Low-Dose Melatonin for Prevention of Delirium in Critically Ill Patients

Abstract

Background

Delirium is a common and serious syndrome of acute brain dysfunction associated with negative outcomes. Melatonin may have a role in delirium prevention for critically ill adults based on data from noncritically ill patient populations. Our objective was to assess the feasibility of a multicenter, randomized, placebo-controlled trial testing the hypothesis that low-dose melatonin prevents delirium in adults who are critically ill.

Research Question

Does low-dose melatonin prevent delirium in critically ill adults?

Study Design and Methods

We conducted this 3-arm feasibility trial in 3 tertiary academic ICUs. Included participants were ≥ 18 years of age, confirmed free of delirium at enrollment, and anticipated to require a > 48-hour ICU stay. We randomized participants to nightly melatonin 0.5 mg, 2 mg, or placebo. Feasibility outcomes were protocol adherence and recruitment rates. Our primary feasibility target was ≥ 85% (± 5% margin of error) of drug doses administered as per protocol. Secondary objectives were to explore adverse drug effects, melatonin pharmacokinetics, and clinical outcomes.

Results

We screened 2,259 patients and excluded 1,863 patients (82.5%), resulting in 396 eligible patients, of whom 71 provided consent, for a recruitment rate of 0.8 patients/mo/site. Median age was 60.5 years (interquartile range [IQR], 48-67 years), and median admission Sequential Organ Failure Assessment score was 7 (IQR, 3-10). Percentage drug administration protocol adherence per patient was a median of 100% (IQR, 92.3%-100%) or a mean of 88.7% (SD, 24.4%). Twenty-five protocol violations occurred, with no differences between groups. Fourteen patients (20%) experienced delirium during the study period and 25 patients (36%) experienced subsyndromal delirium, with no differences between study groups. No serious adverse effects were detected.

Interpretation

Our trial protocol comparing 2 low doses of melatonin and placebo for delirium prevention in critically ill adults demonstrated feasibility for protocol adherence. However, trial eligibility rates were modest primarily because of delirium being present or the inability to screen. Consent rates also were low. This finding suggests that conducting delirium prevention trials that require delirium to be absent on enrollment is particularly challenging in the ICU.

Clinical Trial Registry

ClinicalTrials.gov; No.: NCT02615340; URL: www.clinicaltrials.gov

Take-Home Points.

Study Question: Can we achieve ≥ 1 patient/mo/site recruitment and 85% (± 5%) protocol adherence in a multicenter, randomized feasibility study of low-dose melatonin for prevention of delirium in the ICU?

Results: We achieved a recruitment rate of 0.8 patients/mo/site and a median study drug administration of 100% (interquartile range, 92.3%-100%).

Interpretation: Trial eligibility rates were modest primarily because of the presence of delirium or the inability to screen for delirium. Conducting delirium prevention trials that require delirium to be absent on enrollment is particularly challenging in the ICU.

Melatonin (N-acetyl-5-methoxytryptamine) is a neurohormone that is derived sequentially from tryptophan and serotonin in the pineal gland during the hours of darkness.¹ Melatonin primarily regulates and synchronizes the circadian rhythm.¹ When given exogenously, it acts as a hypnotic, increasing sleep efficiency and total sleep duration, and decreasing sleep onset latency. Melatonin has multiple other biological effects, including antioxidant, antiapoptotic, antiinflammatory, and analgesic activities.²

Delirium is a common and serious syndrome of acute brain dysfunction characterized by fluctuations in attention, awareness, and cognition.³ Numerous patient populations with delirium, such as postoperative older adults, patients with severe sepsis, and those with critical illness, demonstrate sleep disturbances and changes in melatonin secretion.^{4, 5, 6, 7} Medications commonly used during critical illness associated with delirium (eg, corticosteroids, sedatives) are known to alter melatonin secretion or excretion.⁸ Therefore, abnormal melatonin secretion and metabolism resulting in circadian dysregulation may play a role in delirium pathogenesis in critically ill patients.⁹ Exogenous melatonin administration may prevent delirium through its effects on restoring the circadian rhythm and sleep.^10,11 Unlike other hypnotics, exogenous melatonin does not change sleep architecture significantly and has a wide safety margin across adult and pediatric populations.¹² It also may be sedative sparing,^13,14 which may be important for reducing delirium risk in the critically ill.^{15, 16, 17}

