Assessing accessibility to the extracorporeal cardiopulmonary resuscitation pathway for out-of-hospital cardiac arrests in London Authors: Amiyesh Sahay, Caitlin Sears, Mark Faulkner, Rachael Fothergill Published: 2026 ABSTRACT AbstractAimsExtracorporeal cardiopulmonary resuscitation (ECPR) may improve outcomes in selected patients with refractory out-of-hospital cardiac arrest (OHCA), but timely access is essential. In Greater London, a prehospital ECPR pathway enables transport of eligible patients to extracorporeal membrane oxygenation (ECMO)-capable centres within a target arrest-to-arrival interval of 45 min. We quantified accessibility to this pathway and explored how alternative service configurations might influence access.MethodsThis retrospective, observational study analysed five years of routinely collected OHCA data from the London Ambulance Service to identify patients meeting the ECPR eligibility criteria. Using pathway modelling and geospatial analysis we estimated the proportion of patients able to reach an ECMO-capable centre within 45 min and evaluated potential strategies for improving access.Results776 patients were identified as being eligible for the prehospital ECPR pathway, but only 3% (n = 23/776) were estimated to reach an ECMO-capable centre during operating hours and within 45 min of cardiac arrest. In many cases, the threshold was exceeded before patients were ready for transport. Extending the arrest-to-arrival interval to 60 min resulted in a 217% increase (50 additional patients) predicted to access the pathway. Other improvements included extending operating hours (+104%, n = 24), broadening pathway activation criteria (+52%, n = 12), and increasing the number of receiving centres (+30%, n = 7).ConclusionAccessibility to the prehospital ECPR pathway appears limited. Modifications to operational parameters, particularly extending time thresholds and operating hours, could substantially increase access. These findings have important implications for the design and optimisation of ECPR services in urban medical systems.Keywords: Extracorporeal cardiopulmonary resuscitation (ECPR), Extracorporeal membrane oxygenation (ECMO), Emergency Medical Services (EMS), Out-of-hospital cardiac arrest (OHCA), Geospatial analysis Introduction Outcomes from out-of-hospital cardiac arrest (OHCA) remain poor, with registry data from England reporting 30-day survival rates below 10%.1 In this context, interest in extracorporeal cardiopulmonary resuscitation (ECPR) is growing. ECPR involves the intra-arrest application of extracorporeal membrane oxygenation (ECMO) to provide oxygenation and end-organ perfusion while potentially reversible causes are identified and addressed.2 While emerging evidence of the benefits of ECPR for OHCA is mixed and of low certainty,3 positive outcomes, including survival to hospital discharge rates of up to 43%4 and improved neurological outcomes,5 have been reported. Consequently, the European Resuscitation Council supports consideration of ECPR in adults when conventional CPR has failed and when timely initiation is feasible.6 However, ECPR is resource-intensive, requiring specialist equipment, trained personnel and infrastructure2, 7 as well as careful patient selection.8 Although eligibility criteria can vary,9 current guidance prioritises those with the greatest likelihood of good neurological recovery, considering factors such as age, initial cardiac rhythm and comorbidities.2 Shorter low-flow time – the interval between cardiopulmonary resuscitation (CPR) and ECPR10 – is consistently associated with better outcomes, and a threshold of less than 60 min widely accepted.9, 10, 11, 12 While prehospital ECPR has shown feasibility in some major cities, its sustainability and effectiveness require further investigation.13, 14 Potential delivery models, including ECPR initiation both pre- and in-hospital, have been explored using geospatial modelling,15, 16, 17, 18, 19, 20 but generalisability is limited by differences in population density, emergency healthcare systems, and inconsistent adjustments for on-scene and emergency driving times. This study quantified the accessibility of a prehospital ECPR pathway, where eligible patients receive expedited transport to an ECMO-capable centre in Greater London for in-hospital ECPR, using pathway modelling and geospatial analysis. We also explored how alternative delivery configurations might influence access. By addressing a key evidence gap, this work seeks to inform future ECPR service design and optimisation within urban emergency medical service (EMS) systems. Methods Design This retrospective, observational study examined five years’ worth of data for OHCA patients treated by the London Ambulance Service NHS Trust (LAS). Cases meeting ECPR eligibility criteria were identified, and the proportion expected to arrive at an ECMO-capable hospital within 45 min of arrest was estimated. A range of alternative delivery scenarios were modelled to inform future accessibility of the ECPR pathway. As this study involved the secondary use of routinely collected non-identifiable data, NHS Research Ethics Committee approval was not required. Approval was granted under the LAS’s research governance process. Setting and population The LAS provides prehospital emergency and urgent care across