Publication

• Title: Early extracorporeal CPR for refractory out-of-hospital cardiac arrest
• Acronym: INCEPTION (Early Initiation of Extracorporeal Life Support in Refractory Out-of-Hospital Cardiac Arrest)
• Year: 2023
• Journal published in: New England Journal of Medicine
• Citation: Suverein MM, Delnoij TSR, Lorusso R, et al. Early extracorporeal CPR for refractory out-of-hospital cardiac arrest. N Engl J Med. 2023;388(4):299-309.

Context & Rationale

• Background

  • Refractory out-of-hospital cardiac arrest (OHCA) with an initial shockable rhythm carries very high mortality once conventional advanced life support (ALS) is prolonged.
  • Extracorporeal cardiopulmonary resuscitation (ECPR; veno-arterial extracorporeal life support initiated during ongoing CPR) is intended to provide immediate circulatory support to allow time for definitive therapy and post-arrest care pathways.
  • Pre-trial evidence was largely observational and highly heterogeneous in patient selection, timing, and system design, with major risk of confounding by indication.
  • Early randomised evidence existed but was limited in size and setting, leaving uncertainty about effectiveness in broader “real-world” regional systems of care.
• Research Question/Hypothesis

  • In adults with witnessed refractory OHCA and an initial ventricular arrhythmia, does an early ECPR strategy improve 30-day survival with favourable neurological outcome compared with continued conventional CPR?
• Why This Matters

  • ECPR is resource-intensive, high-risk, and system-dependent; robust trial evidence is required before widespread implementation or regional commissioning decisions.
  • INCEPTION tested ECPR as a strategy within an established national EMS-to-hospital network, aiming to inform both patient-level benefit and system-level feasibility.
  • Neurologically intact survival (rather than ROSC alone) is the clinically meaningful endpoint for evaluating such high-intensity rescue strategies.

Design & Methods

• Research Question: Among adults with refractory witnessed OHCA and an initial shockable rhythm, does an early ECPR strategy (vs conventional CPR) increase 30-day survival with favourable neurological outcome (CPC 1–2)?
• Study Type: Multicentre, pragmatic, randomised, open-label trial in the Netherlands; prehospital randomisation during intra-arrest transport; 1:1 allocation with variable permuted block sizes (2, 4, 6) stratified by ECPR centre.
• Population:
  • Setting: OHCA managed by EMS with intra-arrest transport to participating ECPR-capable hospitals.
  • Key inclusion criteria (simplified trial criteria): age 18–70 years; witnessed OHCA; initial rhythm VF/VT (or AED shock delivered); bystander basic life support; no sustained ROSC within 15 minutes after start of conventional CPR.
  • Key exclusion criteria (simplified trial criteria): ROSC within 15 minutes with sustained haemodynamic recovery; terminal heart failure (NYHA III–IV); severe pulmonary disease (COPD GOLD III–IV); disseminated malignancy; suspected/obvious pregnancy; bilateral femoral bypass surgery; known contraindication to ECPR; known pre-arrest CPC 3–4; known advanced directive; multi-trauma (ISS >15); expected time-to-start cannulation >60 minutes from start of cardiac arrest.
  • Operational enrolment feature: randomisation could occur before full eligibility confirmation to allow timely ECPR team mobilisation; final eligibility was confirmed on hospital arrival.
• Intervention:
  • ECPR strategy: activation of an ECPR team during ongoing CPR and intra-arrest transport; cannulation for veno-arterial extracorporeal life support on hospital arrival if arrest persisted.
  • Location of ECPR delivery: predominantly in the emergency department (when performed).
  • Downstream care: post-cannulation management including coronary angiography and percutaneous coronary intervention as clinically indicated, alongside standard post–cardiac arrest care pathways.
• Comparison:
  • Conventional CPR strategy: continued guideline-based ALS without planned initiation of extracorporeal support; hospital-based resuscitation and downstream care at clinician discretion when ROSC occurred.
  • Crossovers/rescue: extracorporeal support was not intended, but occurred in a small minority.
• Blinding: Treatment allocation was not blinded; EMS teams were kept unaware of group allocation to support comparable prehospital management; the primary outcome was assessed by an independent neurologist unaware of trial-group assignment.
• Statistics: A total of 110 patients were initially planned to detect an absolute increase in 30-day survival with favourable neurological outcome from 8% to 30% with 80% power at a two-sided 5% significance level; the trial used an adaptive design with a preplanned interim analysis and sample-size recalculation, resulting in a revised target of 134 patients; the primary analysis used a prespecified regression model producing odds ratios, with analysis in the trial “intention-to-treat” population as defined by post-arrival eligibility confirmation.
• Follow-Up Period: Neurological outcome assessed at 30 days (primary), and at 3 months and 6 months (secondary).

