Sleep and delirium in ICU patients: a review of mechanisms and manifestations
- First Online:
- Cite this article as:
- Figueroa-Ramos, M.I., Arroyo-Novoa, C.M., Lee, K.A. et al. Intensive Care Med (2009) 35: 781. doi:10.1007/s00134-009-1397-4
- 2.4k Views
Sleep deprivation and delirium are conditions commonly encountered in intensive care unit patients. Sleep in these patients is characterized by sleep fragmentation, an increase in light sleep, and a decrease of both slow wave sleep and rapid eye movement sleep. The most common types of delirium in this population are hypoactive and mixed-type. Knowledge about the mechanisms of sleep and delirium has evolved over time, but these phenomena are not yet well understood. What is known, however, is that different areas in the brainstem transmit information to the thalamus and cortex necessary for sleep–wake regulation. Delirium is related to an imbalance in the synthesis, release, and inactivation of some neurotransmitters, particularly acetylcholine and dopamine. The relationship between sleep deprivation and delirium has been studied for many years and has been viewed as reciprocal. The link between them may be ascribed to shared mechanisms. An imbalance in neurotransmitters as well as alteration of melatonin production may contribute to the pathogenesis of both phenomena. A better understanding of the mechanisms and factors that contribute to sleep deprivation and delirium can guide the development of new methods and models for prevention and treatment of these problems and consequently improve patient outcomes.
KeywordsSleepSleep deprivationDeliriumIntensive care unitMechanisms
Sleep deprivation and delirium are conditions commonly encountered in intensive care unit (ICU) patients, but they are not yet well understood. Although several hypotheses concerning their mechanisms have been advanced, the alteration in specific neurotransmitters associated with sleep and delirium is the foundation of current research. The link between sleep deprivation and delirium has been studied for many years. However, it is yet unknown whether delirium causes sleep deprivation or whether delirium is a disorder caused by altered sleep architecture or circadian rhythm desynchrony.
Although sleep functions are not well understood, it is clear that sleep is a dynamic as well as complex physiologic state necessary for life; when lacking, deprivation results in serious physiological consequences . Delirium may also result in consequences that negatively influence patient outcomes, including mortality . Many risk factors have been implicated in the development of both sleep deprivation and delirium. Although some factors are unique to each phenomenon, other factors are shared by both. For example, sedatives and analgesics can contribute to the development of both sleep deprivation and delirium.
The purpose of this article is to review the mechanisms that underlie the regulation of sleep–wake cycles as well as mechanisms of delirium. We also provide general definitions and concepts of both sleep and delirium and their manifestation in ICU patients. Finally, we present the relationship between sleep and delirium as well as the influence of sedatives and analgesics on both.
Characteristics of NREM and REM sleep
% of the TST
Mixed frequency activity
Intermittent sleep spindles and K-complexes
Slow delta waves
Saw-tooth waves high frequency EEG
↓ CBF (brain stem and cerebellum in stages 1 and 2)↓ CBF (cortex in SWS)
↑ GH and ↓corticosteroids and catecholamines (SWS)
↓ HR, ↓ BP, ↓ RR (more regular than REMs)
↓ Brain temperature
Arousal threshold increase through the stages
Cardio-respiratory irregularitiesa (↑ HR, ↑ RR and BP variations)
↑ Brain temperatureb
High arousal thresholdb
Changes in posture
Dreams (at sleep onset and stage 2)
Rapid eye movementa
Components of the wake–sleep-regulatory system
ARAS 1st pathway
Wakefulness, REM sleep
Wakefulness, REM sleep
ARAS 2nd pathway
Wakefulness, suppress NREM
Wakefulness, suppress REM
NREM sleep, wakefulness, suppress REM
Wakefulness, REM sleep
Wakefulness, REM sleep
VLPO cluster and VLPO extended
GABA and galanin
The “flip-flop switch” mechanism is stabilized by orexin (hypocretin), a peptide produced in the lateral hypothalamus . Orexin strengthens the ARAS (thereby maintaining wakefulness) and prevents inappropriate transition to the sleep state . On the other hand, when VLPO is activated, it inhibits both monoaminergic and orexin neurons to maintain sleep .
In addition to the wake–sleep neuron-regulatory system, the Homeostatic Drive for Sleep (Process S) provides an useful explanation for waking and sleeping based on the observation that sleep debt accumulates during wakefulness . It is proposed that a substance that accumulates during prolonged wakefulness activates VLPO neurons and inhibits ARAS neurons, producing a transition to sleep . Although it is not yet determined, it is thought that adenosine could be this sleep-promoting substance which accumulates in the basal forebrain and inhibits wake-promoting neurons in this area .
