Sleep in the Intensive Care Unit: Strategies for Improvement

Abstract

Sleep in the intensive care unit (ICU) is considered to be subjectively poor, highly fragmented, and sometimes referred to as “atypical.”Although sleep is felt to be crucial for patient recovery, little is known about the association of sleep with physiologic function among critically ill patients, or those with clinically important outcomes in the ICU. Research involving ICU-based sleep disturbance is challenging due to the lack of objective, practical, reliable, and scalable methods to measure sleep and the multifactorial etiologies of its disruption. Despite these challenges, research into sleep-promoting techniques is growing and has demonstrated a variety of causes leading to ICU-related sleep loss, thereby motivating multifaceted intervention efforts. Through a focused review of (1) sleep measurement in the ICU; (2) outcomes related to poor sleep in the ICU; and (3) ICU-based sleep promotion efforts including environmental, nonpharmacological, and pharmacological interventions, this paper examines research regarding sleep in the ICU and highlights the need for future investigation into this complex and dynamic field.

Sleep in the intensive care unit (ICU) is subjectively poor, highly fragmented, and sometimes referred to as “atypical.”^1–3 Sleep deprivation is linked to impaired immune function,⁴ hyperalgesia,⁵ and neurocognitive dysfunction including delirium.^6,7 Given the effects of sleep loss, one could assume that poor sleep would adversely impact critically ill patient outcomes. Hence, sleep promotion is gaining popularity as a method to improve ICU outcomes.

Despite advances in design and technology, the ICU is not conducive to sleep. Poor sleep in the ICU has been attributed to sound pollution, inappropriate lighting, patient interactions, medications, and critical illness itself.^8,9 Rising awareness regarding sleep quality has prompted ICUs to develop multifaceted interventions to promote sleep. This interest in sleep has also motivated research focused on novel modes of sleep measurement, the impact of medications and mechanical ventilation on sleep, short- and long-term consequences of ICU-based sleep disruption, sleep environment design, and implementation of sleep-promoting interventions.

This review aims at providing an update on sleep in the ICU, which is focused on the following three specific topics: (1) sleep measurement in the ICU; (2) outcomes related to poor sleep in the ICU; and (3) ICU-based sleep promotion efforts including environmental, nonpharmacological, and pharmacological interventions.

Measuring Sleep in the ICU

Measurement of sleep in critically ill patients is extremely complex and represents a key barrier to ICU investigations involving sleep. Measurement of ICU sleep on a large scale is not only logistically challenging, but common medications, organ dysfunction, and critical illness itself can also affect sleep classification and quantification. For this reason, there is no widely accepted “gold standard” for sleep measurement in the ICU.

Polysomnography

Polysomnography (PSG), the gold standard for sleep evaluation in noncritically ill patients, utilizes electroencephalography (EEG), electrooculography (EOG), and electromyography (EMG) to characterize sleep. In healthy adults, sleep is divided into rapid eye movement (REM) and non-rapid eye movement (NREM), comprising N1 (transition from wakefulness to sleep), N2 (stable sleep), and N3/slow wave sleep (SWS; restorative deep sleep), which cycle over 90- to 100-minute periods over the course of the night.^10,11 PSG demonstrates that critically ill ICU patients exhibit severely deranged sleep, with a preponderance of N1 and significantly diminished or absent REM and SWS (►Fig. 1).^12–15 However, with application of traditional PSG scoring criteria, atypical EEG patterns, seen in up to 85% of nighttime recordings in critically ill patients,³ have no clearly assignable sleep stage.^16,17 For example, sleep spindles, which characterize N2 sleep, can be induced by benzodiazepines, making the distinction between sleep and sedation challenging. Similarly, diffuse EEG slowing, which is common in the ICU, can be difficult to distinguish from the slow wave activity of N3 sleep. Recently, Drouot et al used PSG data from 57 conscious, nonsedated ICU patients receiving mechanical or noninvasive ventilation to propose a new ICU-specific PSG scoring algorithm, adding atypical sleep and pathologic wakefulness stages to the preexisting scoring system.¹⁸ In addition, Watson and colleagues observed a dissociation between PSG-derived sleep staging and observed sleep–wake behavior in 37 critically ill, mechanically ventilated ICU patients. In this study, some PSG-derived epochs were scored as sleep despite observed patient movements consistent with wakefulness. This observation led to a proposed algorithm of PSG scoring among the critically ill and further defined subdivisions of atypical sleep.³

Fig. 1.

Polysomnography-defined sleep stages. Total sleep time in prototypical healthy¹⁰ versus critically ill adults.¹² Percentages based on nocturnal PSG in healthy adults and 24-hour PSG in ICU patients. ICU, intensive care unit; PSG, polysomnography; REM. rapid eye movement; SWS, slow wave sleep.