Greater London, serving a population of over 9 million people across an area of 607 square miles.21 Each year, LAS clinicians treat over 1.2 million patients and initiate resuscitation in over 4000 cardiac arrest patients.22 OHCA is managed in line with the Resuscitation Council UK Guidelines6 with Basic and Advanced Life Support (ALS) provided at the scene typically until return of spontaneous circulation (ROSC) is achieved or resuscitation termination criteria are met. In certain circumstances, clinicians may consider early transportation to hospital for specific interventions such as ECPR.23 LAS employs a cohort of specialist Advanced Critical Care Paramedic Practitioners (APPCCs) with enhanced training and scope of practice. Up to five APPCCs are selectively dispatched each day to the most critically unwell patients. London prehospital ECPR pathway The London prehospital ECPR pathway was implemented in April 2024 to identify patients potentially suitable for ECPR and facilitate rapid conveyance to an ECMO-capable centre. Pathway activation is undertaken by the attending APPCC following confirmation of eligibility based on predefined inclusion criteria. Clinical inclusion criteria were: non-traumatic witnessed OHCA with immediate effective CPR; an initial shockable rhythm; absence of perceived clinical frailty; no active do not attempt resuscitation order or advanced directive; no major comorbidity or pregnancy; estimated weight >40 kg; refractory cardiac arrest (defined as no sustained ROSC after 3 shocks or 15–20 min of cardiac arrest24). Additionally, the anticipated arrest-to-arrival at an ECMO-capable hospital interval must be <45 min. This time threshold is intended to facilitate initiation of ECMO within 60 min of the cardiac arrest. Eligible patients can be transported by LAS clinicians to one of four ECMO-capable centres that are situated within specialist Heart Attack Centres (HACs): St. Bartholomew’s, Harefield, King’s College and St. Thomas’s hospitals. Each centre accepts patients via the pathway between the hours of 08:00 and 17:00 from Monday to Friday and is notified in advance by a telephone alert. Data collection Anonymised data were collected for OHCA patients who had resuscitation attempted by LAS clinicians following a non-traumatic witnessed arrest (bystander or EMS) between 1st January 2018 – 31st December 2019 and 1st January 2022 – 31st December 2024. Cases from 2020 and 2021 were excluded due to the COVID-19 pandemic-related increases in OHCA incidence25 and the unique demands that were placed on ambulance services during that time.26 Data were sourced from the LAS’s OHCA Registry, which collates information from 999 call handling, dispatch systems and electronic clinical patient records. Data were collected according to Utstein definitions,27 and included EMS response timestamps, patient demographics, prehospital interventions, conveyance details, and ROSC both at any time and sustained to hospital arrival. Two datasets were created: 1.ECPR clinically eligible cohort: patients meeting the clinical inclusion criteria were identified to estimate accessibility and support scenario modelling. Due to data limitations, proxy measures were used: patients aged <14 years were excluded as a surrogate for body weight <40 kg, and those aged >70 years as a proxy for clinical frailty. ECPR clinically eligible cohort: patients meeting the clinical inclusion criteria were identified to estimate accessibility and support scenario modelling. Due to data limitations, proxy measures were used: patients aged <14 years were excluded as a surrogate for body weight <40 kg, and those aged >70 years as a proxy for clinical frailty. 2.HAC conveyance cohort: Data for all OHCA patients conveyed to a HAC, irrespective of pathway eligibility, were used to derive a blue-light driving adjustment factor (as described later). HAC conveyance cohort: Data for all OHCA patients conveyed to a HAC, irrespective of pathway eligibility, were used to derive a blue-light driving adjustment factor (as described later). Data analysis Descriptive statistics were used to summarise patient characteristics and time intervals. Continuous variables were summarised as medians with interquartile ranges (IQR), and categorical variables as frequencies and percentages. All analyses, including time interval modelling and geospatial analyses, were undertaken using Python (v3.12). Time interval modelling For each clinically eligible patient, four time intervals were modelled to estimate the time from cardiac arrest to arrival at an ECMO-capable centre (Fig. 1). With the exception of driving times, all time intervals after the first shock were agreed by clinical consensus to reflect realistic clinical and operational processes. 1.Arrest to first shock Arrest to first shock Modelling ECPR pathway timeline. Observed timestamps were used to calculate the interval from cardiac arrest to first defibrillation. The time of arrest was defined as the 999-call time for bystander-witnessed arrests, and as CPR start times for EMS-witnessed arrests. 