Vasopressin versus Norepinephrine Infusion in Patients with Septic Shock

Russel. NEJM 2008; 358(9):877-887. doi:10.1056/NEJMoa067373

Clinical Question

  • In adult patients with septic shock, does the addition of vasopressin infusion to a norepinephrine infusion compared to a norepinephrine infusion alone decrease mortality rate at 28 days?


  • Randomised, controlled trial
  • Stratified by severity of shock
  • Double-blinded (patient, clinician and assessors)
  • Modified intention-to-treat analysis
    • Patients that were randomised but did not receive the study infusion were excluded from analysis
    • This decision was taken during recruitment phase of trial without awareness of data
    • It was felt that this would not introduce bias as allocation should be equally distributed between groups
  • Power calculation:
    • Baseline mortality 60%
    • Powered at 80% to detect absolute 10% difference
    • Two-sided alpha error of 0.05


  • 27 centres across Canada, Australia and United States of America
  • July 2001 to April 2006


  • Inclusion: Over 16 years old with septic shock
    • 2 SIRS criteria
    • Proven or suspected infection
    • New organ dysfunction
    • Fluid-unresponsive shock requiring at least 5 microg/min norepinephrine (or equivalent) for at least 6 hours
  • Exclusion: open-label vasopressin prior to inclusion; unstable coronary syndrome; acute mesenteric ischaemia; hyponatraemia (<130mmol/l); traumatic brain injury (GCS < 8); Raynaud’s phenomena; pregnancy; disease with < 6 month prognosis
  • 6229 patients screened –> 802 randomised –> 778 analysed
    • 3.0% lost to follow up, withdrew or excluded post-randomisation


  • Vasopressin 0.01 to 0.03 U per minute
    • Made as 30 units in 250 ml 5% dextrose (0.12 U/ml)
    • Commenced at 5.0 ml/h
    • Increased by 2.5 ml/hr every 10 minutes until 15 ml/hr reached


  • Norepinephrine 5 to 15 microg per minute
    • Made as 15 mg in 250 ml 5% dextrose (60 microg/ml)
    • Commenced at 5.0 ml/hr
    • Increased by 2.5 ml/hr every 10 minutes until 15 ml/hr reached

Management common to both

  • Target blood pressure defined as 65 to 75 mmHg, but modified at clinician’s discretion
  • During initiation, open-label vasopressor was used to maintain constant mean arterial blood pressure
  • Open label vasopressor was allowed to continue if study drug infusion was at maximum rate allowed
  • Open label vasopressor was preferentially weaned
  • Weaning of study drug infusion was permitted once target mean arterial blood pressure was maintained for 8 hours without open-label vasopressors
  • Cross-over or open-label vasopressin were not allowed
  • If subsequent recurrence of shock required vasopressors, the study drug was preferentially used


  • Primary outcome:
    • No significant difference in rate of death at 28 days
    • Vasopressin group 35.4% vs norepinephrine group 39.3%
    • Absolute risk reduction 3.9% (95% CI -2.9% to +10.7%)
    • P = 0.26 by Fisher’s exact test; Fragility Index -12
  • Secondary outcome: no significant difference in…
    • Mortality at 90 days
    • Days alive and free of organ dysfunction
    • Need for renal replacement therapy
    • Need for vasopressor
    • Need for corticosteroids
    • Length of stay (ICU and hospital)
    • Adverse incidents
  • Other measures:
    • Mean arterial blood pressure was similar between groups
    • Heart rate was lower in vasopressin group
    • Norepinephrine infusion rate was lower in vasopressin group (i.e. vasopressin was open-label norepinephrine sparing)

Authors’ Conclusions

  • Low dose vasopressin commenced at 6 hours after the initiation of shock therapy does not decrease mortality
  • With 95% confidence, this trial excludes a benefit of greater than 10.7% or a harm greater than 2.9%, suggesting that vasopressin use is acceptable in clinical practice but warrants further investigation


  • Good external generalisability – multi-centre trial with appropriate inclusion and exclusion criteria
  • Acceptable internal validity – appropriate methodology and statistical testing


  • Clinical question – the design was to compare vasopressin vs norepinephrine but since the majority of patients in the vasopressin group also received open-label norepinephrine, this trial is actually investigating ‘the addition of vasopressin to standard vasopressors vs standard vasopressors alone’
  • Clinical relevance – the dose of vasopressin was low, which is possibly why the safety profile was not dissimilar to noradrenaline
  • Limited external generalisability – baseline mean arterial pressure was ~72mmHg suggesting the patients in this trial were not refractory to vasopressors before randomisation (i.e. they were not the really sick patients in whom you might consider using additional vasopressors)
  • Some internal validity concerns – exclusion of patients post-randomisation may have unforeseen effects and could introduce bias
  • Statistical weakness – the trial was under-powered as the observed mortality rate was considerably less than the expected mortality rate; consequently a false negative conclusion is possible

The Bottom Line

  • This trial does not demonstrate a mortality benefit from adding vasopressin if mean arterial pressure is adequately maintained with standard vasopressors, however the strength of this evidence is not strong enough to conclude it definitely does not have a benefit
  • I shall continue using vasopressin in vasopressor-refractory septic shock, which this trial has not investigated

External Links


Summary author: Duncan Chambler
Summary date: 8 August 2016
Peer-review editor: Adrian Wong


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.