Cognitive Outcome After On- and Off-Pump Coronary Artery Bypass Grafting Surgery: A Systematic Review and Meta-Analysis Ewan D. Kennedy, BSc (Hons),* Kevin C. C. Choy,* R. Peter Alston, MD, FRCA, FFPMRCA, FFICM,† Shaoyun Chen,* Muhamed M. H. Farhan-Alanie,* Jamie Anderson,* Yun Lin Ang* Deborah E. Moore,* Sam A. MacKenzie,* and Robert A. Sykes* Objective: The aim of this study was to compare cognition following coronary artery bypass grafting (CABG) surgery with or without cardiopulmonary bypass (CPB) (on- or off-pump). Design: Systematic review and meta-analysis of randomized control trials comparing cognitive outcome in patients undergoing CABG surgery on- or off-pump as assessed by continuous measures from a battery of 7 psychometric tests. Setting: Multi-institutional centers performing CABG surgery. Participants: Patients with coronary artery disease requiring CABG surgery. Interventions: CABG surgery with or without CPB. Measurements and main results: A structured literature search identified 13 randomized control trials that included a total of 2,405 patients. Results from 7
psychometric tests were grouped into early (r 3 months) and late (6-12 months) postoperative periods. No significant differences were found between on- and off-pump groups in any of the 7 psychometric tests in either the early (p range 0.21-0.78) or late (p range 0.09-0.93) postoperative period. Conclusion: The results suggested that CPB may not be associated with cognitive decline that is associated with CABG surgery. & 2013 Elsevier Inc. All rights reserved.
T
cognitively impaired may be useful. However, dichotomizing continuous data has a number of serious drawbacks in statistical analysis.11 In particular, there is a loss of statistical power to detect a relationship between a variable and patient outcome.11 MacCallum et al extensively reviewed the practice of dichotomization, finding serious drawbacks including spur ious results, and concluded that the practice rarely was justified.12 In psychometric testing of patients undergoing cardiac surgery, the practice of dichotomizing cognition by defining a threshold for decline has been reviewed and considered to be arbitrary.13 Moreover, the use of different definitions was shown to create a large difference in the rates of cognitive decline reported.14
RADITIONALLY, CORONARY ARTERY BYPASS GRAFTING (CABG) surgery has been performed on pump; that is, with the use of a cardiopulmonary bypass circuit (CPB). However, in the 1990s, off pump CABG surgery, performed without CPB, was introduced. A strong motivation for performing off pump CABG surgery was that avoiding CPB would reduce the incidence of brain damage and, in particular, cognitive impairment.1 Indeed, cognitive impairment long has been asso ciated with CABG surgery and popularly has been called ‘pump head’ because of the widely held belief that it was caused by CPB.2 Over recent years, this belief has been called into question by an increasing number of randomized controlled trials (RCTs) that failed to find a better cognitive outcome associated with off pump compared to on pump CABG surgery. However, many of these RCTs enrolled small numbers of patients, and the lack of a significant difference between on and off pump groups could have been the result of type II statistical error. In 2008, a systematic review and meta analysis by Marasco et al found that there were no significant differences in cognitive outcome when comparing patients undergoing CABG surgery on and off pump.3 Since that study’s publication, 6 additional RCTs have been conducted.4-9 This included the largest RCT conducted to date, published in 2010 by Kozora et al, that compared the cognitive outcome of 1,156 patients undergoing CABG surgery on and off pump.9 These new studies contained additional important data to aid in the understanding of cognitive decline after CABG surgery. The meta analysis reported in this article includes these 6 new studies in addition to 1 study10 not previously included by Marasco’s group. An important limitation of many previous studies in this area has been the varying use of categoric definitions of cognitive impairment, rather than using the actual continuous scores of psychometric tests that were measured. In clinical practice, defining a patient as cognitively impaired or not
KEY WORDS: cardiopulmonary bypass, coronary disease, coronary artery bypass, off-pump, cognition, coronary artery bypass grafting surgery, on-pump, psychometric tests
From the *The University of Edinburgh, College of Medicine and Veterinary Medicine, Edinburgh, UK; and yDepartment of Anaesthesia, Critical Care and Pain Medicine, Royal Infirmary of Edinburgh, Edinburgh, UK. Presented as a poster at the first Annual Conference of the National Student Association of Medical Research London, UK, February 4, 2012. Presented at the Joint Annual Meeting of the Association of Cardiothoracic Anaesthetists and the Society for Cardiothoracic Sur gery in Great Britain and Ireland Manchester, UK, April 18 20, 2012. Presented at the 27th Annual Meeting of the European Association of Cardiothoracic Anaesthesiologists Amsterdam, Holland, May 23 25, 2012. Address reprint requests to Ewan Douglas Kennedy BSc (Hons), The University of Edinburgh, College of Medicine and Veterinary Medicine, The Chancellor’s Building, 2nd Floor, 49 Little France Crescent, Edinburgh EH16 4SB, UK. E mail:
[email protected] & 2013 Elsevier Inc. All rights reserved. 1053 0770/2601 0001$36.00/0 http://dx.doi.org/10.1053/j.jvca.2012.11.008
Journal of Cardiothoracic and Vascular Anesthesia, Vol 27, No 2 (April), 2013: pp 253 265
253
254
KENNEDY ET AL
Table 1. MeSH Terms Used to Search for the Existing Literature Across the 4 Databases Medline and Embase
The Cochrane Library
1. cardiopulmonary bypass/ 2. (coronary adj3* bypass).twy 3. coronary artery bypass/ 4. coronary artery bypass, off pump/ 5. myocardial revascularization/
PsycINFO
1. cardiopulmonary bypass expz 2. coronary artery bypass, off pump exp 3. coronary artery bypass exp 4. myocardial revascularization exp
1. exp heart surgery/ 2. (coronary AND artery AND bypass).ti,aby 3. (cardiopulmonary AND bypass).ti,ab 4. (coronary AND artery AND bypass, AND off pump).ti,ab
AND 1. exp delirium, dementia, amnestic, cognitive disorders/ 2. delirium/ 3. exp dementia/ 4. cognition/ 5. 6. 7. 8. 9.
