Skip to main content

Foundational Trials

AIRWAYS-2

Benger JR, Kirby K, Black S, et al. Effect of a strategy of a supraglottic airway device vs tracheal intubation during out-of-hospital cardiac arrest on functional outcome: the AIRWAYS-2 randomized clinical trial. JAMA. 2018;320(8):779-791

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Effect of a Strategy of a Supraglottic Airway Device vs Tracheal Intubation During Out-of-Hospital Cardiac Arrest on Functional Outcome: The AIRWAYS-2 Randomized Clinical Trial
  • Acronym: AIRWAYS-2
  • Year: 2018
  • Journal published in: JAMA
  • Citation: Benger JR, Kirby K, Black S, et al. Effect of a strategy of a supraglottic airway device vs tracheal intubation during out-of-hospital cardiac arrest on functional outcome: the AIRWAYS-2 randomized clinical trial. JAMA. 2018;320(8):779-791.

Context & Rationale

  • Background
    • Advanced airway management (AAM) during out-of-hospital cardiac arrest (OHCA) aims to secure ventilation/oxygenation while minimising interruption to chest compressions; the optimal strategy in paramedic-led systems was uncertain.
    • Endotracheal intubation (TI) has been treated as a “definitive” airway but is technically demanding, may require pauses in compressions, and success varies widely with operator volume/experience.
    • Supraglottic airway devices (SADs) are easier and faster to place, but concerns persisted about aspiration, inadequate ventilation, and displacement; observational comparisons were inconsistent and plausibly confounded by indication and resuscitation quality.2
    • Before AIRWAYS-2, the evidence base comprised heterogeneous observational data and relatively small/indirect trials, leaving major residual uncertainty for guideline panels and EMS policy.1
  • Research Question/Hypothesis
    • Does an initial strategy using a trial supraglottic airway device (i-gel) improve functional outcome (mRS 0–3) at hospital discharge (or 30 days if still inpatient) compared with an initial strategy using tracheal intubation in adult OHCA managed by paramedics?
  • Why This Matters
    • Airway strategy is a high-frequency, system-level EMS decision with major training, governance, and equipment implications.
    • Even modest absolute improvements in neurologically intact survival (pre-specified as 2% absolute) would be clinically and population-relevant in OHCA.
    • A pragmatic, cluster-randomised design was needed because patient-level randomisation is impracticable at the point of arrest care.

Design & Methods

  • Research Question: In adult OHCA treated by paramedics, does a strategy of initial i-gel insertion (trial SAD) versus initial tracheal intubation improve functional outcome (mRS 0–3) at hospital discharge/30 days?
  • Study Type: Pragmatic, multicentre, cluster randomised clinical trial with paramedics as clusters; four English ambulance services; prehospital setting (OHCA).
  • Population:
    • Setting: UK ambulance services; OHCA attended by a participating study paramedic.
    • Inclusion: Adult (≥18 years) non-traumatic OHCA where resuscitation was attempted and the study paramedic was first or second on scene (and eligible to deliver AAM).
    • Key exclusions (trial eligibility): Traumatic injury; age <18 years; not an OHCA / treated in hospital; mouth opening <2 cm; detained by His Majesty’s Prison Service; resuscitation not commenced or not continued by ambulance staff; study paramedic not first/second on scene or airway management already started.
  • Intervention:
    • Allocated strategy: Trial supraglottic airway device (i-gel) as the initial AAM during OHCA.
    • Escalation/rescue: Standard airway ladder permitted (bag-mask ventilation and adjuncts as needed); if initial AAM unsuccessful, other airway approaches could be used as clinically indicated.
  • Comparison:
    • Allocated strategy: Tracheal intubation as the initial AAM during OHCA.
    • Technique support: Intubation guidance included two-person technique and use of a bougie where appropriate; rescue with supraglottic airway devices was permitted after unsuccessful attempts.
  • Blinding: Treating paramedics were unblinded (procedural intervention); primary outcome was derived from hospital status with mRS assessed at discharge/30 days using a standardised approach, with assessors intended to be independent of prehospital allocation where feasible.
  • Statistics: Sample size planned at 9,070 total (4,535 per group) to detect a 2% absolute increase in good neurological outcome (mRS 0–3) from 8% to 10%, with 90% power at α=0.05, inflating for clustering with assumed ICC 0.005; primary analysis was intention-to-treat with mixed-effects regression accounting for cluster (paramedic) and stratification factors.
  • Follow-Up Period: Primary endpoint at hospital discharge (or 30 days if inpatient); survivors were eligible for longer-term follow-up (mRS and quality of life at 3 and 6 months) contingent on consent.

