Skip to main content

Foundational Trials

Wenzel

Image: Shutterstock / ChaNaWiT

Publication

  • Title: A comparison of vasopressin and epinephrine for out-of-hospital cardiopulmonary resuscitation
  • Acronym: Not applicable
  • Year: 2004
  • Journal published in: The New England Journal of Medicine
  • Citation: Wenzel V, Krismer AC, Arntz HR, Sitter H, Stadlbauer KH, Lindner KH, for the European Resuscitation Council Vasopressor During Cardiopulmonary Resuscitation Study Group. A comparison of vasopressin and epinephrine for out-of-hospital cardiopulmonary resuscitation. N Engl J Med. 2004;350:105-13.

Context & Rationale

  • Background
    • Epinephrine was the default vasopressor in advanced life support for cardiac arrest, supported primarily by physiologic rationale (alpha-adrenergic vasoconstriction to augment coronary perfusion pressure) rather than consistent evidence of improved survival with good neurological outcome.
    • Vasopressin (arginine vasopressin) offered a non-adrenergic vasoconstrictor mechanism (V1 receptors), theoretical advantages in severe acidaemia/hypoxia, and a longer half-life than epinephrine.
    • Preclinical work and small human studies suggested potential improvements in haemodynamics and short-term outcomes (return of spontaneous circulation), with uncertainty about patient-centred endpoints (survival to discharge, neurological recovery).
    • There was clinical equipoise about whether vasopressin should replace epinephrine (as first-line), or be reserved as rescue therapy, and whether effects differed by presenting rhythm (particularly asystole).
  • Research Question/Hypothesis
    • In adult out-of-hospital cardiac arrest requiring intravenous vasopressor therapy, does vasopressin (40 IU, repeat once) compared with epinephrine (1 mg, repeat once) improve key resuscitation and survival outcomes (hospital admission and discharge), and are effects rhythm-dependent?
  • Why This Matters
    • Vasopressors are near-universal in advanced cardiac arrest care; even small absolute effects on meaningful survival could translate into substantial population benefit.
    • Replacing (or supplementing) epinephrine would represent a major change in resuscitation algorithms, drug stocking, and training across EMS and hospital systems.
    • Clarifying rhythm-specific effects could enable more targeted vasopressor strategies and inform the design of subsequent confirmatory trials.

Design & Methods

  • Research Question: Whether vasopressin (40 IU, repeat once) is superior to epinephrine (1 mg, repeat once) as the initial vasopressor during adult out-of-hospital cardiopulmonary resuscitation, with outcomes assessed overall and by presenting rhythm.
  • Study Type: Randomised, multicentre, investigator-initiated, double-blind, controlled prehospital trial conducted across 44 advanced life-support units in Austria, Germany, and Switzerland.
  • Population:
    • Setting: Out-of-hospital cardiac arrest managed by advanced life-support units (physician-staffed mobile ICUs) operating under contemporaneous advanced life-support algorithms.
    • Inclusion criteria: Adults with out-of-hospital cardiopulmonary arrest due to ventricular fibrillation, pulseless electrical activity, or asystole requiring cardiopulmonary resuscitation and intravenous vasopressor treatment.
    • Key protocol timing rule: Patients with ventricular fibrillation were randomised only after three unsuccessful defibrillation shocks.
    • Exclusion criteria: Age <18 years; pregnancy; trauma; primary intracerebral haemorrhage; contraindication to vasopressor administration; hypothermia; terminal illness; do-not-resuscitate order; arrest due to haemorrhagic shock; participation in another study.
  • Intervention:
    • Vasopressin 40 IU intravenously as the initial study drug.
    • If there was no return of spontaneous circulation after 3 minutes, a second 40 IU vasopressin injection was given (maximum 80 IU study drug).
    • If return of spontaneous circulation did not occur after two study-drug injections, additional epinephrine could be administered according to the standard algorithm at the treating physician’s discretion.
    • All drug injections were followed by a 20 mL saline flush.
  • Comparison:
    • Epinephrine 1 mg intravenously as the initial study drug.
    • If there was no return of spontaneous circulation after 3 minutes, a second 1 mg epinephrine injection was given (maximum 2 mg study drug).
    • If return of spontaneous circulation did not occur after two study-drug injections, additional epinephrine could be administered according to the standard algorithm at the treating physician’s discretion.
    • All drug injections were followed by a 20 mL saline flush.
  • Blinding: Double-blind; study drugs were provided in coded, indistinguishable form; treating clinicians, patients, and outcome assessors were intended to remain unaware of allocation.
  • Statistics: A total of 571 patients per group were required to detect a 25% improvement in the primary endpoint (survival to hospital admission) with 80% power at the 5% significance level (two-sided); with a planned 30% safety margin, the target sample size was 1500 patients; analyses were primarily intention-to-treat, with logistic regression used for odds ratios and 95% confidence intervals.
  • Follow-Up Period: To hospital admission and hospital discharge (including cerebral performance category assessment among survivors at discharge).