Complicating PSG further, this mode of measurement is expensive, labor intensive, and not feasible for large-scale use in the ICU setting.^12,19–22 A single-center medical ICU feasibility study found that only 2 of 29 patients were able to complete 24 hours of unattended portable PSG monitoring.¹⁹ Alternatives to PSG have been attempted, including application of EEG alone. However, compared with PSG, differentiation of wake versus N1 and N1 versus N2 is difficult with EEG only, especially in the setting of highly fragmented sleep.^23,24

Bispectral Index

Bispectral index (BIS) involves forehead placement of a single-multichannel EEG lead, with integrated processed data to produce a single integer ranging from 0 (deep coma) to 100 (wide awake).²⁵ Traditionally used to monitor sedation depth during general anesthesia, recent studies, involving BIS to measure sleep in critically ill patients observed overlapping values for various sleep stages.²⁶ In addition, misleading recordings were noticed in the presence of common ICU derangements, such as traumatic brain injury, polypharmacy, delirium, and dementia. Hence, despite its ease of use, BIS is not validated for sleep measurement in critically ill patients.^25,27,28

Actigraphy

Actigraphy has gained popularity as a minimally invasive, low-cost option for estimating sleep. Utilizing a wristwatch-like accelerometer, actigraphy measures patient activity levels corresponding to wake and rest states. Actigraphy is well tolerated for long-term, unattended use and is validated for measuring sleep in community populations^29–32; however, its use in critically ill patients is problematic as activity levels are impacted by common ICU factors, such as sedating medications, weakness, delirium, and routine staff interventions (i.e., turning and bathing). Moreover, for mostly sedentary ICU patients, traditional actigraphy scoring algorithms can overestimate total sleep time (TST) and sleep efficiency (SE).^33–35 Despite these limitations, the feasibility and accessibility of actigraphy support its potential use in the ICU setting, both as an objective measure of circadian rest–activity rhythms, and pending the development and validation of ICU-specific scoring algorithms, a surrogate marker of sleep and wake.^36,37

Questionnaires

Subjectively, sleep questionnaires are easy to complete, low cost, and feasible for large-scale use. In the ICU, the most commonly used questionnaires are the Richards–Campbell Sleep Questionnaire (RCSQ) which was previously validated against PSG, and the Verran and Snyder-Halpern Sleep Scale (VSH).^38–40 While both questionnaires utilize easy-to-administer visual–analogue scales evaluating various sleep domains, they have limited reliability due to recall bias, altered patient cognition, rater fatigue, and inability to capture daytime sleep. In the setting of cognitive impairment (i.e., delirium and dementia), bedside nurses can complete questionnaires on their patients’ behalf; however, patient–nurse interrater reliability of the RCSQ is modest at best.⁴¹ Finally, the Sleep in Intensive Care Unit Questionnaire (SICUQ), which involves patient-reported ratings of causes of ICU-related sleep disruption, can be used to highlight areas for improvement and the effectiveness of sleep-promoting interventions.^42–45

Physiological Effects of ICU-Related Sleep Disruption

Although sleep is felt to be crucial for patient recovery, little is known about the association of sleep with physiologic function in critically ill patients, or those with other clinically important outcomes in the ICU (►Fig. 2). Below is a systems-based review of sleep’s potential role in critical illness. Studies involving ICU patients are reviewed, although sleep data in this population are lacking. Additional non-ICU data pertinent to the topics are also included.⁴⁶

Fig. 2.

Causes and consequences of sleep-wake rhythm disturbance in the ICU. ICU, intensive care unit (adapted from Kamdar et al⁴⁶).

Delirium

Delirium, a mental disturbance characterized by acute onset of fluctuating degrees of confusion, inattention, and levels of consciousness, is a common ICU syndrome believed to be closely linked to poor sleep.^47–50 Delirium affects 60 to 80% of mechanically ventilated and 20 to 50% of nonventilated ICU patients,^51–53 and is independently associated with prolonged length of stay, long-term cognitive impairments, and early death.^48,53–58 Sleep deprivation is widely considered to be associated with delirium, given their similar symptoms, and shared characteristics including imbalances in neurotransmitters dopamine and acetylcholine and circadian misalignment, as evidenced by altered diurnal melatonin secretion.^48,59,60 Despite these mechanisms, data supporting a sleep–delirium relationship are lacking. One study involving single-night PSG in 29 mechanically ventilated surgical ICU patients demonstrated higher incidence of delirium among patients experiencing < 6% REM.⁶¹ Furthermore, observational studies describe increased delirium incidence in patients reporting poor sleep following thoracic surgery,⁶² as well as in patients, with nurse-described fragmented sleep.⁶³

