Sleep quality and quantity determined by polysomnography in mechanically ventilated critically ill patients randomized to dexmedetomidine or placebo

Abstract

Background

Abnormal sleep is commonly observed in the ICU and is associated with delirium and increased mortality. If sedation is necessary, it is often performed with gamma‐aminobutyric acid agonists such as propofol or midazolam leading to an absence of restorative sleep. We aim to evaluate the effect of dexmedetomidine on sleep quality and quantity.

Methods

Thirty consecutive patients were included. The study was conducted as a double‐blinded, randomized, placebo‐controlled trial with two parallel groups: 20 patients were treated with dexmedetomidine, and 10 with placebo. Two 16 h of polysomnography recordings were done for each patient on two consecutive nights. Patients were randomized to dexmedetomidine or placebo after the first recording, thus providing a control recording for all patients. Dexmedetomidine was administered during the second recording (6 p.m.–6 a.m.). Objective: To compare the effect of dexmedetomidine versus. placebo on sleep ‐ quality and quantity. Primary outcome: Sleep quality, total sleep time (TST), Sleep efficiency (SE), and Rapid Eye Movement (REM) sleep determined by Polysomnography (PSG). Secondary outcome: Delirium and daytime function determined by Confusion Assessment Method of the Intensive Care Unit and physical activity. Alertness and wakefulness were determined by RASS (Richmond Agitation and Sedation Scale).

Results

SE were increased in the dexmedetomidine group by; 37.6% (29.7;45.6 95% CI) versus 3.7% (−11.4;18.8 95% CI) (p < .001) and TST were prolonged by 271 min. (210;324 95% CI) versus 27 min. (−82;135 95% CI), (p < .001). No significant difference in REM sleep, delirium physical activity, or RASS score was found except for RASS night two.

Conclusion

Total sleep time and sleep efficiency were significantly increased, without elimination of REM sleep, in mechanically ventilated ICU patients randomized to dexmedetomidine, when compared to a control PSG recording performed during non‐sedation/standard care.

Editorial Comment.

Sleep amount and sleep quality affect health and recovery for the critically ill. Using polysomnography recording in mechanically ventilated critically ill patients, this trial demonstrates benefit when treated with dexmedetomidine related to total sleep time and sleep efficiency without affecting Rapid Eye Movement sleep, compared to standard care (non‐sedation).

Abbreviations

AASM

American academy of sleep medicine

APACHE‐II

acute physiology and chronic health evaluation

CAM‐ICU

confusion assessment method for the ICU

CI

confidence interval

COPD

chronic obstructive pulmonary disease

CRF

case report form

DCSM

danish center for sleep medicine

ECG

electrocardiogram

FiO₂

fraction of inspired oxygen

GCP

good clinical practice

HR

heart rate

ICU

intensive care unite

IQR

inter quartile range

LOS

length of stay

MV

mechanical ventilation

NREM

non‐rapid eye movement sleep

NS

non‐significant

PaCO₂

partial pressure of carbon dioxide

PEEP

positive end expiratory pressure

PS

pressure support

PSG

polysomnography

RASS

richmond agitation sedation scale

REM

rapid eye movement sleep

SAE

serious adverse event

SAPS

simplified acute physiologic score

SOFA

sepsis‐related organ failure assessment score

SUSAR

suspected unexpected serious adverse reaction

SWS

slow wave sleep

UTA

unable to assess

1. INTRODUCTION

Abnormal sleep is commonly observed in mechanically ventilated critically ill patients. Patients experience sleep–wake disorganization, sleep fragmentation, circadian rhythm disruption, and abnormal sleep architecture. Even after discharge, patients report sleep disturbance and continue to suffer from poor sleep quality. ¹ , ² , ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ Sleep deprivation is associated with dysfunction of the immune and cardiovascular systems, disturbed metabolism, impaired memory, delirium, and in turn even increased mortality. Hence sedatives are often administered to increase comfort, decrease anxiety, and promote the perception of sleep in critically ill patients. ² , ⁴ , ⁵ , ¹⁰ , ¹¹ , ¹²

