Improving access to extracorporeal membrane oxygenation for out of hospital cardiac arrest: pre-hospital ECPR and alternate delivery strategies

We defined three strategies for ECPR delivery for refractory OHCA within Sydney, Australia and then applied transport accessibility metric analysis methods to determine the effective patient catchment of each strategy. The study was completed and reported in line with the Revised Standards for Quality Improvement Reporting Excellence (SQUIRE 2.0) guidelines 2015 [11].

In-hospital ECPR: ECMO cannulation and ECPR are delivered at an ECPR-capable hospital (current status). Assuming that time of arrest coincides with time of call, arrest to flow time is calculated by: response time + on-scene time + travel time + cannulation time.

Rendezvous ECPR: The patient is transferred to an emergency department, which may not be ECPR-capable, in order to rendezvous with the ECPR team. The rendezvous hospital was selected based on minimisation of the greater of: ECPR team travel time to the rendezvous hospital, and: paramedic response time + on-scene time + travel time to rendezvous hospital. Arrest to flow time is the sum of the maximum of these two intervals + cannulation time. The ECPR team rendezvous with the patient at that hospital, establishes ECMO support and transfers the patient on ECMO back to a central ECMO hospital. Modelling assumes the ECPR team is notified of the OHCA at the same time of the initial cardiac arrest call and begins movement to the rendezvous hospital.

Pre-hospital ECPR: A pre-hospital ECPR team is dispatched and ECMO cannulation is completed at the scene of cardiac arrest, with subsequent transfer of the patient back to an ECMO-capable hospital. Modelling assumed that the pre-hospital ECPR team is dispatched at the same time of the cardiac arrest emergency call. Arrest to flow time is calculated by: response time + cannulation time.

The activation point for the mobile ECMO teams in Rendezvous and Pre-hospital ECPR of time of initial EMS call, was chosen as: (a) a number of current trials [7, 12,13,14], ON-Scene (NCT04620070), currently utilise this approach and (b) previous studies have reported that a majority of OHCA are recognised by emergency dispatchers between 50 s to approximately 2 min, [15,16,17,18].

Transport accessibility metrics analyses

The comparison of the three cardiac arrest strategies was addressed with transport accessibility metrics [10]—Table 1. In this approach, the study area is divided into zones with a known number of potential patients, xi. ECPR facilities (in-hospital, rendezvous or prehospital) can be allocated to each zone, and yj represents the number of facilities in zone j. Usually yj would be zero or one. The ability of a patient in zone i being able to access the ECPR facilities in zone j, requires knowledge of the complete travel time matrix, tij. Since the success of ECPR in zone i depends on the time from arrest to ECMO flow, Ti, the travel time is added to other relevant time intervals for that ECPR delivery strategy. The components are defined as:

Table 1 Components of the time from arrest to ECMO flow under each ECPR delivery strategy for a patient in zone i accessing ECPR located in zone j

Response time (thi): The time from the location of the ambulance in zone h to the location of the patient in zone i. This is the time between the call to emergency medical services (EMS) and arrival of EMS paramedics at the scene of cardiac arrest. For pre-hospital ECPR, the response time is from the location of the mobile ECMO unit in zone j (tji).

Scene time (ts): Time interval between arrival of paramedics on scene and patient departure to hospital, includes patient access, treatment and extrication.

Travel time: Transfer time from location of cardiac arrest to ECPR-capable hospital (tij for in-hospital ECPR) or intermediate emergency department (tik for rendezvous ECPR).

Cannulation time (tc): Time from arrival of ECPR team at the patient to establishment of ECMO flows. This is expected to be longer in a pre-hospital environment.

For an arrest occurring in zone i with ECPR facilities in zone j and a suitable emergency department in zone k, this interval is the minimum across all facilities of the sum of the time components—Table 1.

Determining population coverage by ECPR strategy

To reflect the 60 min cut-off of eligibility, the time from arrest to ECMO flow for patients in zone i, Ti, is compared to the threshold \(\tau = 60\) and that zone is indicated to be either above or below the threshold with a binary variable:

$$b_ = \beginl} 1 \hfill & \hfill & \le \tau } \hfill \\ 0 \hfill & \hfill & > \tau } \hfill \\ \end$$

If a zone is covered by any facility, (i.e., \(b_ = 1\)) then the potential patients in that zone, xi contribute to the total coverage, Ac.

$$A_ = \mathop \sum \limits_^ x_ b_$$

Higher values of Ac. indicate that the strategy or ECPR-facility location offers an advantage in the number of potential patients that can receive ECPR.

Survival benefit modelling

Patients who are commenced on ECMO flow earlier after arrest are more likely to survive [2, 19], therefore we supplemented the population coverage metric with another measure that weights each covered patient by their probability of survival. Probability of survival in zone i, pi, ranges from 0 to 1, and is estimated by evaluating a decreasing survival function at Ti. As defines the population-weighted survival probability below.

$$A_ = \frac^ x_ }} \mathop \sum \limits_^ x_ p_$$

For estimating population-weighted survival probability, we modelled the relationship between survival probability and resuscitation time using the aggregated data reported by Bartos et al. [1]—Fig. 1. The average survival outcomes from that paper are fitted with a logistic curve with time to resuscitation (low-flow time) as the only predictor using the statsmodels package in Python. The fit is evaluated at the arrest to ECMO flow times, ti to provide a probability of survival, pi, at each meshblock i in each scenario.