brain injuries/ exp neuropsychological tests/ (neurocognit J adj2z outcome ).tw (neurocognit adj2 impair ).tw (cognitive adj3 outcome ).tw
1. delirium exp
1. exp neuropsychological assessment/
2. dementia exp 3. amnesia exp 4. delirium, dementia, amnestic, cognitive disorders exp 5. neuropsychological tests exp 6. brain Injuries 7. cognition exp
2. exp brain damage/ 3. exp cognitive ability/ 4. exp cognitive impairment/ 5. 6. 7. 8. 9. 10.
exp performance tests/ exp memory/ exp cognitive processes exp brain damage/ exp brain/ (neurocognit adj2 impair ).ti,ab
Abbreviations: MeSH, medical subject headings. * three or fewer words adjacent to y search in title or abstract z explode all trees y search in title or abstract; J indicates wildcard search; z two or fewer words adjacent to.
For these reasons, the authors aimed to conduct a systematic review of the literature and perform a meta analysis of continuous measures of cognition to determine whether or not cognitive impairment was associated with the use of CPB for CABG surgery. METHODS
All published RCTs that compared cognitive outcome in on versus off pump CABG surgery were considered. A literature search was undertaken on December 1, 2011 across Medline, Embase, The Cochrane Library, and PsychINFO. The search was designed to be sensitive rather than specific, and included the medical subject headings listed in Table 1. Scottish Intercollegiate Guidelines Network devised search protocols and filters,15 specifically for RCTs, were applied to aid in accuracy and reproducibility of the Medline and Embase searches. The Cochrane Collaboration search webpage pro vided a filter for RCTs, as did the National Health Service Evidence website used for searching PsychINFO. In addition, reference lists of reviews were searched by hand to identify any relevant papers not found in the electronic search. Identified studies were imported into Mendeley Desktop reference management software (Version 1.1.2, 2011, Mendeley Ltd., London, UK) and any duplicate studies were merged. Identified studies were assessed for eligibility, first by title and abstract and then by more detailed review of the whole study. Studies were reviewed by all authors, and confirmed by 2 authors who checked for concurrence with inclusion and exclusion criteria. RCTs comparing on versus off pump
Fig 1. Summary of phases of the literature search strategy and applying inclusion and exclusion criteria. *Database search terms were used in this step.
No.
Year
1
2002
2
2002
3 4 5
Author
Van Dijk et al35 Zamvar et al36
Total Patient
Allocation
Intention to Treat
Concealed
Analysis?
(% of Group)
Cognitive testing
Computer generated block randomization/telephone
Yes
Yes (15 crossovers)
19 (14%) on pump 14 (10%) off pump
60
Cognitive testing
Computer generated/sealed envelope
Yes
No crossover
4 (29%) on pump
Sealed envelope
Yes
No crossover
17 (12%) on pump 3 (10%) off pump
Not described
Not clear
Number
281
Outcome Measures
Method of Randomization
2003
Lee et al37
60
Cognitive testing, whole brain SPECT*, transcranial Doppler to assess HITSy
2003
Rankin et al10
39
Cognitive testing
2005 Lund et al38
6
2006
7
2006
8
2007
9
2007
10 2008
Ernest et al39 Vedin et al40 Hernandez et al4 Yin et al5 Jensen et al6
11 2008 Tully et al7 Kozora et al9 Sousa Uva 13 2010 et al8 12 2010
120
Cognitive testing and cerebral MRI
Block randomization
Yes
107
Cognitive testing
Computer generated/sealed envelope
Yes
70
Cognitive testing
Not described
Not clear
201
Cognitive testing
Computer generated/sealed envelope
Yes
40
Cognitive testing, melatonin & cortisol levels
Yes
206
Cognitive testing
66
Cognitive testing and quality of life measures
2,203
Cognitive testing
Not described External press button telephone voice response system Random number generator/sealed envelope Central telephone voice response system
145
Cognitive testing, graft patency, quality of life measures
Computer generated/sealed envelope
Yes Yes Not clear Yes
Yes (1 crossover) Yes (7 crossovers) Yes (15 crossovers) Yes (3 crossovers) Yes (8 crossovers) No crossover Yes (5 crossovers) No crossover Yes (180 crossovers) Yes (3 crossovers)
Attrition Rate (No. of patients)
5 (28%) off pump 4 (13%) on pump 6 (10%) off pump 12 (14%) on pump 14 (23%) off pump 5 (14%) on pump 3 (9%) off pump
COGN T VE OUTCOME AFTER CORONARY ARTERY BYPASS GRAFT NG
Table 2. Randomized Controlled Trial Characteristics of the Included Studies
1 (1%) on pump None 16 (27.1%) on pump 14 (22.9%) off pump 4 (11%) on pump 2 (7%) off pump 518 (47%) on pump 529 (48%) off pump 26 (36%) on pump 32 (44%) off pump
* Single Photon Emission Computer Tomography; y High Intensity Transient Signals
255
Education, y
Previous stroke, %
Hypertension, %
Diabetes, %
Sex (ma e), %
* Not Reported Remarks y comb ned stroke and trans ent schem c attack; z on- and off-pump pat ents comb ned as reported n the or g na pub shed study
2010