Midazolam and propofol are commonly used sedatives. These sedatives are both GABA (gamma‐aminobutyric acid) receptor agonists and they both induce the perception of sleeping by inhibiting neurons in locus coeruleus. However, sedation with GABA agonists does not promote restorative sleep, which must be present to prevent the above‐mentioned complications. These sedatives suppress arousals, SWS (Slow Wave Sleep), and REM (Rapid Eye Movement) sleep stage which further worsens an often already impaired sleep quality. GABA agonists mask the symptoms of sleep deprivation which are then left untreated. ⁴ , ¹³ , ¹⁴ This might explain why GABA agonists have been reported to cause delirium, prolong ICU (Intensive Care Unit) length of stay, and weaning from MV (mechanical ventilation). Especially benzodiazepine has been reported to increase the risk of delirium. ⁸ , ¹⁰ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹

For several years the sedation strategy for mechanically ventilated patients has been a combination of opioids and intravenous infusion of benzodiazepines and/or propofol with consequences as described. Hence recent guidelines have advocated for a revision of ICU sedation practices towards optimal patient comfort with minimal sedation to improve clinical outcomes in mechanically ventilated adult ICU patients. ²⁰ , ²¹ A strategy of light or non‐sedation has proven feasible to avoid affecting restorative sleep and in turn delirium in a recent multicenter study published in the New England Journal of Medicine. ²² This non‐sedation strategy has been applied for several years in the ICU of the present study. The strategy is supported by data demonstrating that good patient comfort can be achieved during MV with no or very light sedation, which is associated with lower incidences of delirium and shorter LOS (Length Of Stay) in the ICU. ²² , ²³ , ²⁴ , ²⁵ However, when sedation for various reasons is inevitable dexmedetomidine has shown to be promising and it has been reported to reduce the risk of delirium in critically ill patients. ²⁶ , ²⁷ , ²⁸ , ²⁹

Dexmedetomidine is a lipophilic imidazole derivative, with an affinity for α₂‐adrenoceptors 1600 times greater than for α₁‐adrenoceptors and an affinity 8 times greater than the prototype α₂ agonist drug, clonidine. ³⁰ Because of the sedative and analgesic properties, it can be used as an alternative in cases where non‐sedation is not applicable. Dexmedetomidine has been reported to inhibit the release of norepinephrine in the locus coeruleus and consequently enhance SWS by mimicking the endogenous NREM sleep pathway. ³¹ Thus, activating central pre‐ and post‐synaptic α₂‐receptors in the locus coeruleus induces a state of unconsciousness like natural sleep that resembles NREM sleep, but without the arousal dynamics as during natural sleep. ³² The action involved is an inhibition of the reticular activation system primarily involving the brainstem, midbrain, and cortex, but to a lesser degree the cardiopulmonary function. This unique site of action lends dexmedetomidine an equally unique sedative profile, conferring the ability to sedate while at the same time allowing the patient to experience arousals during sleep and interaction with relatives and staff during waketime. Dexmedetomidine as a sedative is notable for the lack of respiratory and circulatory suppression. ³³

No study has yet established the effect of dexmedetomidine on sleep quality, and quantity in a randomized study design, but previous studies reported a decreased risk of delirium in patients treated with intravenously infused dexmedetomidine and a causal explanation could be increased TST (Total Sleep Time) and SE (Sleep Efficiency). ¹⁹ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁴ , ³⁵ Hence the objective of this study was to compare sleep quality and quantity in critically ill ICU patients when randomized to treatment with dexmedetomidine or placebo (non‐sedation/standard care).

2. METHOD

2.1. Trial design

The study was conducted as a double‐blinded, randomized, placebo‐controlled trial with two parallel groups (2:1 allocation dexmedetomidine: placebo, (Figure 1)). To reduce the number of patients needed to demonstrate a possible significant result a control recording along with an allocation ratio of 2:1 was applied. This method was preferred since PSG (Polysomnography) performed on critically ill ICU patients is highly time and resource‐consuming.

FIGURE 1.

Intervention timeline during the 2‐day study period

2.2. Approvals, ethics, and data storage

The study protocol was published by the British Medical Journal Open (BMJ Open) ³⁶ and was approved by the National Committee on Health Research Ethics (Approval number: S‐20180214). Registration with EudraCT (European Union Drug Regulating Authorities Clinical Trials Database) registration number: 2017–001612‐11DK (October 27, 2017. URL: https://www.clinicaltrialsregister.eu/ctr-search/trial/2017-001612-11/DK) was performed, upon approval from the Danish Medicines Agency. Monitoring was performed by Good Clinical Practice (GCP). Registration with the Danish Data Protection Agency was done, and all data collected were stored online using REDCap, which complies with international confidentiality and GCP guidelines.