Key Results

This trial was stopped early. Recruitment ended when funding ceased (planned N=1500; randomised N=1219; analysed N=1186 after exclusion of 33 patients with missing study-drug codes).

Outcome Vasopressin Epinephrine Effect p value / 95% CI Notes
Return of spontaneous circulation with study drugs (overall) 145/589 (24.6%) 167/597 (28.0%) OR 1.2 95% CI 0.9 to 1.5; P=0.19 Effect measure is odds ratio; OR<1 indicates advantage for vasopressin (trial convention).
Survival to hospital admission (overall; primary endpoint) 214/589 (36.3%) 186/597 (31.2%) OR 0.8 95% CI 0.6 to 1.0; P=0.06 Borderline; did not meet conventional 0.05 threshold.
Survival to hospital discharge (overall) 57/578 (9.9%) 58/588 (9.9%) OR 1.0 95% CI 0.7 to 1.5; P=0.99 11 (vasopressin) and 9 (epinephrine) were lost to follow-up before hospital discharge.
Survival to hospital admission (asystole subgroup) 76/262 (29.0%) 54/266 (20.3%) OR 0.6 95% CI 0.4 to 0.9; P=0.02 Pre-specified rhythm stratum; multiplicity considerations apply (no adjustment reported).
Survival to hospital discharge (asystole subgroup) 12/257 (4.7%) 4/262 (1.5%) OR 0.3 95% CI 0.1 to 1.0; P=0.04 Absolute numbers small; confidence interval includes values close to no effect.
Survival to hospital admission (no ROSC with study drug + received additional epinephrine) 96/373 (25.7%) 59/359 (16.4%) OR 0.6 95% CI 0.4 to 0.8; P=0.002 Post-randomisation subset defined by treatment response and subsequent management.
Survival to hospital discharge (no ROSC with study drug + received additional epinephrine) 23/369 (6.2%) 6/355 (1.7%) OR 0.3 95% CI 0.1 to 0.6; P=0.002 Neurological outcomes among survivors in this subgroup were reported with small denominators (20 vs 3).
Cerebral performance at discharge among survivors (overall; CPC categories) Good CPC: 15/46 (32.6%) Good CPC: 16/46 (34.8%) Not reported P=0.35 CPC data were missing for 11 (vasopressin) and 12 (epinephrine) survivors.
  • The trial did not demonstrate an overall improvement in survival to hospital discharge (9.9% in both groups), despite a numerically higher hospital admission rate with vasopressin (36.3% vs 31.2%; P=0.06).
  • A rhythm-specific signal was observed in asystole for both hospital admission (29.0% vs 20.3%) and discharge (4.7% vs 1.5%), but absolute event counts were small.
  • Analyses restricted to patients who failed initial study-drug treatment and then received additional epinephrine showed better outcomes in those originally assigned to vasopressin; this is not a randomised comparison of a “combination strategy”.