To maintain homeostasis between sleep and wakefulness, it is important that sleep–wake mechanisms work adequately. However, several factors can negatively influence these mechanisms by provoking an alteration in sleep–wake cycles that subsequently reduce quantity or quality of sleep. The consequences of sleep deprivation that will be addressed have been studied in non-ICU patients. Thus, further studies are necessary to elucidate the effect of sleep deprivation on ICU patient outcomes.
Consequences of sleep deprivation
The consequences of total or partial sleep deprivation have been categorized as physiological and behavioral. Physiological consequences include increase in pain sensitivity [12, 13], reduction in forced expiratory volume and forced vital capacity , increases in sympathetic and decreases in parasympathetic cardiac modulation , impaired immune response [16, 17] and alteration in metabolic and endocrine systems . Behavioral consequences of sleep deprivation include impaired attention and psychomotor performance, increased daytime sleepiness, and impaired mood that includes fatigue and irritability .
The consequences of REM deprivation are similar to total sleep deprivation and include mood and memory alterations. Of importance to ICU patients is that REM deprivation due to CNS depressant medications can be followed by a REM sleep rebound phenomenon if the medication is suddenly discontinued. REM rebound is defined as an above-normal percentage of REM sleep, often a 300% increase, after a period of suppressed REM sleep and includes exacerbations of autonomic activity normally seen during phasic REM periods [20, 21]. Thus, REM rebound may cause an increase in heart rate, hypoxemia, cardiac arrhythmias, and hemodynamic instability [22, 23]. Because of the cardiac and respiratory variability observed during this event, REM rebound can be dangerous for ICU patients .
Sleep in ICU patients
Two primary sleep disorders have been found in ICU patients: parasomnias and dyssomnias. Parasomnias include undesirable physiological or behavioral events occurring during specific sleep or sleep–wake transition phases, which are not associated with abnormalities of the sleep–wake cycle itself . REM sleep behavior disorder, one of the parasomnias which is characterized by loss of atonia, increase in musculoskeletal activity, and vivid dreams , has been reported in ICU patients with Guillian–Barré syndrome . It is suggested that REM sleep behavior disorder may be associated with decreased blood flow in the brain, loss of dopaminergic neurons, or motor system alterations .
According to the American Academy of Sleep Medicine , dyssomnias include disorders related to the inability to initiate or maintain sleep. Specifically, circadian rhythm sleep disorder is a dyssomnia with an irregular sleep–wake pattern that can affect ICU patients. Sleep in ICU patients is often fragmented due to frequent arousals and awakenings. Studies show that their sleep architecture is altered with an increase in light sleep and less SWS and REM sleep; total sleep time averages between 2.1 and 8.8 h and is not continuous [27–32]. Indeed, sleep has been noted to occur in 50–67% of the night [27, 28, 31] and 54–57% of the day in ICU patients [28–31], suggesting that both circadian rhythms and sleep quality are affected. One of the predisposing factors for developing this type of dyssomnia is prolonged bed rest .
Disturbance of the light–dark cycle might also contribute to alteration in circadian rhythms . Light exposure is the main external cue for maintaining circadian rhythm, but ICU patients have limited natural light exposure. In addition, alteration in circadian rhythm has also been linked to melatonin secretion impairment in ICU patients [33, 34]. For example, systemic inflammatory response, hormone interactions, medications, acuity of illness, mechanical ventilation, and environmental factors, could influence melatonin excretion rhythm [34–37].
The methods utilized to measure sleep are classified into three categories: physiologic, behavioral, and self-report. Polysomnography (PSG) is the gold standard to measure sleep; however, this physiologic method is expensive and time-consuming. Several studies have been conducted with PSG in ICU patients [27–29, 31, 39, 50], but Watson and colleagues  found several limitations in applicating the standardized Rechtschaffen and Kales’ criteria to analyze PSG data in the seven ICU patients they studied. Thus, PSG data in this population must be interpreted with caution. Processes electroencephalogram (EEG) such as bispectral index (BIS) has been used to measure sleep; however, there has been and identified overlap of BIS values between light sleep and REM sleep  that could affect its validity.
