Effect of Vasopressin Initiation Threshold on Septic Shock Mortality

Peltan. JAMA 2026. First author. doi:10.1001/jama.2026.21333

Clinical Question

• In adults receiving vasopressors for septic shock, does a hospital strategy recommending vasopressin initiation at a lower or higher dose of other vasopressors, reduce 28-day mortality?

Background

• Vasopressin provides vasoconstriction through a non-adrenergic mechanism, and may reduce catecholamine requirements in septic shock
• Although vasopressin is commonly added when noradrenaline requirements are escalating, the optimal point at which it should be introduced remains uncertain
• Earlier vasopressin initiation might reduce exposure to high-dose catecholamines and potentially improve outcomes
  • Noradrenaline has been associated with a number of adverse effects such as worsening AKI and  immunosuppression
• VASST did not demonstrate a mortality benefit from vasopressin. In a subgroup analysis of those with less severe shock there was a trend in favour of mortality benefit, generating the hypothesis that earlier vasopressin might be beneficial
• VANISH examined the use of early vasopressin in patients with septic shock. Vasopressin did not reduce the number of renal replacement free days
• The OVISS study utilised a reinforcement learning model that recommended more frequent and earlier use of vasopressin than average care patterns and this  was associated with reduced mortality. However, there are no trial directly comparing initiation thresholds
• VASSPR, therefore, directly tested whether an early hospital-level vasopressin implementation strategy (at a lower background vasopressor dose), improves outcomes

Design

• Pragmatic, multicentre, open-label, cluster-randomised, cluster-crossover clinical strategy trial
• Hospitals, rather than individual patients, were randomised to a vasopressin initiation strategy
• Hospitals crossed over to the alternative strategy on the 1st day of each month with no washout period
• Trial duration fixed at shorter of 22 months or period feasible before hospital transferred to a new electronic health record
• Power calculation for cluster, cross-over RCT based on simulations and historical local data
  • 2050 patients would provide 80% power, with 5% false positive rate to detect a 5.4% reduction in 28 day mortality from a baseline of 33.0%
• Assessment of separation based on subgroup of patients who had a vasopressor dose 0.07 mcg/kg/min or greater
• Registered at ClinicalTrials.gov

Setting

• 13 hospitals in  USA
• Recruitment: 1 February 2024 to 31 July 2025

Population

• Inclusion criteria:
  • Adults aged ≥18 years
  • Receiving vasopressors for septic shock
• Exclusion criteria: none
• 3959 patients randomised
• Comparing baseline characteristics of intervention (lower threshold) vs. control group:
  • Median age: 67 vs 68 years
  • Female: 49% vs 48%
  • Chronic kidney disease: 29% vs 30%
  • Chronic cardiovascular disease: 53% vs 52%
  • Median non-cardiovascular SOFA score: 5 vs 5
  • Median lactate: 2.5 vs 2.5 mmol/L
  • Invasive mechanical ventilation: 26% vs 26%
  • Kidney replacement therapy at enrollment: 7.1% vs 7.9%
  • Pneumonia: 29% vs 28%

Intervention

• Lower-threshold vasopressin strategy
  • recommended starting vasopressin when the total dose of other vasopressors reached 0.1 µg/kg/min of noradrenaline or equivalent
  • Among patients reaching a maximum noradrenaline-equivalent dose of at least 0.07 µg/kg/min:
    • 68.0% received vasopressin
    • Median noradrenaline-equivalent dose before vasopressin initiation: 0.14 µg/kg/min

Control

• Higher-threshold vasopressin strategy
  • Hospitals recommended starting vasopressin when the total dose of other vasopressors reached 0.4 µg/kg/min of noradrenaline or equivalent
  • Among patients reaching a maximum noradrenaline-equivalent dose of at least 0.07 µg/kg/min:
    • 42.7% received vasopressin
    • Median noradrenaline-equivalent dose before vasopressin initiation: 0.34 µg/kg/min

