Delirium in the intensive care unit
© BioMed Central Ltd 2008
Published: 14 May 2008
Skip to main content
© BioMed Central Ltd 2008
Published: 14 May 2008
Delirium, an acute and fluctuating disturbance of consciousness and cognition, is a common manifestation of acute brain dysfunction in critically ill patients, occurring in up to 80% of the sickest intensive care unit (ICU) populations. Critically ill patients are subject to numerous risk factors for delirium. Some of these, such as exposure to sedative and analgesic medications, may be modified to reduce risk. Although dysfunction of other organ systems continues to receive more clinical attention, delirium is now recognized to be a significant contributor to morbidity and mortality in the ICU, and it is recommended that all ICU patients be monitored using a validated delirium assessment instrument. Patients with delirium have longer hospital stays and lower 6-month survival than do patients without delirium, and preliminary research suggests that delirium may be associated with cognitive impairment that persists months to years after discharge. Little evidence exists regarding the prevention and treatment of delirium in the ICU, but multicomponent interventions reduce the incidence of delirium in non-ICU studies. Strategies for the prevention and treatment of ICU delirium are the subjects of multiple ongoing investigations.
Recent advances in critical care medicine have improved survival in patients cared for in intensive care units (ICUs) worldwide, and in doing so they have revealed a major public health concern that previously had been under-appreciated. Critical care clinicians have historically been attuned to pulmonary, cardiac, and renal dysfunction as a source of morbidity and mortality in ICU patients but have underestimated the impact of brain dysfunction . Delirium, a common manifestation of acute brain dysfunction in critically ill patients, is associated with poor short-term outcomes and may result in adverse sequelae years after ICU discharge [2–4], and an executive summary on preventable medical injuries commissioned by the American Association of Retired Persons identified delirium as one of six leading causes of injuries associated with hospitalization in patients over 65 years of age . In this article we provide a general overview of the research to date regarding the epidemiology, diagnosis, and pathophysiology of ICU delirium, its association with health outcomes, and possible options for prevention and treatment.
Delirium is defined in the American Psychiatric Association's (APA) Diagnostic and Statistical Manual of Mental Disorders (DSM)-IV  as a disturbance of consciousness and cognition that develops over a short period of time (hours to days) and fluctuates over time. Many different terms have been used to describe this syndrome of cognitive impairment in critically ill patients, including ICU psychosis, ICU syndrome, acute confusional state, encephalopathy, and acute brain failure [1, 7]. However, the critical care literature has recently conformed to the recommendations of the APA and other experts that the term 'delirium' be used uniformly to describe this syndrome of brain dysfunction.
The prevalence of delirium reported in medical and surgical ICU cohort studies has varied from 20% to 80%, depending upon severity of illness observed and diagnostic methods used [3, 8–12]. Despite high prevalence rates in the ICU, delirium often goes unrecognized by clinicians  or its symptoms are incorrectly attributed to dementia, depression, or ICU syndrome (considered an expected, inconsequential complication of critical illness).
Delirium can be categorized into subtypes according to psychomotor behavior, and the high prevalence of hypoactive delirium in critically ill patients probably contributes to clinicians' lack of recognition of delirium. Hypoactive delirium is characterized by decreased responsiveness, withdrawal, and apathy, whereas hyperactive delirium is characterized by agitation, restlessness, and emotional lability . Peterson and coworkers  examined delirium subtypes in a cohort of ventilated and nonventilated medical ICU patients, and they observed that purely hyperactive delirium was rare (1.6%). In contrast, 43.5% of patients had purely hypoactive delirium and 54.1% had mixed delirium. Among non-ICU patients, hyperactive delirium has been associated with a better prognosis than hypoactive delirium , but this relationship has not been evaluated thoroughly among ICU patients. Because it is clear, however, that delirium is associated with poor clinical outcomes in critically ill patients, routine monitoring using valid and reliable delirium diagnosis instruments is recommended in all ICUs so that the prognostic significance of delirium does not go unnoticed.