Among the behavioral methods, observation and actigraphy have been utilized with ICU patients [27, 33, 53]. The validity of the observation method has not been well established. This technique is time-consuming and could be subject to observer bias and observer fatigue. Good accuracy between actigraphy and PSG has been demonstrated in non-ICU patients. However, overestimation of total sleep time and underestimation of awakenings were found with the use of actigraphy . In addition, ICU patients are likely to decrease movement that may not be associated with sleep problems due to the use of sedation/analgesia, neuromuscular blockers, restraint, and weakness. To overcome these limitations, complementary tools, such as sleep diaries or video recording have been suggested. However, sleep diaries would be difficult to perform by the majority of ICU patients.
Self-report questionnaires to assess sleep in the ICU population (Verran and Snyder–Halpern Sleep Scale , Richards–Campbell Sleep Questionnaire , and Sleep in the ICU Questionnaire ) did not show adequate psychometric properties and had several limitations. Recall bias is one potential limitation of these questionnaires. Their use is limited to conscious and stable patients, thereby important ICU populations are excluded and generalizability of the study results is compromised. Moreover, some of these questionnaires fail to assess characteristics of daytime sleep in ICU patients.
Physiologic and behavioral characteristics according to two delirium sub-types
% in ICU
Level of consciousness
Lethargy, ↓ alertness
Speech loud, incomprehensible, rapid and disorganized
Lack of perceptual disturbance
Low-voltage fast EEG
↑ or normal cerebral metabolic activity
↓ GABA activity
↓ cerebral metabolic activity
↑ GABA activity
↑ psychomotor activity
↓ psychomotor activity
↓ stimuli response
Benzodiazepine withdrawal, alcohol/drug withdrawal, drug intoxication
Differences between sleep deprivation, delirium, depression, and dementia (modified from )
Level of consciousness
Disrupted memory consolidation
Impaired (immediate and short-term memory)
Usually intact (short-term memory deficit)
Impaired (immediate and recent events)
Inability to concentrate
Intact (inability to concentrate, negative thoughts)
Difficulty with abstractions, finding words, decreased judgments
Disoriented (time and place)
Intact in early dementia (worse with progression)
The mechanisms of delirium are not fully understood. Nevertheless, it is suggested that they are related to an imbalance of neurotransmitters . A neuroanatomical pathway has been proposed for delirium that involves the thalamus, prefrontal cortex, fusiform cortex, posterior parietal cortex, and basal ganglia .
The most prevalent hypothesis suggests that imbalances in acetylcholine and dopamine neurotransmitters are involved in the development of delirium . More specifically, levels of acetylcholine are low and levels of dopamine are high . However, a literature review performed by Trzepacz  referred to some studies that suggested just the opposite: that either excess acetylcholine or deficiency in dopamine can provoke delirium. The relation of dopamine to delirium is based on the therapeutic effect of haloperidol, which is a potent dopamine blocker . A retrospective study that selected the use of haloperidol as an indicator of delirium occurrence found that dopamine administration was strongly associated with the need for haloperidol, suggesting that dopamine administration could be a risk factor for delirium .
Neurotransmitters other than acetylcholine and dopamine are also implicated in delirium, but their mechanism of action is not well established. These include serotonin, GABA, glutamate, histamine, and noradrenaline [65, 66]. A study conducted with cardiac surgery patients with delirium found a significant decrease in plasma tryptophan, the precursor of serotonin, as well as a significant increase in phenylalanine, a precursor of dopamine and noradrenaline . The authors suggest that alteration in these amino acids may contribute to the development of delirium by a decrease in serotonin and increase in dopamine and noradrenaline.
Another delirium mechanism was explored by Lewis and Barnett  based on the “abnormal tryptophan metabolism” model of delirium from an earlier study . Balan and colleagues  showed that patients with hyperactive delirium had low levels of urinary 6-sulphatoxymelatonin (SMT), a melatonin metabolite, and patients with hypoactive delirium had higher levels of urinary 6-SMT. This model suggests the existence of two metabolic pathways for tryptophan’s ability to enhance either hypoactive or hyperactive delirium.
Delirium in ICU patients
Delirium is common in ICU patients, reported to affect 11–87% of ICU patients [75–79]. This prevalence varies according to patient’s severity of illness and delirium measure or criteria used. Medical ICU patients predominantly develop a mixed-type delirium (55%), followed by hypoactive delirium (43.5%); only 1.6% showed hyperactive delirium . In contrast, surgical and trauma ICU patients who developed delirium showed more hypoactive delirium (64 and 60%, respectively) than mixed-type (9 and 6%, respectively) or hyperactive delirium (0 and 1%, respectively) .