Internal Validity

  • Randomisation and Allocation
    • Randomisation used coded study-drug boxes prepared centrally and distributed across participating units; allocation concealment was intended through sequential use of indistinguishable packs.
    • Randomisation was stratified by centre, supporting balance across sites in a multi-national prehospital setting.
  • Drop out or exclusions (post-randomisation)
    • Randomised: 1219 patients; analysed: 1186 patients after exclusion of 33 patients (2.7%) due to missing study-drug codes.
    • Loss to follow-up before hospital discharge: 11/589 (1.9%) vs 9/597 (1.5%).
    • Cerebral performance category missing among survivors: 11 vs 12 patients.
  • Performance/Detection Bias
    • Blinding was designed to minimise differential co-intervention and outcome assessment bias during resuscitation and hospital follow-up.
    • Primary and key secondary outcomes (hospital admission, discharge) are objective; neurological categorisation (CPC) is more subjective and incomplete.
  • Protocol Adherence
    • Drug delivery timing was similar: time from initiation of basic life support to first injection 9.6 ± 4.1 min vs 10.2 ± 4.3 min; to second injection 13.3 ± 5.6 min vs 13.9 ± 5.8 min.
    • Key co-interventions during advanced life support were broadly similar between groups (e.g., intubation 84.7% vs 86.1%; fibrinolysis 7.1% vs 7.0%; sodium bicarbonate use 86.4% vs 85.9%).
  • Baseline Characteristics
    • Groups were broadly comparable: age 66.9 ± 15.7 vs 66.0 ± 16.0 years; witnessed arrest 75.4% vs 74.2%; bystander CPR 45.2% vs 45.1%.
    • Presenting rhythms were similar: ventricular fibrillation 37.9% vs 41.7%; pulseless electrical activity 17.7% vs 13.7%; asystole 44.5% vs 44.6%.
  • Heterogeneity
    • Conducted across 44 units in three countries, increasing clinical heterogeneity; stratification by centre supports balance but does not eliminate site-level variation in processes of care (including termination and transport).
  • Timing
    • Ventricular fibrillation patients were randomised only after three failed defibrillation shocks, which can delay the first vasopressor dose relative to protocols that introduce vasopressors earlier.
  • Dose
    • Study-drug exposure was protocolised (max two injections): mean number of study-drug injections 1.4 ± 0.6 vs 1.4 ± 0.7.
    • Rescue epinephrine was available in both groups if no response after study drug, affecting the “dose contrast” between strategies.
  • Separation of the Variable of Interest
    • Time from initiation of basic life support to initiation of the standard protocol with epinephrine was 17.5 ± 6.3 min (vasopressin group) vs 17.6 ± 5.5 min (epinephrine group), indicating relatively early epinephrine exposure even in the vasopressin arm.
    • Total epinephrine dose was similar: 7.3 ± 5.2 mg vs 7.2 ± 5.1 mg.
  • Key Delivery Aspects
    • High proportion of out-of-hospital arrests were witnessed and received bystander CPR (~45%), supporting a population in which prehospital pharmacologic effects might plausibly translate into downstream outcomes.
    • The primary endpoint (hospital admission) depends on successful resuscitation and downstream decisions about transport/termination, which can vary across systems and clinicians.
  • Outcome Assessment
    • Survival outcomes (ROSC, admission, discharge) were clearly defined and objective.
    • Neurological outcomes were reported by CPC category at discharge; missingness among survivors (11–12 patients) limits interpretability.
  • Statistical Rigor
    • The trial did not reach its planned sample size (target 1500; analysed 1186), reducing power for plausible effect sizes in discharge survival.
    • Multiple subgroup analyses were reported without adjustment for multiple comparisons.

Conclusion on Internal Validity: Overall, internal validity appears moderate: randomisation, concealment strategy, and blinding were strong for a prehospital drug trial, but early termination, limited separation due to subsequent epinephrine exposure, and unadjusted multiple subgroup analyses constrain causal inference (particularly for subgroup signals).

External Validity

  • Population Representativeness
    • Adults with non-traumatic out-of-hospital cardiac arrest were studied; key exclusions (trauma, haemorrhagic shock, pregnancy, terminal illness, DNR orders) appropriately narrow applicability to “medical” arrests.
    • Screening/enrolment yield was low: 5967 eligible patients were assessed and 4748 were excluded as ineligible; this reflects pragmatic field constraints and eligibility rules, and may reduce representativeness relative to an all-comers arrest registry.
  • Applicability
    • Conducted in physician-staffed prehospital systems (Austria/Germany/Switzerland); results may not translate directly to paramedic-only systems, different termination-of-resuscitation rules, or settings with delayed intravenous access and drug delivery.
    • The protocol delayed randomisation in ventricular fibrillation until after three defibrillation attempts, which may not match contemporary or alternative ALS sequencing in all systems.
    • Vasopressin availability and scope-of-practice constraints vary internationally, influencing implementability even if efficacy were established.

Conclusion on External Validity: Generalisability is moderate: findings are most applicable to adult, non-traumatic out-of-hospital arrests managed by advanced prehospital teams with early intravenous access; applicability to other EMS models and healthcare systems is less certain.

