Introduction
The term postoperative cognitive dysfunction (POCD) has been used to describe an objectively measurable decline in cognition at some time point following anesthesia and surgery. The classification of POCD has been determined by purely objective criteria as assessed by performance on neuropsychological tests administered before and after surgery, and not by signs or symptoms. The presence of a substantial decline in an individual’s performance, relative to performance of the whole group or the change seen in nonsurgical individuals (controls), has been used to imply that POCD is present. POCD has been the subject of considerable investigation, not least because POCD affects the elderly, who are increasingly the demographic presenting for surgery and anesthesia, the numbers of which are projected to increase greatly with the world’s aging population, and because POCD is associated with adverse consequences in the short term, which include increased length of hospital stay and increased mortality. POCD contrasts with cognitive decline in the elderly assessed by other medical disciplines, because of the association with a specific intervention (anesthesia and surgery), and the lack of consideration of magnitude, duration, or a subjective complaint. POCD is undergoing a reassessment to now align with the DSM-5 cognitive disorders to resolve these shortcomings, and as articulated elsewhere in this book, has a new nomenclature. Nevertheless, this chapter deals primarily with the objective component of the diagnosis, which has been intimately associated with the term POCD and will not be substantially different for the different forms of perioperative neurocognitive disorders (PND). Thus the term POCD will continue to be used in this chapter, largely to provide the literature context, while PND will be added as we consider future studies.
Assessing POCD requires a preoperative (baseline) level of cognition to be measured before surgery and then retested at one or more specific time intervals following anesthesia and surgery. Studies of postoperative cognitive change have followed individuals at a variety of time points including 1 week, 6 weeks, 3 months, 12 months, and long term (5–7.5 years). Postoperative testing times appear to have been chosen for convenience rather than based on any knowledge of the postoperative cognitive course. That is, around the time of discharge (4–7 days postoperatively); 6 weeks or 3 months, which possibly coincided with surgical follow-up appointments and was long enough to assume the direct effects of drugs, hospitalization, and surgery had worn off; and 12 months in longer-term studies. There are a small number of studies investigating POCD as far as 5 years and 7.5 years postoperatively, but these have several limitations including large attrition of patient numbers.
Studies of POCD/PND utilize neuropsychological tests grouped together into a test battery. In early studies of POCD the composition of these test batteries varied enormously between investigators. These differences in test batteries, in addition to variability in follow-up assessment time points, made comparison of results between studies and centers difficult. The tests were also administered in a variety of surroundings (rather than a reproducible, quiet, and controlled environment) and analyzed by a variety of statistical methods and criteria, creating further discrepancies between studies. As a response to the diversity in methodology which had emerged, a meeting of experts was convened in 1994 to discuss the issues involved. The result was a Statement of Consensus recommending a core battery of neuropsychological tests which satisfied specific criteria and outlined the manner in which they should be administered (1). The test battery incorporated cognitive tests to assess patients over time and determine change in cognition. It should be noted that these paper-and-pencil neuropsychological tests were not specifically designed for multiple administrations, thus practice effects, time effects, and possibly other unknown effects were inevitably incorporated into the assessment of cognitive change. Importantly, anesthesiologists have relied on neuropsychologists for guidance in measuring this decline in cognition by applying appropriate test batteries and analyses to study designs.
To date there remains a large diversity in the tests used, and more particularly, diversity in the methods of analysis. Additionally, the issue has been complicated not only by variability in the neuropsychological assessments for identification of POCD but also by published research using inappropriate tools such as the Mini-Mental State Examination (MMSE) (2) and the Montreal Cognitive Assessment (MoCA) (3) for attribution of POCD. The MMSE and MoCA are insensitive to subtle impairments in cognition and while useful tools for screening cognition, they are not for the assignment of a clinical diagnosis of cognitive impairment, nor for the classification of POCD/PND. More recently, computerized batteries have been used and have potential to expedite assessment of PND (4).
Cognitive Tests
Neuropsychologists have been primarily concerned with the assessment of brain injury in their clinical environment with the use of comprehensive testing of patients under controlled conditions. This contrasts sharply with the assessment of cognitive change for which anesthesiologists in particular have led research. Neuropsychologists were interested in identifying the cognitive domain affected, and in doing this, identifying the area of the brain which had incurred damage. This type of testing took many hours to perform, included a large battery of tests over multiple cognitive domains, and represented a single point in time for an individual which was subsequently compared to corresponding population normative data. In contrast, in the perioperative environment, there was a requirement to undertake testing which was relatively quick to administer and which could be undertaken in less than ideal environmental conditions. Additionally, anesthesiologists were interested in identifying change in an individual from a preoperative level rather than static performance compared to population data. Neuropsychological tests utilized in the assessment of POCD have been designed to detect subtle impairment in cognitive function which is often unidentified by the individual. As mentioned above, some studies have attempted to assess POCD using the MMSE and the MoCA, although due to the insensitive nature of these assessments such work should be interpreted with caution.
More recently computerized cognitive assessment batteries have been developed which avoid practice effects, offer easier administration, are quicker, standardized, and have the potential to overcome cultural and language difficulties. To date, however, computerized test batteries have received limited attention in POCD/PND research mainly due to a lack of relevant validation studies.