Systematic reviews of melatonin and melatonergic agonists for delirium prevention in mixed populations of critically ill adults report no difference in delirium-related and other outcomes.^18,19 Yet, a meta-analysis of 7 trials conducted in patients admitted to the ICU after cardiac surgery found a reduction in delirium incidence, although with low evidence certainty.²⁰ Furthermore, some evidence is available to support that melatonin and melatonergic agents are effective for delirium prevention in hospitalized noncritically ill patients in whom delirium-negative status before study inclusion is confirmed more easily.²¹

We hypothesized that exogenous low-dose melatonin can reduce the risk of delirium in critically ill adults. To inform our ability to test this hypothesis, we conducted a multicenter, randomized placebo-controlled feasibility trial. Our primary objective was to determine trial feasibility in terms of participant recruitment and protocol adherence for study drug administration. Secondary objectives were to describe adverse drug effects, melatonin pharmacokinetics, and exploratory clinical outcomes of enrolled participants.

Study Design and Methods

Study Design, Setting, and Participants

We conducted a 3-arm, triple-masked, multicenter randomized controlled feasibility trial at 3 ICUs with mixed patient populations in 3 university-affiliated hospitals (ClinicalTrials.gov Identifier: NCT02615340). Trial arms comprised 2 low doses of exogenous melatonin, 0.5 mg or 2 mg, and a placebo. The study protocol was published previously,²² and the CONSORT checklist was used for trial reporting (e-Appendix 1).²³

We screened daily each morning (Monday through Friday) and enrolled adults aged ≥ 18 years with an anticipated ICU stay of at least 48 hours who did not meet any of the exclusion criteria. We excluded patients based on the following criteria: (1) admitted to the ICU > 48 hours before screening; (2) unable to assess for delirium on screening with Sedation Agitation Scale (SAS)²⁴ score of 1, no response score on the Intensive Care Delirium Screening Checklist (ICDSC),²⁵ receiving neuromuscular blockade, history of severe cognitive or neurodegenerative disease (eg, dementia, Parkinson disease), or severe structural brain injury (eg, traumatic brain injury); (3) positive delirium findings at time of screening (defined as ICDSC score of ≥ 4); (4) anticipated withdrawal of life-sustaining therapy; (5) unable to communicate in English or French; (6) absolute contraindication to enteral nutrition (eg, gastrointestinal obstruction, perforation, recent gastrointestinal surgery, deemed strict nil by mouth, and lack of enteral access); (7) active seizures; (8) pregnancy; (9) legally blind; (10) documented allergy to melatonin; and (11) unable to obtain informed consent.

Randomization, Allocation Concealment, and Masking

Participants were assigned randomly in a 1:1:1 manner to 1 of 3 trial arms using a computer-generated, allocation-concealed, permuted-block randomization schedule, stratified by site.²² After obtaining informed consent from participants, clinical trial pharmacy personnel at each site who were not involved in the clinical management of participants accessed the randomization schedule via a Research Electronic Data Capture system to obtain the allocation group. All study personnel, ICU clinical team members, participants, and family members were masked to group assignment. Unmasking was permitted only in the event of an emergency if deemed crucial to the provision of medical care after approval by the principal investigator.

