Skip to main content

Foundational Trials

Prague Cardiac Arrest Study

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Effect of Intra-arrest Transport, Extracorporeal Cardiopulmonary Resuscitation, and Immediate Invasive Assessment and Treatment on Functional Neurologic Outcome in Refractory Out-of-Hospital Cardiac Arrest: A Randomized Clinical Trial
  • Acronym: Prague OHCA Study
  • Year: 2022
  • Journal published in: JAMA
  • Citation: Belohlavek J, Smalcova J, Rob D, Franek O, Smid O, Pokorna M, et al; for the Prague OHCA Study Group. Effect of intra-arrest transport, extracorporeal cardiopulmonary resuscitation, and immediate invasive assessment and treatment on functional neurologic outcome in refractory out-of-hospital cardiac arrest: a randomized clinical trial. JAMA. 2022;327(8):737-747.

Context & Rationale

  • Background
    • Refractory out-of-hospital cardiac arrest (OOHCA) with ongoing advanced life support and no sustained ROSC is associated with extremely poor survival and poor neurological outcomes in conventional systems.
    • Extracorporeal cardiopulmonary resuscitation (ECPR) offers physiological “time-buying” by restoring systemic perfusion during arrest, enabling diagnosis and treatment of reversible causes (notably acute coronary occlusion), but is logistically complex and time-critical.
    • Prior evidence for ECPR in refractory OOHCA was dominated by observational cohorts and centre-level programmes with substantial confounding by indication and survivorship bias.
    • Randomised evidence was sparse and largely restricted to highly selected shockable-rhythm populations, with small sample sizes and early stopping. 2
    • The “hyperinvasive” concept integrates system-level elements: early identification, mechanical CPR to permit transport, intra-arrest transfer to a cardiac arrest centre, rapid cannulation for ECPR, and immediate invasive assessment (including coronary angiography/PCI) to address reversible pathology.
  • Research Question/Hypothesis
    • Whether a bundled hyperinvasive strategy (intra-arrest transport + ECPR + immediate invasive assessment/treatment) improves long-term functional neurological outcome compared with a conventional on-scene strategy with transport after sustained ROSC.
    • The trial was conceived and protocolised prior to widespread contemporary adoption of organised ECPR pathways, with published rationale and sequential monitoring assumptions. 1
  • Why This Matters
    • ECPR is resource-intensive and potentially exposes patients to major harms (bleeding, vascular injury, futile ICU care); robust trial evidence is essential before wide implementation.
    • Prague OHCA tested not a device, but a system-of-care intervention delivered under time pressure—directly relevant to EMS governance, regionalisation, and cardiac arrest centre design.
    • The primary endpoint (CPC at 180 days) targets what matters most: long-term survival with meaningful neurological recovery.

Design & Methods

  • Research Question: In adults with refractory OOHCA of presumed cardiac cause, does an invasive “hyperinvasive” strategy (intra-arrest transport with mechanical CPR, ECPR when indicated, and immediate invasive assessment/treatment) increase 180-day survival with favourable neurological outcome (CPC 1–2) compared with a standard on-scene strategy with transport after sustained ROSC?
  • Study Type: Randomised, single-centre, parallel-group, investigator-initiated clinical trial with prehospital enrolment/randomisation; open-label delivery; blinded neurological outcome assessment; sequential (group-sequential) monitoring with a DSMB; conducted in Prague, Czech Republic (prehospital EMS + cardiac arrest centre).
  • Population:
    • Setting: Prehospital physician-staffed EMS in Prague with direct access to a dedicated cardiac arrest centre (catheterisation laboratory + ICU).
    • Key inclusion (published criteria): Age 18–65 years; witnessed OOHCA of presumed cardiac cause; ongoing resuscitation with no sustained ROSC after ≥5 minutes of advanced life support; unconscious; ECPR team and ICU bed capacity available.
    • Key exclusion (published criteria): Suspected non-cardiac cause (e.g., trauma, drowning, intoxication); unwitnessed collapse; pregnancy; early sustained ROSC; severe comorbidity/advanced chronic organ failure; pre-arrest poor neurological status (CPC ≥3); known DNR/anticipated futility.
  • Intervention:
    • Prehospital (intra-arrest phase): Early initiation of mechanical CPR and immediate intra-arrest transport to the cardiac arrest centre, rather than prolonged on-scene resuscitation.
    • In-hospital: Immediate invasive assessment and treatment in the catheterisation laboratory (including coronary angiography and PCI when indicated) with consideration of ECPR (veno-arterial extracorporeal life support) for persistent arrest or refractory shock, plus post-resuscitation ICU care including targeted temperature management (TTM) as per contemporary practice.
  • Comparison:
    • Standard strategy: Conventional advanced life support delivered predominantly on scene, with transport to hospital after sustained ROSC, followed by post-resuscitation care (including TTM and invasive evaluation when clinically indicated after ROSC).
  • Blinding: Unblinded intervention delivery (logistically unavoidable); neurological outcomes were assessed by a neurologist blinded to treatment allocation.
  • Statistics: A total sample size of 285 patients was planned to detect a 15% absolute increase in survival with favourable neurological outcome (from 10% to 25%) with 90% power at a 5% (two-sided) significance level, using a prespecified sequential monitoring plan; primary analysis was by assigned group (intention-to-treat) using complete-case analysis.
  • Follow-Up Period: Primary endpoint at 180 days; key secondary endpoints at 30 days; last follow-up completed March 30, 2021.