Pulmonary

Of specific concern for critically ill patients is the influence of poor sleep on respiratory muscle function, respiratory drive, and upper airway collapsibility. In a study of mechanically ventilated patients, atypical (vs. more typical) sleep was associated with delayed ventilator liberation, despite a lack of substantial differences between groups in maximal inspiratory pressure and negative airway pressure.⁶⁴ A recent study failed to detect differences in pathological wakefulness or atypical sleep between patients who were liberated from the ventilator and those who were not. However, compared with patients who remained intubated, patients liberated from the ventilator demonstrated less right-left hemispheric discordance, and more wakefulness 15 hours prior to extubation.⁶⁵ In addition, in patients treated for acute hypercarbic respiratory failure, poor sleep predicted late (>48 hours after initiation) noninvasive ventilation failure.⁶⁶ To date, no studies have linked sleep improvement efforts with ventilator or other respiratory outcomes.

Metabolic

Normally, growth hormone (GH), an anabolic hormone synthesized by the anterior pituitary gland, is cyclically secreted in response to sleep or physiologic stress.⁶⁷ In acute illness, GH levels rise; however, in chronic illness, levels decrease to normal or subnormal levels signaling “wasting syndrome” in which anabolic activity is impaired.⁶⁸ Interestingly, GH secretion is delayed until after sleep onset, which is markedly altered in states of sleep deprivation. Thus, sleep promotion may be proposed as a method to attenuate ICU-related myopathy.⁶⁹

Cardiovascular

No studies have evaluated the association of ICU-related sleep loss with cardiovascular function. In community-dwelling adults, acute sleep deprivation increases sympathetic tone,^70–73 and the risk of cardiovascular complications including hypertension, tachyarrhythmias, and QT interval prolongation.^74–76 Multiorgan derangements, medications, and stress also contribute to cardiac events and complicate research in this area.

Immunologic

Both innate and humoral immune systems are negatively impacted by sleep deprivation; however, longitudinal ICU studies involving clinically important endpoints are lacking. In healthy adults, sleep loss is associated with decreased natural killer cell activity and suppressed interleukin (IL)-2 levels.⁷⁷ In addition, sleep deprivation impairs normal humoral responses to vaccination.^78,79 Taken together, in states of critical illness such as sepsis where host defense is vital to recovery, sleep loss and promotion may play an important role.

Strategies for ICU-Based Sleep Promotion

The variety of causes of ICU-related sleep loss have motivated numerous sleep-promoting intervention efforts (►Table 1). In the ICU setting, sound and light reduction, behavioral modification, bundles, and pharmacologic interventions are becoming increasingly utilized.

Table 1.