Study Drugs

Study drug (placebo, melatonin 0.5 mg, or melatonin 2 mg) commenced on the day of enrollment. It was administered by the bedside nurse once daily between 9 pm and 11:59 pm by mouth or feeding tube and continued until ICU discharge, death, or a maximum of 14 days. Melatonin doses were selected based on published delirium prevention trials.^{26, 27, 28} The clinical trial pharmacy personnel at each site compounded melatonin oral suspension 1 mg/mL in accordance with Good Manufacturing Procedure guidelines with synthetic melatonin (NPN 80015041 immediate-release tablets; Jamieson Laboratories). The compounded 1-mg/mL oral solution was diluted further with Oral Mix Sugar-Free flavored suspending vehicle (product code 2600; MEDISCA Pharmaceutique, Inc.) to a 5-mL volume. Participants randomized to the placebo group received an identical 5-mL syringe containing only Oral Mix sugar-free suspending vehicle. Study drugs were identical in appearance (milky white color suspension) and volume (5 mL). Each dose was dispensed by clinical trial pharmacy personnel in accordance with Health Canada regulations. In all study arms, administration of other pharmacologic and nonpharmacologic interventions for sedation, analgesia, delirium, or sleep were not protocolized, but rather were directed by the clinical team. Use of open-label melatonin or melatonin agonists was not permitted in any study arm.

Study Outcomes

The primary feasibility outcome was protocol adherence, defined as the overall proportion of study doses administered as prescribed between 9:00 pm and 11:59 pm divided by the total number of study nights. Feasibility was defined as 85% (SD, 5%) of study drug doses administered as intended. A secondary feasibility outcome was ability to recruit to the trial. We calculated the proportion of screened patients meeting inclusion criteria, the number of and reasons for patient exclusions, and the consent rate of eligible patients. Feasibility was defined as ≥ 1 enrollment per site per month. Other secondary feasibility outcomes were the amount of time required to screen and complete study procedures; rates of missing data related to baseline characteristics and potentially deliriogenic and delirium-protective exposures; exploratory clinical outcomes; and compliance with completion of self-reported sleep quality measure (Richards-Campbell Sleep Questionnaire).^29,30 Time for screening and study procedures was assessed using a previously published time-in-motion template with the first 10 participants per site.³¹

We collected the following exploratory delirium-specific outcomes: (1) delirium (ICDSC score, ≥ 4) and subsyndromal delirium (ICDSC score, 1-3) incidence, (2) day of delirium onset, (3) delirium duration, and (4) delirium-free days (up to ICU discharge). Other clinical outcomes explored were duration of mechanical ventilation, length of ICU and hospital stays, and ICU and hospital mortality.

Our approach to serious adverse events was in accord with the published guidelines for academic drug trials in critical care.^32,33 We documented any serious adverse events judged by the investigator team to be a consequence of study participation and unrelated to the patient’s underlying disease or an unexpected complication of critical illness. We recorded morning drowsiness defined as SAS of < 3 between 7:00 am and 12:00 pm. We had planned to measure pharmacokinetic outcomes at 1 site by assessing plasma melatonin levels,²² but were unable to obtain consent from any participants to collect the 8 additional blood samples required.

Data Collection

Bedside clinical staff (nurses, physicians, and pharmacists) assessed SAS at least every 4 hours and screened for delirium at least once daily (ICDSC score) until ICU discharge. These assessment tools were used as routine practice at the 3 sites. Participants were asked to complete the Richards-Campbell Sleep Questionnaire each morning of study participation, independently or with the assistance of their bedside nurse.

Data recorded by the research team on study enrollment included patient demographics, comorbidities, admission diagnosis, use of visual and auditory aids, medication history, and the Sequential Organ Failure Assessment score.³⁴ Data recorded daily included Glasgow Coma Scale scores, SAS scores, psychoactive drug exposure (ie, type and dose of sedatives, analgesics, antianxiety medications, and antipsychotics), physical restraint use, indwelling device (eg, Foley catheter), mobilization, and nonpharmacologic delirium prevention strategies (eg, visible clock, orientation strategies, television, and family or friend visit).³⁵

Ethical Considerations

This study was approved by Health Canada (Identifier: HC6-24-c197507) and each institutional research ethics board (Sunnybrook Health Science Centre [Identifier: REB086-2016]; Mount Sinai Hospital [Identifier: REB2016-0097-A]; and Hôpital du Sacre Coeur de Montreal [Identifier: REB2017-1385]). All aspects of the study were conducted in accordance with the ethical principles outlined by the Canadian Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans and the Declaration of Helsinki.³⁵ Written informed consent was obtained from competent eligible participants or substitute decision-makers and was documented in the medical record.