2.3. Study hypotheses and endpoints

Primary: Intravenous dexmedetomidine infusion improves sleep quality and quantity, compared to placebo. Secondary: Intravenous dexmedetomidine infusion during the night decreases the frequency of delirium development while increasing alertness, wakefulness, and daytime function compared to placebo.

2.4. Selection of participants

Thirty consecutive patients were enrolled at the Department of Anesthesia and Intensive care at the Hospital of Southwest Jutland, Denmark.

Inclusion criteria: admission to the ICU. 18 years old or over. Anticipated stay in the ICU for another day after the first sleep recording. Mechanically ventilated patients. Hemodynamically stable patients. Acceptable PSG recording during the first night. Conscious non‐sedated patients with the Danish language.

Exclusion criteria: SOFA score above 12. Post‐operative patients. Trauma patients. Patients with structural (e.g., stroke, tumor) or degenerative neurologic diseases (e.g., Parkinson's disease, dementia). Patients with known epilepsy, seizures, cerebral infections, or other disorders affecting the brain. Patients with a major psychiatric disorder (e.g., schizophrenia, severe depression). Former or active drug abuse. Patients with second‐ or third‐degree atrioventricular block (unless pacemaker implanted). Patients of childbearing potential with positive pregnancy test or currently lactating/known pregnancy. Patients with severely agitated delirium. Patients with a high risk of death in the study period. Patients using other alpha‐2 agonists (clonidine) during ICU stay. Patients with limitations in therapy. Patients participating in other studies involving the use of pharmacologically active substances. Patients in need of anti‐psychopharmaceutic or other sedating drugs.

3. INTERVENTION

3.1. Recruitment procedures

All mechanically ventilated patient was screened from 8 a.m. to 4 p.m. every day during the study period and candidates fulfilling the inclusion and exclusion criteria were found using clinical observation and the hospital medical record. Patients that were eligible for inclusion were contacted by the investigator (MD) and informed written and oral consent was obtained, with the contact nurse present. The patients were screened every day and included a few days into their ICU treatment to meet the inclusion and exclusion criteria such as non‐sedation, consciousness, and without non‐protocolled drugs. This was to increase data quality and reliability.

3.2. Study treatments for both groups

Patients were screened for inclusion on day one from 8 a.m. to 4 p.m. and if eligible for inclusion, baseline characteristics as presented in Table 1. were obtained. During this period the patients were mounted with PSG electrodes and so forth as described below. The first 16 h of PSG recording were performed from 4 p.m. on day one to 8 a.m. on day two. From 8 a.m. to 4 p.m. on day two, while no recording was performed the first recording was transferred and validated by qualified staff at the DCSM, Rigshospitalet, Copenhagen. If the quality of the first recording were acceptable (impedance <10 KΩ, electrode placement, calibration, etc.) patients were randomized, otherwise they were excluded before randomization. Randomized patients were prepared for the second recording with an electrode check and the screener was programmed to start the second recording at 4 p.m. on day two. This recording ended at 8 a.m. on the third day, ending of the study. Dexmedetomidine/placebo was administered as described below during the second recording from 6 p.m. on day two to 6 a.m. on day three (Figure 1), and RASS (Richmond Agitation Scale Score) was evaluated every half hour and presented.

TABLE 1.

