Skip to main content

Foundational Trials

UK-REBOA

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Emergency Department Resuscitative Endovascular Balloon Occlusion of the Aorta in Trauma Patients With Exsanguinating Hemorrhage: The UK-REBOA Randomized Clinical Trial
  • Acronym: UK-REBOA
  • Year: 2023
  • Journal published in: JAMA
  • Citation: Jansen JO, Hudson J, Cochran C, et al; UK-REBOA Study Group. Emergency department resuscitative endovascular balloon occlusion of the aorta in trauma patients with exsanguinating hemorrhage: the UK-REBOA randomized clinical trial. JAMA. 2023;330(19):1862-1871.

Context & Rationale

  • Background
    Haemorrhage is a leading cause of preventable early death after major trauma; non-compressible torso haemorrhage is particularly lethal because definitive control requires operative or endovascular intervention.
  • Research Question/Hypothesis
    Does an emergency department strategy that includes resuscitative endovascular balloon occlusion of the aorta (REBOA), in addition to standard major trauma centre care, reduce all-cause mortality compared with standard care alone in trauma patients with exsanguinating torso haemorrhage?
  • Why This Matters
    REBOA had rapidly diffused into practice despite conflicting observational evidence and substantial training/organisational demands; UK-REBOA was the first randomised clinical trial intended to provide definitive effectiveness and safety evidence in real-world major trauma centres.

Design & Methods

  • Research Question: In adult trauma patients with suspected or confirmed life-threatening torso haemorrhage, does a treatment strategy including REBOA plus standard care reduce 90-day all-cause mortality versus standard care alone?
  • Study Type: Pragmatic, multicentre (16 UK major trauma centres), open-label, Bayesian, group-sequential, registry-enabled randomised clinical trial (1:1 allocation); enrolled October 2017 to March 2022; 90-day follow-up (primary).
  • Population:
    • Setting: UK major trauma centres; emergency department enrolment (research without prior consent with deferred consent for ongoing participation/data use).
    • Inclusion: age ≥16 years; confirmed or suspected life-threatening torso haemorrhage judged amenable to adjunctive REBOA (zone I or III).
    • Exclusion: known/suspected pregnancy; age ≤15 years; injuries deemed unsurvivable on clinical grounds.
  • Intervention:
    • Strategy including REBOA plus standard major trauma centre care (device choice not mandated; zone I or III at clinician discretion; partial REBOA permitted); delivered by trained clinicians with site implementation package.
  • Comparison:
    • Standard major trauma centre care alone (including intubation/ventilation as required, balanced blood product resuscitation, rapid surgical/endovascular haemorrhage control; open aortic occlusion by thoracotomy/laparotomy permitted if clinically indicated).
  • Blinding: Unblinded (procedural strategy; blinding not feasible).
  • Statistics: Recruitment target 120 (pragmatic feasibility-constrained); Bayesian group-sequential design with interim analyses after 40 and 80 randomised participants and a prespecified stopping rule for harm/futility at interim looks; primary analysis intention-to-treat using Bayesian logistic regression with odds ratios (OR) and 95% credible intervals (CrI). Protocol published in advance.1
  • Follow-Up Period: Primary: 90 days; key secondary mortality endpoints included 3 hours, 6 hours, 24 hours, in-hospital, and 6 months.

Key Results

This trial was stopped early. Planned n=120; stopped after the second interim analysis (80 participants) because the prespecified stopping rule for harm was met (posterior probability of OR>1 for 90-day mortality exceeded the prespecified threshold). A further 10 participants had already been enrolled by the time interim outcome data were available (final n=90 randomised).