2.First shock to pathway activation First shock to pathway activation Pathway activation requires APPCC presence on scene. Therefore, the activation time was defined as the later of the ‘refractory arrest time’ or the APPCC arrival on scene time, with an additional 4 min allowed for decision-making (Fig. 1). Refractory arrest time was based on the presenting cardiac rhythm: •Shockable rhythm (persistent): For patients who remained in a shockable rhythm after the third shock, 6 min were added to the time of the first EMS shock to account for the delivery of three shocks with 2 min of CPR after each shock (Figs. 1, 3a). Shockable rhythm (persistent): For patients who remained in a shockable rhythm after the third shock, 6 min were added to the time of the first EMS shock to account for the delivery of three shocks with 2 min of CPR after each shock (Figs. 1, 3a). •Shockable to non-shockable conversion: For patients who converted to a non-shockable rhythm following up to three shocks, 15 min from the time of arrest was applied to reflect continued resuscitative efforts before classification as refractory arrest (Figs. 1, 3b). Shockable to non-shockable conversion: For patients who converted to a non-shockable rhythm following up to three shocks, 15 min from the time of arrest was applied to reflect continued resuscitative efforts before classification as refractory arrest (Figs. 1, 3b). 3.Pathway activation to ready for conveyance Pathway activation to ready for conveyance Readiness for conveyance (n = 776). Following pathway activation, three minutes were added for application of the mechanical chest compression device (if not already in position), and a further 12 min to account for patient extrication and readiness for conveyance. 4.Remaining transport time Remaining transport time The combined time intervals above were subtracted from 45 min to provide an estimate of the time remaining within the threshold for transport to an ECMO-capable centre. For each case, an isochrone was generated using the MapBox Isochrones API28 centred on the incident location and adjusted for time and date of incident. The resulting isochrone represented the maximum travel distance achievable within the remaining blue-light adjusted transport time. If hospital coordinates fell within the isochrone, the patient was predicted to arrive in time and considered eligible. Blue light adjustment Using data from the HAC conveyance cohort, known (observed) driving times for each patient taken to a HAC during the five-year study period were calculated. Observed HAC transport times were compared with modelled non-emergency driving times, using the start and end location for each case, to derive a blue light adjustment factor (using MapBox Directions API29) that was applied to all transport estimates to reflect faster emergency driving times. Missing data Missing CPR and defibrillation times data were handled using multiple imputation by chained equations (MICE) with predictive mean matching, stratified by witnessed type. Prior to analysis, 3 defibrillation and 12 CPR times were imputed for EMS-witnessed arrests and 89 defibrillation times for bystander-witnessed arrests. Ten datasets were imputed using the MICE package in R (v4.3.2). Time interval modelling and pathway eligibility classification were performed separately on each, with intermediate results presented as means across the datasets. A Wilson score 95% confidence interval was calculated on the pooled eligible proportion. Scenario modelling Four operational scenarios were modelled to assess their impact on pathway accessibility. Each scenario was compared with the current pathway configuration: 1.Extension of the arrest-to-arrival time from 45 to 60 min Extension of the arrest-to-arrival time from 45 to 60 min 2.Extension of operating hours to 24 h a day, 7 days a week Extension of operating hours to 24 h a day, 7 days a week 3.Expansion of pathway activation to other clinical managers with advanced decision-making capabilities, who can attend an OHCA instead of, or before, an APPCC Expansion of pathway activation to other clinical managers with advanced decision-making capabilities, who can attend an OHCA instead of, or before, an APPCC 4.Expansion of receiving centres to allow coverage by all London HACs that have the cardiac surgery capabilities to initiate ECMO. Expansion of receiving centres to allow coverage by all London HACs that have the cardiac surgery capabilities to initiate ECMO. Results During the five years, 58,860 patients experienced an OHCA in London. Resuscitation was attempted in 22,017 cases (37.4%). Of these, 8.6% (n = 1884/22,017) were in refractory arrest, and 5.3% (n = 1164/22,017) met the clinical eligibility criteria for ECPR. Twenty-nine were excluded due to missing data, leaving 1135 cases. An APPCC attended 776 patients (68.4%; 155.2 per year), and it was these cases that formed the pathway-eligible cohort for the primary analysis (Fig. 2). Study pathway eligibility. Cohort characteristics Patient characteristics are presented in Supplementary Table 1, stratified by witnessed status. The majority of pathway-eligible patients were male (83.5%; n = 648/776). The median age was 55 years. Just over half of OHCAs occurred at home (51.5%; n = 400), and most were bystander-witnessed (89.6%; n = 695). The vast majority of arrests were presumed cardiac aetiology (95.1%; n = 738). Return of spontaneous circulation (ROSC) was achieved in 53.4% of cases (n = 414), and sustained to hospital for 35.4% (n = 275). The median number of shocks delivered (including public access