Another important categorization scheme was recently studied in the critical care literature. Ouimet and colleagues  evaluated 600 ICU patients for symptoms of delirium and categorized them according to the number of symptoms present. Patients with no symptoms were considered to have 'no delirium', those with four or more symptoms to have 'clinical delirium', and those with one to three symptoms to have 'subsyndromal delirium' (because the full syndrome, according to DSM-IV criteria, was not present). This important study corroborated in the ICU setting a finding that had previously been reported from non-ICU studies; subsyndromal delirium represents an intermediate state that is different from both clinical delirium and a normal neurologic state.
Among medical ICU patients, delirium is associated with multiple complications and adverse outcomes, including self-extubation and removal of catheters , failed extubation , prolonged hospital stay , increased health care costs , and increased mortality [3, 20, 21]. Ely and coworkers  studied 275 mechanically ventilated medical ICU patients and determined that delirium was associated with a threefold increase in risk for 6-month mortality after adjusting for age, severity of illness, co-morbidities, coma, and exposure to psychoactive medications. The association between ICU delirium and increased mortality was subsequently confirmed in two other cohort studies [20, 21].
Delirium may be a predictor of long-term cognitive impairment in survivors of critical illness. Jackson and colleagues  reviewed nine prospective studies that included nearly 1,900 non-ICU patients who were hospitalized for medical and surgical treatments, and reported that delirium was associated with cognitive decline over 1 to 3 years after hospital discharge. The relationship between ICU delirium and long-term cognitive impairment is the subject of ongoing investigations, but preliminary data suggest that the association is significant. Jackson and coworkers  recently examined this association in 98 patients who were mechanically ventilated for acute respiratory failure in medical ICUs. Prolonged periods of ICU delirium were associated with an increased risk for long-term cognitive impairment at 3 months post-discharge after adjusting for covariates (P < 0.001).
The pathophysiology of delirium is poorly understood but multiple promising hypotheses are subjects of ongoing research. The majority of studies supporting the hypotheses reviewed here were conducted in non-ICU patients. Thus, significant research is needed to elucidate the complex interplay between the mechanisms of critical illness and ICU delirium.
Delirium is theorized to be a neurobehavioral manifestation of imbalances in the synthesis, release, and inactivation of neurotransmitters that normally control cognitive function, behavior, and mood . Derangements of multiple neurotransmitter systems have been implicated in the pathophysiology of delirium, with the greatest focus being given to dopamine and acetylcholine. These neurotransmitters work in opposition, with dopamine increasing and acetylcholine decreasing neuronal excitability. An imbalance in one or both of these neurotransmitters results in neuronal instability and unpredictable neurotransmission. Specifically, an excess of dopamine  or depletion of acetylcholine  have been associated with delirium. Other neurotransmitters are likely to play a role in the pathogenesis of delirium as well, including γ-aminobutyric acid (GABA), serotonin, endorphins, and glutamate .
Inflammation plays a significant role in the dysfunction of multiple organs caused by critical illness , and inflammatory abnormalities induced by endotoxin and cytokines probably contribute to the development of ICU delirium. Inflammatory mediators produced during critical illness (for example, tumor necrosis factor-α, interleukin-1, and other cytokines and chemokines) initiate a cascade of endothelial damage, thrombin formation, and microvascular compromise . Studies in animal models have revealed that these inflammatory mediators cross the blood-brain barrier , increase vascular permeability in the brain , and result in changes on electroencephalography (EEG) that are consistent with those seen in septic patients with delirium . Inflammation may incite brain dysfunction by decreasing cerebral blood flow via the formation of microaggregates of fibrin, platelets, neutrophils, and erythrocytes in the cerebral microvasculature; by constricting cerebral vasculature through activation of α1-adrenoceptors ; or by interfering with neurotransmitter synthesis and neurotransmission .
Early hypotheses that remain relevant today attempted to explain delirium as a behavioral manifestation of a 'widespread reduction of cerebral oxidative metabolism resulting in an imbalance of neurotransmission' . Engel and Romano  conducted classic experiments by evaluating delirious patients using EEG. These studies showed that delirium is associated with diffuse slowing on EEG, a finding that is believed to represent a reduction in brain metabolism. Thus, they hypothesized that delirium is the result of 'cerebral insufficiency' (a global failure of cerebral oxidative metabolism), a factor that is known to be important in the pathogenesis of multiple organ dysfunction in critical illness .