62.2 61.7 99.5 99.3 39.1 44.1 84 85 7.1 7.6 NR NR
66.1 ⫾ 9.5 64.6 ⫾ 9.8 82.2 85.1 35.6 36.5 83.6 78.4 6.8 5.5 NR NR
2010 2008
75.5 ⫾ 5.1 75.0 ⫾ 4.2 62 60 17 19 68 58 19y 21y NR NR 63.6 ⫾ 10.0z 63.6 ⫾ 10.0z 81.8z 81.8z 13.3 22.2 56.7 58.3 NR NR 11.1 ⫾ 3.3z 11.1 ⫾ 3.3z
2008 2007
58.2 ⫾ 6.5 56.8 ⫾ 5.8 100 100 NR NR 25 20 NR NR NR NR NR NR 80.8 79.4 35.4 30.4 NR NR NR NR NR NR
2007 2006
65.0 ⫾ 9.1 65.0 ⫾ 9.1 78 84 18 19 52 46 3 0 13.3 12.7 63.2 ⫾ 9.0 63.7 ⫾ 10.7 78 81 27 31 79 79 7.1 3.6 11 11.9
2006 2005
64.8 ⫾ 7.8 65.2 ⫾ 8.4 85 72 NR NR 42 43 8.3y 6.6y 9.9 8.3 60.2 ⫾ 9.15 62.0 ⫾ 10.4 71 83 NR NR NR NR NR NR 13.2 ⫾ 1.6 13.0 ⫾ 3.4
2003 2003
65.5 ⫾ 9.6 66 ⫾ 11.2 80 73 20 37 70 87 7 3 11.7 ⫾ 3.3 13 ⫾ 2.1 63.5 ⫾ 9.1 61.6 ⫾ 10 83 90 NR* NR NR NR Exc uded Exc uded NR NR
2002 2002
61.7 ⫾ 9.2 60.8 ⫾ 8.8 66 71 9 17 40 44 4 3 9.3 ⫾ 2.4 9.7 ⫾ 2.8
Year
Off-pump On-pump Off-pump On-pump Off-pump On-pump Off-pump On-pump Off-pump On-pump Off-pump On-pump Age, y (mean ⫾ SD)
Kozora et a Jensen et a Tu y et a Yin et a Hernandez et a Vedin et a Ernest et a Lund et a Rankin et a Lee et a Zamvar et a Van Dijk et a Study
Table 3 Patient Demographics Reported in Included Studies
CABG surgery, in which patients underwent psychometric testing preoperatively and postoperatively, were included. The surgical procedure had to be isolated CABG surgery. All patient populations and all publication languages were con sidered eligible. Exclusion criteria were nonrandomization into on and off pump groups, assessment of brain injury by means other than psychometric tests, and duplicate publication of data. Authors were contacted by email if possible when clarification of their methodology was required, or when insufficient data were reported to calculate the mean and standard deviation of psychometric test scores. Continuous measures of psychometric tests were used. With permission, data also were taken from a previously published meta analysis by Marasco et al.3 Studies were withdrawn if discrepancies could not be resolved, or if the psychometric tests or time periods reported in a study were not those nominated for the meta analysis. The published information of all studies was evaluated to assess for quality and possibility of bias. Included studies were evaluated independently by 3 authors in accordance with the Cochrane Handbook.16 A core test battery recommended in the consensus statement of psychometric testing after cardiac surgery was used.17 Any other psychometric test used in 3 or more studies that produced an overall sample size larger than 500 was added to the meta analysis test battery. The RCTs included in this meta analysis typically conducted their own selection of psychometric tests. In these instances, the results of tests relevant to this meta analysis were extracted. Advice from psychologists was sought on merging test versions. Only tests that were highly correlated (r 4 0.6) and without significant between test difference in scores would be combined. Endpoints for the meta analysis were divided into early (r 3 months) and late (6 12 months) postoperative time periods to coincide with endpoints defined in the majority of included studies. The lack of consistent time points between the RCTs when psychometric testing was applied necessitated the adoption of 2 periods to allow amalgamation of the studies. The early period was chosen to identify short term cognitive impairment that might resolve and the late period to identify patients who might have permanent cognitive decline. If patients were tested more than once during the early period, results of the later test were used to minimize confounding effects of analgesia or sleep disturbance. To determine the extent of cognitive change from baseline, preoperative test scores were subtracted from postoperative test scores while accounting for a variation by a calculation of standard deviation.18 The meta analysis was conducted using RevMan 5.1 (Review Manager. Version 5.1. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2011) in accordance with the Cochrane Handbook.18 The mean values and standard deviations of test scores were analyzed as continuous variables using inverse variance. The fixed effects analysis model was used because each psychometric test measures a similar intervention effect, with variation being solely due to inter patient differences. Statistical significance was set at p o 0.05 and corresponding 95% confidence intervals were calculated.
KENNEDY ET AL
Sousa Uva et a
256
257
COGNITIVE OUTCOME AFTER CORONARY ARTERY BYPASS GRAFTING
The I2 statistic was derived to give an estimate of the degree of heterogeneity across studies not attributed to chance alone. An I2 value above 50% was considered to represent substantial heterogeneity18 requiring exploration. Additionally, publication bias was explored through the use of funnel plots. Post hoc sample size calculation was carried out.
