Postoperative cognitive dysfunction (POCD) is a deterioration of intellectual function presenting as impaired memory or concentration. According to the North American Diagnostic and Statistical Manual of mental disorders, 4th edn , cognitive disorders are divided into several conditions that are characterized by clinically significant deficits in cognition and memory, representing a significant change from a previous level of functioning:
- Delirium is an acute disturbance of consciousness and a change in cognition that tends to fluctuate during the course of the day.
- Dementia is characterized by multiple cognitive deficits including memory impairment. The condition causes impairment in occupational or social functioning.
- Amnestic disorders are characterized by memory impairment in the absence of other significant cognitive impairments.
- Mild neurocognitive disorder is one of the cognitive disorders not otherwise specified and for which research criteria have been suggested. This presentation is characterized by cognitive dysfunction presumed to be caused by either a general medical condition or substance use that does not meet the criteria for any of the other disorders. An essential feature is that the impairment in cognitive functioning is evidenced by neuropsychological testing.
In the vast majority of patients with POCD, it must be regarded as a mild neurocognitive disorder and, accordingly, it requires neuropsychological testing for detection. Neuropsychological testing evaluates several aspects of cerebral function, such as problem solving, speed of information processing, flexibility and memory . In surgical patients, several problems connected with the use of neuropsychological tests have to be anticipated:
Selection of neuropsychological tests
Most neuropsychological tests have been developed in English-speaking countries, and they may not be very useful in other cultures. Specific language problems and even possible illiteracy must be taken into consideration when word lists or letters are used. Sensory deficits may seriously impair the performance because of problems with instructions and their execution. The usual tests for the assessment of psychomotor speed are dependent on normal hand function. This may be a problem for patients undergoing orthopaedic surgery.
A suitable test is characterized by a documented high sensitivity. This means that test results should reflect the effects of well-known factors with impact on cerebral functions such as medication and age. Other tests may be selected for specific use in a particular setting, for example tests assessing delayed recall, as this aspect of cerebral function has been found to be sensitive to brain damage after cardiac arrest or climbing at extreme altitude.
Some of the most used test batteries are the Wechsler Adult Intelligence Scale or the Wechsler Memory Scale. They have not been developed for repeated administration in surgical patients in order to detect subtle cognitive deterioration. In fact, the sensitivity in this respect seems to be quite low, and this probably explains why so many studies of POCD using these batteries have failed in detecting cognitive decline post-operatively or detecting differences between study groups .
Neuropsychological test results are subject to a large variability. This is related to the differences between individual subjects, the differences between test administrators and also the variability in performance of the individual subjects from one session to another. Patients are influenced by numerous factors, of which only some can be controlled. This variability should be minimized in studies of POCD in order to separate normal performance clearly from a deterioration. First, the testing conditions should be identical at each session: this means that the same tester is administering the material, the same room is used, preferably at the same time of the day. External distraction must be avoided, and this may be a problem when testing is carried out on the ward. Test results obtained in that setting should be interpreted with caution, as a very large variability must be expected. In the early post-operative phase, pain, sleep deprivation and the residual effects of analgesics or hypnotics can affect the performance.
'Bottom/ceiling effects' and the definition of change in performance
The results of neuropsychological testing can be reported as whether a deterioration has occurred or has not occurred (±deficit). When this principle is used, the pre-operative score will have a substantial influence on the probability of a patient having a deficit. Test results expressed as scores have a lower and upper limit, and if a score should deteriorate by 10 points, only patients obtaining at least a score of 10 points pre-operatively can have deficits. This is the so-called 'bottom effect', which is apparent in several studies.
The 'ceiling effect' is a problem associated with very easy tasks. If a patient obtains the maximum score pre-operatively, it will be difficult to assess changes in performance on another occasion. If the same score is obtained, changes in either direction may be unnoticed. If a small deterioration is registered, the size of this may be underestimated.
A practice effect can be found for most neuropsychological tests for volunteers as well as for patients. Failure to take the practice effect into consideration tends to underestimate the incidence and severity of POCD. The impact of the learning effect is dependent on the performance level of the patient and also on the particular test. Generally, less able patients gain most from practice. This tends to level out the change in performance for the patients and decrease the possibility of detecting a harmful effect of anaesthesia.
Intervals between test sessions and drop-out
The interval between the operation and the postoperative test can be chosen in advance but, unfortunately, no general consensus exists concerning the length of this interval. It has been realized that testing in the early phase 3-14 days postoperatively probably reflects other problems than follow-up after 2-3 months  but, in fact, the time of postoperative testing is mostly dictated by local practice, such as the usual time of discharge or the follow-up visit at the surgeon's clinic. Patients with postoperative complications, fatigue or depression are most likely to refuse or delay the postoperative test. Absence of postoperative neuropsychological test (drop-out), constitutes a substantial methodological problem, because POCD may be much more common in patients unable or unwilling to undergo testing at the time of usual discharge.
Each test in a battery can be considered independently by:
- comparing before and after anaesthesia for one group;
- comparing differences between pre-operative and postoperative tests (postoperative minus preoperative) between patient groups;
- comparing only postoperative test results between two or more groups.
These statistical tests each carry a risk (typically 2.5% in a particular direction) of finding a statistically significant difference by chance. A statistically significant change in this context means that the difference between the pre-operative and the postoperative test result exceeds approximately twice the standard deviation of the mean values (standard error of the mean). Multiple comparisons may be handled by using Bonferroni's correction, which is conservative however. For the statistical analysis, the number of outcome measures should be limited, and this can be accomplished by, for instance, using a composite measure.