There are many risk factors associated with delirium in ICU patients. In a study of 818 surgical ICU patients, 11% developed delirium diagnosed by a psychiatrist using the Diagnostic and Statistical Manual of Mental Disorders (DSM-III) delirium criteria . They found many predisposing factors for delirium including hyperamylasemia, hypocalcemia, respiratory disease, hypotension, infection, fever, hyperbilirubinemia, hyponatremia, anemia, azotemia, metabolic acidosis, and increase in hepatic enzymes. In another study, hypertension was a risk factor for delirium in 198 medical and surgical ICU . Both the Intensive Care Delirium Screening Checklist (ICDSC) and a psychiatric assessment were used to identify delirium in this study. They also found that a smoking history, hyperbilirubinemia, epidural route of analgesia, and morphine use were risk factors for delirium. Risk factors for delirium in thoracic postoperative patients include diabetes mellitus, length of operation time, age, chemical imbalance, and sleep deprivation .
Pandharipande and colleagues  performed a study with 198 medical and coronary patients using the Confusion Assessment Method for the ICU (CAM-ICU) to identify delirium. They showed that lorazepam was an independent risk factor for daily transition to delirium (i.e., patients who received lorazepam were more likely to develop delirium the following day); every additional year above 65 years of age and an increase in Acute Physiology and Chronic Health Evaluation (APACHE) II score were also risk factors after adjusting for many covariates. A multivariate analysis revealed that an admission APACHE II score >14, history of hypertension, or alcoholism were significant risk factors for delirium in medical-surgical ICU patients . In another study, daily and cumulative doses of lorazepam were significantly higher in patients with delirium identified by CAM-ICU compared with non-delirious patients in 275 mechanically ventilated medical and coronary ICU patients . Micek and colleagues  did not find differences in total doses of infusions of midazolam and fentanyl in 44 delirious versus 22 non-delirious patients identified by the CAM-ICU. However, they found that a significant number of delirious patients received continuous IV midazolam and fentanyl compared with non-delirious patients. In addition, the use of physical restraints was a risk factor for delirium. Recently, Pandharipande and colleagues  using the CAM-ICU, found that midazolam was an independent risk factor for delirium in trauma and surgical ICU patients. In addition, fentanyl was also found as an independent risk factor for delirium but only in surgical ICU patients.
Another possible risk factor for delirium in ICU patients is alteration in melatonin secretion. In 41 ICU patients who underwent thoracic esophagectomy, the association between delirium and serum melatonin concentration was explored . Serum melatonin levels were measured every 6 h over 4 days. Eleven patients (26.8%) developed delirium, and they were significantly older than those who did not develop delirium. Irregular patterns of melatonin secretion were associated with the development of delirium in the 11 patients. Although delirious patients tended to have abnormally low melatonin levels compared with non-delirious patients, differences were not significant, a finding that could have been due to sample size.
Among the delirium instruments, ICDSC and the CAM-ICU have good psychometrics properties and more feasible for use with ICU patients. Both are based on DSM-IV delirium criteria. The ICDSC developed by Bergeron and colleagues  consists of eight domains (altered level of consciousness, inattention, disorientation, hallucination-delusion-psychosis, psychomotor agitation or retardation, inappropriate speech or mood, sleep–wake cycle disturbance, and symptom fluctuation) with descriptions that facilitate its application. However, some domains can be difficult to assess or can be misinterpreted; and the ICDSC may be subject to variability in its interpretation. It has been shown to have an excellent sensitivity (99%); however, its specificity was lower at 64% , allowing other conditions to be identified incorrectly as delirium.
The CAM-ICU was developed to identify delirium in mechanically ventilated and non-ventilated ICU patients . This instrument uses an algorithm system with four domains: acute onset of mental status changes or fluctuating course, inattention, disorganized thinking, and altered level of consciousness. The CAM-ICU has been validated in larger ICU population than the ICDSC and includes tools and questions that reduce subjectivity. While it requires training to use, and is an easy instrument that takes approximately 2 min to administer.