Neurotransmitter levels and function can be affected by changes in the plasma concentrations of various amino acid precursors, and some investigators have proposed that altered availability of large neutral amino acids contributes to the development of delirium . Amino acid entry into the brain is regulated by a sodium-independent large neutral amino acid transporter type 1 (LAT1) . Tryptophan, an essential amino acid and precursor for serotonin, competes with several large neutral amino acids (for example, tyrosine, phenylalanine, valine, leucine, and isoleucine) for transport across the blood-brain barrier via LAT1 transporters. Phenylalanine also competes with large neutral amino acids for transport across the blood-brain barrier. Increased cerebral uptake of tryptophan and phenylalanine, compared with that of other large neutral amino acids, leads to elevated levels of dopamine and norepinephrine (noradrenaline), two neurotransmitters that have been implicated in the pathogenesis of delirium .
Risk factors for delirium can be divided into predisposing factors (host factors) and precipitating factors . Although predisposing factors are present before ICU admission and are difficult to alter, precipitating factors occur during the course of critical illness. They may involve factors of the acute illness or be iatrogenic; these factors represent areas of risk that are potentially modifiable by preventive or therapeutic intervention.
Risk factors for delirium in ICU patients
Factors of critical illness
Immobilization (for example, catheters, restraints)
Medications (for example, opioids, benzodiazepines)
Metabolic disturbances (for example, sodium, calcium, BUN, bilirubin)
High severity of illness
In the only study to date to examine whether there is a genetic predisposition to ICU delirium in some patients, Ely and coworkers  evaluated the association between apolipoprotein E (APOE) genotype and duration of delirium among 53 mechanically ventilated medical ICU patients. Patients with the APOE4 polymorphism (a risk factor for Alzheimer's disease) were delirious for twice as long as those without the APOE4 polymorphism (median [interquartile range]: 4 [3 to 4.5] days versus 2 [1 to 4] days; P = 0.05). Larger studies are ongoing to confirm this association.
Other factors associated with delirium in the ICU include older age, baseline cognitive impairment, metabolic disturbances (for instance, derangements in sodium, calcium, and blood urea nitrogen), acute infection, respiratory disease, acidosis, anemia, and hypotension [10, 39, 40]. Some factors that are clearly associated with delirium in hospitalized patients without critical illness may carry similar risks among ICU patients, although studies to date have not yet proven associations with ICU delirium. These potential risk factors include hearing or vision impairment, depression, and immobilization resulting from placement of catheters or restraints [37, 41]. Critically ill patients are typically exposed to numerous factors that may precipitate delirium. In a study of 53 medical ICU patients, Ely and coworkers  identified an astounding average of 11 delirium risk factors per patient. Two delirium risk factors nearly universally experienced by ICU patients are exposure to sedative and analgesic medications and sleep deprivation. The risk associated with both of these factors is potentially modifiable, as is discussed in greater detail below.
Sedative and analgesic medications are routinely administered to patients receiving mechanical ventilation to reduce pain and anxiety, as recommended by the Society of Critical Care Medicine (SCCM) . These medications, however, are not without detrimental effects. Continuous intravenous sedation, for example, is associated with prolonged mechanical ventilation as compared with sedation via intermittent boluses .
Multiple studies have demonstrated an association between delirium and exposure to sedative and analgesic medications. In a mixed surgical/medical ICU, Dubois and coworkers  determined that morphine was the strongest predictor of delirium in a multivariable model. Ouimet and colleagues  observed that sedative and analgesic medications used to induce coma were associated with delirium (delirium among patients with drug-induced coma: odds ratio [OR] = 3.2, 95% confidence interval [CI] = 1.5 to 6.8). Marcantonio and coworkers  reported that patients treated with benzo-diazepines were more likely to have postoperative delirium than were those not treated with benzodiazepines (OR = 2.7, 95% CI = 1.3 to 5.5). Meperidine also increased the likelihood of postoperative delirium (OR = 3.0, 95% CI = 1.3 to 6.8).
