Extracorporeal cardiopulmonary resuscitation in adult patients with out-of-hospital cardiac arrest: a retrospective large cohort multicenter study in Japan

1The utilization of extracorporeal membrane oxygenation (ECMO) during conventional cardiopulmonary resuscitation (CPR) is termed extracorporeal CPR (ECPR), and the prevalence of ECPR in patients with out-of-hospital cardiac arrest (OHCA) has been rapidly increasing worldwide. In a recent study, promising data regarding the efficacy of ECPR in patients with OHCA younger than 75 years who have refractory ventricular fibrillation (VF) was reported, and ECPR will continue to be considered a “life-saving device” around the world. Unlike with conventional CPR, there are no practice guidelines on ECPR; therefore, ECPR protocol varies across hospitals.

2Initial cardiac rhythm, age, and time (low flow time or no flow time) are commonly used as principal parameters of inclusion criteria in clinical ECPR practice. Several clinical studies on ECPR with small sample sizes have been published; however, real-world data on ECPR, such as patients with OHCA and non-shockable rhythm, old age, long time from cardiac arrest to initiation of veno-arterial extracorporeal membrane oxygenation (VA-ECMO), have not been thoroughly evaluated. Further, details of ECPR, such as its complications, remain unknown.

3Our group designed the retrospective large cohort study known as the Study of Advanced life support for Ventricular fibrillation with Extracorporeal circulation in Japan (SAVE-J II) to provide real-world data on ECPR performed on about 2000 patients. The aim of this study was to determine the association between the above-mentioned parameters and outcomes of patients with OHCA who received ECPR.

4SAVE-J II is a retrospective multicenter registry study with 36 participating institutions in Japan (Additional file 1: Fig. S1). The study was pre-registered at the University Hospital Medical Information Network Clinical Trials Registry, the Japanese clinical trial registry (registration number: UMIN000036490). This study was approved by the institutional review board of Kagawa University (approval number: 2018-110) and of each participating institution. In all the participating institutions, the requirement for patient consent was waived due to the retrospective nature of this study.

5ECPR was defined as resuscitation from cardiac arrest using VA-ECMO. To assess the indications, management, and neurological outcomes, SAVE-J II included consecutive patients’ ≥ 18 years of age who were admitted to the emergency department with OHCA between January 1, 2013 and December 31, 2018 and received ECPR. The current registry excludes patients with OHCA who were transferred to the participating institutions after receiving treatment in another hospital, patients with in‐hospital cardiac arrest (IHCA), and patients who declined to participate through family or other agents. For this study, the following data were collected: patient characteristics, prehospital information, information on hospital arrival, diagnosis and intervention, mechanical support information, time course, body temperature management, intensive care unit (ICU) information, and outcomes. The characteristics of the participating institutions, including inclusion and exclusion criteria and initial resuscitation management, are described in our previous paper.

6We selected patients from SAVE-J II who met the inclusion criterion of initiation of VA-ECMO before ICU admission. The exclusion criteria included non-cardiac conditions such as acute aortic dissection/aortic aneurysm, hypothermia, primary cerebral disorder, infection, drug intoxication, trauma, suffocation, and drowning. Hypothermia was defined as diagnosed by a physician or a body temperature at admission of less than 30 °C. Furthermore, we excluded patients who achieved return of spontaneous circulation (ROSC) on hospital arrival and at ECMO initiation. In addition, patients with unknown outcomes were excluded.

7The following patient data were collected from SAVE-J II: age, sex, incidence of witnessed cardiac arrest and bystander CPR, initial cardiac rhythm at the scene and on hospital arrival, cardiac arrest location, use of adrenaline and defibrillation, prehospital airway management, cardiac rhythm before ECMO initiation, treatment related factors, time course, cause of cardiac arrest, ROSC after hospital arrival, and ECMO information. Initial shockable rhythm was defined as VF, pulseless ventricular tachycardia, or rhythm for defibrillation in automated external defibrillator used by emergency medical staff. Location of cardiac arrest at ambulance was defined as patients who developed cardiac arrest after emergency medical staff (EMS) arrival with the presence of spontaneous circulation on initial EMS evaluation. ROSC was defined as at least one minute of continuing confirmation of pulsation. The length of hospital and ICU stay, in-hospital mortality, and neurological outcomes were also collected. Time from call ambulance to arrival was defined as time from emergency medical services call to hospital arrival, time from arrival to ECMO was defined as time from hospital arrival to establishment of ECMO support, time from call ambulance to ECMO was defined as time from emergency medical services call to establishment of ECMO support, estimated low flow time was defined as the time from cardiac arrest to the establishment of ECMO if the location of cardiac arrest was ambulance and the time from calling an ambulance to the establishment of ECMO if the location of cardiac arrest was other than ambulance. Cause of cardiac arrest was classified as acute coronary syndrome, arrhythmia, myopathy, myocarditis, other cardiac cause, pulmonary embolism, and other non-cardiac cause. Data on complications during ECPR were also collected. Cannula malposition was defined as cannulation requiring correct the position, or cannulation of wrong vessel such as arterial-arterial and veno-veno cannulation. Unsuccessful cannulation was defined as failure to complete cannulation. Cannulation-related bleeding included cannulation site bleeding and retroperitoneal hemorrhage requiring blood transfusion or surgical intervention/interventional radiology (IVR), and other forms of hemorrhage included intracerebral hemorrhage confirmed on computed tomography (CT), mediastinal hemorrhage, intra-abdominal organ hemorrhage, and gastrointestinal hemorrhage requiring blood transfusion or surgical intervention/IVR.