Nonpharmacological sleep-promoting interventions

Intervention	First author	Year	Study type	Sample size	ICU Type	Measurement tool	Outcome
Behavior modification	Kahn et al80	1998	Pre–post	All medical ICU staff	Medical	Sound levels	↓ Mean peak sound
	Olson et al81	2001	CCT	843	Neuro	Nurse observation	↑ Sleep quantity
	Monsén and Edéll-Gustafsson82	2005	CCT	23	Neuro	Sound Levels	↓ mean peak sound
Behavior modification + noise/light reduction + clustering of care	Walder et al83	2000	Pre–post	17	Surgical	Sound levels Light levels	↓ peak sound No change in light
Behavioral modification + earplugs + eye mask	Kamdar et al45	2013	Pre–post	300	Medical	RCSQ	No difference in sleep quality ↓ subjective sound ↓ delirium
	Patel et al44	2014	Pre–post	59	Mixed	RCSQ Sound levels Light levels	↑ Sleep quality ↓ mean nocturnal sound + light ↓ delirium
	Boyko et al84	2017	RCT	17	Mixed	PSG Sound levels	No differences in sleep or mean sound level
Earplugs	Haddock85	1994	RCT	18	Gynecology	Custom questionnaire	↑ Sleep quality
	Scotto et al86	2009	Quasi-experimental	88	Mixed	VHS
	Neyse et al87	2011	CCT	60	Cardiac	PSQI
	Van Rompaey et al88	2012	RCT	136	Unspecified	Custom questionnaire
Earplugs + eye mask	Richardson et al89	2007	Quasi-experimental	64	CT Surgery	Custom questionnaire	↑ Sleep quality in some patients
	Ryu et al90	2011	Quasi-experimental	58	Cardiac	VHS	↑ Sleep quality
	Jones and Dawson91	2012	Pre–post	100	Surgical	Custom questionnaire
	Le Guen et al92	2014	RCT	41	PACU	SSQ MOSS Actigraphy
	Yazdannik et al93	2014	CCT	50	Unspecified	VHS
	Dave et al94	2015	Crossover	50	Cardiac	RCSQ
	Mashayekhi et al95	2013	Crossover	90	Cardiac	VHS
	Bajwa et al96	2015	RCT	100	Unspecified	VHS
Earplug + eye mask + music	Hu et al97	2015	RCT	45	CT Surgery	RCSQ	↑ Sleep quality
Eye mask	Babaii98	2015	RTC	60	Cardiac	PSQI
	Mashayekhi et al99	2012	Crossover	60	Cardiac	VHS
	Daneshmandi et al100	2012	CCT	60	Cardiac	PSQI
Eye mask + headphones + melatonin	Foreman et al101	2015	RCT	12	Neuro	PSG	UTD due to abnormal EEG
Music versus noise cancelling headphones	Chlan et al102	2013	RCT	373	Unspecified	VAS-A Sedative dosage and frequency	↓ Anxiety and sedative use in music group only
Music + daytime rest period	Hansen et al103	2017	RCT	37	Mixed	RCSQ	↑ Subjective sleep
Music	Su et al104	2012	RCT	28	Medical	PSG VSH	↑ SWS ↑ subjective sleep
White noise	Gragert105	1990	CCT	40	Elderly cardiac	RCSQ	↑ Subjective sleep
	Williamson106	1992	CCT	60	General ward postCABG	RCSQ
	Afshar et al107	2016	Quasi-experimental	60	Cardiac	PSQI
Massage	Richards108	1998	RCT	71	Unspecified	PSG	↑ Sleep efficiency in older men
	Nerbass et al109	2010	RCT	57	Cardiac	VAS sleep	↑ Sleep quality
	Oshvandi et al110	2014	RCT	60	Cardiac	SMHSQ
	Shinde and Anjum111	2014	Prepost	60	Unspecified	Modified GSQS Observation
	Hsu et al112	2019	Quasi-experimental	60	Unspecified	VHS Actigraphy Observation
Aromatherapy	Moeini et al113	2010	RCT	64	Cardiac	SMHSQ
	Cho et al114	2013	Prepost	56	Cardiac	VHS
	Hajibagheri et al115	2014	RCT	60	Cardiac	PSQI
	Karadag et al116	2015	RCT	60	Cardiac	PSQI
	Cho et al117	2017	CCT	60	Unspecified	VHS
Valerian acupressure	Chen et al118	2012	RCT	85	Unspecified	Actigraphy SSS Nurse Observation	↑ Actigraphic sleep ↓ waking frequency
Valerian acupressure oil versus massage	Bagheri-Nesami et al119	2015	RCT	90	Cardiac	SMHSQ	↑ Sleep quality w/valerian acupressure oil
Foot reflexology massage ± foot bath versus control	Rahmani et al120	2016	Quasi-experimental	140	Cardiac	VSH	↑ Subjective sleep with foot re- flexology massage foot bath

Abbreviations: CCT, controlled clinical trial; CT, cardiothoracic; EEG, electroencephalogram; GSQS, Groningen’s sleep quality scale; MOSS, Michigan opiate safety score; PACU, postanesthesia care unit; PSG, polysomnography; PSQI, Pittsburgh sleep question index; RCSQ, Richards-Campbell sleep questionnaire; RCT, randomized control trial; REM, rapid eye movement; SMHSQ, St. Mary’s Hospital sleep questionnaire; SSQ, Spiegel sleep questionnaire; SSS, stanford sleepiness scale; UTD, unable to determine; VAS-A, visual analog scale for anxiety; VHS, Verran and Snyder-Halpern sleep scale.

^aExcludes studies performed in simulated ICU environments.

Nonpharmacological Interventions

Sound in the ICU

Excessive environmental sound is associated with 17 to 20% of awakenings in ICU patients.^9,15 The World Health Organization (WHO) recommends that continuous background sound in patient treatment areas not exceed 30 dB (dB), and peak nocturnal sounds remain below 40 dB.¹²¹ Notably, ventilator (51 dB), suction (53 dB), and syringe pump alarms (63 dB) breach these sound limits.⁸ Unsurprisingly, all ICU studies involving sound report levels exceeded recommended limits ^8,122,123 without diurnal variation.^9,15,124,125 Studies have proposed that characteristics of sound, including sources (i.e., electronic vs. environmental), timing and levels, affect sleep.¹²⁶ In addition, the degree of sound level variation, as compared with peak and average sound levels, may be particularly disruptive to sleep.^123,127