Sedative agents that are GABA receptor sparing, such as opioids and dexmedetomidine (a novel α2-receptor agonist), may reduce the risk for delirium in ICU patients as compared with the risk attributable to benzodiazepines. Although studies have consistently identified lorazepam and midazolam as risk factors for delirium, the data regarding opioids are less consistent. For example, Ouimet and coworkers  noted that mean daily opioid doses were higher among ICU patients without delirium than among ICU patients with delirium. Similarly, in 541 hip fracture patients, Morrison and colleagues  determined that those treated liberally with opioid analgesics (>10 mg/day parenteral morphine sulfate equivalent) were less likely to develop delirium than patients who received less analgesia. Treatment with meperidine was an exception, because this drug increased the risk for delirium as compared with other opioids. These investigations point to the importance of judicious use of these psychoactive medications, with focus on adequate analgesia.
Although the association between sleep disturbances and ICU delirium has not been studied, sleep deprivation impairs cognition  and the relationship between sleep during critical illness and delirium is a promising area of ongoing research. On average, ICU patients sleep only 2 hours per day , and less than 6% of their sleep is random eye movement sleep. Cooper and coworkers  studied 20 mechanically ventilated ICU patients using polysomnography and observed that every patient had severely disrupted sleep or none at all. Such disturbances may detrimentally affect protein synthesis, cellular and humoral immunity, and energy expenditure, and ultimately they may contribute to organ dysfunction such as delirium. Excessive noise and patient care activities account for only a minority of arousals in ICU patients, whereas metabolic derangements, mechanical ventilation, and exposure to sedative and analgesic medications probably play significant roles in disturbing sleep in the ICU . Because sedative and analgesic medications may lead directly to delirium or may contribute to impaired cognition via disruption of sleep, and because traditional methods of sedation and analgesia are amenable to change, some investigators have prioritized studies seeking to understand the relationships between sedative and analgesic exposure, sleep disturbance, and brain dysfunction in critically ill patients.
Delirium assessment instruments, such as the Intensive Care Delirium Screening Checklist (ICDSC)  and the Confusion Assessment Method for the ICU (CAM-ICU) [50, 51] allow nonpsychiatric physicians and other ICU personnel to diagnose delirium in ICU patients rapidly and reliably, even when the patient cannot speak because of endotracheal intubation. In fact, the SCCM  recommends that all ICU patients be routinely evaluated for delirium.
Bergeron and colleagues  validated the ICDSC in a study of 93 ICU patients who were evaluated by a consulting psychiatrist who served as the reference standard rater. Using the ICDSC, each patient is assigned a score from 0 to 8; a cut-off score of 4 has a sensitivity of 99% and specificity of 64% for identifying delirium. Thus, because of its very high sensitivity, the authors recommend that the ICDSC be used as a screening tool for ICU delirium.
The Intensive Care Delirium Screening Checklist
Altered level of consciousnessa
Response to intense and repeated stimulation
Response to mild or moderate stimulation
Exaggerated response to normal stimulation
Difficulty following instructions or easily distracted
To time, place, or person
Clinical manifestation or suggestive behavior
Psychomotor agitation or retardation
Agitation requiring use of drugs or restraints, or slowing
Inappropriate speech or mood
Related to events or situation, or incoherent speech
Sleep/wake cycle disturbance
Sleeping <4 hours/day, waking at night, sleeping all day
Symptoms above occurring intermittently
0 to 8
The CAM-ICU, adapted from the Confusion Assessment Method , was designed for use in intubated patients and validated against a reference standard rater in mechanically ventilated ICU patients. Originally validated by Ely and coworkers [50, 53] in two cohorts of 38 and 111 medical ICU patients, the CAM-ICU has a high sensitivity (93% to 100%) and specificity (89% to 100%) for delirium as well as high inter-rater reliability (κ = 0.96, 95% CI = 0.92 to 0.99). Lin and colleagues  subsequently validated the CAM-ICU in a separate cohort of 102 mechanically ventilated medical ICU patients and reported similarly high sensitivity (91% to 95%), specificity (98%), and inter-rater reliability (κ = 0.91).
