Should we think before we treat delirium?

In a systematic review in this issue, Weber et al.1 have drawn together several studies, backed up by many others,2−8 to show that delirium is common and associated with a variety of adverse outcomes, including a higher risk of death,4,8 dementia,4,5,8 instit­utionalisation2,3,6−8 and prolonged hospital stays.2,8 They show, as have others,9−11 that delirium is often treatable10 and preventable.11 Still, delirium remains under-recognised by many physicians and other health-care professionals, existing information to prevent and treat delirium remains woefully under-used and a critical review of delirium has concluded that delirium research largely has stalled.12 Why should this be? At its most basic level, the knowledge to be used seems simple enough. Successful treatment of delirium requires specific treatment of the underlying precipitant and non-specific management of the accompanying behavioural disturbances. This is perhaps more than a convenient fiction, but is itself inadequate. It is a long distance from a scientifically sensible account, a pragmatic plan of action or a pedagogically principled agenda. The problem facing anyone wishing to undertake a systematic review of interventions in delirium is not so much the lack of methodologically sound studies − although Weber et al. show that this is a real problem − but that such studies often have difficulty in getting to grips with what we would argue are more fundamental obstacles to the better care of patients with delirium. The first is its very low level of recognition. Even at a hospital devoted to studies of delirium, with intake most often in the Emergency Room, ER physicians failed to detect delirium in two-thirds of the patients so affected.13 Although it might be argued that this is quite an improvement over the usual standard (approximately five of six patients go undetected),14 it still represents a pragmatic challenge. Why is recognition so poor? How can recognition be improved? Will recognition lead to better outcomes? The Montréal experience again offers some hints. In a study in which patients with delirium were detected systematically and randomised to usual versus consultant care, each group had similar outcomes − approximately half the patients improved, as Weber et al. have noted. The rate of recognition of delirium in the usual care group, however, was higher than normal (27 vs 10−15%).15 Also working against the intervention's effectiveness was that many patients (18%) in the usual care group received specific consultations for delirium from a geriatrician or geriatric psychiatrist. What can we learn from this well-executed but apparently ineffective intervention trial? One methodological consequence of under-recognition is that some thought must be given to the unit of randomisation. Rather than simply accepting that randomisation at the level of the patient accords to some mythical ‘gold standard’ (in ­economics, a basis for the denomination of currency used by no modern country for at least a generation), alternatives must be considered. Thus, for example, some studies of delirium intervention have randomised at the level of the care providers (say, by hospital, ward or care team) rather than at the level of the patient,16,17 or have chosen prospective matching instead of ran­domisation.18 In this way, the threat to validity from non-comparability of groups is traded off for the threat to validity from contamination bias (i.e. that health-care workers will apply their intervention to all patients, whether in the usual care of the intervention groups).19,20 Even this strategy, however, runs the risk of assuming that we know too much. Delirium, according to the latest consensus definition, is chiefly a disorder of consciousness.21 We do not disagree, but we note that consciousness has been historically difficult to describe in a nonarbitrary way. In consequence, delirium has been conceptualised as a disorder of attention22 and often operationalised chiefly as a disorder of attention, as in the third revised edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-III-R),23 the Confusion Assessment Method24 or the Delirium Rating Scale-revised-98.25 This emphasis has persisted even though the concept of attention is itself not unitary. For some time, psychologists have distinguished between visuospatial and auditory attention wherein attentional capacity is restricted within but not between sensory modalities.26 Indeed, in a review, Trzepacz has emphasised the particular importance of impaired visuospatial attention in delirium.27 Delirium is also held to be distinct in its essential features from dementia, even though many dementing disorders that are first seen in delirium28 begin acutely, or, as in stroke or dementia with Lewy bodies, share features that completely overlap the dementia syndrome.29 In short, in reflecting on the Montréal experience, we must get to grips with the facts that many physicians can proceed professionally without showing evidence of a working knowledge of delirium management and that many delirium experts are prepared to live with flawed concepts in their area of expertise. How do we then move ahead? A key step forward, in our view, is to abandon the notion of the brain being usually a specific victim in delirium, in favour of the view of the brain mostly as an innocent bystander. In other words, we recognise, accept and advocate that it is always important to be sure that delirium is not the result of a specific brain injury (as in viral encephalitis, for example). Equally, however, we would promote the idea that the usual causes of delirium need not be the result of specific brain injury.30 Rather, delirium commonly occurs in the setting of congestive heart failure, infection outside the central nervous system or metabolic abnormalities in which the brain is affected secondarily. Why would we insist on this as a starting point? We offer two reasons. The first, though banal, is to undercut the common practice, when physicians have recognised that something bad has happened to a patient's cognition, of defaulting to the notion that the patient ‘must have had a stroke’ and to then proceed without specific delirium management. The second reason is deeper. We believe that there is merit in viewing delirium as the high-order failure of a complex system. There is merit in this approach because it can lead physicians and researchers to what, in our estimation, will be more fruitful areas of activity. One of these will be the recognition of the need for decision-making in the face of complexity. Decision-making in the face of complexity requires some means of conceptualising and working with the complex phenomenon being considered. In medicine, we often encounter complex phenomena and we are generally taught that the best way to proceed is to break the complex problem down into its component parts, prioritise them and work them through skilfully, but one at a time. In medical teaching, we prize such parsimony. Anyone who has had the intensely satisfying experience of explaining many symptoms and signs by one diagnosis − especially if it is a rare one − will understand this deeply. Something similar obtains in research settings, where conventional analytical techniques also prize parsimony, enshrined in the doctrine of Ockham's razor (that ‘explanation should not be multiplied beyond necessity’). Why would anyone advocate otherwise? We advocate otherwise because we believe that the parsimonious account usually is untrue. Delirium chiefly occurs in very old people who have multiple and interacting medical and social problems − in other words, in elderly people who are frail.31 Such people typically do not have one thing wrong and the ‘one-thing-wrong-at-once’ approach appears to do worse than the approach of comprehensive geriatric assessment, in which multiple, interacting problems are accepted and even sought out.32 Such an approach is rooted in the realisation that the consequences of biological redundancy extend to disease presentation.31 In other words, frail elderly patients can be conceptualised as complex systems at the edge of failure and when they fail, they will fail with their highest order functions first. For many patients, this highest order function will be consciousness and, thus, their presentation will be delirium. This, in our view, can account for why delirium is such a sensitive, if non-specific, sign of acute illness in elderly people who are frail. The observation that delirium frequently occurs among the frail is not new, and indeed harks back to the view of ‘acute confusion’ as one of the so-called ‘geriatric giants’.33 An early generation of geriatricians celebrated the intellectually satisfying necessity of making decisions that embraced the complexity of their patients. They were not wrong to have done so. Indeed, a practical alternative to (mis)recognising delirium through opprobrium (‘train wrecks’, ‘poor historians’, ‘gomers’) is to celebrate the clinical challenges presented when they occur. In this way, patients with delirium can go from being objects of denigration to being seen as presenting the opportunity to practise at a high level of clinical skill (Anyone can make a diagnosis when they’re given a history!). Systematic reviews must necessarily work with the evidence at hand and reject practices based on opinion. But the evidence is only as good as the concepts that underlie it. To the extent that simple problems are easier to examine than complex ones, we must be wary of the data before us. But such wariness is not all a matter of methodology. Evidence-based medicine can be equally as decerebrate as mere opinion. Evidence-biased medicine is no substitute for scholarly inquiry in the face of complexity.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.