Management common to both groups

• Vasopressin was administered at 0.03 units/min (1.8u/hr)
• Steroid use post randomisation slightly greater in lower threshold group (33 vs 28%) but otherwise largely balanced co-interventions (resuscitation IV fluid, new CVC, eTable 18) and MAP (eFigure7)

Outcome

• Primary outcome: no significant difference in 28-day mortality
  • Comparing intervention (lower threshold) vs control: 31.8% vs 33.0%
    • Adjusted marginal OR 0.98 (95% CI 0.87–1.06)
    • Adjusted risk difference −0.4 percentage points (95% CI −2.7 to 1.3)
• Secondary outcome:
• Kidney replacement therapy free days through day 28 – no significant difference
  • Adjusted marginal OR 1.01 (95% CI 0.91-1.14)
• Exploratory outcomes
• Comparing intervention vs. control group
  • No significant difference in
    • 90 day mortality: 37.9% vs. 38.4%
    • In hospital mortality: 26.0% vs 26.5%
    • New kidney replacement therapy: 8.1% vs 8.3%
    • Median ICU-free days to 28: 22 vs 22
  • Significantly greater in intervention group
    • Cardiogenic shock
      • 7.4% vs 6.3%, (adjusted OR 1.19, 95% CI 1.01–1.42)

Authors’ Conclusions

• A hospital strategy recommending initiation of vasopressin at a lower dose of other vasopressors did not reduce 28-day mortality

Strengths

• Cluster-crossover randomisation provides a powerful method for evaluating a hospital-level treatment strategy
• Diverse hospital type (tertiary, regional and community) and staffing improve generalisability
• Diverse strategies cross sites to facilitate adherence
• Monthly crossover reduces the influence of fixed differences between participating hospitals
• Robust sensitivity analyses challenging different assumptions all consistent with primary result (eTable 14)
• Highly pragmatic design reflects how a change in vasopressor policy would actually be implemented in an ICU
• Meaningful separation between strategies (proportion administered vasopressin, lower maximum exposure to norepinephrine prior to vasopressin initiation in lower threshold group)

Weaknesses

• Septic shock definition pragmatic (a clinician’s documented indication as sepsis within the vasopressor EMR order) but may include other pathologies and differs from other trials that use Sepsis-3 criteria (EMR question asked “Is infection a possible contributor to shock?)
• Open-label treatment creates potential performance bias
• Substantial crossover/overlap between strategies and strict adherence to strategy only 57.5% in lower threshold group compared to 73.2% in higher threshold group
• Dilution of treatment effect: all patients with septic shock were included, even those whose vasopressor requirements never reached the intervention threshold. Approximately 23% never reached 0.07 µg/kg/min which was pre-defined as the value require to result in differential management in strategies; however the cluster power calculation did account for this
• Small benefits cannot be excluded: The adjusted 95% confidence interval was compatible with up to a 2.7 percentage-point absolute mortality reduction with the lower-threshold strategy. In  my opinion introducing vasopressin when norad is 0.14 vs 0.34 would be unlikely to achieve a mortality benefit of >2.7%. Larger number of patients would be required to get adequate power for a study to pick up smaller differences

The Bottom Line

• This randomised clinical trial reported no significant mortality benefit when vasopressin was introduced when noradrenaline was 0.14 vs 0.34 µg/kg/min. However, confidence intervals mean that a small benefit of up to 2.7% cannot be excluded. As any benefit from this intervention is likely to be small and VASST and OVIS trials have suggested potential benefit from an earlier initiation of vasopressin, I will continue to use an early approach for the introduction of vasopressin in septic shock

External Links

• • article Effect of Vasopressin Initiation Threshold on Septic Shock Mortality
  • further reading CCR Trial Summary & Critique

 

Metadata

Summary author: David Slessor
Summary date: 23rd October 2026
Peer-review editor: George Walker

Picture by: unsplash