Sample Size and Statistical Analysis

We required a sample size of 69 patients based on adherence to the prescribed study drug regimen defined as administration of ≥ 85% of prescribed doses within the drug administration window of 9:00 pm to 11:59 pm, with ± 5% margin of error, an estimated 4 days of drug administration, and an intracluster correlation of 0.115. We report continuous variables as means and SDs or medians and interquartile ranges (IQRs) and categorical data as frequencies and percentages. We analyzed data using the intention-to-treat principle. We did not impute missing data because we intended to assess the volume of missing data only. We used χ² tests to compare categorical feasibility outcomes and Kruskal-Wallis tests for exploratory continuous variables. We calculated the percentage adherence per patient and reported the median (IQR) patient adherence (ie, median percent per patient across all patients). To compare proportion adherence over all days, we used generalized estimating equations with binomial distribution and log-link to account for clustering within patients. All data analyses were conducted by an independent statistician using SAS version 9.4 software (SAS Institute).

Results

Our final analytic sample was 70 patients, with 23 receiving melatonin 0.5 mg, 23 receiving melatonin 2 mg, and 24 receiving placebo (Fig 1). The median age was 60.5 years (IQR, 48-67 years), median admission Sequential Organ Failure Assessment score was 7 (IQR, 3-10), and 35.7% were female (Table 1).

Figure 1.

Flow diagram showing patient screening, enrollment, and randomization. SDM = substitute decision-maker.

Table 1.

Demographics^a

Variable	All (N = 70)	Melatonin		Placebo (n = 24)
		0.5 mg (n = 23)	2 mg (n = 23)
Age, y	60.5 (48-67)	62 (50-67)	56 (46-62)	63.5 (50-67)
Female sex	25 (35.7)	7 (30.4)	9 (39.1)	9 (37.5)
Weight, kgb	77.8 (70-91.3)	76.3 (70-85)	82 (72.5-101.3)	75.5 (70-89)
Admission SOFA score	7 (3-10)	6 (2-12)	7 (3-9)	7 (5-10)
Admission category
Medical	45 (64.3)	17 (73.9)	14 (60.9)	14 (58.3)
Surgical or trauma	25 (35.7)	6 (26.1)	9 (39.1)	10 (41.7)
Primary reason for ICU admission
Pneumonia or aspiration	5 (7.1)	3 (13)	2 (8.7)	...
Sepsis to septic shock	9 (12.9)	5 (21.7)	2 (8.7)	2 (8.3)
Trauma (non-TBI)	12 (17.1)	4 (17.4)	3 (13)	5 (20.8)
TBI (mild-moderate)	3 (4.3)	1 (4.3)	1 (4.3)	1 (4.2)
Cardiopulmonary failure or arrest	6 (12.9)	1 (4.3)	2 (8.7)	3 (12.5)
Otherc	35 (50)	9 (39.1)	13 (56.5)	13 (54.2)
Exposures before hospitald
Benzodiazepine	13 (27.1)	6 (40)	3 (17.6)	4 (25)
Other sedative	8 (16.7)	3 (20)	5 (29.4)	...
Antipsychotic	4 (8.3)	0 (0)	1 (5.9)	3 (18.8)
Antidepressant	14 (29.2)	5 (33.3)	4 (23.5)	5 (31.3)
Opioid	13 (27.1)	6 (40)	5 (29.4)	2 (12.5)
Herbal sleep agents	2 (4.2)	...	2 (11.8)	...
Corticosteroids	6 (12.5)	...	3 (17.6)	3 (18.8)
Illicit drug use	2 (4.2)	...	...	2 (12.5)
Alcohol (≥ 2 alcohol drinks/d)	6 (12.5)	1 (6.7)	1 (5.9)	4 (25)
Tobacco (≥ 10 cigarettes/d)	9 (18.8)	4 (26.7)	4 (23.5)	1 (6.3)