Interventions for the prevention and treatment of ICU delirium
Non-ICU clinical trials [56,58]
Non-ICU clinical trials 
Reduced benzodiazepine use
Prospective ICU cohort studies 
Placebo-controlled, randomized, non-ICU trial 
Non-randomized ICU trial 
Prospective non-ICU cohort studies 
Randomized ICU trials [68,69]
In a landmark trial, Inouye and coworkers  evaluated 852 hospitalized older patients who were nonrandomly assigned to usual care or management with a multicomponent intervention consisting of multiple protocols designed to minimize the effects of delirium risk factors. These protocols directed repeated reorientation of the patient and provision of cognitively stimulating activities; a nonpharmacologic sleep protocol; early mobilization activities and range of motion exercises; timely removal of catheters and physical restraints; use of eyeglasses, magnifying lenses, and hearing aids; and early correction of dehydration. The intervention significantly reduced the incidence of delirium (15.0% in the usual care group versus 9.9% in the intervention group; matched OR = 0.60, 95% CI = 0.39 to 0.92). These results were confirmed in randomized trials studying the efficacy of multicomponent delirium prevention programs in postoperative patients  and geriatric inpatients .
Although the strategies emphasized in these studies (for instance, frequent reorientation, restoration of sleep/wake cycles, timely removal of catheters, early mobilization, and minimization of unnecessary noise/stimuli) are recommended for ICU patients, such strategies have not been studied in the ICU, and the nature and number of delirium risk factors in ICU patients are different from those observed among other inpatients. Preventive interventions, therefore, must address those risk factors that are most common in the ICU, including electrolyte abnormalities, infection, and exposure to sedative and analgesic medications. These drugs play a vital role in maintaining patient comfort, and adequate pain control is likely to reduce the risk for delirium , but practitioners must strive to avoid over-sedation by administering the drugs only as necessary. In fact, sedation protocols that use validated sedation scales , emphasizing intermittent boluses rather than continuous infusions , and promoting daily interruption of sedatives and analgesics  significantly improve multiple clinical outcomes. The effect of these methods of sedation on delirium is the subject of ongoing investigations.
After the ICU team has used the multicomponent prevention strategies described above to minimize risk factors and after acute, life-threatening complications of critical illness that may lead to delirium (for example, hypoxia, hypercapnia, hypoglycemia, and shock) have been sought out and addressed, pharmacologic therapy for the prevention and treatment of delirium should be considered (Table 3). When using such medications, clinicians should be aware that any drug intended to improve cognition may have adverse psychoactive effects, paradoxically exacerbating delirium or causing excessive sedation in some patients. Also, evidence proving the efficacy of pharmacologic strategies for delirium is lacking. All psychoactive drugs should therefore be used judiciously in critically ill patients, in the smallest effective dose for the shortest time necessary.
Although no placebo-controlled clinical trials have been conducted to evaluate its efficacy in ICU patients, haloperidol is recommended as the drug of choice for the treatment of ICU delirium by the SCCM  and the APA . A typical antipsychotic, haloperidol blocks D2 dopamine receptors, resulting in amelioration of hallucinations, delusions, and unstructured thought patterns. The optimal dose and regimen have not been defined in clinical trials, but the SCCM guidelines suggest that patients with hyperactive delirium should be treated with 2 mg intravenously, followed by repeated doses (doubling the previous dose) every 15 to 20 minutes while agitation persists. Once the agitation subsides, scheduled doses (every 4 to 6 hours) may be continued for a few days, followed by tapered doses for several days. Common doses for ICU patients range from 4 to 20 mg/day, but higher doses are frequently used for the treatment of acute agitation. Also, haloperidol has been evaluated as a prophylactic therapy for the prevention of postoperative delirium in elderly hip surgery patients. Kalisvaart and coworkers  conducted a randomized, double-blind, placebo-controlled trial in 430 elderly hip surgery patients and determined that 1.5 mg/day haloperidol significantly reduced the severity and duration of delirium. The incidence of delirium was not altered by haloperidol compared with placebo. A similar trial is currently ongoing in ICU patients.
