Daily Interruption of Sedative Infusions in Critically Ill Patients Undergoing Mechanical Ventilation

Kress et al. N Engl J Med 2000; 342:1471-1477.

Clinical Question

  • In medical patients who are intubated, mechanically ventilated and sedated by a continuous infusion of drugs, do daily sedative interruptions reduce the duration of mechanical ventilation or length of stay?


  • Randomised, controlled clinical trial
  • 2×2 factoria design: sedation breaks (yes/no) and sedative agent (midazolam+morphine / propofol+morphine)
  • Randomised by computer and sealed in envelopes
  • Blinding:
    • Sedation break – single-blinded (investigators aware);
    • Sedative agent – open label, non-blinded
  • Intention-to-treat basis for analysis
  • Method of recruitment (sequential, random or selective etc) not specified


  • A single university-associated medical intensive care unit (ICU) in USA
  • Dates of recruitment not specified


  • Inclusion: Intubated and ventilated patients deemed to require sedation by continuous infusion
  • Exclusion: Pregnant women, transferred from outside institutions, or post- cardiac arrest
  • 150 patients


  • From 48 hours after commencing sedation, daily interruption of sedative and morphine were performed by an investigator (who was not directly providing patient care) until:
    • patient was awake and could follow instructions;
    • or they became uncomfortable or agitated and were deemed to require the resumption of sedation
  • Sedation restarted at half previous rate and titrated
    • Target of sedation Ramsay Sedation Scale 3 or 4 by nurse adjustment of infusion


  • Adjustments were left to the discretion of the intensive care team
    • Target of sedation Ramsay Sedation Scale 3 or 4 by nurse adjustment of infusion


  • Primary outcomes: favoured sedation breaks or showed no difference
    • Duration of mechanical ventilation was significantly reduced in intervention group. Median days 4.9 [IQR 2.5–8.6] vs 7.3 [3.4–16.1]. RR of extubation 1.9 (95% CI 1.3–2.7; p<0.001)
    • Length of ICU stay was significantly shorter in the intervention group. Median days 6.4 vs 9.9. RR of discharge 1.6 (95% CI 1.1–2.3; p=0.02).
    • Length of hospital stay showed no difference: median days 13.3 vs 16.9 (p=0.19)
  • Secondary outcome:
    • Patients were ‘awake’ on more days in the intervention group: 85.5% versus control 9.0% [p<0.001]
    • Fewer diagnostic tests were required for the intervention group: 6 (4.4%) vs control 16 (27%) [p=0.88]
    • No difference in accidents (ETT or CVC removal): intervention 3 (4.4%) vs control 4 (6.7%)
    • No difference in re-intubation: intervention 12 (18%) vs control 18 (30%) [p=0.17]
    • No difference in tracheostomy: intervention 12 (18%) vs control 16 (27%) [p=0.31]

Authors’ Conclusions

  • Daily interruption of the infusion of sedative drugs is a safe and practical approach to treating patients receiving mechanical ventilation. It decreases the duration of mechanical ventilation, length of ICU stay, the dose of benzodiazepines used and improves the ability to perform clinical neurological examination (reducing special investigations).


  • Pragmatic study of a common problem
  • Appropriate randomisation method


  • Medical ICU only, so may not be generalisable to surgical or trauma patients
  • Lack of blinding of physician will strongly bias result toward positive outcome
  • Usual practice on this ICU is not stated. If sedation breaks are usual practice, this may produce a more positive outcome since it is familiar to nursing staff
  • Although “intention-to-treat” stated, this was modified: patients that died or were extubated in initial 48 hours (i.e. did not reach start of intention / control period) were excluded from analysis. Pragmatic and unlikely to bias outcome in either direction. However, it reduces generalisability to the very sick (died within 48 hours) or very well (extubated within 48 hours) cohorts of patients.
    • n=7 (9%) in intervention group; n=16 (21%) in control group.
  • Some control patients had interruptions in sedation on days other than the final day of administration (i.e. a break in sedation). This will bias the result toward no difference seen.
    • n=18 (30%)
  • There was a difference between groups in the numbers that did not awaken from coma
    • Intervention group 7 (10%) versus control group 15 (25%) [p=0.05].
    • This statistical level would suggest unlikely (5%) to be due to random chance and therefore likely (95%) due to intervention / control difference. This is a striking conclusion and is clinically unlikely: do sedation breaks reverse ‘irreversible’ coma?!
    • This emphasises the problem of drawing conclusions from probability statistics! The authors don’t comment on this.
  • Target sedation (Ramsay Sedation Scale 3 or 4) was responsive to commands or stimulus, not calm, orientated and cooperative (Ramsay 2).
    • This target may be deeper than current clinical practice.
    • Results and conclusion may not be generalisable to ICUs that target Ramsay 2 or RASS 0.

The Bottom Line

  • In ICUs (especially medical ICUs) that target moderate depths of sedation (Ramsey ≥ 3 or RASS ≤ -1) with continuous infusions or drugs, daily sedative interruptions will probably reduce the period of mechanical ventilation and ICU length of stay.
  • I believe the methodological flaws in this study exaggerate the positive effect size, and it is underpowered to say that daily sedative interruptions are safe and without psychological harm.


Full text pdf / abstract / doi: 10.1056/NEJM200005183422002

Editorial, Commentaries or Blogs


Summary author: @DuncanChambler
Summary date: 8th May 2014
Peer-review editor: @stevemathieu75

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.