Extracorporeal cardiopulmonary resuscitation for refractory OHCA: lessons from three randomized controlled trials—the trialists’ view

Abstract

Extracorporeal cardiopulmonary resuscitation is a promising treatment for refractory out-of-hospital cardiac arrest. Three recent randomized trials (ARREST trial, Prague OHCA study, and INCEPTION trial) that addressed the clinical benefit of extracorporeal cardiopulmonary resuscitation in out-of-hospital cardiac arrest yielded seemingly diverging results. The evidence for extracorporeal cardiopulmonary resuscitation in out-of-hospital cardiac arrest, derived from three recent randomized controlled trials, is not contradictory but rather complementary. Excellent results can be achieved with a very high level of dedication, provided that strict selection criteria are applied. However, pragmatic implementation of extracorporeal cardiopulmonary resuscitation does not necessarily lead to improved outcome of refractory out-of-hospital cardiac arrest. Centres that are performing extracorporeal cardiopulmonary resuscitation for out-of-hospital cardiac arrest or aspire to do so should critically evaluate whether they are able to meet the pre-requisites that are needed to conduct an effective extracorporeal cardiopulmonary resuscitation programme.

Introduction

The application of veno-arterial extracorporeal membrane oxygenation (V-A ECMO) in refractory cardiac arrest, referred to as extracorporeal cardiopulmonary resuscitation (ECPR), has been regarded a promising therapy to restore circulation.^1–4 Several randomized controlled trials (RCTs) evaluating the therapeutic effect of ECPR in refractory OHCA have been initiated during the past years.^5–9 Three of these trials using clinically relevant outcomes as a primary outcome have been published recently. Chronologically, these trials comprise the Advanced Reperfusion Strategies for Patients with OHCA and Refractory Ventricular Fibrillation Trial (ARREST),⁶ the Prague OHCA study,¹⁰ and the Early Initiation of Extracorporeal Life Support in Refractory OHCA Trial (INCEPTION).¹¹

ARREST trial

The ARREST trial was a single-centre trial performed by the Cardiovascular Division of the University of Minnesota (Minneapolis, MN, USA), randomizing patients with pulseless ventricular tachycardia or ventricular fibrillation (VT/VF) who presented at the Emergency Department without return of spontaneous circulation (ROSC) after at least three defibrillation attempts, to either ECPR or conventional cardiopulmonary resuscitation (CCPR, Table 1).¹² At the first interim analysis, after including 30 patients, the study was prematurely terminated for proven benefit (Table 2).⁶

Table 1.

Study characteristics

	ARREST6,12	Prague OHCA5,10	INCEPTION11,13
Setting	Single-centre University of Minnesota, USA	Single-centre Charles University, Prague, Czech Republic	Multi-centre the Netherlands
Studied intervention	ECPR and early revascularization	Invasive bundle consisting of intra-arrest transport with mechanical CPR, ECPR, and immediate invasive assessment (coronary angiography)	ECPR
Comparator	CCPR and early revascularization	CCPR with encouraged immediate invasive assessment	CCPR
Number of participating centres	1	1	10
Inclusion Period	August 2019 to June 2020	March 2013 to October 2020	May 2017 to February 2021
Inclusion Criteria
Age	18–75 years	18–65 years	18–70 years
Only witnessed OHCA	No	Yes	Yes
Rhythm	VF or pulseless VT	Presumed cardiac aetiology, all rhythms	VT/VF or AED-shock admitted
ROSC	No ROSC after 3 shocks	No ROSC after 5 min of ALS	No ROSC within 15 min of ALS
Other	Body habitus able to support mechanical CPR, estimated transport time <30 min	ECPR team available at cardiac centre	Only bystander ALS
Exclusion criteria
Mechanisms	Drowning, blunt or penetrating trauma, burns, overdose	Presumed non cardiac cause	Non-shockable, multi-trauma
Co-morbidities	Terminal cancer	Obvious life-limiting co-morbidities, known CPC > 3, suspected/confirmed stroke	Oncological disease, bilateral femoral vessel surgery, CPC score >3, NYHA III/IV, COPD GIII/IV
Bleeding	Active GI or internal bleeding	Bleeding diathesis	—
Other	Prisoner or nursing home resident Cath lab unavailable	Suspected or confirmed pregnancy Conscious patient	Expected initial cannulation >60 min after arrest, ROSC with sustained haemodynamic recovery within 15 min
ECMO cannulation
Location	Catheterization lab	Catheterization lab	Emergency department
Randomization
Timing	Upon hospital arrival	On scene in collaboration with trial coordinator	During transport by hospital staff
Blinding	EMS blinded	None	EMS blinded
Crossover	Allowed both directions	Allowed both directions	Allowed both directions
Consent	As soon as feasible following randomization (patient or family)	Legal representative or patient after regaining neurological function	Deferred consent14
Primary outcome	Survival to hospital discharge	180-day survival with favourable neurological status (CPC1-2)	30-day survival with favourable neurological status (CPC1-2)