Key Results

This trial was not stopped early. Recruitment was paused temporarily during COVID-19, and the trial completed enrolment to the revised target after adaptive sample-size recalculation.

Outcome	Extracorporeal CPR strategy	Conventional CPR strategy	Effect	p value / 95% CI	Notes
Survival with favourable neurological outcome at 30 days (CPC 1–2)	14/70 (20%)	10/62 (16%)	OR 1.4	95% CI 0.5 to 3.5; P=0.52	Model-based risk ratio also reported: 1.05; 95% CI 0.97 to 1.13
Survival with favourable neurological outcome at 3 months (CPC 1–2)	13/70 (19%)	9/62 (15%)	OR 1.5	95% CI 0.6 to 3.8; P=Not reported	Secondary endpoint
Survival with favourable neurological outcome at 6 months (CPC 1–2)	13/70 (19%)	10/62 (16%)	OR 1.3	95% CI 0.5 to 3.3; P=Not reported	Secondary endpoint
Survived to hospital discharge	14/70 (20%)	13/64 (20%)	OR 1.0	95% CI 0.4 to 2.4; P=Not reported	In-hospital outcome
Admitted to ICU	57/70 (81%)	23/64 (36%)	OR 0.1	95% CI 0.1 to 0.3; P=Not reported	Marked pathway separation in early post-arrest care access
Major bleeding (serious adverse event)	11/70 (16%)	2/64 (3%)	Not reported	Not reported	Serious adverse events table
Lower limb ischaemia (serious adverse event)	4/70 (6%)	0/64 (0%)	Not reported	Not reported	Vascular complication consistent with femoral cannulation risk

• Primary endpoint: 30-day CPC 1–2 occurred in 14/70 (20%) with the ECPR strategy vs 10/62 (16%) with conventional CPR; OR 1.4; 95% CI 0.5 to 3.5; P=0.52.
• Favourable neurological survival at 3 and 6 months remained numerically higher with ECPR (19% vs 15% at 3 months; 19% vs 16% at 6 months), with wide confidence intervals.
• Protocol separation was substantial in delivery (ECPR initiated in 52/70 [74%] vs 3/64 [5%]), but this did not translate into a demonstrable improvement in favourable neurological survival.

