the groups on any of the continuous outcome measures using two-tailed Student’s t-tests, at the 0.10 significance level with 80% power. Approximately one third of the way through the study period, the decision was made to forego the cognitive function testing at the 1-wk postoperative point (IPO) because of poor patient compliance and patient reluctance to agree to another health care encounter during the early postoperative period. We also observed that elderly patients were generally still substantively affected by mobility problems and family-related issues that made it difficult for them to cooperate with cognitive function testing. We classified patients affected by this change in the design of our study as being in either Wave 1 (i.e., those tested preoperatively, at 7 days, and at 4–6 wk postoperatively) or Wave 2 (i.e., those tested only preoperatively and at 4–6 wk postoperatively). We examined and accounted for possible differential test-retest practice effects resulting from Wave classification (presence or absence of the IPO assessment) in our preliminary and main analyses.
As a check of the effectiveness of the BIS manipulation, we compared the randomized groups on procedural and emergence parameters (day of surgery) using the Wilcoxon’s ranked sum test (MAC hours) or log-rank test for interval-assessed data (every 5 min for time to Aldrete Score ≥9 and time to return to baseline MMSE. We assessed the treatment effect of low versus high BIS on cognitive PSI, WMI, and VMI scores at LTPO (4–6 wk follow-up) adjusted for baseline performance using analysis of covariance. Because previous research indicates that the demographic variables of age, education, and gender affect test and retest performance on the WAIS–III and WMS–III (18–22), these variables were also entered as covariates.
Because the evaluation of treatment effects on cognition within a test-retest paradigm can be confounded by factors such as practice effects (i.e., repeated test exposure), measurement error, and regression to the mean, we also examined the impact of BIS level on our cognitive variables using reliable change methods (23–27). Specifically, we used standardized regression based norms for change derived from the test-retest samples obtained during the standardization of the WAIS–III (16) and WMS–III (17).2 Using multiple regression techniques, equations were derived for predicting retest scores for the WAIS–III (18) and WMS–III (19) cognitive variables using baseline scores, age, education, and sex as potential predictor variables. Table 1 summarizes these prediction equations as well as their standard errors of regression (SEreg). These equations were applied to our randomized groups to determine whether their observed retest scores on the cognitive variables deviated in a reliable and clinically meaningful way from normal expectations. Observed minus predicted differences for each cognitive variable were expressed on a common z-score metric by dividing the difference scores by the appropriate standard error of the regression estimate (SEreg). The effect of BIS level for the randomized groups was analyzed using the z-scores for PSI, WMI, and VMI, adjusting for Wave in separate analyses of covariance and on Wave 2 patients only using analysis of variance.
A total of 100 patients were recruited and randomized. Fifteen subjects could not complete the protocol because of equipment difficulties and an additional 11 subjects were excluded from data analysis because of failure to complete the follow-up protocol. The latter was almost always the result of patient unavailability or unwillingness to be retested. The remaining 74 patients (n = 36 LOBIS; n = 38 HIBIS) provided complete data. Comparison of those who completed testing (n = 74) with those who dropped out (n = 26) revealed no difference in intention-to-treat on our primary outcome variables of PSI (P = 0.529), WMI (P = 0.133), and VMI (P = 0.659).
The LOBIS and HIBIS groups were comparable with respect to age, education, handedness, and gender (Table 2). BIS values and ranges during the surgical procedure, and MAC hours and early postoperative recovery milestones appear in Tables 3 and 4, respectively. The median BIS values for the two groups were separated by 12 points (95% confidence interval [CI], 8–14), Table 3. Furthermore, comparison of MAC hours between the LOBIS and HIBIS groups was statistically significant (P = 0.045; one-tailed test),3 verifying that the two groups were managed differently with regard to depth of anesthesia. The median LOBIS interquartile range was somewhat smaller, suggesting less within-patient variability for the deeper anesthetic group. HIBIS patients had a significantly faster time to Aldrete Score ≥9 (P = 0.032; one-tailed test). The groups were not different with regard to the time to achieve baseline MMSE.