AED, automated external defibrillator; ALS, advanced life support; CCPR, conventional cardiopulmonary resuscitation; COPD, chronic obstructive pulmonary disease; CPR, cardiopulmonary resuscitation; CPC, cerebral performance category; EMS, emergency medical services; ECPR, extracorporeal cardiopulmonary resuscitation; ICU, intensive care unit; MRS, modified Rankin scale; NA, not applicable; NYHA, New York Heart Association class of dyspnoea; QoL, quality of life; RCT, randomized controlled trial; ROSC: return of spontaneous circulation; SF-36. 36-item short-form survey; TTM, targeted temperature management; VF, ventricular fibrillation; VT, ventricular tachycardia.

Table 2.

Outcomes

Outcomes	CCPR/standard treatment	ECPR/invasive strategy	P-value
ARREST	n = 15	n = 14
Survival to hospital discharge	1 (7%)	6 (43%)	0.023
3-month survival	0	6 (43%)	0.006
6-month survival	0	6 (43%)	0.006
Prague OHCA	n = 132	n = 124
30-day survival (CPC1-2)	24 (18%)	38 (31%)	0.02
6-month survival (CPC1-2)	29 (22%)	39 (32%)	0.09
INCEPTION	n = 64	n = 70
30-day survival (CPC1-2)	10 (16%)	14 (20%)	0.518*
3-month survival (CPC1-2)	9 (14%)	12 (17%)	0.600*
6-month survival (CPC1-2)	10 (16%)	14 (20%)	0.537*

CCPR, conventional cardiopulmonary resuscitation; ECPR, extracorporeal cardiopulmonary resuscitation; NA, not applicable; NR, not reported.

62 patients with available data in CCPR group for 30-day survival; 63 patients with available data in CCPR group for 3- and 6-month survival; and 68 patients with available data in ECPR group for 3-month survival.

Prague OHCA study

The Prague OHCA study was a single-centre RCT that randomized 256 patients with an OHCA of presumed cardiac origin between an invasive strategy and standard care.^5,10 Patients became eligible if there was no ROSC within 5 min of advanced life support (ALS). The invasive strategy consisted amongst other things of a bundle of early intra-arrest transport using mechanical chest compression devices and application of ECPR in patients arriving at the hospital without ROSC. Standard care consisted of continued ALS on scene (Table 1). The trial was stopped prematurely after inclusion of 256 patients, when a pre-specified between-group difference of 15% could no longer be achieved.⁵ At that time, 6-month survival with cerebral performance category (CPC) (1–2) was 31% in the invasive strategy group vs. 22% in the control group (P = 0.09, Table 2).¹⁰ Of note, the survival rate in the control group was twice as high as was initially expected from prior experiences and literature.

INCEPTION trial

The INCEPTION trial was a multi-centre RCT performed in the Netherlands in 10 hospitals. It randomized patients with a primarily shockable rhythm (VT/VF), where ROSC was not achieved within 15 min of ALS, to either ECPR or CCPR. All patients underwent intra-arrest transport (Table 1).¹³

The study was completed as planned after inclusion of 134 patients. The primary study outcome, 30-day survival in CPC1-2, was 20% in the ECPR group vs. 16% in the control group (P = 0.518, Table 2).¹¹

Aim and outline of the review

These seemingly diverging results leave the field with persistent uncertainty about the efficacy and effectiveness of ECPR for OHCA. Often, such discrepancies urge the performance of a meta-analysis in order to summarize and reconcile existing evidence. However, a critical appraisal and comparison of individual studies may reveal important explanations for apparent discrepancies that may be lost by simply pooling outcomes in a meta-analysis.

