Early Sedation with Dexmedetomidine in Critically Ill Patients

Shehabi, Y. NEJM. May 19 2019. doi:10.1056/NEJMoa1904710

Clinical Question

  • In ventilated, critically ill patients, does the use of dexmedetomidine as the primary sedative agent compared with usual sedative agents effect 90 day mortality?


  • The ideal agent to keep patients comfortable and safe whilst ventilated in ICU is not known
  • There are several options for sedation including opioids, benzodiazepines, propofol and centrally-acting alpha-agonists such as dexmedetomidine or clonidine
  • The balance of safety and harm is always the priority in the ICU population
  • Previous studies examining dexmedetomidine have suggested shorter times to extubation and a higher number of days without coma or delirium
    • However, there are limitations to these trials, including small sample sizes, underpowering to detect clinically meaningful outcomes and inconsistency with methodology
  • The MIDEX-PRODEX trial examined the feasibility of using propofol or midazolam vs dexmedetomidine to achieve desired Richmond Agitation Sedation Scale (RASS) levels
  • Skrobik’s trial used low dose nocturnal dexmedetomidine and showed a reduced incidence of delirium
  • The DAHLIA trial supported the use of dexmedetomidine when used in delirious patients as a means of earlier liberation from the ventilator


  • Randomised Controlled Trial with 1:1 ratio
  • Allocation concealment achieved via a computerised randomisation program
  • Modified Intention to Treat Analysis (excluding those who withdrew consent and where primary outcome was not known); all but 2.4% of patients
  • Unblinded
  • Variable block sized randomisation; achieves balance in the allocation of participants and the allocation is difficult to predict due to the variable block size
  • Stratified to site and presence of sepsis
  • 90% power to detect a 4.5% difference in mortality with 4000 patients (allowing a 5% loss to follow-up)
  • 6 prespecified subgroups


  • 74 ICUs in 8 countries (Australia, Ireland, Italy, Malaysia, New Zealand, Saudi Arabia, Switzerland, UK)
  • November 2013 to February 2018


  • Inclusion: ICU patients receiving mechanical ventilation, expected to continue to at least the day after next and needing sedation for safety and/or comfort.
  • Exclusion:
    • <18 years
    • Ventilation > 12 hours before enrolment
    • Acute primary brain injury
  • 29502 patients were screened, with 4000 patients randomised, 2001 to receive dexmedetomidine and 1999 to receive usual care
    • 47 patients in the dexmedetomidine group and 35 in the usual group did not give consent, and 6 more in the dexmedetomidine group and 8 in the usual care group were lost to follow-up, leaving all but 96 (2.4%) to be analysed
  • The baseline demographics/characteristics were well matched between groups


  • Dexmedetomidine was given by continuous infusion (1µg/kg/hr) without a loading dose and adjusted as needed up to 1.5µg/kg/hr
  • The infusion was titrated to a prescribed RASS (Richmond Agitation and Sedation Scale, which is between -5 for deep sedation to +4 for severe agitation)
  • If the dexmedetomidine alone did not achieve the desired level of sedation, the treating team could add other sedative agents
    • 64% of patients also received propofol (median daily dose 9.51mg/kg), 3% received midazolam and 7% received both


  • Patients were given propofol, midazalam or other sedatives as prescribed by the treating team
    • The median daily dose of propofol was 17.9mg/kg
  • Dexmedetomidine use was discouraged but could be used as a rescue therapy if the other agents were ineffective
  • Agitated delirium could be managed with anti-psychotics

Management common to both groups

  • Remifentanil and Clonidine use was not allowed
  • Other opioids and analgesics were used at the discretion of the ICU team
  • The treatments were open label


  • Primary outcome: There was no difference
    • 90 day mortality dexmedetomidine vs usual care; 29.1% (566/1848) and 29.1% (569/1956) [95%CI, -2.9 – 2.8, p = 0.98]
  • Secondary outcome: There was no difference
    • Dexmedetomidine vs usual care
      • Death at 180 days 31.5% vs 31.3%
      • Institutional dependency at 180 days 6.7% vs 7%
      • IQCODE (assessment of cognitive decline) 3.14 vs 3.08
      • EQ-5D-3L (assessment of Quality of Life) 69.8 vs 70.2
    • There was a difference in
      • Median days free from coma to day 28; 24 vs 23 (adjusted risk difference 1.0, 95% CI 0.5-1.5)
      • Median ventilator free days at day 28; 23 vs 22 (adjusted risk difference 1.0, 95% CI 0.4-1.6)
  • Adverse Outcomes
    • Dexmedetomidine vs usual care
      • Bradycardia 5.1% vs 0.5%
      • Hypotension 2.7% vs 0.5% (the trial did not examine differences in vasopressor uses between groups)
      • Asystole 14/1954 (0.7%) vs 2/1964 (0.1%) and 7 cardiac massage events in dexmedetomidine group
  • Subgroup analysis; prespecified subgroups were age (above or below median), APACHE II (above or below median), geographic region, PaO2:FiO2 (above or below median), sepsis, admission type (operative or non-operative)
    • The subgroup did not impact the result although there was a divergence in effect observed only for age with more advanced years slightly favouring the dexmedetomidine group (-4.4 [95% CI -8.7 to -0.1])

Authors’ Conclusions

  • The use of dexmedetomidine as the primary or sole sedative in ventilated ICU patients did not result in lower 90-day mortality


  • Excellent internal and external validity
  • The authors have posed a very important question, one that impacts so many of our ICU patients and has been poorly studied previously
  • The trial carefully examined adverse effects of dexmedetomidine and highlighted the increased incidence of bradycardia, hypotension and asystole
  • Important secondary outcomes such as ventilator-free days and delirium were examined


  • I am curious about the primary outcome – it would seem biologically implausible that the use of dexmedetomidine could possibly reduce day-90 death by 4.4%
    • Whilst it’s important to study clinically meaningful outcomes, trial design needs to be careful so that we don’t miss a signal of benefit by searching for the impossible
  • 11.5% of patients in the usual care group received dexmedetomidine which may have diluted the final result
  • The dexmedetomidine group did have one less day on the ventilator and one less day with delirium and coma but these results were unadjusted for multiple comparisons
  • Dexmedetomidine is more expensive than the usual care sedatives
    • For example, it would cost around AUS$500/day to keep an 80kg person asleep with this in my unit
    • This is an important consideration for ICU resource use
    • The trial does not examine the economic impact of using dexmedetomidine

The Bottom Line

  • I would not use dexmedetomidine as my sole or primary sedative agent in ventilated ICU patients
  • I would consider using it in agitated or delirious patients as a means to wean usual sedation and liberate these patients sooner from the ventilator and reduce the duration of delirium
  • I will be very mindful of the adverse effects of dexmedetomidine and wean or stop it if patients are becoming bradycardic with the hope to avoid asystoles

External Links


Summary author: Celia Bradford
Summary date: May 22 2019
Peer-review editor: Duncan Chambler


  • Waleed Tharwat Aletreby

    I believe the dexmedetomidine group had one day MORE not less on ventilator and coma/delirium free.

    • Duncan Chambler

      Thank you for your comment. Having checked again to be sure, the Dexmed group were ventilated for LESS time. Shehabi and his team chose to count “ventilator-free” days up to day-28, which is the inverse of “ventilator days”. Therefore a higher number represents less time ventilated.
      I’ve never really understood this strategy, but it’s common. I think it makes the biostatistics easier. Doesn’t really make it more understandable though.

      • Paul Young

        It’s because death is ranked as zero ventilator free days. That way , VFDs is effectively a composite end point combining mortality and duration of ventilation for survivors.

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.