Data are presented as No. (%) or median (interquartile range). SOFA = Sequential Organ Failure Assessment score; TBI = traumatic brain injury.

^aNo significant difference between groups.

^bNot reported for 10 participants.

^cExamples include gastrointestinal bleeding, postoperative monitoring, and hypotension.

^dNot reported for 22 participants.

Feasibility Outcomes

Our primary feasibility outcome of overall adherence (percentage per patient across all patients) was a median of 100% (IQR, 92.3%-100%), 100% (IQR, 100%-100%) with melatonin 0.5 mg, 100% (IQR, 66.7%-100%) with melatonin 2 mg, and 100% (IQR, 89.3%-100%) with placebo (Table 2). Based on the generalized estimating equation model, the estimates for protocol adherence were not different among study arms: melatonin 0.5 mg, 91.8% (95% CI, 85.2%-99.0%); melatonin 2 mg, 94.6% (95% CI, 90.2%-99.2%); and placebo, 91.0% (95% CI, 85.0%-97.5%; P = .85). Twenty-five protocol violations (of 335 possible study days [7.5%]) occurred, with no differences between groups. Protocol violations primarily were the result of patients declining the dose (n = 9 [36%]) or missing drug administration (n = 8 [32%]); no differences between groups were found.

Table 2.

Trial Outcomes^a

Variable	All (N = 70)	Melatonin		Placebo (n = 24)
		0.5 mg (n = 23)	2 mg (n = 23)
Feasibility outcomes
Protocol adherence
Median (IQR)	100 (92.3-100)	100 (100-100)	100 (66.7-100)	100 (89.3-100)
Mean (SD)	88.7 (24.4)	94.0 (15.6)	81.2 (34.9)	90.8 (17.1)
Estimate (95% CI)b	92.0 (88.4-95.8)	91.8 (85.2-99.0)	94.6 (90.2-99.2)	91.0 (85.0-97.5)
Exploratory outcomes
Delirium (ICDSC score ≥ 4) ever	14 (20%)	5 (21.7%)	5 (21.7%)	4 (16.7%)
Duration, d	1.5 (1-2)	2 (1-2)	2 (1-2)	1 (1-1.5)
Onset, d	2 (1-3)	2 (2-2)	2 (1-3)	2.5 (1.5-3)
SSD (ICDSC score 1-3) ever	25 (35.7%)	6 (26.1%)	8 (34.8%)	11 (45.8%)
Days with SSD	1 (0-2)	0 (0-3)	0 (0-2)	1 (0-2)
Coma (ICDSC = A or B) ever	7 (10.0%)	2 (8.7%)	3 (13.0%)	2 (8.3%)
Days with coma	1 (1-10)	6 (1-11)	1 (1-4)	5.5 (1-10)
Morning drowsinessc ever	7 (10.0%)	3 (13.0%)	2 (8.7%)	2 (8.3%)
Days	2 (1-10)	2 (1-10)	3.5 (2-5)	5.5 (1-10)
RCSQ completed	68 (97.1%)	21 (91.3%)	23 (100%)	24 (100%)
Eligible days completed, %	100 (79.3-100)	100 (100-100)	100 (50-100)	100 (77.5-100)
IMV, d
Received	15 (21.4%)	5 (21.7%)	6 (26.1%)	4 (16.7%)
Duration	4.4 (0.9-11.2)	9.2 (5.7-11.2)	3.6 (0.9-6.2)	1.3 (1-7.8)
LOS, d
ICU	4 (3-7)	4 (3-9)	4 (3-11)	4 (3-5.5)
Hospital	13.5 (8-21)	20 (11-27)	13 (6-21)	10.5 (7.5-17)
Mortality
ICU	6 (8.6%)	1 (4.3%)	2 (8.7%)	3 (12.5%)
Hospital	9 (12.9%)	1 (4.3%)	3 (13.0%)	5 (20.8%)