Because of scheduling difficulties and poor patient compliance, the IPO cognitive assessment was discontinued after completion of the first 26 patients (Wave 1: 12 LOBIS and 14 HIBIS subjects). The remaining 48 patients (Wave 2: 24 LOBIS and 24 HIBIS subjects) had only one follow-up assessment (LTPO). Although there were comparable numbers of Wave 1 and 2 patients in each of the randomized treatment groups (Χ2(1) <1.0, not significant), we expected that Wave 1 patients may have better cognitive performance at LTPO than Wave 2 patients, who did not experience an additional test exposure. Further, LTPO cognitive performance could differ between treatment groups because of possible differential test-retest practice effects resulting from Wave classification (presence or absence of the IPO assessment). Therefore, we computed 2 × 2 BIS-Level by Wave analyses of covariance for each of the cognitive indexes at LTPO, co-varying the subjects’ baseline score, age, education, and sex. As expected, we found directional differences with Wave 1 subjects having better follow-up scores than Wave 2 subjects on PSI [F(1,66) = 3.75; P = 0.0285] and VMI [F(1,65) = 3.17, P = 0.04]. The main effect of Wave for WMI was nonsignificant, and there were no significant BIS-level by Wave interactions for PSI, WMI, or VMI. Because there were no interactions, we collapsed Wave within the treatment groups in our subsequent main analyses of the LTPO scores but retained it as a covariate to provide greater control of within group variance. For the reliable change analyses, we computed between-group analyses of covariance for the entire sample with Wave as a covariate and separate analyses of variance for only those subjects in Wave 2.
To assess the effects of BIS level on cognitive function at LTPO follow-up, separate analyses of covariance for PSI, WMI, and VMI were computed with the subjects’ baseline cognitive scores, demographic variables, and Wave condition entered as covariates. Figure 1 depicts the adjusted mean scores for the cognitive variables at LTPO follow-up, and Table 5 summarizes the statistical comparisons between the LOBIS and HIBIS groups. As can be seen in Table 5, there was a significant group difference between the LOBIS and HIBIS groups for PSI, with PSI being significantly higher in the LOBIS versus the HIBIS group (113.7 versus 107.9; P = 0.006) by 4–6 wk after surgery. This difference represents a large moderate treatment effect (Cohen’s d = 0.686) (28) and accounted for 10.5% of the variance (η2) in PSI scores. There were no significant treatment effects of BIS level for WMI or VMI.
Reliable Change Analyses
For each cognitive variable, we performed two analyses. The first analysis compared LOBIS and HIBIS groups collapsed across Wave conditions (n = 74), using their baseline performances and demographics to predict their scores at LTPO. Because the Wave 1 subjects had the additional benefit of the IPO assessment, Wave was entered as a covariate. The second analysis examined only Wave 2 patients because these subjects had only a single retest assessment and were most analogous to the control subjects from whom standardized regression-based norms were derived.
The results of the reliable change analyses are summarized in Table 6. Adjusting for Wave differences across all subjects (n = 74), the only significant treatment effect again was for PSI (P = 0.004), with BIS level accounting for 11% of the variance (η2) in the standardized deviations from expected PSI retest performance as before; this represents a large moderate treatment effect (Cohen’s “d” = 0.705). Similar results were observed for the analysis of only those patients who had a single postoperative follow-up (Wave 2: n = 48) and who were most analogous to the control subjects from which the standardized regression-based prediction equations were derived. Even with a much smaller sample size (n = 48), LOBIS subjects again performed significantly better than HIBIS subjects in Wave 2 (P = 0.027), with BIS level accounting for 10.2% of the observed variance (η2) in PSI change scores. No statistically significant effects were observed for WMI or VMI, although BIS level did produce a notable effect size for WMI (Cohen’s “d” = 0.439) and accounted for 4.6% of the variance (η2) in WMI.
Using the reliable change z-scores derived from the test-retest multiple regressions presented in Table 1, we constructed frequency tables for each of the three cognitive outcome variables. POCD was operationally defined as decrements in performance that exceed those expected by chance alone in normal samples at the lower 5th percentile (i.e., a negative z-score ≤ −1.64). The incidence of cases in Waves 1 and 2 and Wave 2 only that exceeded this cutoff for each cognitive variable is presented in Table 7 for the LOBIS and HIBIS groups. As can be seen, the incidence of decrements in PSI was much higher for patients in the HIBIS condition (20.8%) than in the LOBIS condition (0.0%). Decrements in WMI were also more frequent in the HIBIS group (17.4%) compared to the LOBIS group (4.1%).