In the ICU, sound attenuation can be accomplished by (1) behavioral interventions, that is, minimizing conversations, reducing unnecessary alarms; (2) sound masking, that is, white noise; and (3) sound blocking, that is, earplugs. Behavior-based sound reduction studies comparing staff education, education with sound monitoring, and education with environmental modifications had mixed results, with no improvement in sound levels in some studies¹²⁸ and a reduction in others (►Table 1).^80,129 Notably, sound reduction interventions did not evaluate sustainability, and decibel levels commonly remained above recommended limits.^80,130

While sound masking with white noise has been demonstrated to improve sleep quality,^105–107 sound blocking via earplugs has received particular attention due to recent research suggesting improved sleep perception, satisfaction, and delirium in ICU patients wearing earplugs.^86,88 However, patients offered earplugs usually declined them⁴⁵ or found them uncomfortable.¹³¹ More recently, sound blocking via noise cancelling headphones has gained popularity. A recent study involving mannequins exposed to simulated cardiac ICU sounds found that noise cancelling headphones were associated with significant reductions in sound exposure.¹³² Investigations on noise cancelling headphones in critically ill patients included one study yielding uninterpretable results due to highly atypical EEG patterns,¹⁰¹ and another focused on anxiety and sedation levels rather than sleep effects.¹⁰² More research is required to determine the best methods for noise reduction in critically ill patients.

Light in the ICU

It is widely believed that abnormal light levels, whether elevated throughout the night, delivered as bright bursts interrupting nighttime darkness or inappropriately low during the daytime, contribute to sleep–wake rhythm disturbance in the ICU. Although there are no guidelines for light levels in the ICU, nocturnal light levels in this setting are, on average, below the arbitrary maximum value of 20 lux.^133,134 However, severe illness is associated with increased high-lux light exposure, which may contribute to sleep fragmentation.¹³⁵ Studies support improved subjective sleep with the use of eye masks in cardiac ICU patients.^98,100

Recently, inadequate daytime light has gained attention for its disruption of circadian sleep–wake cycles. One study evaluating natural light in the ICU found that, although 84% of patient rooms had windows, almost 100% of this natural light degraded before reaching the patients.¹³⁶ Other studies demonstrated ICU daytime light levels ranging from 42 to 158 lux, corresponding to insufficient daytime light.^123,133,134 Finally, the application of daytime bright light therapy is gaining attention, albeit mostly in the context of delirium prevention and circadian rhythm maintenance.^137,138

Alternative Therapies

Alternative therapies for sleep promotion including music, massage, reflexology, aromatherapy, implementation of relaxation techniques, and therapeutic touch have been shown to reduce stress and improve subjective sleep quality (►Table 1).^139–141 Despite favorable effects on sleep in critically ill patients, the 2018 Clinical Practice Guidelines for the Prevention and Management of Pain, Agitation/Sedation, Delirium, Immobility, and Sleep Disruption (PADIS) in adult patients in the ICU do not recommend the individual use of aromatherapy, massage, or music for the promotion of sleep in the ICU due to limited availability and potential respiratory complications from aromatherapy.¹⁴²

Bundled Interventions

Bundling is a practical approach involving implementation of multiple simple interventions at once. The 2018 PADIS guidelines conditionally recommend sleep promoting “bundles,” based on evidence suggesting that bundles involving sound and light reduction and clustering of care may improve sleep quality, ^{89,91–94,96,142–144} while the addition of music, earplugs, eye masks, pharmacological sleep aid guidelines, room modifications (see section Emerging Interventions) and daytime mobilization practices may improve outcomes further.^{44,45,90,97,142,143} While use of established quality improvement methods have been shown to aid in implementation of such bundles,¹⁴⁵ few studies have evaluated strategies to optimize bundle sustainability.¹⁴⁶

Ventilators and Sleep

Given that up to 40% of ICU patients are mechanically ventilated,¹⁴⁷ optimizing ventilator settings is vital to prevent nocturnal sleep disruptions and facilitate sedation minimization necessary for daytime wakefulness. Controlled modes of mechanical ventilation such as pressure-controlled ventilation (PCV), assist-control ventilation (ACV), or proportional assist ventilation (PAV) are associated with improved sleep quality compared with spontaneous modes such as pressure support ventilation (PSV).^148–151 Additionally, in tracheostomized patients undergoing ventilator weaning, controlled (vs. spontaneous) modes of ventilation are associated with improved sleep quality.¹⁵² However, despite this evidence, studies of ventilator modes and sleep have revealed inconsistent findings.¹⁵³ The 2018 PADIS guidelines conditionally recommend assist control ventilation for nocturnal sleep;¹⁴² however, further research is required, including ventilator settings needed to limit sleep-altering sedative medications and promote consolidative sleep.