Outcome REBOA + standard care Standard care Effect p value / 95% CI Notes
All-cause mortality at 90 days (primary) 25/46 (54%) 18/43 (42%) OR 1.58 95% CrI 0.72 to 3.52 Posterior probability OR>1: 86.9%
Death within 24 hours 17/46 (37%) 10/44 (23%) OR 1.85 95% CrI 0.79 to 4.46 Posterior probability OR>1: 91.8%
Death within 6 hours 13/46 (28%) 4/44 (9%) OR 3.14 95% CrI 1.13 to 9.76 Posterior probability OR>1: 98.6%
Death within 3 hours 11/46 (24%) 2/44 (5%) OR 4.25 95% CrI 1.33 to 15.99 Posterior probability OR>1: 99.3%
Definitive haemorrhage control procedure 14/46 (30%) 19/44 (43%) OR 0.60 95% CrI 0.26 to 1.37 Median time from randomisation to procedure: 83 min (IQR 56–156) vs 64 min (IQR 34–83)
ICU-free days at 90 days Mean 35 (SD 40) Mean 40 (SD 37) MD −4.79 95% CrI −20.75 to 11.31 Deaths assigned 0 ICU-free days
Hospital-free days at 90 days Mean 22 (SD 30) Mean 41 (SD 39) MD −18.58 95% CrI −32.86 to −3.93 Deaths assigned 0 hospital-free days
Red blood cells within 24 hours Mean 10 (SD 9) Mean 11 (SD 9) IRR 0.92 95% CrI 0.66 to 1.29 Median 7 (IQR 4–12) vs 9 (IQR 4–17)
Plasma within 24 hours Mean 8 (SD 8) Mean 11 (SD 10) IRR 0.73 95% CrI 0.49 to 1.08 Median 6 (IQR 3–10) vs 7 (IQR 4–18)
Overall complications 6/46 (13%) 10/43 (23%) OR 0.54 95% CrI 0.19 to 1.48 No serious adverse device events reported
  • The posterior probability consistently favoured harm with REBOA, with larger effect estimates for earlier mortality endpoints (3-hour OR 4.25; 95% CrI 1.33 to 15.99).
  • Only 19/46 (41%) in the REBOA strategy group had successful catheter insertion and balloon inflation; median time from ED arrival to inflation was 32 minutes (IQR 20–47).
  • Deaths attributed to bleeding were more frequent in the REBOA strategy group at 90 days (8/25 [32%] vs 3/18 [17%]), with most bleeding deaths occurring within 24 hours.

Internal Validity

  • Randomisation and allocation: Central web-based randomisation (1:1) using permuted blocks of 2 or 4; allocation concealment prior to randomisation plausible; open-label post-allocation.
  • Dropout/exclusions: 1 participant in standard care withdrew consent for ongoing data collection 4 days after enrolment and was excluded from analyses after that point (primary 90-day analysis n=89).
  • Performance/detection bias: Open-label strategy trial in a high-acuity environment; objective mortality endpoints mitigate detection bias, but care-pathway decisions (timing/procedures) could be influenced by strategy allocation.
  • Protocol adherence/separation: Strategy separation was incomplete: in the REBOA strategy arm, 19/46 (41%) balloon inflated; arterial access attempted but unsuccessful in 8/46 (17%); 17/46 (37%) improved during REBOA preparation and balloon inflation was not pursued; 2/44 (5%) in the standard care arm crossed over and received REBOA with balloon inflation.
  • Baseline characteristics: Overall similar injury severity (median ISS 41 [IQR 29–50] in both groups), but more profound hypotension on ED arrival in REBOA strategy arm (median SBP 84 mm Hg [IQR 58–115] vs 99 mm Hg [IQR 72–115]) and higher head AIS distribution, increasing risk of residual confounding despite randomisation.
  • Heterogeneity: Pragmatic “amenable to REBOA” eligibility relied on clinician judgement; treatment pathways varied substantially, reflecting clinical uncertainty and intercurrent events.
  • Timing: Median ED arrival to balloon inflation 32 minutes (IQR 20–47) among those inflated; time from randomisation to definitive haemorrhage control was longer in REBOA strategy arm (median 83 vs 64 minutes).
  • Dose: Among those inflated, median duration of inflation 29 minutes (IQR 19–64); partial REBOA used in 8/19 (42%); zone I 10/19 (53%) and zone III 9/19 (47%).
  • Outcome assessment: Primary outcome (90-day all-cause mortality) objective; registry linkage used for many outcomes; haemorrhage control procedures were adjudicated using prespecified criteria and independent review.
  • Statistical rigour: Prespecified Bayesian group-sequential design with interim monitoring and stopping for harm; ITT primary analysis appropriate for a strategy question; early stopping increases imprecision (wide CrIs) and susceptibility to random high early harm signals, though the direction was consistent across multiple mortality endpoints.

Conclusion on Internal Validity: Moderate—randomisation and objective mortality outcomes support causal inference, but incomplete intervention delivery, crossover, baseline physiological imbalances, and early stopping with small sample size limit precision and complicate attribution of harm to the balloon occlusion itself versus delays and system-level effects of attempting REBOA.

External Validity

  • Population representativeness: Typical UK major trauma epidemiology (predominantly blunt trauma, very high injury severity); includes patients in extremis, including those requiring CPR on arrival (9% in each group).
  • Health system context: UK major trauma centres with managed implementation/training; procedural volume per centre likely low given condition rarity, potentially limiting generalisability to high-volume systems.
  • Applicability: Findings apply most directly to ED-based REBOA adoption across multiple centres with variable institutional experience and time-to-occlusion around 30+ minutes; may not translate to systems with different trauma patterns (e.g., higher penetrating injury) or highly experienced REBOA programmes with faster access/occlusion and tightly protocolised pathways.