Internal Validity

• Randomisation and Allocation:
  • 1:1 randomisation with variable permuted blocks (2, 4, 6) and stratification by ECPR centre.
  • Allocation via a smartphone-based randomisation application, enabling rapid prehospital assignment while teams mobilised in parallel.
  • EMS teams were kept unaware of allocation to minimise differential prehospital care.
• Dropout or Exclusions:
  • 160 patients were randomised, with 26 excluded after randomisation because eligibility criteria were not met on hospital arrival (reflecting prehospital uncertainty and time-critical enrolment).
  • Primary endpoint assessment was available for 70/70 in the ECPR group and 62/64 in the conventional group (1 not assessed by neurologist; 1 lost/withdrew before primary-outcome analysis).
• Performance/Detection Bias:
  • Open-label delivery was unavoidable once ECPR teams were mobilised and cannulation commenced.
  • Primary outcome (CPC at 30 days) was assessed by an independent neurologist unaware of treatment assignment, reducing risk of biased neurological classification.
• Protocol Adherence:
  • ECPR initiated: 52/70 (74%) in the ECPR group vs 3/64 (5%) in the conventional group.
  • Reasons for non-initiation in the ECPR group included stable ROSC (13/70 [19%]) and logistic failure (3/70 [4%]).
  • Crossovers occurred in 3/64 (5%) assigned to conventional CPR who received extracorporeal CPR.
• Baseline Characteristics:
  • Age: 54±12 years (ECPR) vs 57±10 years (conventional).
  • Male sex: 63/70 (90%) vs 57/64 (89%).
  • Witnessed arrest: 68/70 (97%) vs 63/64 (98%); CPR started ≤5 minutes: 69/70 (99%) vs 61/64 (95%).
  • Likely cause acute myocardial infarction: 51/70 (73%) vs 52/64 (81%).
• Heterogeneity:
  • Multicentre delivery across ECPR-capable centres increases practice variability but improves ecological validity for a national system-level intervention.
  • Eligibility was restricted to a highly selected subgroup (witnessed shockable OHCA with bystander CPR), limiting within-trial clinical heterogeneity.
• Timing:
  • Start of arrest to start of EMS transport: 21±9 minutes (ECPR) vs 25±9 minutes (conventional); treatment effect -4.1 minutes; 95% CI -7.2 to -0.9.
  • Start of arrest to arrival at ED: 36±12 minutes vs 38±11 minutes; treatment effect -2.1 minutes; 95% CI -6.0 to 1.7.
  • Among those cannulated, start of arrest to start of cannulation: 58±13 minutes (n=51).
  • Among those with recorded flow initiation, start of arrest to start of ECLS flow: median 74 minutes (IQR 63–87; n=44).
• Dose:
  • Among those decannulated (n=45), median interval from arrest to decannulation was 26 hours (IQR 9–53).
  • ICU stay (median): 1 day (IQR 1–4) in the ECPR strategy group vs 4 days (IQR 1–9) in the conventional strategy group (reflecting early death/cessation patterns and pathway differences).
• Separation of the Variable of Interest:
  • ECPR delivery separation: 52/70 (74%) received extracorporeal CPR in the ECPR group vs 3/64 (5%) in the conventional group.
  • Access to critical care: ICU admission 57/70 (81%) vs 23/64 (36%).
  • Coronary intervention: PCI 34/70 (49%) vs 14/64 (22%).
• Key Delivery Aspects:
  • Prehospital transport was initiated relatively early (mean ~21–25 minutes from arrest to transport), consistent with an intra-arrest transport strategy.
  • Median time to ECLS flow among those receiving it was 74 minutes (IQR 63–87), defining the effective exposure intensity achieved by this system.
• Crossover:
  • 3/64 (5%) assigned to conventional CPR received extracorporeal CPR.
  • 18/70 (26%) assigned to ECPR did not receive extracorporeal CPR, most commonly due to stable ROSC before planned initiation.
• Outcome Assessment:
  • Primary endpoint used CPC categorisation at 30 days and was assessed by a neurologist unaware of allocation.
  • Follow-up extended to 6 months for neurological outcomes (CPC).
• Statistical Rigor:
  • Adaptive sample-size recalculation was incorporated, and the revised target sample was achieved.
  • Effect estimates are imprecise for the primary endpoint (wide CI), consistent with limited event counts.

Conclusion on Internal Validity: Overall, internal validity is moderate: randomisation methods and blinded neurological outcome assessment support unbiased comparison, but post-randomisation exclusions and incomplete delivery of ECPR (with some crossover) plausibly dilute treatment effects and complicate strict intention-to-treat inference.

External Validity

• Population Representativeness:
  • Participants were a highly selected subset of OHCA: age 18–70, witnessed, shockable rhythm, bystander CPR, and refractory after 15 minutes.
  • The cohort reflects typical ECPR “candidate” selection rather than general OHCA populations (older patients, unwitnessed arrests, and non-shockable rhythms were largely excluded).
• Applicability:
  • Findings are most applicable to mature EMS systems capable of intra-arrest transport with high-quality CPR and to hospitals with established ECPR capability and cath-lab access.
  • Time-to-flow achieved (median 74 minutes among treated) may be difficult to replicate in lower-resource or geographically dispersed systems, limiting generalisability.
  • System-level logistics (prehospital triage, transport times, and ECPR team mobilisation) are integral to the intervention and may differ substantially across regions.

Conclusion on External Validity: External validity is limited to selected systems and patients: the results generalise best to well-resourced regional networks targeting witnessed shockable refractory OHCA with rapid transport and established ECPR pathways, and do not address broader OHCA case-mix.

Strengths & Limitations

• Strengths:
  • Pragmatic multicentre randomised evaluation of an ECPR strategy embedded within a national EMS–hospital network.
  • Clinically meaningful primary endpoint (30-day CPC 1–2) with blinded neurological assessment.
  • Prehospital randomisation and attempt to keep EMS teams unaware of allocation to promote comparable early resuscitation.
  • Granular reporting of timing metrics (transport, cannulation, time-to-flow) enabling mechanistic interpretation.
• Limitations:
  • Open-label design for in-hospital treatment delivery, with potential for performance differences in downstream care.
  • Modified “intention-to-treat” analysis population: 26/160 randomised patients excluded after hospital arrival eligibility confirmation.
  • Incomplete exposure to the intended intervention: 18/70 (26%) assigned to ECPR did not receive extracorporeal CPR; 3/64 (5%) assigned to conventional CPR received extracorporeal CPR.
  • Confidence intervals for the primary endpoint are wide, consistent with limited precision for modest treatment effects.
  • Applicability restricted to a narrow, high-functioning subgroup and to systems capable of implementing intra-arrest transport and ECPR.