The use of a control group for comparison allows investigators to correct for expected practice effects, the effects of time, and unknown effects associated with repeated testing by comparing a patient group undergoing surgery and anesthesia with a group of similar individuals tested at similar times who do not undergo surgery and anesthesia. Without the use of appropriate control groups these issues remain a significant source of confounding in the assessment of POCD/PND.
Issues Associated with Repeated Testing
Identifying cognitive change that is meaningful requires serial neuropsychological testing which introduces a number of challenges. Sources of error in assessing change include variables associated with the particular neuropsychological test, variables associated specifically with the change situation, and finally variances associated with each individual patient. Variables associated with the test include reliability, practice effects, and floor/ceiling effects. Situational factors refer to the test-retest interval and the statistical phenomenon of regression to the mean. Patient factors involve demographics including age, gender, culture and education, and the particular clinical condition the patient presents with. In the surgical environment this includes anxiety and mood especially when preoperative testing is undertaken immediately prior to surgery, and postoperative factors such as medications, pain, sleep disturbance, stress, and fatigue.
Several strategies can be employed to improve the reliability of measures of change, including parallel test versions and control groups. Parallel versions are available for a number of neuropsychological tests which may be advantageous for serial testing; however, such versions do not eliminate practice effects entirely. An example of parallel versions is the use of alternate word lists for the Auditory Verbal Learning Test for which six alternate word lists have been validated against the original Rey Auditory Verbal Learning Test (RAVLT) (5). The use of a suitable control group can significantly improve the validity of the study by adjusting for changes in test performance over similar time periods. The major difficulty is choosing an appropriately matched control group. On the one hand it may be desirable to identify how the intervention group compares with a closely matched control group with similar medical conditions; however, there may be ethical considerations associated with not intervening in the control group or it may be that the outcome is associated with the medical condition, in which case it may be more relevant to compare against healthy controls. This issue should be addressed for each investigation. There is little evidence that practice effects can be eliminated by altering the test-retest interval. Indeed, practice effects have been observed across periods as long as 2.5 years (6).
Thus, the assessment of cognitive change is subject to a number of factors beyond performance on individual tests, some of which can be addressed and some which are difficult to control for. Reliable change is a concept which refers to identification of a true or “reliable” change. Several statistical methods have been developed for assessing reliable change which will be discussed further below. It should also be noted that the evidence addressing errors associated with repeated testing has predominantly been confined to adults aged 18–60 years (7). Thus, there may be further sources of error associated with repeated cognitive testing in older subjects which are as yet unknown. To some extent this can be allowed for by including an appropriate control group, as detailed above, but this remains to be elucidated.
Types of Cognitive Tests
A summary of many of the neuropsychological tests employed for assessment of POCD is shown in Table 11.1. Well-designed studies investigating POCD have used a combination of these tests in order to incorporate a broad range of cognitive domains. According to the 1995 consensus statement these tests should be selected according to
1 the cognitive domain of the test (ensuring the battery of tests covers different domains)
2 the sensitivity of the test (will the test scores be a valid estimate of true function?)
3 the time required to perform the test (lengthy batteries will be difficult logistically)
4 practice effects of the test
5 consideration of whether parallel versions of the test are available
6 assurance that the battery of tests is balanced across cognitive domains
The tests presented in Table 11.1 cover the broad domains of cognitive function, including executive function, memory, attention, visuospatial, psychomotor, and language. Some studies have utilized alternative tests which have been developed using the principles of these tests, for example the concept shifting test is based on the Trail-Making Test.
Table 11.1 Examples of Cognitive Tests Used in the Perioperative Setting
| Test | Neuropsychological Domain | Psychometric Properties |
|---|---|---|
| Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) Auditory Verbal Learninga | Memory | Small range of scores Limited alternate forms Variable test-retest reliability |
| Digit Symbol Substitution Test | Executive | Large range of scores Normally distributed Limited alternate forms Good test-retest reliability |
| Trail-Making Test Part Aa | Executive, attention | Large range of scores Normally distributed No alternate forms Poor test-retest reliability |
| Trail-Making Test Part Ba | Executive, attention | Large range of scores Normally distributed No alternate forms Moderate test-retest reliability |
| Controlled Oral Word Association Test (COWAT) Verbal Fluency | Language | Moderate range of scores Normally distributed Limited alternate versions Good test-retest reliability |
| Semantic Fluency (e.g., Animals) | Language | Moderate range of scores Normally distributed Limited alternate versions Good test-retest reliability |
| Grooved Pegboarda | Psychomotor | Moderate range of scores Normally distributed No alternate version Good test-retest reliability |
| Stroop | Executive | Various score ranges Normally distributed One version, various scoring methods Good test-retest reliability |
| Mini-Mental State Examination (MMSE) | Multiple | Screening test Not sensitive to subtle changes |
| Montreal Cognitive Assessment (MoCA) | Multiple | Screening test Not sensitive to subtle changes |
a One of the four tests from the recommended core neuropsychological battery by Murkin et al. (1)
Related posts:
Chapter 1 – Emergence Delirium
Chapter 2 – Postoperative Delirium
Chapter 4 – Postoperative Cognitive Improvement
Chapter 10 – Comorbidities and Postoperative Neurocognitive Disorder
Chapter 12 – Biomarkers of Postoperative Cognitive Dysfunction: Finding the Signal amid the Noise
Chapter 14 – Preoperative Testing to Identify Vulnerable Subgroups
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
