Depression and delirium: association, prediction, causation, and care

Depression and delirium are common and serious problems for patients having surgery. The prevalence of postoperative delirium is nearly 20% after non-cardiac surgery [1] and 4–14% of the broader surgical population present to surgery with moderate or severe symptoms of depression [2]. Depression and delirium are particularly salient problems in high-risk groups such as older adults or those undergoing major complex surgical procedures, a population at risk of suboptimal surgical outcomes and recovery [3, 4]. Screening for depression in peri-operative settings has gained more focus as surgical patients and clinicians begin to recognise the importance of proactively identifying and tackling barriers to recovery from surgery, one of which relates to the question of how and when mental and cognitive health interventions may be most efficacious in the peri-operative period [5]. Management of postoperative delirium is a foundational component of enhanced recovery pathways for older patients, and clinical risk prediction and stratification leading to risk-reducing strategies have been incorporated successfully in peri-operative care [6, 7]. These progressive steps align well with a paradigm change in the notion of peri-operative care. All clinicians are invited to reconsider the peri-operative period as a window in which to optimise patient health conditions, targeting personalised delivery of interventions that achieve comprehensive goals beyond simply survival and avoidance of complications [8, 9]. To this end, Diep et al. offer concrete evidence that should motivate peri-operative care providers to consider the relevance of depression to postoperative neurocognitive recovery [10]. By performing a meta-analysis comprising nearly 5 million patients from five continents, they present pragmatic data on depression prevalence and its unadjusted association with postoperative delirium. The authors offer the clinically important conclusion that those patients with a depression diagnosis and/or clinically significant depression symptoms are at double the risk of postoperative delirium compared with those without depression. These findings indicate that, at a minimum, depression and delirium are common among surgical patients, and that these two constellations of symptoms are somehow related. In interpreting the article, however, it behooves us to consider carefully whether, and how, we should act upon these findings. Considering their potentially debilitating role in surgical recovery, including patient-centred outcomes like function and cognition, depression and delirium both deserve extra attention in peri-operative care. If depressive symptoms identify a surgical population with increased susceptibility to delirium, then establishing a new diagnosis of depression, considering it a marker of elevated future risk, and possibly treating the depression in order to prevent delirium and its various adverse sequelae, could be a compelling option. However, evidence around traditional medications for the management of depression, including antidepressants and mood stabilisers, has been collected in patients with existing neurocognitive disorders (i.e. dementia and delirium), with evidence of mixed or unpromising effectiveness, or even harm [11-14]. When pain and depression coexist, as may be common in a surgical population, adequate analgesia may theoretically improve mood symptoms; however, apart from one trial investigating patients suffering from dementia [15], evidence for this is lacking. In peri-operative settings, evidence is even more scant, [16] emphasising that it is a relatively novel concept to treat depression as a potentially modifiable risk factor for adverse postoperative outcomes. One exception is in the intra-operative use of sub-anaesthetic doses of the dissociative hypnotic ketamine or its pharmacologically active enantiomer, esketamine, as a means to prevent and mitigate mood problems. A meta-analysis of small trials suggested efficacy [17] but a larger multicentre trial showed no benefit [18]. Emerging protocols for depression treatment targeting behavioural therapy and medication optimisation are in progress [19]. Thus, it is not yet known whether postoperative depression is preventable, or whether there are subpopulations of patients whose depression is amenable to peri-operative treatment in order to improve postoperative mental health outcomes. Similar to depression, efforts to prevent postoperative delirium have been hampered by inefficacy. What makes the problem of delirium particularly intractable is that delirium may be regarded as a syndrome rather than a single diagnostic entity; it is often multifactorial in that predisposing and precipitating factors combine (in addition to the probable existence of a number of confounding factors) making it difficult to know which single factor (or combination of factors) is directly causal [20]. Thus, a causal association between depression and delirium may be present only in combinations of precipitating factors for the delirium ‘syndrome’ and not in others. There are many potentially complex causal and non-causal relationships among factors involved in the development of both delirium and depression (Fig. 1). Of course, it would be helpful to find a causal relationship: that is, that depression somehow causes delirium, either directly or through the initiation of factors called ‘mediators’ which are intermediate on the causal pathway (e.g. medication use). Figure 1 shows a theoretical example of how