However, none of the study biases was sufficient to warrant exclusion from the meta analysis. The included RCTs used variations of tests measuring the same cognitive domain, and most of these variations were different editions of the same test. Correlations were explored between the following tests to assess whether it would be valid to combine them for meta analysis.
RESULTS
The process of study selection is shown in Figure 1. Initially, 563 studies were identified from 4 databases, from which 534 were excluded on review of title and abstract. A further 9 studies19-27 were excluded on detailed evaluation of the full article. Twenty studies were initially appropriate for inclusion; however, 728-34 were later withdrawn, leaving 13 RCTs suitable for inclusion.4-10,35-40 The included RCTs are listed in Table 2. Study population demographics are listed in Table 3. The excluded studies and reasons for exclusion are listed in Table 4. All studies had a low risk of bias (Fig 2). Observation of funnel plots showed no evidence of publication bias. The studies by Kozora et al9 and Sousa Uva et al8 were noted for high attrition bias, as large proportions of patients were lost to follow up. Vedin et al40 conducted psychometric testing without investigators being blind to the treatment arm.
1. Digit Symbol subtests of versions R and III of the Wechsler Adult Intelligence Scale (r 0.82)41 2. Digit Span subtests of versions R and III of the Wechsler Adult Intelligence Scale and version III of the Wechsler Memory Scale (r 0.77)41 Seven psychometric tests were included for meta analysis, outlined in Table 5. Post hoc sample size analysis revealed that the size of each test subset was sufficiently large to prevent a type II statistical error 4 5%. Even the smallest subset (n 428) gave an attributable type II error chance of only 4.69%. Six of the 7 preoperative, psychometric tests (Fig 3) showed no significant difference between on and off pump groups, the only exception being the Stroop Color Word Test that favored off pump patients (p 0.04). This same test showed significant heterogeneity among the studies (I2 55%); otherwise all 6 other tests showed no significant heterogeneity.
Table 4. Studies That Were Excluded Based on the Eligibility Criteria and Appropriate Studies That Were Later Withdrawn With Reason for Exclusion Year
Randomized Study
Author
No. of patients
Primary Endpoints
Reason for exclusion
Excluded Studies 1 2000
19
Dieglar et al
Lloyd et al20 Wandschneider 3 2000 et al21 2 2000
Yes
40
S100b serum levels and cognitive testing
Yes
60
S100b serum level and cognitive testing
Cognitive tests reported not nominated for meta analysis Insufficient data provided
Yes
108
S100b serum levels
No cognitive testing
4 2001
Keizer et al22
Yes
5 2003
Lund et al23
Yes
52
6 2003
Schmitz et al24
No
251
7 2005
Kobayashi et al25
Yes
167
Yes
120
8 2005 Mathisen et al26
81
Patient reported cognitive decline and cognitive Cognitive tests reported not nominated for testing meta analysis Transcranial Doppler, Cerebral MRI, and cognitive Data reported in later paper testing Cognitive testing Not randomized 3 year cardiac events, completeness of revascularization, clinical outcomes, and cognitive No cognitive testing testing Preoperative cerebral MRI and patient reported No cognitive testing cognitive testing Time endpoints for cognitive outcome Cognitive testing and cerebral MRI were outside criteria for the meta analysis Withdrawn Studies Sample numbers overlap with study by Troponin T and cognitive testing Tully et al7 Contained data in previously published Cognitive testing paper 35 Graft patency and cognitive testing No preoperative baseline scores Cognitive testing Data reported in later paper6
9 2009
Puskas et al27
Yes
76
1 2001
Baker et al
28
Yes
26
Yes
281
3 2006 Al Ruzzeh et al 4 2006 Jensen et al31 Motallebzadeh 5 2007 et al 32
Yes Yes
168 206
Yes
212
Cognitive testing and cerebral emboli
Insufficient data provided
6 2007 Van Dijk et al33
Yes
240
Cognitive testing
Time endpoints for cognitive outcome were outside criteria for the meta analysis
Shroyer et al34
Yes
2,203
Mortality, complications, graft patency, and cognitive testing
Data reported in paper by Kozora et al9
2 2004 Van Dijk et al29 30
7 2009
258
KENNEDY ET AL
Table 5. Overview of Included Psychometric Tests, the Cognitive Domain Each Assesses, and the Number of Studies That Adopted Each Test Sample Size Test(s)
Measure
Cognitive
Unit
Domain
Studies Off-
On-
pump pump
1
RAVLT*
2
Grooved Pegboard
3
Trail Making A
4
Trail Making B
5 WAIS Ry Digit Symbol WAIS IIIz Digit Symbol 6 WAIS R Digit Span WAIS III Digit Span WMS IIIy Digit Span 7 Stroop Color Word Test
Total Verbal 6 score memory Time Motor capacity 8 taken Time Attention 11 taken Time Divided 10 taken attention Total Information 10 score processing Total Working 8 score memory Time taken
Executive function
5
363
350
474
465
990 1,000 970
980
1,015 1,016 901
905
295
291
Abbreviations: WAIS, Wechsler Adult Intelligence Scale; WMS, Wechsler Memory Scale * Rey Auditory Verbal Learning Test; y Wechsler Adult Intelligence Scale Revised Edition; z Wechsler Adult Intelligence Scale Third Edition; y Wechsler Memory Scale third edition
Color Word tests. In the late postoperative period, the scores of 6 tests significantly improved compared to preoperative values, the exception being the Grooved Pegboard test (Fig 11). DISCUSSION
Fig 2. Summary diagram of review authors’ judgments about Low risk of bias, each risk of bias item for each included study. high risk of bias, and unclear risk. (Color version of figure is available online.)