Baseline characteristics and admission diagnosis

Baseline data	DEX (N = 20)	Placebo (N = 10)	P
Gender (Female)	7	2	NS
Age (years)	65 (59;72)	70 (65;75)	NS
Height (cm)	174 (170;179)	175 (168;182)	NS
Weight (kg)	84.9 (69.3;98;5)	85 (70.6;99.4)	NS
BMI	27 (24;32)	28 (24;32)	NS
Alcohol consumption (units/week)	7 = 0 3 = 7–14 1 = 15–21 9 > 21	3 = 0 2 = 7–14 5 > 21
APACHE II	24.7 (20.3;29)	24.4 (17.9;30.9)	NS
SAPS 3	75 (66.1;83.9)	74.4 (63.3;85.5)	NS
SOFA day 1	2.8 (2.1;3.6)	4.2 (2.3;6.3)	NS
SOFA day 2	4.1 (3.1;5)	5.1 (3;7.2)	NS
SOFA day 3	3.9 (2.8;4.9)	4.8 (2.8;6.8)	NS
Days from admission	19 a (12;35)	14 a (4;16)	0.02
Airway management	2 Intubated 18 Tracheotomy	3 Intubated 7 Tracheotomy	NS
Temperature (°C) day 2	37.2 (36.9;37.4)	37.2 (36.6;37.6)	NS
Billirubin (μmol/l) day 2	14.8 (8.4;21.1)	10 (6.3;13.7)	NS
Carbamide (mmol/l) Day 2	11 (7.9;14.1)	13.8 (10.7;17.9)	NS
PaCO2 (kPa) b Day 2	6 (5.5;6.6)	6.6 (5.4;7.8)	NS
PS (cm H2O) Day 2	5.5 (4.4;6.6)	8.2 (5.2;11.2)	0.03
PEEP (cm H2O) Day 2	6.3 (5;7.6)	7.2 (4.2;10.2)	NS
FiO2 (%) Day 2	33.9 (30.1;37.8)	32.8 (21.2;44.3)	NS
Ventilator setting PS/CPAP (N)	20	10
Infused study drug (dex/placebo) (ml)	196 (148;145)	245 (170;320)	NS
Admission diagnosis
Respiratory insufficiency	7	7
Pneumonia	4	3
Sepsis	3
COPD	2
Pancreatitis	2
Heart failure	2

Note: Data presented as means with 95% CI.

Abbreviations: COPD, Chronic Obstructive Pulmonary Decease; DEX, Dexmedetomidine.

^a DATA presented in median with IQR.

^b Highest value during the period.

Both study treatments were provided and delivered by Orion Pharma (Ørestads Boulevard 73, 2300 København S) to the Hospital Pharmacy at the Hospital of Southwest Jutland. Written instruction and verbal training for dilution and blinding of the study drug were provided for the pharmacist. The pharmacy personnel were not otherwise involved in the treatment of the patients, and they were not allowed any interaction with the ICU staff. Study drug batch numbers were kept in a log by the pharmacy.

Paracetamol and morphine were allowed for pain relief when clinically needed and in accordance with the department's guidelines. Haloperidol was used in case of delirium determined by CAM‐ICU (Confusion Assessment Method – Intensive Care Unit) and morphine in case of discomfort due to intubation. Sedating drugs for example, benzodiazepines or propofol were not allowed. No sleep‐promoting agents or interventions were allowed (earplugs, sedatives, etc.).

3.3. Polysomnography

Eight EEG electrodes were placed on the patient's scalp at the left/right frontal lobe, left/right central lobe, and left/right occipital lobe, according to the international 10/20 system of electrode placement. Positioning was relative to the two electrodes placed on the mastoid process. Ground and reference electrodes were placed at the top center of the scalp. Eye and eyelid movement was monitored with two EOG (extraocular) electrodes to differentiate wakefulness, REM, and NREM sleep. Two submaxillary and one chin electromyogram (EMG) electrodes were placed to monitor muscle tone. In addition, two EMG electrodes placed on the proximal lateral crus were monitoring leg movement. Mounting The PSG was performed according to the AASM (American Academy of Sleep Medicine) guidelines except for respiration monitoring since this was monitored using the ventilator.

The PSG was performed using hardware and utensils from Somnomedics including Domino software (Somnomedics GmbH Am Sonnenstuhl 63 D‐97236 Randersacker Deutschland). Calibration to ensure acceptable electrode impedance (<10 KΩ, otherwise the electrodes were replaced) was conducted. Bio calibrations were done prior to each PSG recording to identify patterns of eyes‐open/eyes‐closed, wakefulness, and eye movement. PSG scoring was done according to the AASM guidelines. ³⁷

Sleep latency is defined as the time in minutes between “lights out” (the patient attempts to sleep), until falling asleep (the first epoch scored as sleep). Sleep efficiency refers to the percentage of total time in bed spent sleeping. It is calculated as the percentage of total time in bed spent in REM and non‐REM sleep. ³⁸

3.4. Intervention group

The intervention group was administered intravenous dexmedetomidine (100 μg/mL) diluted in glucose 5% to a concentration of 4 μg/mL. Patients were started out on continuous infusion of dexmedetomidine 0.4 μg/kg/h. The attending MD would increase or decrease the infusion rate by 0.2 μg/kg/h every half hour, targeting a RASS of −2, in accordance with the department guidelines. A maximum infusion rate of 1.4 μg/kg/h was allowed.