In this review, we present a comparative analysis of design and execution of the three RCTs, discussing the impact of these features on these diverging results. We will attempt to integrate the unique additional knowledge yielded by each of these pivotal and complementary studies in order to definitively establish the position of ECPR in refractory OHCA.

Eligibility and randomization

Age, presenting rhythm, and duration of low-flow time are amongst the most important factors that affect the outcome of ECPR.^15–18 The maximum allowed age differed slightly between the trials (maximum age of 75, 65, and 70 years for the ARREST trial, Prague OHCA study, and the INCEPTION trial, respectively).^5,12,13 However, the mean ages were comparable (55–59 years, Table 3) and are not likely to have affected diverging trial outcomes.

Table 3.

Patient and cardiac arrest characteristics

	ARREST6,12 (n = 30)	Prague OHCA5,10 (n = 256)a	INCEPTION11,13 (n = 134)
ECPR/invasive group	15	124	70
CCPR group	15	132	64
Age (mean, SD)	59 (10)	57 (14)	55 (11)
Male sex (%)	25 (83%)	212 (83%)	120 (90%)
Hypertension (%)	7 (23%)	89 (35%)	39 (29%)
Diabetes mellitus (%)	6 (20%)	36 (14%)	16 (12%)
Previous stroke (%)	1 (3%)	NR	12 (9%)
Hyperlipidaemia	3 (10%)	NR	25 (19%)
Obesity (%)	1 (3%)	NR	NR
Known CAD (%)	6 (20%)	34 (13%)	13 (10%)
Previous PCI	2 (7%)	NR	10 (7%)
Previous CABG	3 (10%)	NR	6 (4%)
Chronic CHF (%)	1 (3%)	16 (6%)	6 (4%)
Primary rhythm
VT/VF	30 (100%)	156 (61%)	132 (99%)c
Asystole	0	55 (22%)	NA
PEA	0	45 (17%)	NA
Bystander CPR (%)	25 (83%)	252 (98%)	130 (97%)b
Defibrillation attempts (pre-hospital, mean, SD)	6 (3)	4 (3)	9 (6)
Mechanical CPR (%)	30 (100%)	218 (85%)	120 (90%)
Lactate on admission (mmol/L, mean, SD)	11.1 (3.8)	11.5 (4.9)	13.5 (4.6)
pH on admission (mean, SD)	6.95 (0.11)	6.99 (0.23)	6.92 (0.17)
Underwent CAG (%)	15 (50%)	181 (71%)	73 (54%)
CCPR/standard	2 (13%)	66 (50%)	22 (34%)
ECPR/invasive	13 (87%)	115 (93%)	51 (73%)
P-value for trial	0.017	0.002	0.014
PCI	NR	80 (31%)	48 (36%)
Attempted/performed		30 (23%)	14 (22%)
CCPR/standard		62 (50%)	34 (49%)
ECPR/invasive P-value for trial		0.002	0.026
ECMO initiated (%)
Within ECPR/invasive group	12 (80%)	82 (66%)	52 (74%)
Within CCPR group	0	10 (8%)	3 (5%)

AED, automatic external defibrillator; CAD, coronary artery disease; CAG, coronary angiography; CCPR, conventional cardiopulmonary resuscitation; CHF, congestive heart failure; ECMO, extracorporeal membrane oxygenation; ECPR, extracorporeal cardiopulmonary resuscitation; EMS, emergency medical services; IQR, interquartile range; PCI, percutaneous coronary intervention; NA, not applicable; NR, not reported; SD, standard deviation.

For generalizability issues and reporting of overall study numbers, median and IQR were converted to median and SD using Wan’s method.¹⁹

In other cases, the arrest was witnessed by the EMS.

In the remaining cases, shock administered by AED, but first observed rhythm by EMS was non-shockable.