Results of all 7 postoperative psychometric tests (Figs 4 10) showed no significant differences between on or off pump groups in both the early and late postoperative time periods. Additionally, when the overall effect was vertically compared between the early and late time periods for each individual test, no significant differences were found for all 7 tests measured. Five out of the 7 postoperative psychometric tests showed no significant heterogeneity in either time period. Significant heterogeneity was noted in the early postoperative periods for the Trail Making B test (I2 = 51%, Fig 7) and the Digit Span test (I2 74%, Fig 9), but not for the late time periods. Because no significant differences were found between on and off pump, the groups were combined to assess the effect of CABG surgery on cognition. In the early postoperative period, there were significant improvements compared with preoperative scores (Fig 11) in the majority of psychometric tests. The exceptions were the Trail Making B, Digit Span, and Stroop
This systematic review and meta analysis revealed that following CABG surgery, there was no significant difference in change in cognitive outcome from preoperatively to post operatively, between on or off pump patient groups. Changes in all 7 psychometric tests that collectively assessed an array of cognitive domains showed no significant differences between patients undergoing CABG surgery on or off pump either in the early (r 3 months) or late (6 12 months) postoperative period. To the authors’ knowledge, this was the largest meta analysis to date investigating cognitive outcomes using con tinuous measures after on versus off pump CABG surgery, and the first to report any cognitive changes as a change in psychometric test scores from preoperative baseline. Further more, the statistically nonsignificant likelihood of a type II error occurring, revealed by post hoc sample size analysis, indicated that the results of this meta analysis were very unlikely to reflect a false negative result. Previously, Takagi et al undertook a systematic review and meta analysis42 using a dichotomized definition of cognitive decline. While they found off pump CABG surgery to be associated with less cognitive decline in the early postoperative period, they found no difference in cognitive outcome whether surgery was performed on or off pump in the late (6 12 month) postoperative period. Cheng et al43 also under took a meta analysis using a similar approach to Takagi et al, finding similar outcomes. Since completing the literature search for this study, Sun et al have published another meta analysis
COGNITIVE OUTCOME AFTER CORONARY ARTERY BYPASS GRAFTING
259
Fig 3. Preoperative baseline scores. Off-pump versus on-pump patient group comparison of preoperative psychometric scores. No Overall effect—mean bound by 95% confidence significant difference between groups in all but the Stroop Color Word Test (p ¼ 0.04). intervals.
using a dichotomous approach that found similar results.44 Subsequent to publication of Cheng et al’s study, Marasco et al3 rejected the dichotomous approach and undertook the first meta analysis to compare continuous measurements of psycho metric test scores up to one year postoperatively. They found no significant differences between patients undergoing CABG surgery on or off pump in either time period.
With respect to long term cognitive outcome, the results of this meta analysis concurred with all previous meta analyses in spite of the differing methods of analysis, psychometric tests, endpoints, and sample sizes. With respect to short term cogni tive outcome, the present meta analysis and a previous one by Marasco et al3 found no significant difference in cognitive outcome, whereas earlier meta analyses found that patients
Fig 4. Auditory Verbal Learning Test. Off-pump versus on-pump patient group comparison, of change in the early and late postoperative Individual period psychometric score, of the Auditory Verbal Learning Test. No significant differences between groups. overall effect—mean bound by 95% confidence intervals. (Color version of figure is study—mean and 95% confidence intervals, available online.)
260
KENNEDY ET AL
Fig 5. Grooved Pegboard Test. Off-pump versus on-pump patient group comparison, of change in the early and late postoperative period Individual study—mean and psychometric score, of the Grooved Pegboard Test. No significant differences between groups. overall effect—mean bound by 95% confidence intervals. (Color version of figure is available online.) 95% confidence intervals,
Fig 6. Trail-Making A Test. Off-pump versus on-pump patient group comparison, of change in the early and late postoperative period Individual study—mean and 95% psychometric score, of the Trail-Making A Test. No significant differences between groups. overall effect—mean bound by 95% confidence intervals. (Color version of figure is available online.) confidence intervals,
COGNITIVE OUTCOME AFTER CORONARY ARTERY BYPASS GRAFTING
261
Fig 7. Trail-Making B Test. Off-pump versus on-pump patient group comparison, of change in the early and late postoperative period Individual study—mean and 95% psychometric scores, of the Trail-Making B Test. No significant differences between groups. overall effect—mean bound by 95% confidence intervals. (Color version of figure is available online.) confidence intervals,
Fig 8. Digit Symbol Test. Off-pump versus on-pump patient group comparison, of change in the early and late postoperative period Individual study—mean and 95% psychometric scores, of the Digit Symbol Test. No significant differences between groups. overall effect—mean bound by 95% confidence intervals. (Color version of figure is available online.) confidence intervals,
262
KENNEDY ET AL
Fig 9. Digit Span Test. Off-pump versus on-pump patient group comparison, of change in the early and late postoperative period Individual study—mean and 95% psychometric scores, of the Digit Span Test. No significant differences between groups. overall effect—mean bound by 95% confidence intervals. (Color version of figure is available online.) confidence intervals,
having off pump CABG surgery had a better short term cognitive outcome. These differences in findings may reflect methodologic differences, exclusion of RCTs that used only dichotomous analysis and so were unable to provide this study with continuous data, and the use of continuous measures rather than dichotomized definitions of cognitive decline. Two studies reported insufficient data for this meta analysis, and the authors could not be contacted.30,32 Motallebzadeh et al’s RCT reported that initial differences in cognitive outcome between on and off pump groups did not persist
beyond 6 weeks, whereas Al Ruzzeh’s group found some improvement in the off pump patient group. It is unlikely that inclusion of either study strongly would have influenced the results of this meta analysis. For future reference, there is currently an RCT in progress45 that intends to recruit 4,700 patients and monitor outcomes, including cognitive assessment, over a postoperative period of 5 years. The quantity and sensitivity of psychometric tests are important in detecting cognitive decline. The battery of tests presented in this study assessed verbal memory, motor
Fig 10. Stroop Color Word Test. Off-pump versus on-pump patient group comparison, of change in the early and late postoperative period individual study—mean and psychometric scores, of the Stroop Color Word Test. No significant differences between groups. overall effect—mean bound by 95% confidence intervals. (Color version of figure is available online.) 95% confidence intervals
COGNITIVE OUTCOME AFTER CORONARY ARTERY BYPASS GRAFTING
263
Fig 11. Comparison of preoperative and postoperative psychometric test scores. Preoperative test scores versus early (r 3 months) and late (6-12 months) postoperative test scores. The scores of the Rey Auditory Verbal Learning Test and Digit Span and Digit Symbol tests, which used a positive scoring system, have been made negative to allow comparison with the rest of the psychometric tests where scoring Mean bound by 95% confidence intervals. was based on the subject’s timing.
