The Devil Is in the Details: Comparison of Postoperative Delirium and Neurocognitive Dysfunction

Evaluating cognitive change within individuals (and in groups) has bedeviled neuropsychologists, biostatisticians and clinicians for ages. In this issue of Anesthesiology, a study by Daiello et al.¹ illustrates quite well some of the difficulties inherent in how we define and assess “clinically meaningful” postoperative cognitive outcomes and the practical impact of those difficulties can have on determining the association between postoperative delirium and postoperative cognitive dysfunction (POCD) – two conditions frequently associated with heightened risk for mortality within 1 year after surgery, longer-term cognitive decline and dementia.^2–4

In their examination of in-hospital postoperative delirium and POCD at 1-, 2- and 6-months following non-cardiac surgery in SAGES Trial participants,⁵ Daiello et al.¹ report a weak association between in-hospital postoperative delirium and the risk of lingering POCD at 2-months recovery (odds ratio = 1.30, p=0.062) and a lack of association at 6-months recovery (odds ratio = 1.07, p=0.744). Their results suggest that the risk of significant POCD is really highest at 1-month in patients with in-hospital delirium (relative risk = 1.34, p=0.010). Even then, only a small proportion of patients with both conditions at 1-month were found in their large surgery sample (75 of 548 patients; 14%) compared to the total patients with POCD at 1-month (256 of 548 patients; 47%). More patients developed POCD during the early stages of recovery without a history of in-hospital postoperative delirium. Although 12.5% of patients with POCD at 1-month were still evincing moderate to severe cognitive decline at 6 months of recovery, a history of postoperative delirirum was not predictive of persistent cognitive deficits. These findings demonstrate that postoperative delirium is poorly predictive of POCD following 1-month of recovery and beyond, suggesting that postoperative delirium and POCD are separate clinical conditions, rather than shared conditions reflecting two sides of the same coin.⁶

One fundamental conundrum is determining how we define POCD, as differing definitions of POCD can have notable impact on the POCD frequency in a given study. For instance, when Daiello et al.¹ defined POCD as a mild rather than moderate change in cognition from presurgical baseline, the incidence of POCD at 1-month recovery rose from 47% to 60%. These cognitive change considerations, while seemingly arcane, are salient to clinicians’ assessment of patient risk and long-term outcomes and patients’ understanding of their condition and expectations for recovery. Importantly, a newly proposed perioperative neurocognitive disorders diagnostic nomenclature⁷ has started a robust debate about what to call POCD, yet significant work remains on how to best measure, discriminate and define the clinically meaningful cognitive change suggested for mild and major perioperative neurocognitive disorder diagnoses.

The lingering uncertainty in how we characterize the cognitive change associated with POCD is an uncomfortable truth often lost in translation and buried in study methods. Is POCD reflective of global cognitive impairment; more selective to particular cognitive domains (e.g., memory, attention, etc.); or could it be both? It seems sensible to capture both possibilities, but doing so allows for greater heterogeneity in cognitive performance patterns that may, in turn, complicate our efforts to determine if there are critical cognitive deficits that bridge between delirium and POCD. We know that many of the same pre-surgical risk factors for delirium are shared with POCD, particularly those involving frontal/executive functioning,⁸ and it has also been observed that functional declines are greatest in patients with POCD characterized by selective memory and executive dysfunction.⁹ Why then would we expect to observe a global decline in cognitive performance, which is often derived from averaging or summing individual test variables in a battery? Doing so may only serve to dilute and obscure potentially more sensitive indicators of POCD, which could include deficits like those observed in delirium (e.g., inattention). Alternatively, maybe we should consider the possibility that POCD exists on a spectrum and therefore prioritize quantifying cognition change as a continuous variable? Whatever side of the debate on which we land, we should seek to optimize POCD detection for what is unusual for the patient. Depending upon an individual’s age, presurgical cognitive abilities, occupational complexity, and levels of psychosocial support, even a mild cognitive decline from presurgical baseline could have severe functional consequences.

The debate about how to best capture and characterize cognitive change is significant because the clinical impact of postoperative delirium and POCD can be devastating, with both conditions associated with worse cognitive decline 3 to 5 years later,¹⁰ decreased quality of life, and increased 1-yr postoperative mortality risk.² For any patient, this could range from not being able to complete a work project on time, to loss of independent living, or even death 1-yr after surgery/anesthesia. Although the impact of delirium and POCD is significant, we still lack a complete understanding of their pathophysiologic underpinnings; thus, it is unclear whether delirium and POCD truly represent distinct conditions. If delirium and POCD are two separate conditions (but with similar predisposing factors), as the findings of Daiello et al.¹ suggest, one could argue that two clinical strategies should be employed to prevent delirium and POCD. First, potentially modifiable predisposing factors should be identified and addressed before surgery. Known predictors of delirium and POCD such as age and years of education are virtually immutable, but other predictors such as frailty, depression, obstructive sleep apnea, and presurgical cognition may be modifiable to a degree through treatment or “prehabilitation.” We hope that continued research on these potentially modifiable factors will guide preoperative optimization strategies to prevent postoperative delirium and POCD. Second, distinct postoperative strategies to reduce delirium and POCD should be employed. For example, validated strategies to reduce in-hospital delirium, such as medication avoidance (e.g., benzodiazepines, anticholinergics, meperidine), encouraging sleep hygiene, performing early mobilization, preventing sensory deprivation and encouraging family presence might not be effective in preventing POCD later in the postoperative period. Instead, perhaps strategies to reduce POCD severity or duration should focus on cognitive stimulation therapies and aerobic exercise interventions, which have the potential to reduce cognitive decline.

Overall, the study by Daiello et al.¹ reveals the difficulties inherent in ascertaining what is “meaningful” postoperative cognitive change and remind us that often the devil is in the details. The study also suggests that delirium and POCD are distinct clinical entities, for which differential clinical management will be needed to guide anesthesiologists as we move forward to optimize perioperative brain health.