Interpretation & Why It Matters

• Clinical effect estimate

  • INCEPTION did not demonstrate a statistically significant improvement in 30-day favourable neurological survival with an early ECPR strategy (20% vs 16%; OR 1.4; 95% CI 0.5 to 3.5; P=0.52).
  • The confidence interval spans clinically important benefit and no benefit, emphasising uncertainty rather than proof of equivalence.
• System-level implication

  • ECPR should be considered a complex system intervention: feasibility and effectiveness depend on candidate selection, transport strategy, and time-to-flow achieved.
  • Even with major pathway separation (ECPR initiation 74% vs 5%; ICU admission 81% vs 36%), favourable neurological survival was not clearly improved in this setting.
• Practical takeaway for clinicians

  • INCEPTION supports a cautious interpretation of ECPR as a default rescue strategy for refractory shockable OHCA in regional practice, and underscores the need to evaluate local time-to-flow capability and downstream care pathways before adoption.

Controversies & Subsequent Evidence

• Eligibility confirmation after randomisation (modified ITT design):
  • Randomisation prior to full eligibility confirmation enabled rapid mobilisation of ECPR teams but resulted in exclusions after randomisation, raising interpretability concerns for strict intention-to-treat inference.¹
• Intervention dilution and timing-to-flow:
  • A substantial minority assigned to ECPR did not receive extracorporeal support (26%), most commonly because ROSC occurred before initiation; this feature is intrinsic to prehospital randomisation in a time-critical pathway.¹
  • Commentary highlighted that ECPR effectiveness is likely sensitive to system-achieved time-to-flow and to the balance between early transport and ongoing on-scene optimisation of conventional ALS.²
• Post–cardiac arrest care, neuroprognostication, and withdrawal decisions:
  • Correspondence raised concern that non-standardised neuroprognostication and treatment withdrawal practices can materially influence neurologically intact survival in post-arrest trials, particularly when ICU courses are short and death often follows treatment discontinuation.¹
• Outcomes beyond patient survival (organ donation):
  • Correspondence noted that ECPR strategies may change the opportunity for organ donation (e.g., increased ICU admission with severe neurological injury), which is not captured by CPC-only endpoints and may be relevant to societal evaluation of ECPR programmes.¹
• Subsequent evidence:
  • Subsequent trials and syntheses continue to report heterogeneous effects across systems, reinforcing that ECPR is not a uniform “drug-like” intervention but a time- and system-dependent programme.
  • Recent guideline updates have generally retained conditional or selective recommendations for ECPR in carefully chosen refractory cardiac arrest patients within experienced, protocolised systems (see Further Reading).

Summary

• INCEPTION tested an early ECPR strategy vs conventional CPR in adults aged 18–70 with witnessed refractory shockable OHCA receiving bystander CPR.
• Primary outcome (30-day CPC 1–2): 14/70 (20%) with ECPR vs 10/62 (16%) with conventional CPR; OR 1.4; 95% CI 0.5 to 3.5; P=0.52.
• ECPR delivery separation was large (ECPR initiated 74% vs 5%; ICU admission 81% vs 36%), yet neurologically intact survival was not clearly improved.
• Timing metrics define the achieved exposure: among treated patients, median arrest-to-flow was 74 minutes (IQR 63–87).
• Serious harms consistent with ECPR were observed, including major bleeding (16% vs 3%) and lower limb ischaemia (6% vs 0%).

Overall Takeaway

INCEPTION is a landmark multicentre pragmatic RCT because it tested ECPR as a real-world system strategy (not merely a device) in a mature national network, using neurologically intact survival as the endpoint. In this setting, despite substantial pathway separation and a median arrest-to-flow of 74 minutes among treated patients, an early ECPR strategy did not significantly improve 30-day favourable neurological survival, reinforcing that net benefit remains uncertain and highly dependent on system performance and patient selection.

Bibliography

• 1.Lamhaut L, Hutin A, Jouffroy R, et al. Extracorporeal CPR for out-of-hospital cardiac arrest. N Engl J Med. 2023;388(20):1913-1917. Link
• 2.Keaney JF Jr, Münzel T. Extracorporeal CPR in out-of-hospital cardiac arrest—still on life support? N Engl J Med. 2023;388(4):370-371. Link