Key Results

This trial was not stopped early. Recruitment met/exceeded the planned sample size and follow-up for the primary endpoint was near-complete.

Outcome Supraglottic airway strategy (i-gel) Tracheal intubation strategy Effect p value / 95% CI Notes
Primary: mRS 0–3 (“good recovery”) at hospital discharge/30 days 311/4882 (6.4%) 300/4407 (6.8%) OR 0.92 95% CI 0.77 to 1.09; P=0.33 Adjusted risk difference −0.6% (95% CI −1.6 to 0.4)
Time to death (overall) Median 67 min (IQR 41 to 267); n=4871 Median 63 min (IQR 41 to 216); n=4400 HR 0.97 95% CI 0.93 to 1.02; P=0.22 Adjusted for stratification variables; robust SEs for clustering
72-hour survival 664/4872 (13.6%) 575/4395 (13.1%) OR 1.04 95% CI 0.92 to 1.18; P=0.54 Adjusted risk difference 0.4% (95% CI −1.0 to 1.9)
Initial ventilation success (≤2 attempts at AAM) 4255/4868 (87.4%) 3473/4397 (79.0%) OR 1.92 95% CI 1.66 to 2.22; P<0.001 Adjusted risk difference 8.3% (95% CI 6.3 to 10.2)
Any unintended loss of a previously established airway (among those with ≥1 successful AAM) 412/3900 (10.6%) 153/3081 (5.0%) OR 2.29 95% CI 1.86 to 2.82; P<0.001 Adjusted risk difference 5.9% (95% CI 4.6 to 7.2)
Regurgitation at any time (before/during/after AAM combined) 1268/4865 (26.1%) 1072/4372 (24.5%) OR 1.08 95% CI 0.96 to 1.20; P=0.21 Adjusted risk difference 1.4% (95% CI −0.6 to 3.4)
Aspiration at any time (before/during/after AAM combined) 729/4824 (15.1%) 647/4337 (14.9%) OR 1.01 95% CI 0.88 to 1.16; P=0.84 Adjusted risk difference 0.1% (95% CI −1.5 to 1.8)
ROSC at arrival to ED or hospital 1495/4880 (30.6%) 1249/4404 (28.4%) OR 1.12 95% CI 1.02 to 1.23; P=0.02 Adjusted risk difference 2.2% (95% CI 0.3 to 4.2)
Chest compression fraction (subset; two ambulance trusts) Median 86% (IQR 81 to 91); n=34 Median 83% (IQR 74 to 89); n=32 GMR 0.82 95% CI 0.62 to 1.07; P=0.14 Very small analysed subset (returned/usable compression cards)
  • The i-gel strategy improved early procedural endpoints (ventilation success; ROSC on ED/hospital arrival) but did not improve the primary functional outcome (mRS 0–3) or 72-hour survival.
  • “Unintended loss of airway” after successful AAM occurred more often with the i-gel strategy (10.6% vs 5.0%; OR 2.29; 95% CI 1.86 to 2.82; P<0.001).
  • Regurgitation and aspiration, when aggregated across timing, were not significantly different between strategies (regurgitation OR 1.08; aspiration OR 1.01).