Key Results

This trial was stopped early. The DSMB recommended stopping after 256 patients were analysed because a prespecified stopping boundary was met in the futility monitoring scenario.

Outcome Invasive strategy Standard strategy Effect p value / 95% CI Notes
Survival with favourable neurological outcome at 180 days (CPC 1–2) (primary) 39/124 (31.5%) 29/132 (22.0%) OR 1.63 95% CI 0.93 to 2.85; P=0.09 Absolute difference 9.5% (95% CI −1.3 to 20.1).
Neurological recovery within 30 days (CPC 1–2 at any time within 30 days) (secondary) 38/124 (30.6%) 24/132 (18.2%) OR 1.99 95% CI 1.11 to 3.57; P=0.02 Absolute difference 12.4% (95% CI 1.9 to 22.7).
Cardiac recovery within 30 days (no pharmacological and mechanical support ≥24 h) (secondary) 54/124 (43.5%) 45/132 (34.1%) OR 1.49 95% CI 0.91 to 2.47; P=0.12 Absolute difference 9.4% (95% CI −2.5 to 21.0).
Survival to 180 days (any CPC) (post hoc) 39/124 (31.5%) 31/132 (23.5%) Log-rank test P=0.014 Kaplan–Meier analysis; no hazard ratio reported.
Major bleeding (TIMI major) (harm) 36/116 (31%) 10/69 (15%) Not reported Not reported Denominators reflect patients with bleeding assessment available; TIMI major criteria included intracranial haemorrhage, fatal bleeding, or haemoglobin decrease ≥5 g/dL.
  • The invasive strategy produced a clinically important absolute increase in the primary endpoint (9.5%), but the estimate was imprecise and did not meet conventional statistical significance (OR 1.63; 95% CI 0.93 to 2.85; P=0.09).
  • Neurological recovery within 30 days was higher with the invasive strategy (30.6% vs 18.2%; OR 1.99; 95% CI 1.11 to 3.57; P=0.02), while cardiac recovery did not differ statistically (P=0.12).
  • The intervention achieved large “system separation” (e.g., hospital arrival 123/124 vs 87/132; ECLS implanted 82/124 vs 10/132), but major bleeding was more frequent among assessed patients (31% vs 15%).