Presenting rhythm

In contrast to the ARREST and INCEPTION trials, the Prague OHCA study allowed enrolment of patients with ‘non-shockable’ rhythms (Table 1). This led to the inclusion of patients presenting with asystole (22%) and pulseless electric activity (17%) (Table 3) and 40% of the enrolled patients. First, the inclusion of the patients decreased the overall survival throughout the study,²⁰ as is emphasized in a recent post hoc analysis of the Prague OHCA study, demonstrating 40% CPC1-2 survival in patients presenting with shockable rhythms.²¹ Second, it underlined non-shockable rhythms to be a controversial indication for ECPR, reflected by the mere 5% survival in these patients.²¹ Moreover, a recently published individual patient, pooled analysis of the ARREST and PRAGUE-OHCA trials showed that non-shockable rhythms, in the current way of ECPR delivery, may not benefit from ECPR.²²

Low-flow time

All trials took measures in their inclusion criteria to ameliorate or prevent excessive low-flow times, as this has been associated with worse outcome.²³ The INCEPTION trial used an estimated cut-off time of 60 min between arrest and ECPR-related cannulation.¹³ Of note, exceeding this time did not lead to post-randomization exclusion. The ARREST trial mandated an estimated transfer time to the emergency department of <30 min,¹² whilst in Prague OHCA, emergency medical services (EMS) teams were forced not to exceed time between collapse and cardiac catheterization lab (CCL) arrival over 60 min.⁵ Nonetheless, the mean total low-flow time in the INCEPTION trial was 74 min (Table 4),¹¹ which as such was considerably longer than pursued and longer than low-flow times in the ARREST trial and Prague OHCA study (59 and 61 min, respectively). This will be further discussed below.

Table 4.

Timing of procedures and cannulation

	ARREST6,12 (n = 30)	Prague OHCA5,10 (n = 256)a	INCEPTION11,13 (n = 134)
ECPR/invasive group	15	124	70
CCPR group	15	132	64
Time from 911 to EMS arrival (minutes, mean, SD)	6.5 (2.4)	8.9 (3.0)	8 (4)
Time from 911 to randomization (minutes, mean, SD)	50.2 (26.1)	25.2 (8.0)	33.0 (11.0)
Time from collapse to hospital arrival (minutes, mean, SD)	NR	55.5 (13.9)	37.0 (11.5)
Time from 911 to V-A ECMO initiation (minutes, mean, SD)b	59 (28)	62.0 (11.3)	74.7 (18.2)
Time from randomization to V-A ECMO initiation (minutes, mean, SD)b	12 (6)	NR	NR

CCPR, conventional cardiopulmonary resuscitation; ECPR, extracorporeal cardiopulmonary resuscitation; EMS, emergency medical services; IQR, interquartile range; SD, standard deviation; V-A ECMO, veno-arterial extracorporeal membrane oxygenation.

For generalizability issues and reporting of overall study numbers, median and IQR were converted to median and SD using Wan’s method.²⁴

For ECPR, randomized patients only.

Mechanical CPR

Mechanical CPR was a keystone in the invasive strategy of the Prague OHCA study.⁵ Despite not being an inclusion criterion, 90% of the patients in the INCEPTION trial received mechanical chest compressions during intra-arrest transport (Table 3).¹¹ In the ARREST trial, the application of an automated CPR device was mandatory, leading to the additional inclusion criterium of ‘body morphology able to accommodate a Lund University Cardiac Arrest System 2 (LUCAS 2)’.¹²

Randomization

The Prague OHCA study and the INCEPTION trial mandated pre-hospital randomization. In the Prague OHCA study, patients became eligible after 5 min of unsuccessful ALS and in the INCEPTION trial after 15 min. Contrarily, the ARREST trial applied in-hospital randomization. These different strategies are reflected by differences in the mean time between cardiac arrest and randomization, which was substantially shorter in the Prague OHCA study and the INCEPTION trial (Table 4).^10,11 The differences in timing between arrest and randomization led to relevant differences amongst the study populations. Since both the chance of ROSC and the chance of survival decline with prolonged CPR,²⁵ early randomization inevitably increased the proportion of patients achieving ROSC in both study arms, as well as the proportion of patients surviving without ECPR in the intervention arms (Table 3), thereby obscuring the true potential survival benefit of ECPR.

Setting and organization

Both the ARREST trial and the Prague OHCA study were single-centre RCTs performed in expert institutions that have been historical front runners of ECPR, and ECPR was performed by small, highly dedicated and trained expert teams with high-volume operators.^26,27 Contrarily, the INCEPTION trial was performed in 10 centres covering most of the Netherlands, which is a highly developed country with a very well, densely organized healthcare structure. In the INCEPTION trial, sites were allowed to apply ECPR logistics around local protocols for CPR and extracorporeal life support, but without specific protocols for ECPR, leading to considerable practice variation.¹¹

Successful ECPR for OHCA relies on intricate and swift decisions throughout the entire pre-hospital and intra-hospital chain of resuscitation care, aiming at the minimization of low-flow time.^28–30 As such, the time between hospital arrival and the actual start of cannulation is an accurate indicator of efficiency of in-hospital ECPR logistics.²⁸ Similarly, the time between the start of cannulation and the start of ECMO-flow is a good indicator of the skills and experience of the physicians actually performing the procedure.