defibrillation) was 6. Observed timings The median time from cardiac arrest onset to first EMS shock was 11:49 min (IQR 9:34–15:35, n = 627) for bystander-witnessed arrests, and 1:00 min (IQR 0:09–2:00, n = 68) for EMS-witnessed. The median time from cardiac arrest onset to APPCC arrival was 28:39 min (20:39–38:38, n = 766). Pathway accessibility Overall, 33.5% (n = 260/776) of pathway-eligible patients (clinically eligible with an APPCC on scene to activate the pathway) were ready for conveyance within pathway operating hours (Fig. 3). However, the 45-min threshold was already exceeded for 58.5% (n = 152/260) before transportation (Fig. 4). Where the threshold was not exceeded (n = 108; 41.5%), the median time remaining for transport was 7:48 min (IQR 4:40–11:00). This was 7:25 min for bystander-witnessed arrests (IQR 4:32–10:28) and 11:00 min for EMS-witnessed arrests (IQR 5:02–15:22). Distribution of time from cardiac arrest onset to ready for conveyance (n = 260). Geospatial analysis Fig. 5 illustrates the area reachable within the average remaining transport time, rounded up to 8 min, for each ECMO-capable centre. After adjusting each isochrone’s radius by 23.7% to reflect emergency driving conditions, 23 of the 776 patients (3.0%, CI 2.0–4.5%; 4.6 per year) were predicted to reach an ECPR centre within 45 min of cardiac arrest onset. Average remaining driving time of 8 min under emergency conditions.* *Isochrone generated for 6th June 2023 at 12:00 to represent typical incident conditions using modal day and time of arrest in the study dataset. All 23 patients were predicted to either reach St Bartholomew's, St Thomas’ or King’s College Hospitals; none were predicted to reach Harefield. Nine patients were within the catchment of a single hospital, while 14 had access to more than one. Scenarios to improve pathway accessibility All modelled scenarios independently increased the number of patients able to access the ECPR pathway, although the magnitude varied (Table 1). Table 1Modelled impact (from baseline) of operational scenarios on access to the ECPR pathway.Scenario% Increase (n)No. eligible%, (95 CI)Baseline (n = 776)−233.0% (2.0–4.5%)Extend arrest-to-arrival at ECMO centre threshold to 60 min (n = 776)+217.4% (50)739.4% (7.6–11.8%)24/7 operating hours (n = 776)+ 104.3% (24)476.1% (4.6–8.1%)Pathway activation by clinical managers (n = 965)+ 52.2% (12)353.6% (2.6–5.0%)Expansion of receiving centres (n = 776)+30.4% (7)303.9% (2.8–5.5%) Modelled impact (from baseline) of operational scenarios on access to the ECPR pathway. The greatest increase (217.4%; 50 additional patients) was achieved by extending the arrest-to-arrival interval to 60 min. Extending pathway operating hours to 24/7 increased access by 104.3% (24 additional patients), reflecting that 66.5% of cases (n = 516/776) were ready for conveyance out-of-hours. A Clinical Manager arrived before an APPCC in 35.8% of cases (n = 278/776) and attended an additional 189 clinically eligible patients without an APPCC on scene (Fig. 2). Allowing Clinical Managers to activate the pathway was predicted to increase access by 52.2% (an additional 12 patients). Including all HACS with cardiac surgery capability increased accessibility by 30.4% (7 additional patients). Discussion Our analysis revealed constraints to accessing the pre-hospital ECPR pathway in London, driven by operational and geographical factors. While over 5% of patients receiving a resuscitation attempt met the clinical eligibility criteria for ECPR, few were predicted to reach an ECMO-capable centre within the time window. These findings demonstrate a mismatch between theoretical eligibility and real-world access to this potentially lifesaving intervention. Time was the principal barrier to ECPR access. Nearly two-thirds of potentially eligible patients were ready for conveyance outside pathway operating hours. Even when patients were ready during operational hours, the 45-min arrest-to-arrival limit was exceeded before departure from scene in more than half of cases. Furthermore, among those with time remaining within the arrest-to-arrival window, the time available for transport was short, averaging only 8 min. Consequently, OHCA patients nearest to an ECMO-capable hospital were disproportionately more likely to be able to access the pathway. Time constraints and proximity were also reported as precluding factors to accessing hospital-based ECPR in a study by a UK Air Ambulance Service.17 Temporal challenges were compounded by the geographical clustering of ECMO-capable centres within a relatively small area of London. These findings highlight the importance of location for timely ECPR, even within an urban EMS system with short distances to hospital. Our modelling suggests that pre-hospital ECPR access may be highly sensitive to pathway modifications. In particular, expanding the arrest-to-arrival interval from 45 to 60 min increased predicted access by more than 200%. These findings are consistent with those from the US, where broader time windows substantially increased the number of eligible patients.19 However, such increases would need to be balanced against receiving centre capacity and the time-sensitive nature of the intervention. Arrival at an ECMO centre does not mark the end of the low-flow period, as time is needed for patient