Fig. 1figure 1

Fitted logistic survival rate functions for ECPR [20] and conventional cardio-pulmonary resuscitation using data from the Amiodarone, Lidocaine or Placebo study (ALPS) [21]

Unlike the coverage metric, the population weight survival probability varies from 0 to 1 and gives the overall probability of survival for ECPR-eligible arrests in each delivery scenario. Therefore, it can distinguish between two strategies that reach the same number of potential patients, but one reaches them faster and provides better survival outcomes. Furthermore, the population-weighted survival probability allows us to relax the 60 min threshold and quantify the benefit to patients who sit just outside this coverage boundary.

Modelling location and data levels

Modelling was completed for Greater Metropolitan Sydney, Australia with a 2016 census population of 4.8 million and area of 12,368 km2. Hospital-based ECPR services exist at 5 hospitals (Additional file 1: Fig. S1), and rendezvous and pre-hospital ECPR are not offered within Sydney. The analysis zones are Greater Sydney’s approximately 58,000 meshblocks, the finest spatial resolution available in the Australian census data [22]. Patients are assumed to be distributed proportionately to the meshblock resident populations from the 2016 census counts. Historical cardiac arrest cases [23] from 2017 to mid-2021 aggregated to the statistical area level 2 were used to calculate localised ambulance response times. The distribution of on-scene treatment times for CCPR were obtained from the NSW OHCA registry [23]. Meshblock-to-meshblock travel times on the road network were calculated from Compass IoT’s connected vehicle data averaging speeds for every link in the Sydney network from one week in November 2019. Travel times were validated against realised ambulance travel times from the cardiac arrest registry and shown to be consistent to within 2% (Additional file 1: Fig. S2). These data comprise the necessary inputs for calculating Ac as described above.

Base case modelling for the status quo ECPR delivery strategy—Fig. 2, summarised in Table 1, uses locally-appropriate response times, a 27 min on-scene treatment time, travel time to the nearest ECPR capable hospital time, and 15 min of cannulation time. The on-scene treatment time of 27 min was chosen based on published data [5, 24] of expedited transfer of patients from scene as until 2021, mechanical CPR devices were not available in Sydney, NSW. Interim data from our currently recruiting, EVIDENCE study (ACTRN12621000668808), comparing expedited transfer to more extended on-scene resuscitation thus far, has reports a median on scene time of 26 min in the expedited arm.

Fig. 2figure 2

Survival rates subject to the current status quo of in-hospital ECPR offered at 5 hospitals assuming local response times, 27 min on-scene time, and 15 min cannulation. Eligibility is subject to a one-hour threshold. Blue shades indicate areas that are both within the 1 h threshold and offer higher ECPR survival than CCPR

Rendezvous modelling assumed that ECPR teams are located at the five current hospitals but can mobilise to move to any of the 26 emergency departments in the study area. The time from arrest to treatment includes the locally-appropriate response time, 27 min on-scene time, travel time to the rendezvous hospital emergency department and 15 min of cannulation time.

The pre-hospital strategy assumes one optimally-positioned mobile ECPR team thereby the providing the minimum benefit of a pre-hospital ECPR service. The methodology for identifying the optimal position is described below. The time from arrest to ECMO flow is the sum of the travel time from the optimal location to the patient plus 22 min for cannulation, where the additional 7 min accounts for the difficult cannulation context and has been based on published experience thus far [7].

Additional sensitivity analyses (Additional file 1: Table S1 and Figs. S3–S7) included variation key variables to determine changes in the outcome: On-scene treatment was tested at 22 min, 27 min and 32 min to reflect aspirational, reported [5] and historical values [23, 25] respectively. Pre-hospital cannulation time is tested at 22 and 27 min to take into account delays owing to the difficulty of the pre-hospital environment. Additional mobile ECMO teams were also tested Additional file 1: Fig. S7.

Optimal location for basing pre-hospital ECPR team

The optimal location for placement of a pre-hospital ECPR team was determined by enumerating all possible locations, j, and identifying the one with the highest population-weighted survival probability, As. One additional model was completed assuming a mobile ECPR team located at an existing aeromedical base for practicality regarding staffing and restocking—Additional file 1: Fig. S8.

Sensitivity of the timepoint of when mobile ECMO team is dispatched was also assessed. The base case assumes activation of the pre-hospital team at time of EMS call as described in methods above. Modelling was completed assuming the activation of the mobile ECMO team, two minutes after the arrival of the first ambulance to allow for additional review of suitability for the pre-hospital ECPR—Additional file 1: Fig. S9.

Role of the funding source

The funding source, the New South Wales Translational Grant Scheme, had no role in the study design, collection, analysis, or interpretation of data, the writing or editing of the manuscript, or the decision to submit the work for publication.

The work was approved by Sydney Local Health District ethics committee reference: X21-0002.

留言 (0)

沒有登入
gif