3.5. Placebo group

The placebo group was treated as the intervention group, except for receiving a placebo infusion of glucose at 5% without Dexmedetomidine.

3.6. Outcomes

3.6.1. Primary

Sleep quality and quantity during intravenous dexmedetomidine infusion, determined by PSG (TST, SE, and total REM sleep) compared to placebo.

3.6.2. Secondary

The frequency of delirium development determined by CAM ICU. Alertness, wakefulness, and daytime function while treated with dexmedetomidine infusion, determined by RASS (Richmond Agitation Sedation Scale) and physical activity, compared to placebo.

3.7. Randomization method

Randomization was performed electronically using REDCap (Research Electronic Data Capture), which is a browser‐based online case report form. The allocation ratio was 2:1 for dexmedetomidine and placebo respectively in blocks of 10 patients. The randomization numbers with the corresponding allocation were stored by the REDCap data manager. The randomization codes corresponding with the treatment regimen were sealed in non‐transparent envelopes for the pharmacist to prepare either dexmedetomidine or placebo. The envelope was at all times stored by the hospital pharmacy and placed away from the ICU. The medication was left in a refrigerator after preparation and labeling (randomization code/patient identification), for the attending nurse to collect and administer.

3.8. Blinding

Neither the electronic treatment code, randomization envelopes nor the study drug was accessible to the investigators, nor anyone employed in the ICU. No contact with the pharmacy personnel preparing the study drug was allowed at any time. Only the data manager and the GCP monitor had access to the randomization code. Sleep assessments were done randomly by an independent doctor having only the randomization code for identification, thus blinded to treatment and the order of the recordings. A second person checked for the accuracy of the blinding. The randomization code was not broken until all data were collected and analyzed. In case of an emergency unblinding, the data manager could be contacted at any time. An external statistician was used.

3.9. Trial personnel

The trial personnel were doctors and nurses employed in the ICU at the hospital of Southwest Jutland. The personnel were trained in non‐sedation and the use of dexmedetomidine in theory and by supervised practice. Dexmedetomidine was part of the department guidelines for years before the present study. The pharmacy personnel received verbal and written training for preparing, diluting, and blinding the study and placebo. PSG analysis was performed by qualified personnel employed at the DCSM (Danish Center of Sleep Medicine).

3.10. Sample size

The sample size was calculated based on similar work by Alexopoulou and Oto, ¹⁹ , ²⁷ reporting significantly improved TST and SE (100% * TST/total PSG time) in critically ill patients treated with dexmedetomidine. Placebo was assumed to result in 0% of non‐REM sleep with a common standard deviation of 7%. Using parallel design and assuming a 10% difference in time spent in non‐REM sleep, a sample size of 12 subjects per treatment arm would provide 80% power to detect a statistically significant difference at a 5% level. Using a 2:1 random allocation ratio reduces power and approximately 12% more subjects were required. Randomizing a total of 30 subjects allows approximately a 10% dropout rate. The 2:1 randomization ratio was performed as all patients received a control recording on the first day creating almost equally large intervention and control groups.

3.11. Statistical method

The data distribution has been tested for skewness, kurtosis, and standard deviation. T‐test has been used in the case of normally distributed data and the Wilcoxon‐rank‐sum test, for non‐parametric data. Linear regression has been used to determine correlations between the outcome and clinically relevant variables: days from admission (not equalized between groups), gender, age, BMI, APACHE‐II (acute physiology and chronic health evaluation), SAPS3 (simplified acute physiologic score), SOFA, and PS (pressure support) (not equalized between groups). No significant link was found. To assess the difference in sleep parameters between night 1 (control recording) and night 2 (intervention recording), the mean values for each patient, each night were established, then subtracted, and compared between the groups using a t‐test. The difference in CAM‐ICU scores between groups was tested using Fisher's exact test. A mean RASS score was calculated and the difference between the groups was estimated using Wilcoxon rank‐sum test. Data was handled according to the intention to treat principles.