Sleep deprivation and delirium
The relationship between sleep deprivation and delirium has been studied for many years. However, methodological issues related to the studies make it difficult to establish the relationship between these two phenomena. One can ask: does sleep deprivation contribute to delirium, or does delirium contribute to sleep deprivation? Studies conducted with cardiac surgical patients suggest that sleep deprivation is a result of delirium [94, 95]. However, in a review of 17 studies performed with different types of surgical patients who had delirium risk factors, sleep deprivation was not a risk factor for delirium . Yet, Sveinsson  found that sleep deprivation is a potential precipitating factor for delirium in cardiac surgical patients, and Helton and colleagues  found that patients with sleep deprivation were significantly more likely to develop delirium than patients without sleep deprivation.
Sleep deprivation was found to be a risk factor that predicted delirium in postoperative patients . However, this study was a retrospective record review, and investigators did not report how sleep deprivation was measured. A prospective study performed with 27 ICU patients showed a significant association between delirium (measured by CAM-ICU) and altered sleep architecture (measured by PSG) during a one-night recording . They identified longer sleep onset latency, longer REM sleep latency, shorter REM sleep duration, and fewer REM sleep periods in patients with concomitant delirium.
Effect of benzodiazepines and opioids on sleep and delirium
Many medications can influence the wake–sleep-regulatory system by a direct effect on neurotransmitters and hormones . Benzodiazepines and opioids can reduce both SWS and REM sleep via GABA type A and opioid mu receptors stimulation, respectively . On the other hand, opioids can cause delirium by decreasing acetylcholine and increasing dopamine and glutamate activity . Benzodiazepines might play a role in hypoactive delirium by increasing GABA activity . A theory of drug-induced delirium was proposed by Gaudreau and Gagnon  who identified the role of the thalamus in filtering information from cortical and stem regions of the brain. They established that medications like benzodiazepines and opioids interfere with neurotransmitter pathways to cause a transient thalamic filtering dysfunction that contributes to delirium.
In addition to the administration of benzodiazepines and opioids, the sudden discontinuation of these medications may influence sleep and delirium through development of a withdrawal syndrome [103, 104]. Benzodiazepine withdrawal decreases GABA activity which may lead to the development of hyperactive delirium . Sleep disturbance, specifically REM rebound, can result from benzodiazepine withdrawal . In a review of several studies, Wang and Teichtahl  concluded that opioid withdrawal was associated with alterations in sleep architecture. REM rebound may result from discontinuation of sedatives and opioids therapy [22, 49, 107]. Knill and colleagues  found that higher opioid doses correlated with marked SWS and REM sleep suppression; REM sleep reappears when opioid doses are reduced.
As previously mentioned, an alteration in melatonin secretion may contribute to sleep disturbances and delirium in postoperative or critically ill patients . It is important to note that both opioids and benzodiazepines affect melatonin secretion. A study performed with an animal model (bovine pineal glands) showed that morphine significantly increased the activity of N-acetyltransferase, promoting melatonin synthesis . Melatonin levels decrease with benzodiazepines in humans via the GABA system [109, 110]. Researchers noted that chronic benzodiazepine administration reduced melatonin through a reduction in the activity of N-acetyltransferase in a rat model . However, findings related to the effects of opioids and benzodiazepines on melatonin are equivocal. Gogenur and colleagues  did not find a correlation between opioid doses and melatonin level in 11 patients undergoing major abdominal surgery. In addition, Frisk et al.  found a significant difference in excretion of 6-SMT (melatonin metabolite); excretion was higher with benzodiazepine therapy than with opioid or propofol therapy in an analyses of 257 collection periods. Larger studies are needed to better elucidate the effect of opioids and benzodiazepines on melatonin and Process S (sleep) and Process C (circadian) effects as well as delirium.
Despite significant advances in our understanding of the sleep–wake cycle and delirium mechanisms as well as how both influence ICU patient outcomes, significant gaps remain requiring elucidation. Largely unknown are the relationships between sleep deprivation and delirium; the interaction of sedatives and opioid analgesics with sleep and delirium; the effects of long-term continuous sedation and analgesia on sleep and delirium; the importance of sleep in the recovery of ICU patients; the impact of sleep fragmentation and delirium on patient outcomes; and the most valid and reliable method to measure sleep stages in ICU patients. Moreover, most hypotheses for sleep and delirium mechanisms have been established from studies in non-ICU patients. Therefore, studies are needed to test hypotheses in ICU patients. A better understanding of these mechanisms, as well as the factors that contribute to both, can guide the development of new methods and models for prevention and treatment that consequently improve in ICU patient outcomes.
We thank Dr. Richard S. Bourne for his review of a prior rendition of this manuscript.