Prophylactic melatonin for delirium in intensive care (Pro‑MEDIC): a randomised controlled trial

Delirium is a frequent condition in the critically ill population. It is highly stressful for patients and their families and manifests as an acute cerebral compromise affecting attention and cognition. It worsens outcomes, mortality rate, ICU and hospital LOS, duration of mechanical ventilation, use of physical or pharmacological restraints and suppressors, cognitive deterioration and increased medical costs. The pathophysiology is still elusive but is considered multifactorial, hinting at an association between benzodiazepines and dysregulated sleep. The best approach combines a paradigm shift from sedation to improving analgesia and non-pharmacological measures such as early mobilisation, physiotherapy, and family presence.

Melatonin, produced by the pineal gland, modulates sleep patterns, improves sleep quality and duration in noncritically ill patients, and has anti-inflammatory and immunomodulatory properties and possible neuroprotective effects in TBI, encephalopathy, and neurodegenerative disease. Dysregulated sleep and abnormal melatonin secretion have been associated with delirium. Melatonin is safe, readily available, inexpensive, and widely used in the general population, all favourable properties as an intervention for delirium prevention. The authors hypothesised that early melatonin administration could successfully ameliorate delirium in critically ill patients. They designed and performed a multicenter, placebo-controlled, double-blind, intention-to-treat RCT. It enrolled 847 patients in 12 ICUs in Australia, within 48 h of ICU admission, 18 years or older, with anticipated LOS of 72 hours or longer. Pregnant, breastfeeding, non-English-speaking patients with prior administration or hypersensitivity to melatonin, with persistent neurological and/or hepatic impairment or expected death within 48 h, were excluded from the study. 

Patients were randomised by computer to either 4 mg (2 ml) of melatonin or placebo at 21:00 h for 14 consecutive nights, or until ICU discharge, on a 1:1 ratio, with unidentifiable drug characteristics. All personnel except the randomisation statistician, the pharmacy and DSMC Clinicians remained blinded. Protocol violation was defined as an open-label prescription of melatonin, while all other interventions remained at the clinician’s judgement. Melatonin serum levels, 2–3 h post-administration, were measured in a suitable sample of 17 (12 melatonin, 5 placebo) patients. In the advanced stages of the study, the enrolment was narrowed to prospectively consenting individuals because of the juridical withdrawal of the permission to enlist debilitated patients. A pre-arranged evaluation of outcomes before and after this time point was also performed. Delirium was calculated twice daily, with the Confusion Assessment Method for ICU (CAM- ICU) score on essentially de-sedated (as much as possible) patients. 

The primary outcome was the variation in delirium-free assessments in the two groups, within 14 days or before ICU discharge, as an indicator of delirium prevalence. Secondary outcomes calculated for the patients without delirium or coma, delirium and coma-free duration, delirium severity (measured via CAM-S score), rescue therapies implemented, repeated sleep assessments (Richards–Campbell’s Questionnaire at the bedside and Little’s Questionnaire at 2–14 days after discharge), LOS (ICU and hospital) improvement of cognition and attention and mortality with a follow-up of 90 days. Serious adverse events (SAEs) reported mortality, unexpected or morbidity (presumed to be) associated with the intervention.

The primary outcome, variations in delirium-free evaluations, were analysed using Mann–Whitney U test, along with a longitudinal analysis for repeat participant exams and generalised estimating equations (GEE) for fixed effects of treatment, time, and their interplay. Delirium was examined, with univariate logistic regression demarcating patients as with or without manifesting delirium during the study. Medications, non-pharmacological delirium treatments, mechanical ventilation, and differences in LOS and ICU mortality were evaluated with univariate logistic regression. Cox proportional hazards and Kaplan–Meier curves modelled and presented the 90 days hospital mortality. For the statistical analysis, the investigators chose Stata V15 (Stata- Corp LLC).

The median number of assessments was 5 (melatonin group) and 6 (placebo group) for each patient. There were missing assessments in 17.1% of the population. The sample size was increased by 15% to counter the chasm. From all the time points examined, 78.7% were delirium-free, with no characteristic variation between patients under melatonin (79.4%) or placebo (77.9%). Moreover, no significant differences in mechanical ventilation duration, LOS, ICU readmission, or mortality outcomes in the proportion of delirium-free assessments per patient between melatonin (79.2%) and placebo (80%) pools (p = 0.457), and no difference in total patients, free from delirium or coma throughout the study [193 (46.1%) vs 216 (51.2%)] was observed. Additionally, no significant effect of treatment, interaction between treatment and time, with or without adjustment for age and gender, was recorded. The same patterns were observed regarding antipsychotics, benzodiazepines, or physical restraints. Irrespective of the treatment, there was a consistent effect of time on diminishing the chances of delirium presentation.

STUDY STRENGTHS & LIMITATIONS

Strengths

• As of its publication, this is the largest RCT of melatonin for delirium prevention. 
• All analysed patients had extensive follow-up data, including mortality and LOS. 
• The intervention and primary outcome assessment were limited to ICU, which increases homogenisation and specificity.
• Delirium-free assessments were more frequent than anticipated.

Limitations

• The intervention and primary outcome assessment were limited to ICU population, thereby preventing the detection of melatonin’s long-term effects and prohibiting universal assumptions.
• There were missing assessments, and the investigators increased the sample size to offset the omissions, but this may have been deficient. 
• Since there is no specific approved dosing strategy, the dose may have been inadequate. Moreover, a modification in the dosing modality and frequency may produce a more successful outcome, even though the study verified the enteral absorption of melatonin. 

TAKE-HOME MESSAGES

In this large RCT with 847 ICU patients, the early and continued melatonin administration for a maximum of 14 consecutive days did not yield a difference in the prevalence of delirium compared to placebo. The rate of delirium-free assessments was 79.2% in the melatonin group and 80% in the placebo group, producing a non-significant variation. The secondary outcomes were also similar in both groups; therefore, the results of this study do not encourage the preemptive routine administration of melatonin to avert delirium in an ICU population.

This article review was prepared and submitted by Dimitrios Papadopoulos MD, MSc, PhD, Senior Consultant in Intensive Care Medicine, General Hospital of Larisa, Larisa, Greece, on behalf of the ESICM Journal Review Club.