Durable, functionally impactful, and subjectively evident cognitive decline is a feared outcome of anesthesia and surgery, particularly for older adults (Br J Anaesth 2017;118:486-8). We need to look no further than Dr. Daniel Cole, a coauthor of this article and the editor for this column, to find someone impacted by permanent cognitive decline after major surgery. He has written here, and in a Washington Post article, about his father’s experience after coronary artery bypass grafting (CABG). Better handling of the discussion of outcomes like his father’s – unanticipated permanent cognitive decline that produces adverse functional impacts perceived by the patient and/or their loved ones – is, and must be, a focus of our care.
Three main streams of evidence inform our approach (Figure 1) to long-term postoperative neurocognitive disorder (PND), formerly termed postoperative cognitive dysfunction/decline (POCD).
Most anesthesiologists are familiar with primary cohort studies of older patients who undergo detailed cognitive testing in a research setting before, and then at intervals after, surgery. These find that perhaps 10%-25% of older adults meet neuropsychological test criteria – that is, a research definition – for PND/POCD, typically in the short term (<1 year). That definition has only recently expanded to include subjective impact, which is crucially important, because deficits on neuropsychological tests do not correspond well with functional impact (Br J Anaesth 2018;121:1005-12). Alignment between the subjective experience of cognitive decline and research definitions for POCD has historically been poor; most patients whose test performance declines will experience no subjective impact (Br J Anaesth 2019;122:742-50). We are still waiting to fully understand how the recent redefinition of PND will impact our understanding.
One missed opportunity is that these definitions have been applied one-sidedly; we have not found people who experience postoperative cognitive improvement (POCI) because we have not looked. But direct and indirect benefits of surgery – decreased pain, functional improvement, enhancement in quality of life and mental health – are largely unstudied potential factors that plausibly could improve the preoperatively established rate of cognitive decline (Figure 2). It would be a patient-centered outcome to measure POCI as well; decline is only one “side” of the story, and identifying subjects who meet POCI criteria could be done (even retrospectively) without additional data collection in primary cohort studies of PND.
Anecdote is one powerful stream of evidence that must be reconciled. The role of anecdote in our patients’ understanding of POCD/PND is poorly described but is probably highly impactful. Popular news articles on this topic, such as the one featuring Dr. Cole’s father’s experience, often lead with an anecdote of permanent, functionally impactful POCD. In another popular media article published recently in The Guardian, over 80 comments were submitted, including nearly 40 unique anecdotes of POCD (Br J Anaesth 2023;130:567-72). These anecdotes often described permanent cognitive change as well as dissatisfaction with – indeed, abandonment by – the medical establishment, which failed to give warning about POCD symptoms and its management. Anecdotes of POCD were offered across the spectrum of adult ages, surgeries, and preexisting cognition. We are very limited in our understanding of anecdotal cognitive change; there is, for example, no data on how frequently a patient may experience cognitive decline that is both subjectively impactful and permanent after surgery. We have no grasp of risk factors for anecdotal cognitive change. Cognitive change reaching the level of subjective anecdote, while clearly an important patient-centered outcome, must presently be regarded as idiosyncratic, absent better study.
A third stream we must also incorporate is evidence from epidemiological studies. These differ from primary cohort studies in that they can capture long-term preoperative and postoperative cognitive trajectory (i.e., rate of decline) of individuals. Tracking cognitive function several months to years before and after the inciting event of surgery, to capture the trend in cognitive changes over time, may be more informative about risks of permanent, unanticipated cognitive change (as Dr. Cole’s father experienced) than a cognitive measurement at one point in time before surgery, as in primary cohort studies. Across a variety of populations and settings, a constellation of epidemiological studies demonstrates either no or only subtle differences in cognition after major surgical exposures, once the rate of cognitive decline before surgery is taken into account. Even in highly provocative settings – for example, CABG surgery – there is no detectable difference in long-term cognitive outcomes compared with nonsurgical strategies for treating similar conditions (i.e., percutaneous coronary intervention [PCI]) (JAMA 2021;325:1955-64). This implies that, for most patients who undergo CABG, cognitive outcomes are no different than what would have been expected for their preoperative rate of cognitive decline. Still, patients with adverse cognitive outcomes can be identified in epidemiological data sources after CABG and after PCI (JAMA 2021;325:1955-64). Indeed, a medical hospital admission is associated with significantly poorer average cognitive outcomes compared with elective surgery (BMJ 2019;366:l4466). It may be that any serious health perturbation, whether elective or unplanned, has the potential to result in serious cognitive change. Elective surgery, with its attendant opportunities to offer optimization and carefully controlled risk factors, may be inherently less provocative.
Combining these three streams of evidence, we may conclude that anecdotal – subjectively impactful, durable – cognitive change clearly occurs after major surgery and anesthesia, but that it is probably too rare to impact population-level outcomes and that most people, after elective major surgery, will be cognitively unaffected once they have recovered from the acute impact of surgery (e.g., beyond a month or two after surgery). Primary cohort studies using detailed neuropsychological tests have identified that most deficits consistent with POCD/PND are in the domains of memory and executive function. This may help guide our patients about what POCD/PND may feel like and when to seek help. Looking for patients who meet criteria for cognitive improvement, as well as for decline, would help provide a more balanced picture of plausible cognitive test outcomes when research definitions are used. POCI is mathematically necessary for POCD (and anecdotal adverse cognitive outcomes) to exist, while population-average cognitive outcomes, as shown in epidemiological studies, remain unchanged. Note that, in Figure 2, the median outcome is 0 – that is, cognitive performance exactly as expected, as observed in epidemiological studies – but that in order for some participants to decline (POCD), some must improve (POCI) to nonetheless yield a median of 0.
A balanced approach would support patients in understanding that the individual effect of surgery on long-term cognition could be favorable (POCI), unfavorable (POCD), or null. Here is where the need for personalized and patient-centered risk stratification and tailored patient consultation and care is particularly acute. A recognition of the role of anecdote on the patient-centered understanding of POCD, while not being equivalent to research-defined PND, is critical to communicate. Ultimately, older patients do not pursue surgery with a primary interest in cognitive outcomes – instead they are seeking functional improvement, relief of pain, or perhaps extension of life as their primary goal. Cognitive risks must be balanced against those anticipated benefits. But it is plausible that for some patients, those improvements may translate into cognitive benefits as well, or – most probably – cognitive recovery to the expected outcome of their own, individually established cognitive trajectory.