Mean on-scene times for the emergency services were 23 min in the ARREST trial and only 13 min in the INCEPTION trial (Table 4). Despite the considerably shorter pre-hospital time, the mean low-flow time in the INCEPTION trial was 74 min, which is longer than that in the ARREST trial and Prague OHCA study (59 and 61 min, respectively). The median time between hospital arrival and start of cannulation was 16 min in the INCEPTION trial, followed by a median cannulation duration of 20 min, whereas median total ‘door-to-ECMO time’ was 12 min in the Prague OHCA study (Table 4), and mean time from randomization to ECMO initiation was 12 min in the ARREST trial (7 min from CCL arrival). No between-site differences were found in the INCEPTION trial. These data illustrate the challenge of optimizing ECPR logistics and skills in centres lacking certain exposure and expertise.

In the Prague OHCA study, 256 patients were randomized in 92 months of which 92 (82 in the invasive group and 10 crossovers in the standard group) actually received ECPR (= 1 patient/month). More strict selection criteria for ECPR (in terms of presenting rhythm) would likely improve outcome but might lead to a reduction in case load and, hence, expertise. In the ARREST trial, 30 patients were included in 10 months, with only 15 in the ECPR arm (1.5 patients/month). In the INCEPTION trial, the mean duration of active enrolment was 18 months/centre (range: 5–42), with a mean inclusion rate of 0.7 cases/centre/month (range: 0.2–2.0), and the mean number of patients with attempted cannulation per centre per month was 0.29 (range: 0–0.81).¹¹

The Minneapolis/St. Paul metropolitan area, where the ARREST trial was performed, inhabits >3.5 million people.³¹ Local agreements were made that all potential candidates were transferred to the study site. Similar agreements were made in Prague, although the adherence region of the study site was only 1.25 million residents.³² Conversely, the 10 participating centres in the INCEPTION trial served ∼8 million residents, translating to a mean adherence population of 800 000/centre with a considerable spread and overlap. The question arises whether there is a minimum adherence population for a single hospital for which it remains feasible and justifiable to offer ECPR whilst retaining acceptable efficacy and cost-effectiveness.³³

A recent analysis of the Extracorporeal Life Support Organization registry showed that the outcomes of ECPR are strongly associated with the number of procedures annually performed by a centre. It was calculated that the adjusted odds ratio for hospital survival after ECPR was 0.36 in hospitals performing <10 ECPR procedures per year.³³ Of note, only 10 hospitals in that registry exceeded that threshold. This information provides support to the notion that ECPR systems need to be developed in a similar fashion as trauma or burn centres, where concentrated high-volume expertise can be implemented to serve patients. As time-to-ECPR initiation appears to be the most important parameter associated with survival, a special challenge arises on how to deliver optimal care in large cities or metropolitan areas.

Delivery of treatments

The Prague OHCA study evaluated the efficacy of a therapeutic strategy, rather than just ECPR vs. CCPR.^5,10 The treatments in the intervention and control groups diverged immediately after randomization to intra-arrest transport with mechanical CPR and eventual ECPR, vs. continued on-scene CPR, hampering a direct comparison between patients where the application of ECPR was the only discriminating variable.¹⁰ Moreover, the external validity of this study particularly relates to healthcare systems where continued on-scene ALS is a standard of care. Of note, systematic implementation of intra-arrest transport with immediate subsequent ECPR and invasive management might have also positively influenced the logistics in the control arm, leading to unexpectedly high survival (22% vs. 10% presumed when study was designed).⁵ In the INCEPTION trial and ARREST trial, treatments differed only in the intention to apply ECPR or not, enabling an actual evaluation of the addition of ECPR to standard of care. Owing to the in-hospital randomization, the ARREST trial was most successful in assessing the efficacy of the ECPR procedure itself.