4. RESULTS

From May 2018 to June 2020, 853 patients were admitted to the intensive care unit at the Hospital of Southwest Jutland (12‐bed mixed adult medical and surgical ICU). Thirty‐five consecutive patients were found eligible for inclusion Five patients were not included due to failure to obtain informed consent. 30 patients were enrolled and randomized. 10 patients were randomized to receive a placebo and 20 to receive dexmedetomidine. All 30 patients completed the trial as planned and data were analyzed for the outcomes (Figure 1).

Baseline equalization was achieved with regards to airway management, gender, height, weight, BMI, alcohol consumption, APACHE‐II, SAPS3, SOFA (day 1–3) score, temperature, bilirubin, carbamide, highest PaCO₂, PEEP, FiO₂, and Ventilator settings. Days from admission differed significantly by 5 days between the groups and PS (ventilator support) differed significantly by 2.7 cmH₂O, as shown in Table 1. Seven patients in both groups were admitted with respiratory insufficiency. Other admission diagnoses were pneumonia, sepsis, COPD, pancreatitis, and heart failure as shown in Table 1.

Doses of morphine and haloperidol administered as rescue medication exhibited no significant difference between the groups (Table 2).

TABLE 2.

Rescue medication administered during the study period

Day/time	Morphine			Haloperidol
	DEX	Placebo	P	DEX	Placebo	P
D1 8–16:00	3 mg (1.2;4.7)	4 mg (0.8;7.2)	NS	0	0
D1 16–8:00	8.8 a (0;11.3)	2.5 a (0;15)	NS	7.5 mg a (n = 3) (2.5;10)	17.5 mg a (n = 2) (2.5;32)	NS
D2 8–16:00	3.75 mg (1.5;6)	2.1 mg (0.0;4.7)	NS	0	0
D2 16–8:00	1.3 a (0;10)	6.3 a (0;12.5)	NS	0	7.5 mg (n = 1)

Note: Data presented in means with 95% CI.

Abbreviations: DEX, Dexmedetomidine; N, the number of patients receiving the drug; NS, nonsignificant.

^a Median and IQR.

When comparing the control recording (day 1) with the intervention recording (day 2) during the time interval 6 p.m.–6 a.m. patients exhibited non‐REM sleep with significant improvement of TST in the intervention group, with 271 min. (210;324 95% CI) compared to 27 (−82;135 95% CI) in the placebo group (p < .001). SE improved significantly as well as in the intervention group with 37.6% (29.7;45.6) compared to the placebo group with 3.7 (−11.4;18.8 95% CI) (p < .001). As a result, duration of wake decreased significantly in the dexmedetomidine group with −269 min. (−328; −211 95% CI) compared to the placebo group with 15 min. (−125;95 95% CI, p < .001). Latency wake was significantly decreased in the dexmedetomidine group with −146 (−209; −72 95% CI) compared to the placebo group with 52 (−18;123 95% CI) (p < .001). No significant difference was found regarding sleep latency, REM latency, number of awakenings, number of arousals, latency wake, total latency, and duration REM (Table 3). PSG data during a shorter interval (10 p.m.–8 a.m.) are presented in Table 3. to demonstrate the results for a time interval more in accordance with a normal circadian rhythm.

TABLE 3.

PSG results

	6 p.m.–6 a.m.			10 p.m.–8 a.m.
Recording interval	DEX N = 20	Placebo N = 10	p	DEX N = 20	Placebo N = 10	p
Total sleep time (min)	271 (210;324)	27 (−82;135)	p < .001	149 (102;196)	−13 (−133;106)	p = .002
Sleep efficiency (%)	37.6 (30;46)	3.7 (−11.4;18.8)	p < .001	24.9 (17;32.7)	−2.2 (−17.7;22.1)	p = .002
Sleep latency (min)	−124 (−46;‐202)	−78 (−201;45)	NS	−23 a (−75;6)	−1 a (−19;13)	NS
REM latency (min)	−22 (−145;102)	−10 (−321;300)	NS	90 (−66;246)	−67 (−217;83)	NS
Wake latency (min)	−146 (−209;‐72)	52 (−18;123)	p < .001	−89 (−143;‐36)	56 (−34;145)	p < .001
Number of awakenings	3.2 (−8;14)	−10.9 (−35.2;13.4)	NS	4 a (−10;13.5)	2.5 a (−12;18)	NS
Number of arousals	−1.5 (−19;16)	−12.8 (−47;21)	NS	18 (1;35)	23.9 (−14.7;62.5)	NS
Duration wake (min)	−269 (−328;‐211)	15 (−125;95)	p < .001	−144 (−193;‐93)	15 (103;134)	p = .003
Duration REM (min)	−12 a (−34;2)	0 a (−9;8)	NS	−21 a (−62;0)	0 a (−20;2)	NS
Recording time in total (min)	720	720		600	600