Data are reported as No. (%) median (interquartile range). ICDSC = Intensive Care Unit Delirium Screening Checklist; IMV = invasive mechanical ventilation; IQR = interquartile range; LOS = length of stay; RCSQ = Richards-Campbell Sleep Questionnaire; SSD = subsyndromal delirium.

^aAll continuous variables are compared using the Kruskall-Wallis test and categorical variables are compared using the Wilcoxon rank-sum test. No significant differences were detected.

^bProtocol adherence also was analyzed using the generalized estimating equation model with log-binomial link.

^cMorning drowsiness was defined as Sedation Agitation Scale score of < 3 between 7:00 am and 12:00 pm.

Secondary Feasibility Outcomes

We screened and enrolled patients between November 2017 (study initiation at the first active site) and June 2020 (enrollment at the last active site). Of ineligible patients (n = 1,863), 49.9% (n = 930) were excluded because of preexisting delirium (n = 166) or the inability to assess for delirium at the time of screening (n = 764) (Fig 1). Therefore, 325 patients (10.5% or 3.5 patients/mo/site) were eligible for enrollment. The primary reasons for eligible patients not being randomized were the patient or substitute decision-maker declined consent (n = 205 [63.1%]) or lack of availability of research personnel to approach for consent (n = 90 [27.7%]). Therefore, we recruited 0.8 patients/mo/site. Of the 71 patients who consented to the study, none consented to the nested pharmacokinetic study (separate consent was required for blood samples). One patient withdrew consent after randomization and before the first dose administration. Research personnel documented a mean (SD) of 87 (36) min/wk to complete screening activities and 5:06 (1:12) hours per participant to complete study-related procedures including consent and case report form completion. The Richards-Campbell Sleep Questionnaire was completed by 68 patients (97.1%) at least once with or without the support of the clinical nurse, with no difference in completion rates on eligible study days between groups (P = .62) (Table 2). We identified only a small proportion of missing data, all related to exposures before hospital admission (eg, medications before hospitalization) (Table 1).

Adverse Events and Exposures

No serious adverse events were detected. We identified 7 patients who experienced morning drowsiness at least once, or 31 of 380 study days (8.2%) with no difference between groups (Table 2). We found no difference across groups in terms of exposure to potentially deliriogenic psychoactive drugs or physical restraints, as well as no difference in terms of exposure to strategies to prevent delirium, including nonpharmacologic strategies and mobilization (Table 3).

Table 3.

