Diagnosis of Postoperative Cognitive Decline

POCD has been diagnosed with several different approaches, all of which rely on arbitrary statistical thresholds rather than reproducible clinical diagnostic criteria. The most stringent criterion for the diagnosis of POCD that is commonly used is a decline of at least two standard deviations (2 SD) in two cognitive domains or a decline of at least 2 SD in a composite cognitive score. A liberal criterion that has been proposed for POCD diagnosis and has been used in several prominent studies is a decline in at least 1 SD in any cognitive domain or in a composite cognitive score. This “1-SD” technique has been criticized as failing to account for factors that may confound interpretation of serially acquired cognitive test scores, including regression to the mean, measurement error caused by poor test–retest reliability, and practice effects. With this liberal 1-SD diagnostic approach, the probability of detecting POCD purely by chance in just one of four domains, which is the diagnostic criterion used in one prominent study, would be about 33%.

To take into account the learning that occurs with repeated psychometric testing, a correction factor (based on the mean learning divided by the standard deviation of learning in a control population) was subtracted from the follow-up score in the relevant cognitive domain or in the composite cognitive score in several studies. This approach to adjust for learning based on (average) improvement in a control group is termed the reliable change index and is based on several assumptions: (1) that control subjects who are not undergoing surgery learn no more efficiently than patients facing the prospect of surgery, (2) that the control subjects are well-matched with those undergoing surgery, and (3) that it is appropriate to correct for an individual’s learning based on the average learning of a group. A study by Evered and colleagues suggests that these assumptions might not be valid. This study included four groups, two surgical (cardiac and orthopedic surgery) groups and two nonsurgical control groups. One of the control groups was undergoing coronary angiography and the other control group was not undergoing any procedure. Learning in the nonprocedural control group was measured so that a reliable change index could be calculated and applied to the other three groups. When the three procedural groups were evaluated for cognitive decline at 3 months, the group that underwent coronary angiography (with no surgery and no general anesthesia) had the highest incidence of cognitive decline, after learning was corrected for with the nonprocedural control group’s reliable change index. Perhaps it is not surprising that patients who are undergoing either surgical or nonsurgical procedures do not learn as efficiently as control subjects who are not distracted by the prospect of a procedure. Alternative statistical approaches, like mixed effects models, have been used in studies of POCD and are probably more robust than methods that rely on correction for learning based on a nonprocedural control group.

Interestingly, most studies that have followed up postoperative cognition have ignored the fact that there are patients who appear to improve cognitively, just as there are patients who appear to decline. This apparent cognitive improvement might represent artifact, or it might reflect a genuine phenomenon. It has been demonstrated that neuroplasticity occurs throughout life. Pain and inflammation carry a cognitive burden, and successful elective surgery might result in alleviation of pain and resolution of inflammation. Functional recovery is also possible, with resultant enhancement in quality of life and physical fitness. Taken together, these factors could be associated with cognitive improvement.

Delirium

Delirium is a well-recognized state of acute confusion in the elderly; it is described in the DSM-IV classification and has been assigned an ICD-10 code. Delirium is an acute and fluctuating disorder of arousal, attention, and logical thinking. In the nonsurgical setting delirium has been found to occur more commonly in patients with mild cognitive impairment or early dementia and is associated with clinical deterioration and an increased mortality rate. Postoperative delirium is common (10% to 70% incidence) among elderly patients older than 65 years in the early postsurgical period and typically resolves within the first 1 to 2 postoperative weeks. Risk factors for delirium include baseline cognitive impairment and age. An association between postoperative delirium and increased mortality rates has been shown, but a link between postoperative delirium, POCD, and incident dementia has not been definitively established, although the evidence is mounting. A study published in the New England Journal of Medicine found that patients who had delirium after cardiac surgery were more likely than those who did not have delirium to have lower mini-mental status evaluation scores (compared with their baseline scores) 1 year postoperatively. Whether prevention of postoperative delirium is possible and could prevent this cognitive decline is currently unknown.