Emerging Interventions

Staffing models, unit layout, and visitation policies have the potential to impact sleep in critically ill patients. While ICU patients can experience up to 40 interactions each night,^154,155 it is estimated that up to 14% of interruptions can safely be omitted.¹⁵⁶ Thus, clustering of care and quiet hours are becoming commonplace, but further evaluation of staffing and workflow patterns is needed.

Hospital layout can affect sound and light in the ICU, thus impacting patient sleep. Closely spaced beds facing a central nurse station strongly contribute to sound pollution, which may guide future ICU layout decisions.¹⁵⁷ Furthermore, windows and natural light are associated with decreased pain, agitation, depressive symptoms, and improved circadian rhythmicity, but have yet to be evaluated in the context of sleep.¹⁵⁸

Finally, as ICUs liberalize their visitation policies, there is potential for sleep disruption due to nighttime visitors; however, family visitation can reduce patient and visitor stress and anxiety,¹⁵⁹ and did not negatively impact patients’ ability to rest.¹⁶⁰ Some have proposed using family members to engage patients during the day and protect their loved ones’ sleep at night.¹⁶¹ To date, no studies have evaluated the association of visitation policies, or specific visitors, with sleep in the ICU.

Pharmacologic Interventions

Initial pharmacologic management of poor sleep in the ICU involves deprescribing of medications known to perturb sleep. Medications commonly prescribed for sleep are often accompanied by untoward side effects, particularly in older and critically ill patients. The Beers criteria for inappropriate medications in older adults, updated in 2015 by the American Geriatrics Society, discourage the use of most sleep-promoting medications in the aged population.¹⁶² Nevertheless, medications are often administered in the ICU to promote sleep. The most commonly prescribed ICU medications associated with sleep are summarized below and in ►Table 2. Notably, studies involving medication effects on sleep were generally small in size, involved non-ICU based populations, and limited to single night sleep evaluation; therefore, they should be interpreted with caution.

Table 2.

Medications commonly used to promote sleep

Medication	Mechanism of action	Route of administration	Side effects	Sleep effects
Dexmedetomidine	a2-agonist	Intravenous	Bradycardia, hypotension	↑N2 with sleep spindles, ?↑N3/SWS, ↓REM, ↑SE, ↓SL
Propofol	GABA receptor agonist	Intravenous	Bradycardia, hypotension, propofol infusion syndrome, respiratory depression	↓REM, ↓SL, ↑TST, ↓W
Opiates	CNS opioid receptor agonist	Oral or Intravenous	Dependency, delirium-inducing, hypotension, respiratory depression, withdrawal	↓N3, ↓REM, ↓TST, ↑W
Melatonin and melatonin receptor agonists	Melatonin 1 and 2 receptor agonist	Oral	Dizziness, hallucinations, nausea, vivid dreams	↑SE, ↓SL, ↑TST
Atypical antipsychotics	5HT2, D2–receptor antagonist	Oral	Dizziness, extrapyramidal Symptoms, neuroleptic malignant syndrome, orthostatic hypotension	↑N3, ↑REM, ↑SE, ↓SL, ↑TST, ↓W
Typical antipsychotics	Dopamine receptor antagonist	Oral or intravenous	Anticholinergic effects, extrapyramidal symptoms, neuroleptic malignant syndrome, QT prolongation, tardive dyskinesia	↑N2, ↑N3, ↑SE, ↓SL, ↑TST, ↓W
Trazodone	Serotonin reuptake inhibitor, 5-HT1A,1C,2, H1 receptor antagonist	Oral	Anticholinergic syndrome, arrhythmias, orthostatic hypotension	↑N3, ↑↓REM, ?↑SE, ↓SL
Antihistamines	H1-receptor antagonist	Oral or intravenous	Anticholinergic syndrome, dizziness, impaired coordination	?↑N3, ↓REM, ?↑SE, ↓SL
Benzodiazepines	GABA receptor agonist	Oral or intravenous	Dependency, delirium-inducing, dizziness, hypotension, withdrawal	↓N3, ↓REM, ↓SL, ↑TST, ↓W
Nonbenzodiazepine hypnotics	GABA receptor agonist	Oral	Daytime somnolence, dizziness, confusion	↓N2, ↓N3, ↑↓REM, ↓SL ↑TST, ↓W

Abbreviations: CNS, central nervous system; GABA, gamma-aminobutyric acid; N2, deeper sleep; N3/SWS, restorative, slow wave sleep; REM, rapid eye movement; SE, sleep efficiency; SL, sleep latency; TST, total sleep time; W, wake;↓, decreased; ↑, increased; ↑↓, equivocal; ?↑, may increase; ?↓, may decrease.