Potentially Deliriogenic and Delirium Protective Exposures^a

Variable	All (N = 70)	Melatonin		Placebo (n = 24)
		0.5 mg (n = 23)	2 mg (n = 23)
Sedatives, any
Ever	25 (35.7)	9 (39.)	10 (43.5)	6 (25)
Study days, %	0.2 (0.1-0.4)	0.3 (0.1-0.5)	0.3 (0.1-0.4)	0.2 (0.1-0.2)
Benzodiazepines
Ever	10 (14.3)	3 (13)	5 (21.7)	2 (8.3)
Study days, %	0.1 (0.1-0.1)	0.2 (0.1-0.6)	0.1 (0.1-0.1)	0.1 (0.1-0.1)
Nonbenzodiazepine sedativesb
Ever	25 (35.7)	9 (39.1)	10 (43.5)	6 (25)
Study days, %	0.2 (0.1-0.4)	0.3 (0.1-0.5)	0.3 (0.1-0.4)	0.2 (0.1-0.2)
Antipsychotics
Ever	15 (21.4)	4 (17.4)	6 (26.1)	5 (20.8)
Study days, %	0.2 (0.1-0.4)	0.1 (0.1-0.4)	0.4 (0.1-0.4)	0.2 (0.2-0.4)
Opioids
Ever	57 (81.4)	21 (91.3)	17 (73.9)	19 (79.2)
Study days, %	0.3 (0.1-0.5)	0.3 (0.1-0.5)	0.3 (0.2-0.6)	0.2 (0.1-0.4)
Devices, No./dayc	3 (2-4)	3 (2-5)	3 (2-4)	3.5 (2-4)
Physical restraints
Patients ever	11 (15.7)	3 (13)	5 (21.7)	3 (12.5)
Study days, %	33.3 (16.7-66.7)	42.9 (28.6-66.7)	33.3 (14.3-37.5)	25 (16.7-100)
Eye masks at night
Patients ever	4 (5.7)	1 (4.3)	3 (13)	...
Study days, %	58.3 (28.6-83.3)	7.1 (7.1-7.1)	66.7 (50-100)	...
Earplugs at night
Patients ever	5 (7.1)	...	3 (13)	2 (8.3)
Study days, %	25 (25-50)	...	25 (16.7-50)	37.5 (25-50)
Enteral nutrition
Patients ever	55 (78.6)	18 (78.3)	18 (78.3)	19 (79.2)
Study days, %	83.3 (60-100)	78.6 (53.8-100)	85.7 (66.7-100)	83.3 (50-100)
Visitors, family or friends
Patients ever	59 (84.3)	19 (82.6)	21 (91.3)	19 (79.2)
Study days, %	83.3 (60-100)	78.6 (53.8-100)	85.7 (66.7-100)	83.3 (50-100)
Mobilized
Patients ever	61 (87.1)	22 (95.7)	18 (78.3)	21 (87.5)
Study days, %	85.7 (50-100)	70.8 (42.9-100)	61.9 (42.9-100)	100 (80-100)

Data are reported as No. (%) median (interquartile range).

^aNo significant difference between groups.

^bInclude zopiclone, clonidine, dexmedetomidine, and propofol.

^cInclude chest tube, epidural, Foley catheter, IV line, nasogastric tub, and percutaneous drain.

Exploratory Clinical Outcomes

Overall, we detected delirium in 14 patients (20%) after randomization, with a median duration of 1.5 days (IQR, 1-2 days) (Table 2). Subsyndromal delirium was detected (ever) in 25 patients (35.7%), and coma was detected (ever) in 7 patients (10%). Median duration of ICU stay was 4 days (IQR, 3-7 days); median duration of hospital stay was 13.5 days (IQR, 8-21 days).

Discussion

In this feasibility trial of 2 low-dose melatonin regimens and a placebo, we demonstrated good fidelity of protocol delivery in which the prescribed medication was administered as intended with minimal adverse events and excellent data capture, including exploratory clinical outcomes. However, recruitment for the study was challenged by the modest availability of eligible patients and a relatively low consent rate compared with other ICU trials.³⁶ The key reason for limited eligible patients was the mandatory requirement to confirm that delirium was not present before enrollment. This was affected by clinical factors such as sedation level that precluded the ability to administer the ICDSC or the presence of delirium on screening.

Our delirium prevention feasibility study differs in that we required patients to be confirmed as delirium-free before inclusion, and therefore, we are confident that we were exploring the effect of melatonin on the prevention of new-onset delirium as opposed to treatment of preexisting delirium, including delirium masked by sedation or coma. The restriction results in a very small pool of eligible patients for recruitment. However, demonstrating that melatonin prevents delirium might have an important clinical benefit for elective ICU admissions as a prehabilitation intervention. Other trials evaluating the effect of melatonin or melatonergic agents in critically ill patients for delirium or sleep have not made this important distinction. Two recently published systematic reviews of trials of prophylactic melatonin for delirium prevention in critically ill patients¹⁹ identify a total of 7 trials described as reporting on the effect on delirium incidence. Yet none of these trials confirmed that patients were without delirium before study inclusion. As opposed to describing their primary outcome inaccurately as delirium incidence, the most recent trial by Wibrow and colleagues³⁷ used the proportion of delirium-free assessments (within 14 days or before ICU discharge) as a marker of delirium prevalence because 10% showed positive results for delirium at the first assessment after randomization.