Strengths & Limitations

  • Strengths:
    • Large, multicentre, prehospital randomised controlled trial with double-blinding in a high-acuity emergency context.
    • Pragmatic design anchored to operational advanced life-support care, enhancing real-world relevance within similar EMS models.
    • Objective primary and major secondary endpoints (ROSC, hospital admission, discharge) with rhythm-stratified reporting.
  • Limitations:
    • Early stopping due to funding reduced statistical power for discharge survival and neurological outcomes.
    • Primary endpoint (hospital admission) is less patient-centred than neurologically intact survival and may be influenced by local transport/termination practices.
    • Limited separation between groups because most patients in the vasopressin arm subsequently received epinephrine (timing to epinephrine protocol was similar between groups).
    • Subgroup analyses (especially the “additional epinephrine” subset) were not randomised comparisons of treatment strategies and are vulnerable to bias from post-randomisation selection.
    • Neurological outcome reporting at discharge had missing data among survivors, and the distribution of CPC categories suggests substantial residual morbidity among those discharged alive.

Interpretation & Why It Matters

  • Clinical practice signal
    As an overall strategy, replacing epinephrine with vasopressin as the initial vasopressor during out-of-hospital resuscitation did not improve survival to hospital discharge.
  • Rhythm-specific hypothesis
    The observed association with better outcomes in asystole supports a biologically plausible, rhythm-dependent effect, but absolute numbers were small and the finding requires cautious interpretation.
  • Trial design implications
    This study illustrates the methodological challenges of prehospital resuscitation trials: dilution by protocolised rescue therapy, event-driven post-randomisation subgrouping, and reliance on endpoints that are not purely patient-centred.

Controversies & Subsequent Evidence

  • Interpretation tension: the primary endpoint (hospital admission) was numerically higher with vasopressin but did not meet conventional statistical significance (P=0.06), and survival to discharge was identical; this highlights the frequent disconnect between short-term resuscitation metrics and meaningful long-term outcomes.
  • The accompanying editorial emphasised that the overall neutral discharge survival result should dominate interpretation, and that any rhythm-dependent signal (especially asystole) should be treated as hypothesis-generating rather than definitive for guideline change.1
  • The “additional epinephrine” analyses describe a post-randomisation subset (patients who failed to achieve ROSC with study drugs), which precludes a clean causal interpretation of “vasopressin + epinephrine” synergy versus selection effects and downstream clinical decision-making.
  • Subsequent evidence syntheses have not shown a consistent benefit of vasopressin over epinephrine for survival to discharge or neurological outcome across heterogeneous trials, and have underscored limitations driven by small-study effects, rhythm heterogeneity, and non-uniform treatment algorithms.345
  • Guideline evolution: major contemporary resuscitation guidelines do not recommend routine vasopressin as a substitute for epinephrine in adult cardiac arrest, reflecting the absence of robust patient-centred benefit in trials and meta-analyses.67
  • Combination therapy has been tested in a different clinical context (in-hospital cardiac arrest) using vasopressin plus corticosteroid and epinephrine regimens, showing improved ROSC and survival in that setting; however, these trials do not directly validate vasopressin-alone replacement strategies in out-of-hospital arrest.2

Summary

  • Double-blind, multicentre out-of-hospital RCT (analysed N=1186) comparing vasopressin (40 IU, repeat once) vs epinephrine (1 mg, repeat once) as the initial vasopressor during cardiopulmonary resuscitation.
  • Trial stopped early due to funding cessation (planned N=1500); primary endpoint survival to hospital admission was 36.3% vs 31.2% (P=0.06).
  • Overall survival to hospital discharge was identical: 9.9% vs 9.9% (P=0.99).
  • Asystole subgroup showed higher admission (29.0% vs 20.3%) and discharge survival (4.7% vs 1.5%) with vasopressin, but with small absolute event counts and unadjusted multiplicity.
  • Later evidence syntheses and guidelines did not support routine vasopressin use as a substitute for epinephrine in adult cardiac arrest.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Overall Takeaway

Wenzel et al. provided one of the most methodologically robust early tests of vasopressin as a first-line vasopressor in out-of-hospital cardiac arrest, showing no overall benefit in survival to hospital discharge compared with epinephrine. The trial’s rhythm-specific signals and response-defined subgroup findings influenced subsequent research directions, but the broader evidence base and later guidelines ultimately did not support routine replacement of epinephrine with vasopressin in adult cardiac arrest care.

Overall Summary

  • In adult out-of-hospital cardiac arrest, vasopressin did not improve overall survival to discharge compared with epinephrine, despite a borderline increase in hospital admission.

Bibliography