Dexmedetomidine

Dexmedetomidine, an α−2-adrenergic agonist, is a hypnotic medication commonly used in the setting of mechanical ventilation, agitation, or alcohol withdrawal. Studies involving dexmedetomidine for sleep promotion uniformly demonstrate increased N2,^163–165 with inconsistent findings regarding SE and N3/SWS.^{163,164,166,167} Furthermore, one study demonstrated presence of sleep spindles in patients receiving dexmedetomidine, arguing for its role in approximating “natural sleep.”¹⁶⁵ To our knowledge, no study has demonstrated improvement in REM with dexmedetomidine.¹⁶⁶ The 2018 PADIS guidelines make no recommendation on dexmedetomidine for sleep promotion in the ICU; however, it is recommended when agitation precludes ventilator liberation.¹⁴² Notably, two recent randomized controlled trials (RCTs) were performed involving dexmedetomidine in patients requiring sedation; one demonstrating no difference in 90-day mortality with a dexmedetomidine-only (with additional as-needed sedative medications, versus usual care) sedation strategy,¹⁶⁸ and a second demonstrating reduced ICU delirium, but no difference in subjective sleep, when comparing low-dose nocturnal dexmedetomidine with placebo.¹⁶⁹ Larger studies involving dexmedetomidine and its impact on sleep in the ICU are needed.

Propofol

Propofol is a rapidly titratable GABA_A receptor potentiator frequently used fordeep sedation and anesthesia. Although it may produce a clinical state of unconsciousness, propofol has been associated with decreased REM when compared with the same critically ill patients who did not receive propofol¹⁷⁰ as well as when compared with nocturnal sedation with benzodiazepine flunitrazepam.¹⁷¹ Upon an examination of four studies evaluating propofol for sleep promotion in the ICU, a Cochrane review found insufficient evidence to determine whether propofol improved sleep quality or quantity in critically ill patients.¹⁷² The 2018 PADIS guidelines do not recommend propofol for sleep in critically ill patients.¹⁴²

Melatonin and Melatonin Receptor Agonists

Melatonin and melatonin receptor agonists such as ramelteon are gaining popularity for sleep promotion in the ICU (►Tables 2 and 3). Studies demonstrate reduced delirium in ICU patients receiving melatonin or ramelteon.^173,174 Further studies are needed to evaluate melatonin and melatonin receptor agonists for their role in sleep promotion in the ICU, along with its link to delirium prevention. Due to insufficient evidence, a recent Cochrane review was unable to determine whether melatonin improved sleep quality or quantity in critically ill patients,¹⁷⁵ and the 2018 PADIS guidelines make no recommendation regarding their use in sleep.¹⁴²

Table 3.

Trials of melatonin and ramelteon (melatonin-receptor agonist) in critically ill patients

Medication	First author year	Study type	Sample size	Population	Dose (mg)	Medication timing	Assessment	Results
Melatonin	Shilo176 2000	RCT	8	ICU	3	22:00	Actigraphy	↑ Sleep quality ↑ sleep time
	Ibrahim177 2006	RCT	32	ICU tracheostomy	3	20:00	Nurse observation	No change in nocturnal sleep duration
	Bourne178 2008	RCT	24	ICU tracheostomy	10	21:00	BIS	↓ BIS AUC ↑ sleep quality ↑ sleep time
	Mistraletti179 2010	RCT	82	Mixed ICU mechanically ventilated	3 and 3	20:00 and 00:00	Nurse observation	↑ Nocturnal TST
	Huang131 2015	NR	40	Healthy volunteers subjected to ICU environment	1	21:00	PSG	↑ REM ↑ TST ↓ SOL ↓ awakenings
	Foreman101 2015	RCT	12	Neuro ICU	3	20:00	EEG	UTD; only one patient in each group had scorable sleep
Ramelteon	Hatta173 2014	RCT	67	Elderly patients ICU and general wards	8	21:00	DRS-R-98	↓ Delirium
	Nishikimi174 2018	RCT	88	ICU	8	20:00	CAM-ICU nurse observation	↓ Delirium ↓ nocturnal awakenings Trend toward ↓ ICU LOS

Abbreviations: BIS, bispectral index area; CAM-ICU, confusion assessment method-ICU; DRS-R-98, delirium rating scale-revised-98; EEG, electroencephalogram; ICU, intensive care unit; LOS, length of stay; NR, nonrandomized; PSG, polysomnography; RCT, randomized control trial; SE, sleep efficiency; SOL, sleep onset latency; TST, total sleep time; WASO, wake after sleep onset.