Dementia

Unlike delirium and POCD, dementia is thought to be an irreversible, degenerative loss of brain function that occurs with various disorders (e.g., Alzheimer disease, vascular dementia, Lewy body disease, and Huntington disease). The symptoms of dementia include impairments in cognition, especially memory, personality changes, depression, impaired judgment, sleep disturbances, decreased ability to perform daily activities, and, ultimately, inability to recognize loved ones and to function even at a basic level. No conclusive association has been found between POCD and incident dementia. However, epidemiologic research has shown that patients with repeated hospital admissions are more likely to become demented. There is also a suspicion that specific general anesthetic agents, such as isoflurane, might initiate pathologic processes (e.g., the generation in the brain of beta amyloid proteins or phosphorylated tau), which could initiate or accelerate the development of dementia. Surgery might promote neuroinflammation, which could also theoretically increase susceptibility to dementia. However, a potential causal association between surgery or anesthesia and subsequent incident dementia is purely speculative. Although it has been reported that surgery increases the risk of subsequent dementia, the majority of studies that have explored this hypothesized link have been negative.

Uncontrolled Studies

Uncontrolled observational trials have suggested that approximately half of patients undergoing cardiac surgery or major noncardiac surgery have persistent cognitive decline. One of these studies focused on cardiac surgery patients and was published in the New England Journal of Medicine in 2001. This study showed that 41% of patients who underwent cardiac surgery had persistent cognitive decline 5 years postoperatively. This study had a major impact in the medical community and on public opinion and reinforced the perspective that cognitive decline is a major complication of cardiac surgery, potentially attributable to cardiopulmonary bypass. The concern about brain damage associated with cardiopulmonary bypass was a major stimulus for the advent of off-pump cardiac surgery. Another influential study published in Anesthesiology showed that 46% of older patients had persistent POCD 1 year after major noncardiac surgery. The main limitations of these studies have been the lack of appropriate controls and the use of the liberal approach to diagnose POCD (a decline by more than 1 SD in any cognitive domain), which, purely by statistical chance, would be likely to detect a high incidence rate of POCD.

Serial Assessments

Even with a more rigorous diagnostic approach, the methodologic obstacle to reliably diagnosing POCD is reflected in studies that have assessed patients at serial time points. These studies have generally reported poor intrapatient reproducibility in the diagnosis of POCD. For example, in one study, the patients given diagnoses of POCD at 3 months postoperatively had very poor overlap with the patients who were given diagnoses of POCD at 2 years postoperatively.

Control Subjects

Studies that have included control subjects, including the seminal and influential International Study of POCD (ISPOCD), have generally found that POCD appears to resolve with time. The ISPOCD was established as an international research consortium in 1994. This group was founded on the basis that POCD occurred commonly in elderly patients and frequently persisted. Members of the ISPOCD group suggested that POCD after cardiac surgery was a recognized complication, which was probably attributable to cardiopulmonary bypass. Their major purpose was to characterize POCD after noncardiac surgery ( www.sps.ele.tue.nl/ispocd/sub0/main.html ). The main goals of ISPOCD-1 were to determine whether POCD occurred after noncardiac surgery with general anesthesia and to test the hypothesis that intraoperative hypotension and hypoxemia contributed to POCD. The resulting study was published by the ISPOCD group in 1998 in The Lancet and showed that 26% of patients older than 60 years had POCD at 1 week and 10% had POCD at 3 months postoperatively. While age and educational level were found to be risk factors for POCD, counter to the investigators’ hypothesis, hypotension and hypoxemia did not appear to be associated with POCD. A relationship was noted between POCD and impaired functionality as reflected by decrements in Instrumental Activities of Daily Living scores. Two studies that have included control groups have found that POCD might persist up to 1 year postoperatively. One of these studies was hard to interpret; there appeared to be persistent cognitive decline in the visuospatial domain but lasting improvement in language. A study by Ballard and colleagues, in which 256 subjects were assessed at 1 year (roughly balanced between surgical patients and nonsurgical community age-matched control subjects), found that, according to a global composite cognitive score, 11.8% of mostly orthopedic surgical patients experienced cognitive decline 1 year postoperatively compared with only 3.8% of the nonsurgical control patients. In this study, impairments in attention and executive function were particularly noticeable. These are striking results, but their validity rests on the assumption that the control subjects were appropriately matched for the surgical patients and that both groups would learn (or improve on the cognitive test battery) as efficiently.