Irrespective of the timing of randomization, ECPR logistics were initiated in the pre-hospital phase in all three trials.¹² This not only allowed for a timely initiation of ECPR logistics but also led to many instances in which ECPR logistics were activated superfluously due to patients obtaining ROSC or due to contra-indications becoming apparent after the initial call. The number of superfluous activations of ECPR logistics occupying acute care personal and facilities in vain were not reported in the ARREST trial but should be taken into account when assessing total resources allocated to ECPR.³⁴ The intention-to-treat analyses and pre-hospital randomization of the Prague OHCA study and the INCEPTION trial address the efficacy and effectiveness of the broader initiation of ECPR logistics, rather than the efficacy of the procedure itself. But, conversely, not implementing the ‘ECPR readiness’ in patients managed by CCPR still led to forced crossover in highly selected individuals with favourable CPR characteristics to implement ECPR as a rescue strategy anyway. Therefore, being prepared for immediate ECPR initiation might become an integral part of emergency portfolio at the relevant admission points (such as the CCL or the emergency department).

Follow-up and outcome analysis

All three RCTs reported their results based on the intention-to-treat principle. However, the primary outcome variables differed between the three studies. The ARREST trial used survival at hospital discharge, irrespective of neurological outcome,¹² whilst the Prague OHCA and the INCEPTION trial reported survival rate with favourable cerebral performance as the primary outcome measure as is currently regarded best practice in resuscitation research.³⁵ In addition, the ARREST and INCEPTION trials determined their primary outcome after 1 month. However, as improvement in neurological status was observed even up to 6 months, it may be important to determine the primary outcome well beyond these initial 30 days. Evidence regarding the long-term outcomes of ECPR patients treated in Minneapolis/St. Paul recently showed neurologically intact survival rates of 79% at 1 year and 72% at 4 years in successfully resuscitated patients (i.e. hospital survivors).

The ARREST trial used a Bayesian adaptive approach,^12,36 which urged the data safety monitoring board (DSMB) to stop the trial when the probability of any beneficial effect of ECPR exceeded 98.6% after 30 patients already. However, the probability of a minimally relevant difference at the time of study cessation could not be assessed since this was not specified beforehand.^12,37 In contrast, the Prague OHCA study, which used a classical frequentist approach, was stopped by the DSMB as a pre-defined target of a 15% absolute risk difference (relative risk of 2.5) could no longer be reasonably achieved. An optional extension to 474 patients, which was allowed for by the statistical analysis plan,⁵ aimed at proving a possible 10% absolute risk reduction was not pursued. This decision conceivably gave rise to a potential type II error as also stipulated by the authors.

The expected absolute risk difference of 22% in the INCEPTION trial proved to be far too optimistic.¹³ However, the observed odds ratio of 1.4 for the primary endpoint in favour of ECPR may be compatible with a clinically relevant outcome difference.¹¹ As such, the study was underpowered to detect such a small yet relevant difference statistically, and therefore the INCEPTION trial still leaves some uncertainty about the true effectiveness of ECPR.

Reconciling diverging outcomes

Three well-designed and well-performed randomized trials, addressing the same question, yielded apparently diverging conclusions. Although the three trials seem comparable at first glance, their results should be weighed separately, as they each highlight different important aspects of ECPR in refractory OHCA.³⁸

Part of the critical appraisal of a randomized trial is to determine whether its design is rather explanatory (evaluating the efficacy of an intervention in optimal conditions) or pragmatic (assessing the effectiveness of an intervention when broadly implemented in ‘real-world’ circumstances).^39–41 Pragmatic or explanatory trials on the same subject actually may address very different questions and yield very different answers.⁴² The place of a trial along a pragmatic-to-explanatory continuum can be assessed by the validated PRECIS-2 tool that assigns a score of 1–5 ranging from very explanatory (1) to very pragmatic (5) to each of nine domains of study design.⁴³ The tool was completed by four of the authors of the current review, three of whom had no involvement in the original study design (H.U., S.H., A.v.d.K., and M.v.d.P.) (Figure 1, details of the score can be found in Supplementary material online, Material S1).

Figure 1.

Study design assessment of the three randomized trials, according to the PRECIS-2 domains. Of note, the domain flexibility (adherence) was deemed not applicable, as adherence in ECPR treatment is not an issue. The omission of this domain in such interventions is also advocated by the PRECIS-2 authors. PRECIS-2, Pragmatic-Explanatory Continuum Indicator Summary 2.