Internal Validity

  • Randomisation and allocation: Paramedics (clusters) were randomised 1:1 via a secure web-based system using varying block sizes (4–8) with stratification (EMS organisation, paramedic experience, distance from base to usual destination hospital), supporting allocation concealment at the time of paramedic enrolment.
  • Missing data / exclusions: Primary outcome analysis included 4407/4410 (TI) and 4882/4886 (SAD) patients; exclusions were minimal and largely administrative (eg, admitted to nonparticipating hospital / could not be identified), limiting attrition bias.
  • Performance and detection bias: Providers were unblinded; however, the primary outcome was anchored to survival status and structured mRS assessment at discharge/30 days, with death deterministically mapped to mRS=6.
  • Protocol adherence and separation: Among enrolled patients, ≥1 airway management attempt occurred in 3419/4410 (TI) and 4161/4886 (SAD); first AAM differed substantially by allocation (TI arm: 2724 TI-first vs 623 trial SAD-first; SAD arm: 4009 trial SAD-first vs 116 TI-first), indicating meaningful but asymmetric crossover/contamination.
  • Separation of the variable of interest (procedural): Initial ventilation success (≤2 AAM attempts) was 4255/4868 (87.4%) with the SAD strategy vs 3473/4397 (79.0%) with TI (OR 1.92; 95% CI 1.66 to 2.22; P<0.001), demonstrating robust separation in a proximal process measure.
  • Baseline comparability: Baseline Utstein-style descriptors were closely balanced (median age 73–74 years; witnessed arrest ~63%; presenting VF ~23%; bystander CPR ~63–65%), reducing concern for confounding through imbalance.
  • Timing and context at paramedic arrival: Airway management was frequently already in progress when the study paramedic arrived (~30%); this pragmatic feature improves realism but can dilute the contrast in “initial airway” strategy and complicates causal attribution for downstream outcomes.
  • Heterogeneity: The cluster design acknowledges operator-level variability; mixed-effects modelling with paramedic as a random effect appropriately targets this, but residual heterogeneity in TI proficiency remains a plausible effect modifier.
  • Statistical rigour: The planned sample size was achieved; the primary analysis was pre-specified (intention-to-treat) and clustered; however, the observed “good recovery” rate (6.4–6.8%) was lower than the planning assumption (centred on 9%), reducing power for a fixed 2% absolute effect.

Conclusion on Internal Validity: Overall, internal validity appears moderate-to-strong: randomisation and near-complete outcome ascertainment were robust, but unavoidable unblinded delivery, pragmatic pre-arrival care, and substantial crossover (particularly in the TI arm) plausibly diluted any true between-strategy effect on functional outcomes.

External Validity

  • Population representativeness: Large, pragmatic OHCA cohort across four English ambulance services; typical age and rhythm distributions for adult OHCA; exclusions were mainly pragmatic/ethical (trauma, paediatrics, non-OHCA, very limited mouth opening).
  • System dependence: Findings are most applicable to paramedic-led EMS systems with similar TI exposure, governance, and rescue options; generalisability to physician-led EMS (higher TI success, different interruption patterns) is uncertain.
  • Intervention feasibility: The i-gel strategy reflects an implementable equipment/training decision; however, local airway algorithms (attempt limits, use of videolaryngoscopy, supraglottic rescue choices) can materially influence effects.
  • Outcome relevance: Functional outcome at discharge/30 days is clinically meaningful but may miss late neurological recovery; longer-term outcomes depended on survivor consent and are not the primary driver of the headline result.

Conclusion on External Validity: Overall, external validity is strong for UK-style paramedic OHCA care and more limited for EMS models with markedly different TI proficiency, airway devices, or post-arrest pathways.

Strengths & Limitations

  • Strengths:
    • Large, pragmatic, multicentre trial with paramedic-level randomisation aligned to real-world operational constraints.
    • Clinically meaningful primary endpoint (functional outcome) with near-complete ascertainment.
    • Prespecified clustered analyses with mixed-effects modelling and adjustment for stratification factors.
    • Rich process measures (ventilation success, airway loss, regurgitation/aspiration, ROSC timing) that help interpret null functional findings.
  • Limitations:
    • Unblinded procedural delivery and pragmatic pre-arrival care (airway already in progress in ~30%) reduce controllability and potentially dilute between-group contrast.
    • Substantial crossover/contamination, particularly from TI allocation to SAD-first airway (623 received trial SAD first in the TI arm), complicating interpretation of “strategy” effects.
    • Lower-than-anticipated good-outcome event rate (6.4–6.8% vs planning centred on 9%) limits power for modest absolute effects.
    • Compression fraction assessment was limited to a very small analysed subset, constraining inference about interruption-mediated mechanisms.