Note: Reported as the difference between day 1 (control recording) and day 2 (intervention recording). The difference in means for each patient has been established and data presented as means with 95% CI.

Abbreviations: DEX, Dexmedetomidine; NS, Non – Significant (p > .1).

^a Median with IQR.

Time spent asleep, awake, and in REM sleep for the dexmedetomidine and placebo patients are presented in Figure 2.

FIGURE 2.

Sleep/Wake/REM distribution during night 1 (control) versus night 2 (intervention): Means with 95% CI

RASS was evaluated every half hour as mean RASS days 1 and 2 from 8 a.m. to 4 p.m. and from 4 p.m. to 8 a.m. Significantly decreased RASS score was observed in the dexmedetomidine group during intervention with a mean of −1.8 (−2.1;1.4 95% CI) compared to 1.0 (−1.6; −0.4 95% CI) in the placebo group (p = .02). No significant difference was shown during the rest of the study period.

No significant difference between the groups was found at any time with regards to delirium determined by CAM‐ICU nor regarding physical activity.

5. DISCUSSION

The present trial is the first randomized placebo‐controlled trial to demonstrate that TST and SE improve significantly without eliminating REM sleep when treated with dexmedetomidine compared to standard care (non‐sedation) in mechanically ventilated critically ill patients.

Recent research shows that when studying critically ill patients, the AASM sleep criteria, might be inadequate, as ICU patients often present pathologic EEGs (Electroencephalogram) with atypical sleep lacking some of the normal sleep architecture and characteristics, even when non‐sedated. This includes the absence of stage‐2 sleep markers (K‐complexes and sleep spindles), decreased REM sleep, and slow background EEG activity with impaired EEG reactivity during wakefulness. The absence of some of the normal sleep characteristics, as described above is in line with our PSG recordings. Consequently, distinguishing between sleep stages N1, N2, and N3 using AASM was not possible, since a new scoring system for sleep (atypical sleep) in critically ill patients still is under development. ⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² However, distinguishment between wake, REM sleep, and non‐REM sleep was performed with certainty, and data was of high quality.

This study demonstrates the effect of dexmedetomidine in patients exposed to ICU treatment for a number of days, which might influence the result compared to a study designed with patient inclusion immediately upon admission. However, a study design with immediate inclusion would have compromised the data quality and in turn the interpretation. It is not known what impact ICU environment exposure for a number of days has on dexmedetomidine‐induced sleep. Though patients in the intervention group were admitted to the ICU for a longer time and pressure support on the ventilator was lower compared to the controls equalization between the groups was found regarding parameters that stratify severity of illness and possible affection of sleep patterns for example, SOFA, APACHE II, and SAPS 3 and SOFA (day1–3).

The present study reconfirms previous non‐randomized studies reporting that critically ill patients exhibit disorganized and poor sleep quality ⁸ , ⁴¹ , ⁴³ and with the randomized design it states the findings of C. Alexopoulou et al. and J. Oto et al. showing increased sleep efficiency in their pilot studies. ¹⁹ , ²⁷

REM sleep was not eliminated in the dexmedetomidine group but reporting on dexmedetomidine's long‐term effect on the REM sleep stage is not supported by the current study design. However, sedatives, such as propofol used to achieve light sedation have proven to eliminate REM sleep with no effect on sleep efficiency or sleep fragmentation. ¹⁴

Sleep efficiency and total sleep time improved in the present study, thus extending sleep, and possibly establishing a more stable sleep–wake and circadian pattern. This offers an explanation for previous studies showing that dexmedetomidine reduces the risk of delirium when treated with the drug for a longer time. ²⁸ , ²⁹ , ³⁵