Internal Validity

  • Randomisation and Allocation: Web-based 1:1 randomisation with concealed allocation; stratified into 4 strata by sex and age (≤45 vs >45 years) with a block size of 8 (block size not disclosed to investigators).
  • Dropout / exclusions: 264 patients randomised; 8 excluded after randomisation (7 consent not obtained from relatives; 1 randomised after DSMB stop), leaving 256 patients in the primary (complete-case) analysis.
  • Performance/Detection Bias: Open-label treatment delivery; neurological outcome (CPC) assessed by a neurologist blinded to allocation.
  • Protocol Adherence (delivered separation): ECLS implanted 82/124 (66%) vs 10/132 (8%); diagnostic angiography among those arriving to hospital 120/123 (98%) vs 67/87 (77%); TTM among those arriving to hospital 117/123 (95%) vs 61/87 (70%).
  • Baseline Characteristics: Median time from collapse to randomisation 24 (IQR 20–30) min vs 26 (IQR 22–31) min; bystander CPR 123/124 (99%) vs 130/132 (98%); initial ventricular fibrillation 72/124 (58%) vs 84/132 (64%).
  • Timing: Time from randomisation to hospital admission 19 (IQR 14–23) min vs 29 (IQR 21–38) min; time from collapse to hospital admission 49 (IQR 43–60) min vs 60 (IQR 52–75) min.
  • Dose (intensity of the bundle): Mechanical CPR device used 114/124 (92%) vs 104/132 (79%); duration of CPR on scene 13 (IQR 10–18) min vs 20 (IQR 14–26) min; total duration of resuscitation 58 (IQR 50–70) min vs 46 (IQR 34–64) min.
  • Separation of the Variable of Interest: Arrived to hospital 123/124 (99%) vs 87/132 (66%); prehospital death declared 1/124 (1%) vs 45/132 (34%); coronary angiography among those undergoing diagnostic angiography 115/120 (96%) vs 57/67 (85%).
  • Crossover: Accepted crossover occurred in 20 patients (11 standard→invasive; 9 invasive→standard); ECLS implantation occurred in 10 patients in the standard group.
  • Outcome Assessment: Primary endpoint (CPC 1–2 at 180 days) is clinically meaningful and was assessed with blinded neurological evaluation; follow-up completeness for the primary endpoint was reported as complete for analysed patients.
  • Statistical Rigor: Prespecified sequential monitoring and DSMB oversight; early stopping reduced achieved sample size below the planned 285 patients and widened confidence intervals for the primary endpoint.

Conclusion on Internal Validity: Overall, internal validity appears moderate: randomisation and allocation concealment were robust and outcome assessment was blinded, but open-label delivery with crossovers, post-randomisation exclusions, and early stopping for futility reduce precision and may dilute true effects.

External Validity

  • Population Representativeness: Highly selected refractory OOHCA population (witnessed, presumed cardiac cause, age 18–65, high bystander CPR rates) within a single metropolitan EMS system with physician-staffed response and a mature cardiac arrest centre pathway.
  • System Dependence: The intervention requires rapid transport logistics, immediate catheterisation laboratory access, and an experienced ECPR/cannulation team; feasibility and outcomes are likely sensitive to transport time, team availability, and centre volume.
  • Applicability: Most applicable to urban regions with regionalised cardiac arrest care and the ability to deliver cannulation and invasive diagnostics rapidly; generalisability is limited in rural systems, paramedic-only EMS models, and resource-limited settings where intra-arrest transport and ECPR cannot be delivered within similar timelines.

Conclusion on External Validity: Generalisability is limited-to-moderate: findings are most transferable to comparable organised ECPR-capable cardiac arrest centre systems, and less transferable where prehospital staffing, transport times, or ECPR capacity differ materially.

Strengths & Limitations

  • Strengths:
    • Randomised evaluation of a system-of-care intervention delivered during active resuscitation (high pragmatic relevance).
    • Clinically meaningful primary endpoint at 180 days with blinded neurological assessment.
    • Detailed reporting of process measures enabling appraisal of treatment separation (transport, ECLS, invasive diagnostics).
    • Prespecified sequential monitoring with DSMB oversight.
  • Limitations:
    • Single-centre design in a highly organised EMS/cardiac arrest centre network (system-specific generalisability).
    • Early stopping reduced achieved sample size and precision of the primary effect estimate.
    • Open-label delivery with crossovers and protocol deviations in a time-critical context.
    • Bundled intervention limits attribution of benefit/harm to any single component (transport vs ECPR vs invasive assessment vs co-interventions).

Interpretation & Why It Matters

  • Clinical signal, statistical uncertainty
    • The primary endpoint favoured the invasive strategy but did not reach statistical significance; the confidence interval includes both minimal benefit and clinically meaningful benefit, consistent with residual imprecision after early stopping.
    • The significant improvement in 30-day neurological recovery suggests early neurological salvage may occur even when longer-term outcomes remain uncertain, but multiplicity and sequential stopping necessitate cautious interpretation.
  • System-level mechanisms
    • The bundle shifted a substantial fraction of patients from prehospital termination to hospital-based evaluation and ECLS-supported resuscitation, materially changing the care pathway and potential access to definitive therapies.
    • Coronary angiography and PCI rates among those reaching the catheterisation laboratory were relatively high in both arms, emphasising that the principal “active ingredient” tested was enabling hospital-phase resuscitation (including ECPR) when ROSC could not be achieved on scene.
  • Harms and trade-offs
    • Higher major bleeding in the invasive strategy underscores the need for meticulous cannulation technique, anticoagulation discipline, and careful candidate selection within ECPR programmes.