Atypical antipsychotics (for instance, risperidone, ziprasidone, quetiapine, and olanzapine) may also be helpful in the treatment of delirium, but only preliminary data exist supporting their use in the ICU. These medications target dopamine receptors as well as receptors for other neurotransmitters, including serotonin, acetylcholine, and nor-epinephrine (noradrenaline). Skrobik and coworkers  compared olanzapine with haloperidol in a nonrandomized trial that enrolled 73 medical and surgical ICU patients; they reported that resolution of delirium symptoms was similar in both treatment groups, but more side effects were observed in patients treated with haloperidol. Numerous studies have compared different antipsychotics with each other for the treatment of delirium in non-ICU patients, but all are limited by the absence of a placebo-treated control group as well as by other methodologic limitations . Well designed, randomized, placebo-controlled trials are needed to inform critical care clinicians regarding the efficacy and safety of both typical and atypical antipsychotics in the prevention and treatment of ICU delirium.
Patients treated with haloperidol or other antipsychotics should be monitored for adverse effects, including hypotension, dystonia, extrapyramidal effects, laryngeal spasm, malignant hyperthermia, glucose and lipid dysregulation, and anticholinergic effects such as dry mouth, constipation, and urinary retention. Torsades de pointes is a life-threatening arrhythmia that may occur in patients treated with antipsychotics, and these medications should be avoided in patients with prolonged QT intervals.
Recently, investigators observed an association between antipsychotic use (typical or atypical) and increased mortality in elderly patients [65, 66]. The patients studied were not critically ill, and no study of ICU patients has demonstrated an increased risk for death due to antipsychotic use. In fact, Milbrandt and coworkers  studied 989 mechanically ventilated ICU patients and observed that haloperidol was associated with significantly lower hospital mortality.
Although they remain the drugs of choice for the treatment of delirium tremens (and other withdrawal syndromes), benzodiazepines are not recommended for the management of delirium because these medications are themselves risk factors for delirium [21, 40, 44]. Dexmedetomidine, a novel α2-receptor agonist that does not act on GABA receptors, may prove to be an alternative sedative agent that is less likely to cause delirium. In a preliminary report of an unblinded, randomized trial conducted in postoperative cardiac surgical patients, Maldonado and colleagues  described a significant reduction in the incidence of delirium associated with dexmedetomidine; 8% of patients sedated with dexmedetomidine at sternal closure developed delirium as compared with 50% of patients sedated with propofol or midazolam. Similarly, in a recently completed double-blind, randomized controlled trial  it was determined that ICU patients sedated with dexmedetomidine spent fewer days in coma and more days neurologically normal (without coma or delirium) than did those sedated with lorazepam. These pilot studies suggest that larger trials are warranted to evaluate the efficacy and safety of sedation with dexmedetomidine as well as clonidine, a less selective α2-receptor agonist, in ICU patients.
Delirium is a common manifestation of acute brain dysfunction in critically ill patients that is associated with poor short-term outcomes and may result in adverse sequelae years after ICU discharge. The strategies described here for the prevention, diagnosis, and treatment of ICU delirium are subjects of multiple ongoing investigations. Every ICU clinician should be aware of these strategies, institute routine monitoring for delirium in the ICU, seek to reduce the impact of risk factors for delirium when possible, and use treatment options when necessary.
American Psychiatric Association
Confusion Assessment Method for the ICU
Diagnostic and Statistical Manual of Mental Disorders
Intensive Care Delirium Screening Checklist
intensive care unit
large neutral amino acid transporter type 1
Richmond Agitation-Sedation Scale
Society of Critical Care Medicine
Dr Girard is a Hartford Geriatrics Health Outcomes Research Scholar and a Vanderbilt Physician-Scientist Development Program Scholar. Dr Pandharipande is the recipient of a Mentored Research Training Grant from the Foundation for Anesthesia Education and Research. Dr Ely is the Associate Director for Research of the VA Tennessee Valley Geriatric Research, Education and Clinical Center (GRECC), VA Service, Department of Veterans Affairs Medical Center, Nashville, Tennessee, USA. He receives support from the VA Clinical Science Research and Development Service (VA Merit Review Award) and the National Institutes of Health (AG0727201).
This article is part of Critical Care Volume 12 Supplement 3: Analgesia and sedation in the ICU. The full contents of the supplement are available online at http://ccforum.com/supplements/12/S3.
Publication of the supplement has been funded by an unrestricted grant from GlaxoSmithKline.