The definition of POCD based on neuropsychological test results
Neuropsychological tests are quantitative variables and can be used as such in the analyses. If an incidence of POCD is to be calculated, it is necessary to define diagnostic criteria based on the test battery used. This is usually done by considering deficits in the single tests. Several definitions of POCD have been used, as shown in Table 1.
In many studies, a postoperative neuropsychological deficit in a test has been defined as a deterioration of one standard deviation compared with the pre-operative test results (from now on, this principle is called 'the 1 SD criterion'. This SD is calculated from the preoperative test session of all patients and, in fact, reflects the variability of performance between subjects in a population. This value is greater than the standard deviation in a normative population and will differ between studies. This makes it very difficult to compare the reported incidences of POCD.
'The 1 SD criterion' seems to have been introduced on the basis of the experience of clinical neuropsychology in which no baseline performance is known. Instead, population-based estimates are used to achieve mean values and variability for neuropsychological tests. In surgical patients, we are focusing on minor changes in cognitive function in individual subjects. Accordingly, the statistical analysis is applied to changes in performance. In the calculation of deficits, learning effects should be considered, and the estimate of variability should reflect variability in individual subjects. Collection of data from a normative population undergoing repeated testing is the only way to fulfil these requirements . The '1 SD criterion' is associated with a considerable risk of finding deficits after repeated neuropsychological testing just by chance, and the risk is increased significantly when a large test battery is used. Maybe as a consequence of this, the definition of POCD has been based on the occurrence of at least two deficits in neuropsychological tests. But there is still no agreement concerning the number of tests to be considered.
The results of different neuropsychological tests can be combined and expressed as a composite score containing results from all or only some of the tests in the battery. A neuropsychological test may assess a particular function or even a specific brain area, and this information may be lost in a composite score. However, until now, in studies of surgical patients, the aim of using neuropsychological tests has been to evaluate cognitive function more generally and not specific functions. The statistical analysis and final interpretation of results is much simpler when composite scores are used.
The postoperative cognitive function is difficult to evaluate. In the literature, very large differences are apparent in methodology, such as the test batteries, the interval between sessions, the endpoints to be analysed, how the analysis should be done statistically and how neuropsychological deficits and POCD should be defined.
Traditionally, intelligence tests or tests developed for clinical neuropsychology have been used. More appropriately, selection of the tests for detecting POCD should be based on well-described sensitivity and suitability in relation to surgical patients. In tests using scores, bottom/ceiling effects may compromise the evaluation if the tests are either too easy or too difficult. Uncontrolled testing conditions may increase the variability between sessions in the individual subjects. Practice effects are pronounced in neuropsychological tests but have generally been ignored. Only with the use of suitable normative material is it possible to correct for practice effects and variability between sessions. In the analysis of the test results, the evaluation of cognitive function should be based on differences between pre- and postoperative performance. The test parameters considered from the individual tests should reflect cognitive function in a normative material, and the number of parameters should be limited in order to control the number of statistical tests. The definition of cognitive dysfunction should be restrictive, and the criteria should be fulfilled in only a small proportion of volunteers undergoing repeated testing.
1 American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders,
4th edn (DSM-IV). International version. Washington DC: American Psychiatric Association, 1995.
2 Lezak MD. Neuropsychological Assessment.
New York: Oxford University Press, 1995.
3 Moller JT, Svennild I, Johannesen NW et al.
Perioperative monitoring with pulse oximetry and late postoperative cognitive dysfunction. Br J Anaesth
4 Murkin JM, Newman SP, Stump DA, Blumenthal JA. Statement of consensus on assessment of neurobehavioral outcomes after cardiac surgery. Ann Thorac Surg
5 Savageau JA, Stanton B, Jenkins CD, Klein MD. Neuropsychological dysfunction following elective cardiac operation. Early assessment. J Thorac Cardiovasc Surg
6 Shaw PJ, Bates D, Cartlidge NEF et al.
Early intellectual dysfunction following coronary bypass surgery. O J Med
7 Treasure T, Smith PLC, Newman S et al.
Impairment of cerebral function following cardiac and other major surgery. Eur J Cardiothorac Surg
8 Pugsley W, Klinger L, Paschalis C et al.
The impact of microemboli during cardiopulmonary bypass on neuropsychological functioning. Stroke
9 Hammeke TA, Hastings JE. Neuropsychologic alterations after cardiac operation. J Thorac Cardiovasc Surg
10 Patel RL, Turtle MR, Chambers DJ et al.
Alpha-stat acid-base regulation during cardiopulmonary bypass improves neuropsychological outcome in patients undergoing coronary artery bypass grafting. J Thorac Cardiovasc Surg
11 Stump DA, Newman SP, Coker LH, Hilbawi MS, Rogers AT. Age is a risk factor for acute but not persistent neuropsychological deficits after cardiac surgery. Anesthesiology
12 Heyer EJ, Adams DC, Todd GJ et al.
Neuropsychometric changes in patients having surgery for carotid endarterectomy. Anesthesiology
13 Stump DA, Newman SP, Coker LH, Phipps JM, Miller CC. Persistence of neuropsychological deficits following CABG. Anesthesiology
14 Moller JT, Cluitmans P, Rasmussen LS et al.
for the IS-POCD investigators. Long-term postoperative cognitive dysfunction in the elderly: ISPOCD1 study. Lancet