Conclusion on External Validity: Moderate—highly relevant to many civilian trauma systems considering broad implementation, but extrapolation to high-volume, highly experienced centres or markedly different trauma case-mix should be cautious.

Strengths & Limitations

  • Strengths: First randomised trial of REBOA in trauma; pragmatic registry-enabled design across 16 centres; prespecified Bayesian interim monitoring with stopping for harm; objective mortality outcomes with early timepoints (3/6/24 hours) informative for haemorrhage-related effects; explicit capture of real-world implementation constraints.
  • Limitations: Early stopping and small sample size with wide credible intervals; open-label strategy design with potential pathway effects; incomplete delivery of balloon occlusion (41% inflated) and crossover; baseline SBP/head injury imbalances; device type and operator experience not standardised (reflecting pragmatism but increasing heterogeneity).

Interpretation & Why It Matters

  • Clinical implication
    In this multicentre UK implementation context, a REBOA-inclusive ED strategy did not reduce mortality and showed a high posterior probability of increased early death (e.g., 6-hour mortality 28% vs 9%; OR 3.14; 95% CrI 1.13 to 9.76), arguing against routine ED REBOA deployment as a broadly implemented system-level adjunct.
  • Mechanistic inference
    The excess harm signal was strongest at early timepoints and accompanied by longer time to definitive haemorrhage control (median 83 vs 64 minutes) and fewer definitive haemorrhage control procedures (30% vs 43%), consistent with competing time/attention costs and/or failure to achieve timely haemostasis during attempted REBOA.

Controversies & Subsequent Evidence

  • Is the observed harm “REBOA” or “REBOA implementation”? Editorial commentary highlighted that device type/size and credentialling/experience were not standardised, and the observed median time to balloon inflation (32 minutes) was substantially longer than reports from experienced centres; the editorial argued this may reflect system and learning-curve effects rather than a definitive refutation of REBOA’s potential in high-experience settings.2
  • Selection and timing concerns: The trial’s pragmatic, judgement-based enrolment and the frequency of intercurrent events (37% improving before inflation; 17% failed access) suggest many enrolled patients either did not require balloon occlusion or could not receive it in time, complicating interpretation of a “device efficacy” question versus a “strategy” question.
  • Correspondence critique: Subsequent letters raised concerns about generalising the results beyond the participating centres and about the relationship between REBOA attempts and delays to definitive haemorrhage control; these critiques emphasised the need to interpret UK-REBOA as evidence about ED-based, multi-centre implementation rather than “best-case” REBOA performance.3
  • How later evidence integrates UK-REBOA: A 2025 EAST practice management guideline (systematic review and meta-analysis) incorporated evidence up to 2024 and issued conditional recommendations, underscoring ongoing uncertainty, heterogeneity, and the importance of patient selection, rapid haemorrhage control capability, and programme experience.4

Summary

  • First randomised trial of REBOA in trauma; conducted pragmatically across 16 UK major trauma centres.
  • Stopped early for harm after the second interim analysis; final n=90 randomised.
  • 90-day mortality was higher with a REBOA-inclusive strategy (54% vs 42%; OR 1.58; 95% CrI 0.72 to 3.52), with stronger harm signals at earlier timepoints (3-hour OR 4.25; 95% CrI 1.33 to 15.99).
  • Intervention delivery was incomplete (41% balloon inflation); median ED arrival-to-inflation 32 minutes (IQR 20–47), and time to definitive haemorrhage control was longer (median 83 vs 64 minutes).
  • Findings argue against broad ED-based REBOA implementation in similar systems without clear evidence of rapid delivery, robust pathways, and high procedural experience.

Overall Takeaway

UK-REBOA is the first randomised evaluation of a REBOA-inclusive emergency department strategy for exsanguinating torso haemorrhage and found no mortality benefit, with a high posterior probability of harm—particularly early after randomisation. The trial’s pragmatic implementation context, incomplete balloon delivery, and delays to definitive haemorrhage control suggest that in similar systems, attempting REBOA may impose time and pathway costs that outweigh any theoretical physiological advantage.

Overall Summary

  • In UK major trauma centres, a REBOA-inclusive ED strategy did not improve survival and may increase early haemorrhage-related mortality.
  • Only 41% of patients allocated to the REBOA strategy underwent balloon inflation; median time to inflation was 32 minutes, emphasising the primacy of rapid definitive haemorrhage control over complex adjuncts.

Bibliography