Our results suggest that depth of anesthesia during elective surgical procedures of intermediate duration may affect cognitive performance as late as 4–6 weeks postoperatively. Specifically, we showed that patients kept at somewhat lower BIS levels during the majority of the surgical procedure performed better in the information processing speed domain of cognitive function, as measured by PSI, compared with their somewhat more lightly anesthetized counterparts.
The change in cognitive processing speed from the baseline was more favorable in the LOBIS group than in the HIBIS group, whereas there was no change in working memory or verbal memory. Judging from clinical experience with these neurocognitive measurements, the patients in the LOBIS group with mean PSI scores of 113 (versus 107 in the HIBIS group) were likely to be more alert to changes in the environment, process conversations more quickly, and make connections more quickly between different bits of information. Although we observed an average difference of 6 points between the groups, some HIBIS patients had PSI values 10–15 points lower than LOBIS patients; consequently the effect on their lifestyle would have been more substantial. PSI is a subset of the WAIS–III (16) that tests psychomotor speed. It measures speed of information processing defined by the Digit Symbol and Symbol Search subtests (29). Compared with other WAIS–III composite scores, such as WMI, the PSI is more sensitive in detecting abnormal cognitive function in children and adults after traumatic brain injury (30,31). Decreased PSI scores by 6 points have also been associated with unemployment among patients with multiple sclerosis (32). It is interesting that POCD after cardiac surgery also affects very specific cognitive domains but not necessarily all cognitive functions tested (2). Specific domains impaired at 1 month were verbal memory and language, which improved after 1 year in most patients. However, psychomotor speed and motor speed impairment persisted both at 1 month and 1 year (2). Our findings point to a specific cognitive decrement in mental and psychomotor speed (as measured by PSI) that is associated with a relatively “lighter” anesthetic regimen.
There are several limitations inherent in the design and findings of this study. First, of the three main cognitive test components studied, deeper anesthesia was significantly associated with improved long-term cognitive recovery only with respect to PSI. The clinical significance of changes in only one of several neurocognitive tests may be limited. Nevertheless, research among patients with traumatic brain injury (33) and multiple sclerosis (34) demonstrates differential sensitivity. We chose 3 cognitive dimensions that represent relatively independent ability domains and are commonly used as clinical end-points. Furthermore, the power to detect a difference may have been affected by the change in protocol, which resulted in two “waves” of analysis. Although the two groups were comparable with respect to most factors known to affect POCD, it is possible that other, as yet unidentified, confounding factors were still responsible for the observed difference in PSI recovery. It is also possible that combinations of commonly prescribed drugs (35) may have caused central anticholinergic effects sufficient to account for the difference in our elderly patients. Yet, in a large prospective study (ISPOCD-1) (4), age was found to be the most significant factor for late POCD, for which our groups were well matched (Table 2). The other predictive factor was presurgically impaired mental status (36). Because all our patients had MMSE scores more than 23, preoperative mental status was likely not a confounding factor in our study. Also, there is a large variability associated with neuropsychological testing in surgical patients with the possibility that the differences between the groups were attributable to random variation rather than true cognitive differences. However, by randomizing subjects to either the HIBIS or LOBIS groups, the possibility of systematic bias in neuropsychological functioning between groups is minimized. Although cognitive performance is still variable among individuals within a group, the purpose of analysis of variance is to detect whether observed group mean differences are sufficiently large relative to group variance to be most likely (i.e., significant) the result of non-random (i.e., chance) factors. In our study with the observed variability of performance in both BIS groups, mean differences of 6 points in PSI at long-term postsurgical follow-up would only be expected by chance in 6/1000 cases. Lastly, our protocol specified target BIS ranges instead of target BIS values. This resulted in group median BIS values on the upper end of the target range for the LOBIS group and the lower end for the HIBIS group, possibly leading to an understating of the effect of anesthetic depth on postoperative cognitive function.