Using RASS to titrate dexmedetomidine administration is used extensively worldwide and the scale has been validated properly. It has excellent interrater reliability, and it is relatively simple to use, furthermore it was already part of the study site guidelines. ⁴⁴ , ⁴⁵

Two recording intervals have been presented (6 p.m.–8 a.m. and 10 p.m.–8 a.m.). This was to report the sleep quality and quantity during a sleeping interval, that one might consider “normal” along with an extended interval. In non‐delirious critically ill patients, we aim to restore a circadian rhythm as close to normal as possible to prevent delirium. On the other hand, in critically ill delirious patients one might extend the sleeping interval to protect the patient from complications related to delirium. However, no significant difference in CAM‐ICU score during the present study period were shown. The effect on delirium that previous studies report might have been revealed during a prolonged study period, that would however be speculation. ²⁸ , ²⁹ , ³⁴ , ³⁵ In addition, using CAM‐ICU to evaluate delirium, during sedation with a RASS level of −2 might transiently influence the positivity of the test without consequence for the patient on a long‐term basis as described by Patel SB et al. ⁴⁶

Clinical implication: As opposed to other sedatives for example, propofol and midazolam, dexmedetomidine promoted PSG assessable sleep without eliminating REM sleep. In turn, this leads to increased restorative sleep and offers an explanation, however not causal for dexmedetomidine's reported effect on delirium in critically ill ICU patients. However, the present study did not show a reduced incidence of delirium.

6. LIMITATION

To facilitate sleep classification using the standard criteria is not optimal. These criteria have not been developed for critically ill patients who often present atypical sleep patterns in some studies called Watson to sleep or electroencephalogram featuring absence of SWS and typical stage 2 sleep (i.e., absence of K complexes and sleep spindles). ³⁹ , ⁴⁰ , ⁴² In the absence of a suitable sleep classification system for critically ill patients, the results of the present study have been reported without distinguishment between sleep stages 1, 2, and 3. There is a need for an EEG/PSG classification system of sleep in ICU patients. The study was not designed to show a difference in delirium and physical activity. The present study performing PSG on critically ill patients is highly time and resource‐consuming, hence it was performed as a single‐center study. A multicenter study investigating multiple consecutive nights during intervention would have provided the study with more generalizable results and should be considered in the future.

7. CONCLUSION

Total sleep time and sleep efficiency were significantly increased, without elimination of REM sleep, in mechanically ventilated ICU patients randomized to dexmedetomidine. This result was demonstrated when comparing the intervention PSG recording with a control recording performed during non‐sedation (standard care). Future studies are needed for developing a sleep classification system for critically ill patients and to state the result of the present study in a larger multicenter trial.

AUTHOR CONTRIBUTIONS

Poul Jørgen Jennum, Palle Toft, and Mikael Sörberg developed the aim, hypothesis, and study design. Jakob Oxlund executed the study. Poul Jørgen Jennum analyzed the PSG recordings. Poul Jørgen Jennum, Palle Toft, Torben Knudsen, Thomas Strøm, and Jakob Oxlund performed computations and developed the study design. Jakob Oxlund wrote the manuscript and all authors discussed and contributed to the final version.

FUNDING INFORMATION

An unrestricted grant of 70,000 euros and the study drug was provided by Orion Corporation. In addition, a grant of 33,000 euros was provided by Rigshospitalet and Odense University Hospitals' common research fund. The grants have in part covered PSG equipment, utensils, and salaries for study nurses and neurophysiology assistants. The researchers have no economic interests in the company and received no payment. Sponsor: Poul Jørgen Jennum.

CONFLICT OF INTEREST

The company that financially supported the study upstart was otherwise without influence on initiation, execution, data collection, interpretation of the results, data handling/reporting, or decision to publish the study. M. Sörberg authors the study affiliated to Karolinska University Hospital. However, he is also employed by Orion Corporation.

Oxlund J, Knudsen T, Sörberg M, Strøm T, Toft P, Jennum PJ. Sleep quality and quantity determined by polysomnography in mechanically ventilated critically ill patients randomized to dexmedetomidine or placebo. Acta Anaesthesiol Scand. 2023;67(1):66‐75. doi: 10.1111/aas.14154