medication use may be a mediator: where medication use prompted by depression, but not the depression itself, directly affects delirium risk. If this model were correct, treating depression by nonpharmacological means, e.g. cognitive behavioural therapy, would not inherently prevent delirium. Several highly plausible confounding factors also exist, where common underlying pathophysiological conditions increase the risk of both depression and postoperative delirium. For example, depression is sometimes a preclinical harbinger of Alzheimer's disease [21]. Underlying vascular pathology resulting in neural disconnection, inflammation, and hypoperfusion, may yield a ‘vascular depression’ phenotype that plausibly influences the development of both cognitive dysfunction and depression [22, 23] (Fig. 1). If this hypothesis explains the association of depression and postoperative delirium, then treating symptoms of depression would not mitigate the risk of postoperative delirium. These causal relationships may not matter, if the desire is to predict postoperative delirium, rather than treating or preventing it. If prediction is the goal, working out precise theoretically motivated causal pathways is unnecessary, since important predictors can be causal, confounders, mediators, or anything else prompting a mathematical association. Think of the prototypical association between carrying a cigarette lighter and lung cancer. Cigarette lighters are predictive of lung cancer because they are highly associated with tobacco smoking, but not causal. This is easy to imagine for the problem of depression and delirium; there are extensive conceivable, yet tangled, connections among pain, depression, comorbidities, frailty and neurocognitive disorders (Fig. 1). To elaborate on one area, pain may be an underlying source of both depression and delirium that is easily modifiable with interventions [24], or may be a marker of a vulnerable brain, where future neurocognitive disorder risk is agnostic to the presence, or successful treatment, of pain. If prediction of delirium is the goal, it is (somewhat) irrelevant whether pain causes depression, depression causes pain, or one or both causes delirium; the only thing that matters is that depression and delirium coexist in a ‘vulnerable brain’ phenotype. Diep et al. indirectly acknowledge these problems by using unadjusted or crude odds and risk ratios in characterising the association between depression and delirium [10]; that is, the ‘independent’ association between depression and delirium, with potential sources of confounding and mediation mathematically removed on theoretical grounds, was not sought. This approach allows the authors to conclude only that depression is a reproducible indicator of postoperative delirium risk, not why or how to mitigate it [10, 25, 26]; this is still an important conclusion. For example, it offers justification that peri-operative care providers might add depression to the growing list of risk factors for adverse postoperative events, along with baseline cognition, frailty and other comorbidities or vulnerabilities. Depression surveillance may be particularly germane for older patients, who make up a large proportion of the surgical population and who have a higher baseline risk of both depression and delirium. At a minimum, and regardless of the causal significance of the association, depression may be a predictor of delirium (an important and potentially preventable clinical outcome) and offers an opportunity for diagnosis and peri-operative initiation of treatment, potentially improving quality of life for a patient beyond postoperative hospital discharge. This perspective reflects the continued evolution of peri-operative care with the focus expanded beyond merely survival and recovery of patients, to include the identification and triggering of interventions targeted towards maximising postoperative quality of life. The peri-operative period has become an invaluable teaching moment for patients to optimise their health conditions in multiple aspects through prehabilitation [27] and other targeted interventions. Smoking cessation is a prototypical example of how patients and physicians can form a therapeutic alliance to take advantage of surgery to implement long-term health promotion [28]. This can yield short-term improvements in surgical outcomes (which, admittedly, remain speculative with depression at present) and longer-term benefits from recognising the health harms of poor behaviours and untreated comorbid disease. Similar logic can apply to mental health as long as peri-operative physicians are willing to leverage this opportunity to bend the course of a vulnerable patient's trajectory towards optimal physical and mental health. While the work by Diep et al. cannot offer evidence of causality, their analysis definitively establishes an association between depression and delirium which crosses continents, ages, and types of surgery [10]. Prospects for the development of successful peri-operative mental health interventions deployed are hopeful, showing feasibility in combined pharmacological and behavioural programs to address depression [29]. A critical missing puzzle piece will be evidence of the efficacy and safety of those interventions, as well as other experimental approaches in large trials. In the interim, the meta-analysis by Diep et al. [10] provides the impetus to define, using rigorous methodology, how the reproducible association between depression and delirium may be leveraged to improve peri-operative care. No competing interests declared.