The cause of POCD has not been determined. Hypoxemia (37,38), hypotension (39), markedly abnormal preoperative serum sodium, potassium or glucose levels, poor preoperative cognitive status, self-reported alcohol abuse (40), diminished cerebral oxygenation extraction (41), and decreased regional cerebrovenous oxygenation (42) have been associated with the occurrence of POCD. Hypoxia reduces central acetylcholine release (43). Acetylcholine appears to play an important role in the maintenance of mental and intellectual function (44). The loss of cholinergic neurons is believed to be a key factor in the development of the learning and memory deficits characteristic of Alzheimer’s disease (45). Centrally acting anticholinergic drugs can lead to lethargy and impaired short-term memory (46). Even histamine receptor blockers, corticosteroids, and digitalis have central antimuscarinic activity (35).
Given the uncertain pathogenesis of POCD and the ability of certain anesthetics to depress central cholinergic transmission (47), it is difficult to explain our finding of better mental processing function after deeper anesthesia. Furthermore, BIS-guided lighter anesthetic regimens have been found to improve early postoperative recovery (48). Protection against POCD could potentially be a result of the neuroprotective effect of isoflurane (49). It has been shown that decreases in BIS level that occur with deeper levels of anesthesia correlate with magnitude of cerebral metabolic depression. Alkire (50) showed that at a mean BIS level of 54 ± 9, the mean cerebral metabolic rate decreased by 46%; however, at a mean BIS level of 37 ± 6, it decreased by 60%. We speculate that deeper levels of isoflurane anesthesia might have had a neuroprotective effect, perhaps by decreasing the cerebral metabolic rate (50). Important cellular mechanisms of isoflurane neuroprotection occur via calcium-mediated decreases in glutamate release, as well as through the preservation of the important neuronal regulatory enzyme CaMKIIB (51,52). We wish to emphasize, however, that neither ischemia nor any other injurious events were observed in our study and that invoking protective mechanisms in the explanation of our results is speculation.
In summary, deeper general anesthesia, as defined by a median BIS level of 39 compared with 51, was associated with somewhat better recovery of cognitive function 4–6 weeks postoperatively, particularly with respect to the ability to process information. To our knowledge this is the first study to link intraoperative anesthetic level to remote postoperative cognitive performance. Our observations highlight the need for further studies to better understand the contribution of perioperative management to POCD.
The authors would like to thank Mrs. Carrie Beven, RN, Ms. Christine Cribbs, RN and Mrs. Stephanie Ziegman, RN for their dedicated unflagging support of this project, as well as Mrs. Tanya D. Smith, BA, PA for her editorial assistance.
1. Savageau JA, Stanton BA, Jenkins CD, Frater RW. Neuropsychological dysfunction following elective cardiac operation. II. A six-month reassessment. J Thorac Cardiovasc Surg 1982;84: 595–600.
2. McKhann GM, Goldsborough MA, Borowicz LM Jr, et al. Cognitive outcome after coronary artery bypass: a one-year prospective study. Ann Thorac Surg 1997;63:510–5.
3. Townes BD, Bashein G, Hornbein TF, et al. Neurobehavioral outcomes in cardiac operations: a prospective controlled study. J Thorac Cardiovasc Surg 1989;98:774–82.
4. Moller JT, Cluitmans P, Rasmussen LS, et al. Long-term postoperative cognitive dysfunction in the elderly: ISPOCD1 study. Lancet 1998;351:861.
5. Williams-Russo P, Sharrock NE, Mattis S, et al. Cognitive effects after epidural vs general anesthesia in older adults: a randomized trial. JAMA 1995;274:44–50.
6. Lipowski ZJ. Delirium in the elderly patient. N Engl J Med 1989;320:578–82.
7. Lipowski ZJ. Transient cognitive disorders (delirium, acute confusional states) in the elderly. Am J Psychiatry 1983;140: 1426–36.
8. Bedford PD. Adverse cerebral effects of anaesthesia on old people. Lancet 1955;269:259–63.
9. Murray AM, Levkoff SE, Wetle TT, et al. Acute delirium and functional decline in the hospitalized elderly patient. J Gerontol 1993;48:M181–M186.
10. O’Keeffe S, Ni Chonchubhair A. Postoperative delirium in the elderly. Br J Anaesth 1994;73:673–87.
11. Dodds C, Allison J. Postoperative cognitive deficit in the elderly surgical patient. Br J Anaesth 1998;81:449–62.
12. Ritchie K, Polge C, de Roquefeuil G, et al. Impact of anesthesia on the cognitive functioning of the elderly. Int Psychogeriatr 1997;9:309–26.