capacity, attention, divided attention, information processing, working memory, and executive function, and they surpassed the recommendations from the statement of consensus and thus provided a valid and holistic assessment of cognition.17 Six out of 7 preoperative psychometric test scores showed no significant difference between on and off pump groups, suggest ing effective randomization, and both groups had comparable preoperative cognitive function. Preoperative scores of the Stroop Color Word Test favored the off pump group. Significant heterogeneity (I2 55%) was apparent, suggesting a discrepancy between studies. The results from the study by Yin et al5 were an apparent outlier. When removed, the overall effect lacked statistical significance, favoring neither on nor off pump, and lacked heterogeneity. This study was noted to be unclear in its randomization procedure and the measures to ensure allocation concealment. However, bias assessment failed to highlight a sufficient level of bias to warrant exclusion of this study from the meta analysis. Moreover, any effect of differences in scores between groups at baseline was minimized by using change in test scores from baseline for statistical analysis. On examining heterogeneity across the amalgamated studies, no significant heterogeneity was observed in the results of the late time period (6 12 months) for all 7 tests. The results of this time period were more reflective of permanent changes in cognition after transient disturbances after surgery have settled. Even in the early time period (r 3 months), 5 of the 7 psychometric test results showed no significant heterogeneity. The anomalies may be accounted for by a wide variation in exact testing time after surgery, varying from 4 10 days to 3 months. Notably, the outlying studies4,5,36 had psychometric testing conducted relatively early after surgery (r 1 week). Psycho metric testing at less than 6 weeks may not be meaningful because patients may not have recovered after surgery.17 Testing less than 1 week postoperatively may be subject to the effects of analgesia, pain, and sleep deprivation.46 Although these factors
may have accounted for heterogeneity in the 2 tests, they were unlikely to have impacted on the overall results. Studies often insufficiently clarified methods of randomiza tion and blinding that they had used, making bias assessment difficult. This raises concerns about possible bias when analyzing the results, particularly when they are unclear on fundamental methods, such as how patients were randomized to treatment groups. In the largest study, by Kozora et al, attempts were made to blind patients, but at follow up 20% of patients believed they knew their treatment group, 75% being correct.9 Moreover, the authors conceded difficulties in main taining patient blinding when a patient was transferred to another clinician’s care. Overall, although the flaws inherent in each study should be taken into account in interpretation of the results, none of the studies was assessed as having reached a bias level that warranted exclusion on these grounds. Attrition rates across all studies overall were unremarkable. The largest RCT, reported by Kozora et al,9 was found to have a high attrition rate (54% completed psychometric testing). Reasons for patient attrition were similar in both groups, and, consequently, limitations due to attrition may be considered minor. In addition, the number of patients who completed the study was still large (n 1,156), reducing the role of chance in influencing the conclusions drawn. Although the meta analysis was unable to detect any significant differences between groups, it was sensitive enough to detect significant postoperative improvements in 4 of the 7 psychometric tests in the early postoperative period and all but 1 of the tests in the late postoperative period. The most likely explanation for this was a learning effect47 whereby patients became more skilled at undertaking the psychometric tests, despite alternate forms of the tests being used, with repeated exposure. Potentially, these learning effects could have masked cognitive impairment associated with CPB. However, the authors examined postoperative changes in psychometric scores
264
KENNEDY ET AL
from preoperative values, and therefore their methodology would have detected any differences in improvement of test scores between the on and off pump groups that might have been caused by cognitive impairment. Identification of a learning effect in this study emphasized the need to include nonsurgical control groups in the design of future studies examining cognitive outcome after surgery to account for this effect. This systematic review and meta analysis cannot exclude the fact that the process of CABG surgery itself causes cognitive decline, as again no comparison was made with nonsurgical control groups. However, longer term studies suggest that cognitive decline may be secondary to progression of underlying cerebrovascular disease, hypertension, or other changes related to aging rather than CABG surgery48,49
In conclusion, systematic review and meta analysis of RCTs using continuous measures of psychometric tests found no significant difference in cognitive outcome between patients undergoing CABG surgery on and off pump, suggesting that CPB may not be associated with cognitive decline. ACKNOWLEDGMENTS The authors would like to thank Mrs Sheila Fisken for her assistance identifying database search terms. Many thanks to Professor Silvana Marasco, Dr Christian Lund, Professor Kjetil Sundet, Professor Jeremiah Brown, Professor Robert Baker, Dr Phil Tully, Dr Katherine Rankin and Dr Miguel Sousa Uva for providing clarification or data or both from their studies. This study was undertaken as part of a second year Student Selected Component at the University of Edinburgh College of Medicine and Veterinary Medicine supervised by Dr R. Peter Alston.