assessment, cannulation and the initiation of extracorporeal flow. More favourable outcomes are observed when the low-flow time is under 60 min, with the likelihood of survival and good neurological recovery declining significantly as low-flow times increase.30 In line with this, current ECPR guidelines support a prolonged low-flow duration only in the presence of favourable prognostic features, such as signs of life during CPR, hypothermia or intermittent ROSC.2 Although extending time thresholds may improve equity of access, it should not compromise the likelihood of achieving a positive outcome. Beyond transport times, pathway accessibility was also strongly influenced by ECMO-capable centre availability. Providing a 24/7 service was estimated to more than double pathway availability, while increasing the number of ECMO-capable centres produced a more modest, but meaningful gain. Evidence from established ECPR systems demonstrates that a substantial proportion of patients present outside routine daytime operating hours.31 However, providing a fully operational 24/7 service has considerable cost implications, and alternative delivery models may warrant consideration. One approach, successfully implemented elsewhere with a reported 63% increase in ECMO availability, is the rendezvous strategy, which involves expedited transfer to a local hospital for treatment by a mobile ECMO team.18, 31 Leveraging existing emergency departments and specialist cardiac centres to facilitate such a model could increase coverage without the need to establish additional ECMO centres. Pre-hospital ECMO presents another potential solution. The SUB30 feasibility study showed that pre-hospital ECPR delivery is safe and effective in London, with a mean time to full flow of 47 min (range 37–59).13 More recently, London’s Air Ambulance established the Endovascular Cardiac Arrest Team (ECAT), which can provide ECMO at the scene of an OHCA.32 Bringing ECMO to the patient may mitigate the geographical barriers identified in this study. However, a single mobile ECMO team is still likely to face availability and coverage limitations. Future research should examine the impact of hospital-based, rendezvous strategies and pre-hospital ECPR on access, low-flow duration and outcomes. Finally, changes to the activation process may provide an opportunity to improve pre-hospital ECPR access by reducing delays between recognition of refractory cardiac arrest and pathway initiation. Expanding activation beyond APPCC clinicians to include other senior clinicians already on scene, in particular Clinical Managers, was predicted to increase pathway access by more than 50%. Optimising clinical decision-making and activation may represent a pragmatic and lower cost intervention compared with major service reconfiguration. Limitations of this study include the use of proxy measures, the assumed effectiveness of bystander CPR, and modelling assumptions related to clinical decision-making and extrication times, which may have introduced imprecision to estimated patient eligibility and pathway timings. However, they were based on clinical consensus and applied consistently across all cases, so are unlikely to alter the overall conclusion that pre-hospital ECPR access in London is predominantly constrained by operational and geographical factors. Additionally, this study was conducted within a single highly urban EMS system, which may limit generalisability to other regions. Conclusion As ECPR services continue to evolve, improving accessibility must remain a priority. Our findings demonstrate the challenges of delivering timely ECPR within an urban EMS system, highlighting operational barriers and meaningful improvements that could be achieved through targeted changes to the pathway design. Developing an effective and equitable ECPR system requires multidisciplinary collaboration to ensure access is maximised while preserving the time-sensitive benefits of this intervention. Funding sources No specific funding was received from public, commercial, or not-for-profit funding bodies. CRediT authorship contribution statement Amiyesh Sahay: Writing – original draft, Methodology, Formal analysis, Data curation, Conceptualization. Caitlin Sears: Writing – original draft, Methodology, Formal analysis, Data curation, Conceptualization. Mark Faulkner: Writing – review & editing, Methodology, Data curation. Rachael Fothergill: Writing – review & editing, Writing – original draft, Methodology, Formal analysis, Conceptualization. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements We would like to acknowledge and thank members of the London Ambulance Service’s Clinical Audit and Research Unit who were involved in populating and managing the LAS OHCA registry during the study period. Footnotes Supplementary material to this article can be found online at https://doi.org/10.1016/j.resplu.2026.101468. Supplementary material The following are the Supplementary material to this article: Supplementary Data 1 Supplementary Table 1: Incident information and characteristics of ECPR eligible patients in the study, stratified by witness type. References Associated Data Supplementary Materials Supplementary Data 1 Supplementary Table 1: Incident information and characteristics of ECPR eligible patients in the study, stratified by witness type.