13. Rasmussen LS, Steentoft A, Rasmussen H, et al. Benzodiazepines and postoperative dysfunction in the elderly. Br J Anaesth 1999;83:585–9.
14. Folstein MF, Folstein SE, McHugh PR. Mini-mental state: a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189–98.
15. Schubert A, Mascha EJ, Bloomfield EL, et al. Effect of cranial surgery and brain tumor size on emergence from anesthesia. Anesthesiology 1996;85:513–21.
16. Wechsler D. WAIS-III: Administration and Scoring Manual. (1997a). San Antonio, TX, The Psychological Corporation. 1997.
17. Wechsler D. WWS-III test administration and scoring manual (1997b). San Antonio: The Psychological Corporation, 1997.
18. Barrett J, Chelune GJ, Naugle RI. Test-retest characteristics and measures of meaningful change for the Wechsler Adult Intelligence Scale–III. J Int Neuropsychol Soc 2000;6:147–8.
19. Chelune GJ, Sands K, Barrett J. Test-retest characteristics and measures of meaningful change for the Wechsler Memory Scale III. J Int Neuropsychol Soc 1999;5:109.
20. Heaton RK, Taylor MJ, Manly J. Demographic effects and use of demographically corrected norms with the WAIS-III and WMS-III, Clinical interpretation of the WAIS-III and WMS-III. New York: Academic Press, 2003.
21. Lineweaver TT, Gordon J, Chelune GJ. Use of the WAIS-III and WMS-III in the context of serial assessments: Interpreting reliable and meaningful change, clinical interpretation of the WAIS-III and WMS-III. New York: Academic Press, 2003.
22. Taylor MJ, Heaton RK. Sensitivity and specificity of WAIS-III/WMS-III demographically corrected factor scores in neuropsychological assessment. J Int Neuropsychol Soc 2001;7:867–74.
23. Chelune GJ, Naugle RI, Lüders H, et al. Individual change after epilepsy surgery: practice effects and base-rate information. Neuropsychology 1993;7:41–52.
24. Chelune GJ. Assessing reliable neuropsychological change, prediction in forensic and neuropsychology: sound statistical practices. Mahwah: Lawrence Erlbaum Associates, 2003.
25. Kneebone AC, Andrew MJ, Baker RA, Knight JL. Neuropsychologic changes after coronary artery bypass grafting: use of reliable change indices. Ann Thorac Surg 1998;65:1320–5.
26. McSweeny AJ, Naugle RI, Chelune GJ, Luders HO. “T-scores for change”: an illustration of a regression approach to depicting change in clinical neuropsychology. Clin Neuropsychol 1993; 7:300–12.
27. Sawrie SM, Chelune GJ, Naugle RI, Luders HO. Empirical methods for assessing meaningful neuropsychological change following epilepsy surgery. J Int Neuropsychol Soc 1996;2:556–64.
28. Zakzanis KK. Statistics to tell the truth, the whole truth, and nothing but the truth: formulae, illustrative numerical examples, and heuristic interpretation of effect size analyses for neuropsychological researchers. Arch Clin Neuropsychol 2001;16:653–67.
29. Donders J, Warschausky S. A Structural Equation Analysis of the WISC-III in Children with Traumatic Head Injury. Child Neuropsychol 1996;2:185–92.
30. Hoffman N, Donders J, Thompson EH. Novel learning abilities after traumatic head injury in children. Arch Clin Neuropsychol 2000;15:47–58.
31. Axelrod BN, Fichtenberg NL, Liethen PC, et al. Performance characteristics of postacute traumatic brain injury patients on the WAIS-III and WMS-III. Clin Neuropsychol 2001;15:516–20.
32. Marrie RA, Miller DM, Chelune GJ, Cohen JA. Validity and reliability of the MSQLI in cognitively impaired patients with multiple sclerosis. Mult Scler 2003;9:621–6.
33. Donders J, Tulsky DS, Zhu J. Criterion validity of new WAIS-III subtest scores after traumatic brain injury. J Int Neuropsychol Soc 2001;7:892–8.
34. DeLuca J, Chelune GJ, Tulsky DS, et al. Is speed of processing or working memory the primary information processing deficit in multiple sclerosis? J Clin Exp Neuropsychol 2004;26:550–62.