Controversies & Subsequent Evidence

  • Early stopping vs effect direction: The trial stopped early for futility monitoring despite a point estimate favouring the invasive strategy for the primary endpoint; the resulting wide confidence interval leaves persistent uncertainty regarding clinically meaningful benefit.
  • Post-randomisation exclusions and consent dynamics: Exclusions after randomisation due to delayed consent processes are uncommon in arrest trials and raise interpretive challenges for strict intention-to-treat inference in emergency research contexts.
  • Protocol deviations and crossovers: Crossovers (including ECLS use in the standard arm) and protocol deviations (including intra-arrest transport in the standard arm without crossover approval, and age-estimation issues) may have reduced between-group separation and diluted effect estimates.
  • Bundled intervention attribution: Because the intervention combined transport strategy, ECPR, and immediate invasive assessment (with differential use of mechanical CPR and intranasal cooling), the trial cannot isolate which component(s) drove any observed benefit or harm.
  • Correspondence focus: The published correspondence interrogated enrolment fraction, baseline process differences, and interpretation of outcomes; the author reply clarified resuscitation timelines, selection logic, and crossover outcomes. 34
  • Subsequent RCT evidence: The multicentre INCEPTION trial did not demonstrate a statistically significant improvement in survival with favourable neurological outcome with early ECPR compared with conventional resuscitation in refractory OOHCA. 5
  • Pooled patient-level evidence: A pooled individual patient data analysis of ARREST and Prague OHCA concluded that an ECPR-facilitated strategy improved 180-day survival and neurological outcomes compared with conventional CPR, with suggested benefit emerging predominantly in the hospital phase. 6
  • Long-term follow-up: Long-term follow-up of Prague OHCA reported higher long-term survival in the invasive arm (34/123 [27.6%] vs 26/132 [19.7%]; log-rank P=0.01) with similar severe neurological disability rates among survivors (CPC 3: 1/34 vs 1/26). 7
  • Meta-analytic uncertainty: Updated meta-analysis with trial sequential methods highlights persistent imprecision and sensitivity to trial inclusion in estimating ECPR benefit across arrest settings. 8
  • Guideline position: Contemporary guidelines remain cautious and selective, supporting consideration of ECPR only within organised systems capable of rapid delivery and appropriate patient selection. 910

Summary

  • Prague OHCA tested a bundled hyperinvasive system strategy for refractory OOHCA, including intra-arrest transport, ECPR, and immediate invasive assessment/treatment.
  • The trial stopped early for futility at 256 analysed patients; the primary endpoint favoured the invasive strategy (31.5% vs 22.0%) but was not statistically significant (OR 1.63; P=0.09).
  • Neurological recovery within 30 days was significantly higher with the invasive strategy (30.6% vs 18.2%; OR 1.99; P=0.02), while 30-day cardiac recovery was not (P=0.12).
  • The intervention achieved large care-pathway separation (hospital arrival 99% vs 66%; ECLS implantation 66% vs 8%), but increased major bleeding among assessed patients (31% vs 15%).
  • Interpretation hinges on system feasibility, early stopping/precision limits, and the inability to isolate which bundle components drive benefit; subsequent trials and pooled analyses remain mixed.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • The primary endpoint used Cerebral Performance Category (CPC) at 180 days, assessed by a blinded neurologist; “neurological recovery within 30 days” was defined as CPC 1–2 at any time within 30 days.
  • Complication denominators (e.g., bleeding) vary because outcomes were reported among patients with relevant assessments available; values are reproduced as published without recalculation.

Overall Takeaway

Prague OHCA was a landmark attempt to randomise a time-critical, system-level ECPR-enabled cardiac arrest centre pathway in refractory OOHCA, using long-term functional outcome as the primary endpoint. Although stopped early and statistically inconclusive for the primary outcome, it demonstrated substantial pathway separation with a consistent direction-of-effect signal and important bleeding trade-offs, helping define the modern evidence base for selective, programme-based ECPR implementation.

Overall Summary

  • Bundled hyperinvasive strategy for refractory OOHCA improved 30-day neurological recovery but did not achieve a statistically significant improvement in 180-day favourable neurological survival; early stopping and system-specific delivery drive ongoing debate.

Bibliography