REFERENCES 1. Murkin JM, Boyd WD, Ganapathy S, et al: Beating heart surgery: Why expect less central nervous system morbidity? Ann Thorac Surg 68:1498 1501, 1999 2. Alston RP: Pumphead or not! Does avoiding cardiopulmonary bypass for coronary artery bypass surgery result in less brain damage? Br J Anaesth 94:699 701, 2005 3. Marasco SF, Sharwood LN, Abramson MJ: No improvement in neurocognitive outcomes after off pump versus on pump coronary revascularisation: A meta analysis. Eur J Cardiothorac Surg 33: 961 970, 2008 4. Hernandez F Jr, Brown JR, Likosky DS, et al: Neurocognitive outcomes of off pump versus on pump coronary artery bypass: a pro spective randomized controlled trial. Ann Thorac Surg 84:1897 1903, 2007 5. Yin YQ, Luo AL, Guo XY, et al: Postoperative neuropsy chological change and its underlying mechanism in patients under going coronary artery bypass grafting. Chin Med J 120:1951 1957, 2007 6. Jensen BØ, Rasmussen LS, Steinbru¨chel DA: Cognitive outcomes in elderly high risk patients 1 year after off pump versus on pump coronary artery bypass grafting. A randomized trial. Eur J Cardiothorac Surg 34:1016 1021, 2008 7. Tully PJ, Baker RA, Kneebone AC, et al: Neuropsychologic and quality of life outcomes after coronary artery bypass surgery with and without cardiopulmonary bypass: A prospective randomized trial. J Cardiothorac Vasc Anesth 22:515 521, 2008 8. Sousa Uva M, Cavaco S, Oliveira AG, et al: Early graft patency after off pump and on pump coronary bypass surgery: A prospective randomized study. Eur Heart J 31:2492 2499, 2010 9. Kozora E, Kongs S, Collins JF, et al: Cognitive outcomes after on versus off pump coronary artery bypass surgery. Ann Thoracic Surg 90:1134 1141, 2010 10. Rankin KP, Kochamba GS, Boone KB, et al: Presurgical cognitive deficits in patients receiving coronary artery bypass graft surgery. J Int Neuropsychol Soc 9:913 924, 2003 11. Altman DG, Royston P: The cost of dichotomising continuous variables. Br Med J 332:1080, 2006 12. MacCallum RC, Zhang S, Preacher KJ, et al: On the practice of dichotomization of quantitative variables. Psychol Methods 7: 19 40, 2002 13. Abu Omar Y, Taggart DP: Pitfalls of neuropsychometric assess ment and alternative investigative approaches, in Bonser RS, Pagano D, Haverich A (eds): Brain Protection in Cardiac Surgery. London, Springer, 2010, pp. 57 64 14. Mahanna EP, Blumenthal JA, White WD, et al: Defining neuropsychological dysfunction after coronary artery bypass grafting. Ann Thorac Surg 61:1342 1347, 1996
15. Scottish Intercollegiate Guidelines Network (SIGN): Search filters, 2009. Available at: http://www.sign.ac.uk/methodology/filters. html. Accessed May 15, 2012. 16. Higgins JPT, Altman DG, Sterne JAC: Assessing risk of bias in included studies, in Higgins JPT, Green S (eds). Cochrane Handbook for Systematic Reviews of Interventions 5.1.0. Chichester, John Wiley & Sons Ltd, 2008 17. Murkin JM, Newman SP, Stump DA, et al: Statement of consensus on assessment of neurobehavioral outcomes after cardiac surgery. Ann Thorac Surg 59:1289 1295, 1995 18. Deeks JJ, Higgins JPT, Altman DG: Analysing data and under taking meta analyses, in Higgins JPT, Green S (eds). Cochrane Hand book for Systemic Reviews of Interventions 5.1.0. Chichester, John Wiley & Sons Ltd, 2008 19. Diegeler A, Hirsch R, Schneider F, et al: Neuromonitoring and neurocognitive outcome in off pump versus conventional coronary bypass operation. Ann Thorac Surg 69:1162 1166, 2000 20. Lloyd CT, Ascione R, Underwood MJ, et al: Serum S 100 protein release and neuropsychologic outcome during coronary revas cularization on the beating heart: A prospective randomized study. J Thorac Cardiovasc Surg 119:148 154, 2000 21. Wandschneider W, Thalmann M, Trampitsch E, et al: Off pump coronary bypass operations significantly reduce S100 release: An indicator for less cerebral damage? Ann Thorac Surg 70:1577 1579, 2000 22. Keizer AM, Hijman R, van Dijk D, et al: Cognitive self assessment one year after on pump and off pump coronary artery bypass grafting. Ann Thorac Surg 75:835 838, 2003 23. Lund C, Hol PK, Lundblad R, et al: Comparison of cerebral embolization during off pump and on pump coronary artery bypass surgery. Ann Thorac Surg 76:765 770, 2003 24. Schmitz C, Weinreich S, Schneider R, et al: Off Pump versus on pump coronary artery bypass: Can OPCAB reduce neurologic injury? Heart Surg Forum 6:127 130, 2003 25. Kobayashi J, Tashiro T, Ochi M, et al: Early outcome of a randomized comparison of off pump and on pump multiple arterial coronary revascularization. Circulation 112:1338 1343, 2005 26. Mathisen L, Andersen MH, Hol PK, et al: Preoperative cerebral ischemic lesions predict physical health status after on pump coro nary artery bypass surgery. J Thorac Cardiovasc Surg 130:1691 1697, 2005 27. Puskas JD, Stringer A, Hwang SN, et al: Neurocognitive and neuroanatomic changes after off pump versus on pump coronary artery bypass grafting: Long term follow up of a randomized trial. J Thorac Cardiovasc Surg 141:1116 1127, 2011