35. Tune L, Carr S, Hoag E, Cooper T. Anticholinergic effects of drugs commonly prescribed for the elderly: potential means of assessing risk of delirium. Am J Psychiatry 1992;149:1393–4.
36. Goldstein MZ. When older persons undergo anesthesia and elective surgery. Am J Geriatr Psychiatry 2000;8:35–9.
37. Berggren D, Gustafson Y, Eriksson B, et al. Postoperative confusion after anesthesia in elderly patients with femoral neck fractures. Anesth Analg 1987;66:497–504.
38. Rosenberg J, Kehlet H. Postoperative mental confusion–association with postoperative hypoxemia. Surgery 1993;114:76–81.
39. Gustafson Y, Brannstrom B, Berggren D, et al. A geriatric-anesthesiologic program to reduce acute confusional states in elderly patients treated for femoral neck fractures. J Am Geriatr Soc 1991;39:655–62.
40. Marcantonio ER, Goldman L, Mangione CM, et al. A clinical prediction rule for delirium after elective noncardiac surgery. JAMA 1994;271:134–9.
41. Yoshitani K, Kawaguchi M, Sugiyama N, et al. The association of high jugular bulb venous oxygen saturation with cognitive decline after hypothermic cardiopulmonary bypass. Anesth Analg 2001;92:1370–6.
42. Edmonds HL, Thomas MA, Sehic A, et al. Cerebral oxygen desaturation during myocardial revascularization is associated with frontal lobe injury. Anesth Analg 1998;86:SCA13.
43. Hirsch JA, Gibson GE. Selective alteration of neurotransmitter release by low oxygen in vitro
. Neurochem Res 1984;9:1039–49.
44. Gibson GE, Blass JP, Huang HM, Freeman GB. The cellular basis of delirium and its relevance to age-related disorders including Alzheimer’s disease. Int Psychogeriatr 1991;3:373–95.
45. Fernandez CR, Fields A, Richards T, Kaye AD. Anesthetic considerations in patients with Alzheimer’s disease. J Clin Anesth 2003;15:52–8.
46. Wetherell A. Some effects of atropine on short term memory. Br J Clin Pharmacol 1980;10:627–8.
47. Violet JM, Downie DL, Nakisa RC, et al. Differential sensitivities of mammalian neuronal and muscle nicotinic acetylcholine receptors to general anesthetics. Anesthesiology 1997;86:866–74.
48. Burrow B, McKenzie B, Case C. Do anaesthetized patients recover better after bispectral index monitoring? Anaesth Intensive Care 2001;29:239–45.
49. Warner DS. Isoflurane neuroprotection: a passing fancy, again? Anesthesiology 2000;92:1226–8.
50. Alkire MT. Quantitative EEG correlations with brain glucose metabolic rate during anesthesia in volunteers. Anesthesiology 1998;89:323–33.
51. Miao N, Frazer MJ, Lynch C 3rd. Volatile anesthetics depress Ca2+ transients and glutamate release in isolated cerebral synaptosomes. Anesthesiology 1995;83:593–603.
52. Blanck TJ, Haile M, Xu F, et al. Isoflurane pretreatment ameliorates postischemic neurologic dysfunction and preserves hippocampal Ca2+/Calmodulin-dependent protein kinase in a canine cardiac arrest model. Anesthesiology 2000;93:1285–93.
1 The IPO assessment was discontinued approximately one third of the way through the study as a result of procedural difficulties. Twenty-six of 74 patients were affected. See text for further elaboration.
2 Norms for change have previously been developed by administering the WAIS–III (n = 392) and the WMS–III (n = 281) twice approximately 6–7 wk apart to normal controls.
3 The one-tailed Student’s t-test used here represents a manipulation check on the fact that deeper anesthesia should lead to more MAC hours. This was a matter of considerable debate in our group. The variables described in Tables 3 and 4 were not outcome variables but were included as manipulation checks to verify the efficacy of the BIS manipulation. These tests were conducted to verify the effectiveness of experimental manipulations and are by definition unidirectional (one-tailed) and not exploratory; they are considered effective or not effective. A negative result would be considered the same as a non-significant result, as it would mean that the manipulation did not achieve its intended purpose. We believe that one-tailed tests in this situation are appropriate.© 2006 International Anesthesia Research Society