Delirium on the intensive care unitBMJ 2014; 349 doi: https://doi.org/10.1136/bmj.g7265 (Published 28 November 2014) Cite this as: BMJ 2014;349:g7265
- 1Intensive Care Unit, Watford General Hospital, Watford WD18 0HB, UK
- 2Inserm Research Center for Epidemiology and Biostatistics, Team Neuroepidemiology, Bordeaux, France
- Correspondence to: V J Page
Over the past few years the interest in delirium in patients on the intensive care unit has gathered pace, seemingly exponentially. Given that delirium is common, particularly in ventilated patients, a recurring question is, “does delirium increase mortality directly or is it an epiphenomenon?” In a linked paper (doi:10.1136/bmj.g6652) Klein Klouwenberg and colleagues try to answer this question.1 In their prospective cohort study the authors used sophisticated statistical modelling to show that delirium probably does not increase mortality directly, overturning findings from a large body of previous observational research.2 3
Dealing with clinical situations that change over time (such as severity of illness) is one of the key challenges when exploring associations between delirium and death in critically ill patients. Traditional analysis methods may result in incorrect estimates if the main exposure (delirium) influences important covariates such as treatment decisions or clinical complications during follow-up, and also if the main exposure itself (severity of delirium) changes over time.4 Both these problems occur in cohort studies linking delirium with mortality on the intensive care unit. Other complexities include the fact that patients who are discharged alive from intensive care may be in a different health state from those who remain on the unit (for whom further information is available), and that delirium preferentially develops in the most severely ill patients, who already have an increased risk of mortality.
Klein Klouwenberg and colleagues’ analysed their data using both standard methods and a more sophisticated method, called a marginal structural model, which overcomes some of these difficulties. The different results illustrate clearly how inadequate adjustments can bias findings and may produce misleading conclusions. Once the analysis has been done correctly—adequately accounting for complex time varying relations between clinical conditions and the consequences and for the competing risks of discharge—the link between delirium and increased mortality disappears.
Mortality has well known shortcomings as an outcome in studies on intensive care. In the 2014 BBC Reith lecture, “The Idea of Wellbeing,” Harvard surgeon Atul Gawnde encourages a shift of focus from “mere survival” to protecting quality of life (www.bbc.co.uk/podcasts/series/reith). Delirium is essentially about the brain and is thought to increase the risk of long term cognitive impairment after critical illness, a concern for patients and their families. The BRAIN-ICU trial enrolled 821 critically ill adults and assessed the global and executive function of survivors at three and 12 months after enrolment.5 Twelve months after the critical illness, one in four patients had cognitive impairment comparable to mild Alzheimer’s disease and one in three had impairment comparable to a moderate traumatic brain injury. The longer patients were delirious, the worse their cognitive outcome. This association with cognitive impairment may be subject to the same issues as mortality. However, the results of a sister study6 in the same cohort showed no association between delirium and poor functional outcomes, suggesting the link with cognitive outcomes would not disappear if the data were analysed differently.
Delirium is a syndrome, not a diagnosis like smallpox. It is a constellation of signs and symptoms suggestive of a malfunctioning brain. Regardless of the cause, delirium is associated with neuroinflammation, alterations in blood flow, and electroencephalographic changes, indicating damage that could result in cognitive impairment. Neuroimaging studies are difficult to carry out on intensive care units, and rarely do patients have a baseline magnetic resonance image for comparison. However, researchers have reported an association between longer duration of delirium and smaller brain volumes7 and between longer duration of delirium, white matter disruption, and later cognitive impairment.8 Delirium alone may not directly cause death; however, ongoing pathological changes in the brain, manifested as delirium, cannot easily be dismissed.
Delirium in patients on the intensive care unit is at the extreme end of a spectrum of disease, and studies that exclude elective patients (27.8% of patients in Klein Klouwenberg and colleagues’ study) may not generalise well to less intensive settings where patients have more reserve and less disease. Some interventions, for example, seem to decrease the risk of delirium in patients undergoing elective cardiac procedures but not of those admitted as emergencies to intensive care units.9 Perhaps the burden of disease among emergencies is simply too great.
What does the future hold for research on delirium in critically ill patients? In the United Kingdom, the James Lind Alliance—a non-profit making organisation—brings together patients, carers, and doctors to ensure that funders invest in research questions that matter to patients and the professionals who care for them. A recent priority setting partnership with the UK Intensive Care Foundation started with a survey and review that generated over 1300 suggestions. After two years of work, the identification of delirium and how to monitor and manage its effects emerged as one of the top three priorities for research in intensive care.10
Many patients on the intensive care unit die with delirium. We now know delirium may not cause death directly but it does result in longer hospital stays,1 complications, and anguish to patients, families, and carers in both homes and hospitals.11 12 13 Good evidence suggests that many patients are left with delirium related cognitive impairment.
Delirium on the intensive care unit has now come of age. It may not be deadly, but it is still an extremely serious complication that richly deserves its priority status for action and research.
Cite this as: BMJ 2014;349:g7265
Competing interests: We have read and understood the BMJ Group policy on declaration of interests and declare the following interests: none.
Provenance and peer review: Commissioned; not externally peer reviewed.