The ARREST trial was assessed to be predominantly explanatory.⁶ The trial demonstrated exceptional outcomes for ECPR in refractory OHCA but was performed in strictly selected patients by a team of seasoned ECPR experts within a single centre of expertise. The Prague OHCA study was pragmatic on eligibility criteria for ECPR, but overall, the study had numerous characteristics that should be considered rather explanatory.¹⁰ In particular, similar to the ARREST trial, it was a single-centre study, and ECPR was performed by a small, highly dedicated team, which is reflected by the low-flow times and cannulation times that were similar to the ARREST trial.^6,10 However, the pragmatic selection criteria including non-shockable initial rhythms most likely contributed to the lower overall success rate of ECPR as compared with the ARREST trial.⁴⁴ Moreover, the external validity of the Prague OHCA study is dependent on the comparability of the treatment protocols in the control group and the standard of care in other healthcare systems.

Finally, the multi-centre INCEPTION trial was predominantly pragmatic in design, as it was performed in multiple centres without imposing a specific ECPR protocol on these institutions. Its primary aim was not to provide evidence for the efficacy of ECPR itself but to assess the effectiveness of implementation of ECPR for OHCA in the real world.

Cost-effectiveness

The implementation and maintenance of an effective ECPR system require a great investment of human and financial resources. In addition, healthcare costs for patients receiving ECPR are high; most patients undergo computed tomography scanning, coronary angiography, and percutaneous coronary interventions and may go through a complex prolonged intensive care unit (ICU) stay. Survivors often require intensive rehabilitation programmes. Healthcare costs of patients treated with conventional CPR are considered to be much lower since, in most instances, they demise before hospital admission or at the emergency department. Patients surviving to ICU admission through ROSC often have a faster recovery than patients undergoing ECPR.¹ Moreover, the mean costs per ECPR patient will further increase with increasing survival rates.² Consequently, the incremental cost-effectiveness ratio of ECPR vs. CCPR is the resultant of a complex interaction between several variables, such as the amount of resources invested in the ECPR system and the clinical results of both conventional CPR and ECPR. Thus far, no trial-based cost-effectiveness studies are available. Of note, preliminary, though unpublished, data from the Prague OHCA study suggest similar hospitalization costs for favourably surviving patients in both invasive and standard arms, however, vast difference in terms of incremental cost of ECPR patients with unfavourable neurological outcome. Anyway, several studies based on Markov modelling have suggested that ECPR can be cost-effective.^3,45 Cost-effectiveness may vary amongst systems, but a high survival rate and probably a high caseload seem necessary to achieve a cost-effectiveness ratio that is acceptable from a health economic perspective.

Future research priorities

Based on the presented evidence with its limitations and current understanding of the science, the field needs to focus on research questions to advance patient care and ECPR indications. Priorities for future research should include definition of patient groups that better benefit from ECPR including development of assessment tools for rapid identification of candidacy; optimization of the timing for ECMO initiation and time constraints including use of smart applications and alerts; ethical and timely transitions from attempted survival and recovery to organ donation; development of logistically simplified and effective ECPR systems using hub-spoke systems vs. pre-hospital methods of implementation; physiologic evaluation of prolonged CPR in patients without ROSC; cost-effectiveness research for ECPR; and finally the research in post-arrest care including accurate neuromonitoring, haemodynamic targets, temperature control, and early neuroprognostication.

Conclusion

The ARREST trial has demonstrated the efficacy of ECPR for OHCA in a highly controlled and dedicated environment. The Prague OHCA study showed that in a similarly dedicated environment, outcomes of CCPR may exceed presumed estimations by implementing advanced logistics and ECPR effectiveness itself may become impaired if inclusion criteria are widened to non-shockable rhythms. Finally, the INCEPTION trial illustrated the complexity of implementing ECPR in a real-world setting even in a highly developed country and underlines that implementation of ECPR for OHCA does not necessarily lead to replication of the excellent results obtained in single centres of excellence in metropolitan areas. Centres that perform ECPR or aspire to do so should critically assess their setting, logistics, performance, and outcomes to assure that their resources are well spent.

Supplementary material

Supplementary material is available at European Heart Journal: Acute Cardiovascular Care online.

Funding

None declared.

Data availability

All reasonable requests for data sharing will be considered and should be emailed to Prof. Jan Belohlavek at jan.belohlavek@vfn.cz