Cardiac Surgery

Recent evidence from cardiac surgery studies has challenged the broadly accepted perspective that persistent and severe POCD is common, especially when there is a period of cardiopulmonary bypass. Using an elegant research design, Selnes and colleagues followed up four age- and education-matched cohorts. The first had coronary artery disease and underwent cardiac surgery, the second had coronary artery disease and had percutaneous coronary intervention, the third had coronary artery disease and was treated medically, and the fourth did not have heart disease. Their findings were surprising. The three cohorts with coronary artery disease all declined cognitively over 6 years, whereas the cohort without heart disease did not decline. This study suggested that specific comorbidities, like vascular disease, are likely to be much more potent drivers of cognitive decline than cardiac surgery or general anesthesia. In an article published in the New England Journal of Medicine, Selnes and colleagues commented, “it is now increasingly apparent that the incidence of both short- and long-term cognitive decline after CABG has been greatly overestimated, owing to the lack of a uniform definition of what constitutes cognitive decline, the use of inappropriate statistical methods, and a lack of control groups.” They also proposed that “Most patients in whom new cognitive symptoms develop during the immediate postoperative period can be reassured that these symptoms generally resolve within 1 to 3 months.”

Randomized Trials

In the last 15 years, major studies have randomly assigned patients with coronary artery disease to receive either surgery or percutaneous coronary intervention. These trials have provided an important opportunity to judge whether cardiac surgery and general anesthesia are really potent independent agents of cognitive decline and decrements in quality of life. The trials have not demonstrated that patients randomly assigned to surgery had worse cognitive outcomes, and generally, quality of life was improved whether patients underwent surgical treatment or percutaneous coronary intervention. Taken together, the evidence suggests that persistent POCD is not a common phenomenon, and surgery and anesthesia are, at worst, very minor culprits in relation to lasting cognitive decline.

General Anesthesia

One approach to teasing out the relative contribution of general anesthesia to POCD is to randomly assign surgical patients to general or regional anesthesia and track postoperative cognition in both groups. Randomized trials that have followed this approach have usually not found that regional anesthesia was associated with a decrease in persistent POCD. A meta-analysis of 21 trials published in the Journal of Alzheimer’s Disease showed that general anesthesia was marginally but nonsignificantly associated with POCD (odds ratio, 1.34; 95% confidence interval, 0.93 to 1.95). If, despite the current negative evidence, general anesthesia does independently contribute to POCD, it is likely that its contribution is minor.

Cardiopulmonary Bypass

Recent rigorously conducted randomized controlled trials have been instrumental in dispelling the popular myth that cardiopulmonary bypass is a major independent cause of cognitive decline. The Octopus trial randomly assigned 281 patients to cardiac surgery with or without the use of cardiopulmonary bypass. Both 1-year and 5-year cognitive outcomes have been published for this trial in the Journal of the American Medical Association. At 5 years, the investigators found that about one third of patients in both the on-pump and the off-pump groups had cognitive decline. The 2200-patient ROOBY trial, published in the New England Journal of Medicine, also randomly assigned largely male patients to cardiac surgery with or without cardiopulmonary bypass. Patients underwent baseline and follow-up neuropsychological tests that were designed to evaluate dysfunction in attention, memory, and visuospatial skills. Similar to the Octopus trial and against prevailing views, no difference in cognitive outcomes was found between groups. Perhaps even more intriguing was that, with comprehensive follow-up of about 1150 patients at 1 year postoperatively, the long-term postoperative changes in individual neuropsychological test scores were similar to or improved from baseline for both treatment groups.

Genetic Risk Factors

It has been hypothesized that genetic risk factors for POCD would probably overlap with those for neurodegenerative disorders, such as Alzheimer disease. The epsilon4 allele of the apolipoprotein E gene is a known risk factor for Alzheimer disease, poor outcome after cerebral injury, and accelerated cognitive decline with normal aging. No association has been demonstrated between the apolipoprotein E genotype and POCD. It remains possible that some people have a genetic predisposition for POCD. Because no agreed-on diagnostic criteria exist for POCD, detecting an association between candidate genotypes and the phenotype (i.e., POCD) is a major challenge.