COGNITIVE OUTCOME AFTER CORONARY ARTERY BYPASS GRAFTING
28. Baker RA, Andrew MJ, Ross IK, et al: The Octopus II stabilizing system: Biochemical and neuropsychological outcomes in coronary artery bypass surgery. Heart Surg Forum 4:S19-S23, 2001 29. Van Dijk D, Moons KG, Keizer AM, et al: Association between early and three month cognitive outcome after off-pump and on-pump coronary bypass surgery. Heart 90:431-434, 2004 30. Al-Ruzzeh S, George S, Bustami M, et al: Effect of off-pump coronary artery bypass surgery on clinical, angiographic, neurocognitive, and quality of life outcomes: Randomised controlled trial. Br Med J 322:1365, 2006 31. Jensen BØ, Hughes P, Rasmussen LS, et al: Cognitive outcomes in elderly high-risk patients after off-pump versus conventional coronary artery bypass grafting: A randomized trial. Circulation 113: 2790-2795, 2006 32. Motallebzadeh R, Bland JM, Markus HS, et al: Neurocognitive function and cerebral emboli: Randomized study of on-pump versus off-pump coronary artery bypass surgery. Ann Thorac Surg 83: 475-482, 2007 33. Van Dijk D, Spoor M, Hijman R, et al: Cognitive and cardiac outcomes 5 years after off-pump vs on-pump coronary artery bypass graft surgery. J Am Med Assoc 297:701-708, 2007 34. Shroyer AL, Grover FL, Hattler B, et al: On-pump versus off-pump coronary-artery bypass surgery. N Engl J Med 361:1827-1837, 2009 35. Van Dijk D, Jansen EW, Hijman R, et al: Cognitive outcome after off-pump and on-pump coronary artery bypass graft surgery: A randomized trial. J Am Med Assoc 287:1405-1412, 2002 36. Zamvar V, Williams D, Hall J, et al: Assessment of neurocognitive impairment after off-pump and on-pump techniques for coronary artery bypass graft surgery: Prospective randomised controlled trial. Br Med J 325:1268, 2002 37. Lee JD, Lee SJ, Tsushima WT, et al: Benefits of off-pump bypass on neurologic and clinical morbidity: A prospective randomized trial. Ann Thorac Surg 76:18-25, 2003 38. Lund C, Sundet K, Tennøe B, et al: Cerebral ischemic injury and cognitive impairment after off-pump and on-pump coronary artery bypass grafting surgery. Ann Thorac Surg 80:2126-2131, 2005
265
39. Ernest CS, Worcester MU, Tatoulis J, et al: Neurocognitive outcomes in off-pump versus on-pump bypass surgery: A randomized controlled trial. Ann Thorac Surg 81:2105-2114, 2006 40. Vedin J, Nyman H, Ericsson A, et al: Cognitive function after on or off pump coronary artery bypass grafting. Eur J Cardiothorac Surg 30:305-310, 2006 41. Kaufman AS, Lichtenberger EO: Assessing Adolescent and Adult Intelligence (ed 3). New Jersey, John Wiley & Sons, 2006 42. Takagi H, Tanabashi T, Kawai N, et al: Cognitive decline after off-pump versus on-pump coronary artery bypass graft surgery: Metaanalysis of randomized controlled trials. J Thorac Cardiovasc Surg 134: 512-513, 2007 43. Cheng DC, Bainbridge D, Martin JE, et al: Does offpump coronary artery bypass reduce mortality, morbidity, and resource utilization when compared with conventional coronary artery bypass? A meta-analysis of randomized trials. Anesthesiology 102: 188-203, 2005 44. Sun HJ, Wu XY, Wang WJ, et al: Cognitive dysfunction after off-pump versus on-pump coronary artery bypass surgery: A metaanalysis. J Int Med Res 40:852-858, 2012 45. Lamy A, Devereaux PJ, Prabhakaran D, et al: Rationale and design of the coronary artery bypass grafting surgery off or on pump revascularization study: A large international randomized trial in cardiac surgery. Am Heart J 163:1-6, 2012 46. Alston RP: On-pump and off-pump coronary artery surgery and the brain, in Bonser RS, Pagano D, Haverich A (eds): Brain Protection in Cardiac Surgery. London, Springer, 2010, pp. 191-205 47. Benedict RH, Zgaljardic DJ: Practice effects during repeated administrations of memory test with and without alternate forms. J Clin Exp Neuropsycho 20:339-352, 1998 48. Selnes OA, Royall RM, Grega MA, et al: Cognitive changes 5 years after coronary artery bypass grafting: Is there evidence of late decline? Arch Neurol 58:598-604, 2001 49. Hlatky MA, Bacon C, Boothroyd D, et al: Cognitive function 5 years after randomization to coronary angioplasty or coronary artery bypass graft surgery. Circulation 96:11-15, 1997