Atypical Antipsychotics

The sedative–hypnotic properties of atypical antipsychotic medications appear to be dose-dependent and related to histamine 1 receptor blockade.¹⁸⁰ The use of atypical antipsychotic agents for sleep alone is not Food and Drug Administration (FDA) approved; however, their use is common in the ICU in the setting of delirium or agitation. Olanzapine and quetiapine are the most studied and efficacious sleep-promoting atypical antipsychotics due to their moderate sedative properties.^180,181

No studies have evaluated atypical antipsychotics for sleep in critically ill patients. In healthy subjects, quetiapine increased TST, SE, and subjective sleep quality when compared with placebo.^182,183 In addition, N2 and REM sleep increased when quetiapine (25 mg) was administered, but REM subsequently decreased at higher doses (100 mg).¹⁸² Olanzapine increased nocturnal TST, SE, and SWS.^184–186 Olanzapine’s effect on REM is unclear, with some studies showing increased REM¹⁸⁴ and others showing decreased REM and increased REM latency.^185,186 Despite improvements in PSG and subjective sleep, the use of atypical antipsychotics for sleep remains off-label and requires further investigation.

Typical Antipsychotics

Haloperidol, a typical antipsychotic medication commonly used in the ICU for treatment of hyperactive delirium and agitation, has sedating properties. Few studies evaluating its efficacy as a sleep aid in this population exist; however, one study evaluating haloperidol in pediatric burn patients found increased TST and N2 compared with control, with no effect on SWS or REM.¹⁸⁷ For delirium treatment and prevention, haloperidol does not decrease incidence or duration of delirium when compared with placebo.^188–192

Trazodone

Trazodone, a selective serotonin reuptake inhibitor and histamine receptor type-1 channel blocker, is an antidepressant–hypnotic. Despite its off-label use, trazodone remains among the most commonly prescribed sleep aids in the United States.^193,194 Trazodone’s effect on sleep in critically ill patients has not been studied. Various small studies demonstrate increased SWS following administration of trazodone in healthy^195–198 and depressed populations.^199–203 One systematic review supported the use of trazodone for insomnia,²⁰⁴ while another felt the benefits did not outweigh the risks.²⁰⁵ Trazodone is generally well tolerated; however, side effects of arrhythmias, orthostatic hypotension and anticholinergic syndrome have been reported, especially in older patients.²⁰⁴

Benzodiazepines

Benzodiazepines bind to gamma-aminobutyric acid (GABA)_A receptors and enhance the effect of GABA, producing anxiolysis and sedation. There are five FDA-approved benzodiazepines used to treat short-term insomnia as follows: estazolam, flurazepam, quazepam, temazepam, and triazolam.¹⁹⁴ While benzodiazepines may improve sleep onset latency (SOL) and TST, and decrease nocturnal arousals in healthy patients, these drugs decrease SWS and REM.^206,207 Additionally, in vulnerable populations including older adults, benzodiazepines are associated with increased delirium.^208,209 Given the detrimental effects of benzodiazepines, including their risk of delirium and dependence, as well as counterproductive effects on sleep, these drugs are not encouraged for sleep promotion in critically ill patients.

Nonbenzodiazepine Hypnotics

Nonbenzodiazepine sedative–hypnotics, commonly known as “z-drugs,” such as zolpidem, eszoplicone, and zaleplon are similar to benzodiazepines in that they bind GABA_A receptors and potentiate the effect of GABA. In a meta-analysis of “z-drugs” in community-based insomniacs, SOL was only mildly decreased in the treatment group compared with the control.²¹⁰ Recently, the FDA issued a boxed warning for “z-drugs” due to multiple reports of dangerous side effects.²¹¹ In hospitalized patients, these side effects include dizziness, daytime somnolence, falls, and confusion, particularly in the older population.^212–214 Notably, following hip or knee replacement, postoperative use of zolpidem failed to increase REM or SWS compared with placebo.²¹⁵ Such results, combined with safety warnings, raise the question concerning the utility of these drugs for sleep, particularly in at-risk critically ill patients.

Future Directions

Improving our understanding of the ill effects of poor sleep in the ICU and the effects of sleep-promoting interventions on outcomes is an important priority in critical care research. The lack of objective, practical, reliable, and scalable methods to measure sleep poses a major hindrance to large-scale, ICU-based sleep investigations. Novel modes of sleep measurement are needed, in addition to the development of ICU-specific sleep scoring algorithms. Research on sleep in the ICU is growing and has demonstrated a variety of causes leading to ICU-related sleep loss, including common medications, thereby motivating numerous multifaceted intervention efforts. Among these interventions, bundled protocols have been shown to be feasible, and efficacious; however, their generalizability and sustainability require further evaluation.