The inclusion of critically ill patients without confirmation of delirium status is not exclusive to melatonin or melatonergic agent trials, and can be found in trials of other interventions evaluated for delirium prevention. Inclusion of patients receiving sedation or who are comatose and therefore unable to confirm delirium-negative status before enrollment is somewhat pragmatic, but this may lead to misinterpretation of the effect of melatonin on delirium prevention.³⁸ This is important given that we have not identified any pharmacologic or nonpharmacologic interventions for delirium treatment that consistently and effectively improve important patient-centered outcomes for critically ill patients.^{15,39, 40, 41} For patients with an elective ICU admission, delirium absence should be confirmed and documented before anaesthesia; however, delirium related to iatrogenic risk factors such as anaesthesia may have developed by the time a patient is admitted to the ICU.⁴²

The melatonin doses evaluated in our feasibility trial were smaller than doses evaluated in previous delirium or sleep trials in the critically ill: 3 mg,^{13,38,43, 44, 45, 46} 10 mg,^47,48 and 21 mg.⁴⁹ We selected the dose regimens based on the physiologic dose of melatonin, approximately 0.3 mg/d,¹ and the published delirium prevention trials in hospitalized patients not in the ICU. In a small series of 12 patients in the ICU, Mistraletti and colleagues⁵⁰ found peak serum values were reached approximately 15 minutes after a 3-mg dose of oral melatonin, with high pharmacologic melatonin concentrations observed for a further 10 hours. A small study evaluating melatonin for sleep in the ICU found that 10 mg administered orally at night resulted in excessive morning drowsiness.⁴⁷ These authors suggested doses of < 3 mg likely would be most appropriate for critically ill patients. Therefore, we selected 2 different low doses to test which may explain the low incidence of adverse effects detected.

Our feasibility trial demonstrated challenges with participant enrollment both in terms of eligibility and of consent. Our consent rates may have been influenced by junior research staff, despite receiving training on informed consent processes, and multiple active trials competing for enrollment. Using novel strategies such as an animated video or infographic to provide participant information may improve consent rates for this relatively simple and low-risk intervention. Other limitations included that we were unable to recruit any patients to the planned analysis of plasma melatonin levels, resulting in an inability to understand the pharmacokinetics of the 2 melatonin dosing regimens in critically ill patients. We hypothesize that those participating in the trial did not consent to the 8 blood samples given a relatively low acuity of illness, with only a few patients having arterial catheters that are used commonly for easier blood draws. We also did not protocolize other interventions that could impact delirium such as the A2F bundle. However, based on detailed data capture of factors associated with delirium, we had similar management of all 3 groups.

Interpretation

Our trial protocol comprising 2 low doses of enteral melatonin and a placebo for delirium prevention in critically ill adults demonstrated feasibility for our primary outcome of protocol adherence in terms of drug administration. In terms of secondary outcomes, trial eligibility rates were modest, primarily because of delirium present on screening or inability to screen for delirium, and consent rates were low. No serious adverse events occurred, with few participants experiencing morning sleepiness. This finding showcases true delirium prevention trials, that is, those that confirm delirium is absent on enrollment, are particularly challenging in critically ill patients in a reasonable time frame and should be considered when designing future delirium prevention trials.

Funding/Support

This work was supported by the University of Toronto and in-kind funding from each site.

Financial/Nonfinancial Disclosures

None declared.