Prophylactic melatonin for delirium in intensive care (Pro-MEDIC): a randomized controlled trial

B Wibrow. Intensive Care Medicine 2022. doi: 10.1007/s00134-022-06638-9

Clinical Question

  • In adults with an expected ICU stay of > 72 hours, does the provision of enteral melatonin compared to placebo increase the number of delirium-free assessments within 14 days or before ICU discharge?


  • Delirium is highly prevalent in ICU with associated increased mortality and morbidity
  • Melatonin concentrations are lower in those patient with delirium compared to those without
  • Melatonin has been shown to speed up sleep initiation and improve sleep efficiency without changing the sleep cycle
  • Small and predominantly single centre studies have shown a benefit of melatonin improving delirium rates; however these results haven’t been reproduced in large, multi-centre RCTs


  • Multi-centre, randomized, placebo-controlled, double-blind trial
  • Randomised in 1:1 ratio via compute generated list
  • Stratified by site in blocks of 6
  • Melatonin and placebo manufactured in liquid form with identical physical characteristics
  • CAM-ICU assessment tool performed twice daily on all assessable patients
  • Sub-set of patients had serum melatonin concentrations measured 2-3 hours post administration
  • Local ethical approval
  • DMSB included a sleep physician
  • Sample size
    • Based on prospective data from two main hospital sites involved
    • 54% assessments deemed delirium free (SD 45%)
    • To detect a 10% absolute increase in delirium free assessments with 80% power and alpha of 0.05 required 319 patients in each arm
    • Rounded to 425 patients in each group to allow for non-parametric tests, missing data and loss to follow-up


  • 12 ICUs in Australia
  • July 2016 – September 2019


  • Inclusion:
    • Adults (>18 yo)
    • Admitted to ICU with an expected LOS > 72 hours
    • Within 48 hours of ICU admission
  • Exclusion:
    • Already receiving melatonin
    • Pregnancy / Breastfeeding
    • Expected death within 48 hours
    • Treating clinician believed that a delirium assessment within the next 14 days would not be possible due to persistent neurological dysfunction
    • No enteral route
    • Hepatic impairment
  • 2764 met inclusion criteria
    • 1478 excluded –> 1286 eligible
    • 847 enrolled
  • Baseline demographics similar (comparing melatonin vs placebo)
    • Age: 62 vs 62
    • Regular alcohol use: 34% vs 39%
    • Past psychiatric presentation: 16 vs 18%
    • Pre-DELIRIC score: 45.3 vs 42.7
    • CAM +ve at baseline: 10% vs 8%
    • Emergency ICU admission: 82% vs 80%
    • Sepsis as diagnosis: 24% vs 23%


  • Melatonin
    • 4mg (2mls) at 21:00 for 14 nights or until ICU discharge


  • Placebo
    • 2mls of placebo at the same time

Management common to both groups

  • All other care at treating clinicians discretion
  • Staff advised to de-sedate patients as much as possible and allow flexibility in time of CAM-ICU assessment to minimise un-assessable patients


  • Primary outcome:
    • Number of delirium free assessments per patient: 79.2% (melatonin) vs. 80.0% (placebo); p = 0.547
      • Comparing melatonin group vs placebo group:
        • Number of assessments performed: 2975 vs 2846
        • Number of assessments per patient (median): 5 vs 6
        • Number of non-assessable time points: 19.8% vs 17.7%
        • Number of missed time points: 16.6% vs 17.7%
  • Secondary outcomes:
    • Comparing melatonin vs placebo
    • No significant difference in
      • Average percentage of delirium free days: 79.5% vs. 80.2%
      • Average percentage of delirium and coma free days: 64.2% vs 67.6%
      • Participants with delirium: 35.1% vs 32.7%
      • No difference in sleep quantity or quality
        • > 40% in both groups had sleep quality and quantity rated as poor or very poor
      • Requirement for anti-psychotics: 21.2% vs 19.4%
      • ICU LOS: 5 vs 5 days
      • Hospital LOS: 14 vs 12 days
      • 90 day mortality: 15.5% vs 15.6%
  • Subgroups:
    • No difference in any subgroup (Age, Sex, Baseline delirium state, ICU diagnosis and risk of delirium by pre-DELIRIC)
  • No adverse events in any group

Authors’ Conclusions

  • Enteral melatonin within 48 hours of ICU admission did not reduce the prevalence of delirium compared to placebo


  • Multi-centre
  • Randomised
  • Double-blind and placebo controlled
  • Pre-published statistical analysis plan
  • Local data used in sample size calculation
  • Minimal loss to follow up (increases power of study)
  • Plausible rationale for use of melatonin and data collected to show that melatonin levels increased post administration
    • Median serum concentration was 19.4 ng/ml in melatonin group vs < 0.1 ng/ml in placebo
  • Excellent collection of data (pharmacological and non-pharmacological therapies) that could act as confounders
  • Largest study to date regarding a treatment which is widely used with a mixed, limited and heterogenous evidence base


  • Large numbers excluded. Of those screened:
    • ~10% were not expected to improve within 14 days
    •  ~7% unable to perform CAM-ICU for preceding reason
    • ~11% other
  • Rates of missed assessments were marginally higher than expected, and allowed for in the sample size calculation
  • ~9% were CAM +ve at baseline
  • Only one dosing regime used for all patients – would a higher dose or SR release formulation be more appropriate?
    • The authors provide a solid rationale for the decision to use 4mg – 10mg has been shown to have supratherapeutic day time level

The Bottom Line

  • This trial provides high quality evidence that although safe, 4mg melatonin does not affect rates of delirium
  • Interestingly sleep quantity and quality was not improved (albeit as a secondary outcome); this will make me re-consider my use of melatonin routinely

External Links


Summary author: George Walker @hgmwalker89
Summary date: 5th December 2022
Peer-review editor: David Slessor

Picture by: nambitomo/iStock


Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.