Interpretation & Why It Matters

  • Clinical practice
    In a paramedic-led system, choosing i-gel as the initial AAM strategy improved ventilation success and modestly increased ROSC on ED/hospital arrival, but did not improve neurologically favourable survival at discharge/30 days; airway strategy alone is unlikely to be a dominant determinant of functional outcome in OHCA.
  • Mechanistic signal
    The i-gel strategy reduced “failed ventilation” but increased unintended airway loss after successful AAM (10.6% vs 5.0%), highlighting that ease of initial placement does not equate to sustained airway security.
  • System implication
    These results support airway algorithms that prioritise rapid, high-success ventilation while tightly governing confirmation, fixation, and ongoing airway surveillance, with escalation pathways matched to provider skill and minimising CPR interruptions.

Controversies & Subsequent Evidence

  • “Strategy” vs “device effect”: Cluster allocation to a strategy (rather than forced device delivery) improved pragmatic relevance but introduced dilution through crossover and variable opportunities to deliver the allocated initial AAM; the TI arm had substantial SAD-first use (623 patients), while the SAD arm had less TI-first use (116 patients).
  • Event rate and detectable effect: Good neurological recovery occurred in 6.4–6.8%, lower than the planning assumption (centred on 9%), meaning AIRWAYS-2 was well-powered for its prespecified 2% absolute target but less informative for smaller true effects.
  • Comparative interpretation alongside PART: A contemporaneous editorial interpreting AIRWAYS-2 and the US PART trial emphasised that both studies reduce uncertainty but shift the focus toward system proficiency, CPR quality, and pragmatic rescue pathways rather than “one definitive airway”.3
  • Subsequent RCTs: PART (laryngeal tube strategy vs TI) showed improved 72-hour survival with a supraglottic strategy in a different EMS context, underscoring potential effect modification by provider skill and local airway governance.4
  • Bag-mask vs TI evidence: The CAAM trial (bag-mask ventilation vs TI during OHCA) further supported the notion that “advanced airway first” is not invariably superior when TI success and interruption patterns are unfavourable.5
  • Synthesis evidence: Recent systematic reviews/meta-analyses integrating AIRWAYS-2, PART and other evidence conclude that airway choice influences process measures (success, ROSC) more consistently than long-term neurological outcome, and that heterogeneity across EMS models remains substantial.6
  • Guideline convergence: Major guideline documents after AIRWAYS-2 recommend matching airway approach to provider skill and emphasise minimising compression interruptions, with SAD or TI acceptable where competently delivered.789

Summary

  • AIRWAYS-2 was a large, pragmatic, cluster randomised UK OHCA trial comparing initial i-gel strategy vs initial tracheal intubation strategy delivered by paramedics.
  • Primary outcome (mRS 0–3 at discharge/30 days) was similar: 6.4% (i-gel) vs 6.8% (TI); OR 0.92; 95% CI 0.77 to 1.09; P=0.33.
  • Initial ventilation success (≤2 AAM attempts) was higher with the i-gel strategy: 87.4% vs 79.0%; OR 1.92; 95% CI 1.66 to 2.22; P<0.001.
  • Unintended loss of airway after successful AAM was more frequent with i-gel strategy: 10.6% vs 5.0%; OR 2.29; 95% CI 1.86 to 2.82; P<0.001.
  • ROSC at ED/hospital arrival was slightly higher with i-gel strategy (30.6% vs 28.4%; OR 1.12; 95% CI 1.02 to 1.23), without translating into better functional outcome.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • When interpreting AIRWAYS-2, distinguish “allocated strategy” from “delivered airway”, and interpret sensitivity/per-protocol analyses cautiously given strong risk of resuscitation-time selection bias.

Overall Takeaway

AIRWAYS-2 is a landmark pragmatic cluster trial because it moved OHCA airway strategy from opinion-driven practice to randomised evidence in a paramedic-led system. It showed that while an i-gel-first strategy improves ventilation success and modestly increases ROSC on hospital arrival, it does not improve neurologically favourable survival at discharge/30 days, reinforcing that airway choice must be integrated with CPR quality, resuscitation choreography, and system proficiency.

Overall Summary

  • In UK paramedic OHCA care, i-gel-first improved airway process metrics but did not improve functional outcome compared with tracheal intubation-first.

Bibliography