8The primary outcome was favorable neurological outcome, evaluated based on the cerebral performance category (CPC) at hospital discharge. A favorable outcome was defined as a CPC of 1 or 2, whereas an unfavorable outcome was defined as a CPC of 3, 4, or 5. The secondary outcomes were survival rate at hospital discharge and complications during ECPR.

9Descriptive statistics were used to summarize data on baseline characteristics, outcomes, and complications during ECPR. Categorical variables were counted and presented as proportions. Continuous variables were expressed as medians and interquartile ranges (IQRs). We compared baseline characteristics, outcomes, and complications according to favorable and unfavorable neurological outcomes or according to survival and mortality at hospital discharge using Mann–Whitney U test for continuous variables and Fisher’s exact test or chi-square test for categorical variables, as appropriate. Univariate and multivariable logistic regression analyses were performed for favorable neurological outcome and survival to hospital discharge. We fit logistic regression models with generalized estimating equation to account for patients clustering within the hospital. In the multivariable models, we adjusted for age, sex, witnessed cardiac arrest, bystander CPR, initial cardiac rhythm (shockable rhythm, pulseless electrical activity [PEA], asystole), location of cardiac arrest, and estimated low flow time. Next, we described the baseline characteristics, outcomes, and complications according to initial cardiac rhythm at the scene. We used locally estimated scatterplot smoothing (LOESS) curves with 95% confidence interval to illustrate the relationship between age and the proportion of favorable neurological outcome and survival rate and the relationship between estimated low flow time and the proportion of favorable neurological outcome and survival rate, in all patients and separately for each initial cardiac rhythm. We also used the LOESS curves to illustrate the relationship between age or estimated low flow time and complications. Statistical analyses were performed using R version 3.4.4. and JMP version 12 statistical software (SAS Institute, Cary, NC, USA). Missing data were not replaced or estimated.

10Of the 2157 adult patients with OHCA who received ECPR in SAVE-J II, 1644 were included for analysis (Fig. 1). The distribution of participating institutions and ECPR cases is shown in Additional file 1: Fig S2. Four of the 36 institutions had more than 100 cases of ECPR.

11Patient age was 18–93 years (median = 60 years). Overall, 84.6% of patients were men, 78.7% had witnessed cardiac arrest, and 58.2% received bystander CPR. The most common location of cardiac arrest was at home (39.9%). The initial cardiac rhythm at the scene was shockable rhythm, PEA, and asystole in 69.4%, 22.6%, and 8.0% of cases, respectively. Regarding prehospital intervention, 64.9% of patients were defibrillated, while 34.5% of patients were administered epinephrine. The median times (and IQRs) from call ambulance to arrival, arrival to ECMO, call ambulance to ECMO, and estimated low flow time were 32 (26–39), 22 (15–32), 56 (47–68), and 55 (45–66) minutes, respectively. ROSC was observed after hospital arrival in 79.0% of patients, and it was observed before ECMO pump on in 17.6% of patients and after ECMO pump on in 82.4% of patients (Table 1). The distributions of age and estimated low flow time are shown in Additional file 1: Figs. S3A and S3B.

12Details of the outcomes are shown in Table 2. Favorable neurological outcome at hospital discharge was observed in 14.1% of the 1644 patients. Survival rate at hospital discharge was 27.2%, and in-hospital mortality occurred at a median of 2 (1–4) days. ECPR cases every three months, neurological outcome, and survival to hospital discharge are plotted and shown in Additional file 1: Figs. S4A and S4B. Cannula malposition was observed in 4.9% of patients, and unsuccessful cannulation occurred in 0.7% of patients. Cannulation site bleeding and other types of hemorrhage were observed in 16.4% and 8.5% of patients, respectively. Overall, complications were observed during ECPR in 32.7% of patients.

13Comparisons of patient characteristics and complications according to favorable and unfavorable neurological outcomes and according to survival and mortality are shown in Additional file 1: Tables S1, S2 and S3, respectively. Unadjusted and adjusted associations with favorable neurological outcomes and with survival to hospital discharge are shown in Table 3. Multivariable analysis revealed that age, sex, initial shockable rhythm at the scene, and location of cardiac arrest were significantly associated with both favorable outcome and survival to hospital discharge (P < 0.01), and estimated low flow time was significantly associated with survival to hospital discharge (P < 0.001).

14Patient characteristics, outcomes, and complications according to initial cardiac rhythm at the scene are shown in Additional file 1: Tables S4 and S5. The proportions of favorable neurological outcome and survival rate at hospital discharge in terms of shockable rhythm, PEA, and asystole were 16.7%, 9.2%, and 3.9% and 32.0%, 18.5%, and 10.8%, respectively (Fig. 2A, B). The LOESS curve of all patients shows negative relationships between age and the proportion of favorable neurological outcome and survival rate and negative relationships between time and the proportion of favorable neurological outcome and survival rate (Fig. 3). The LOESS curve for each initial cardiac rhythm shows high variability of the proportion of favorable neurological outcome and survival rate (Fig. 3). At a short time interval, patients who had PEA as the initial cardiac rhythm at the scene had a higher survival rate and a lower proportion of favorable neurological outcome than at a longer time interval (Fig. 3B, D). Additional file 1: Fig. S5 shows the LOESS curve of complications. Age and time have weak relationships with complications during ECPR (Additional file 1: Figs S5A and S5E). High age and longtime are associated with high incidence of cannula malposition (Additional file 1: Figs. S5C and S5G) and hemorrhage (Additional file 1: Figs. S5D and S5H).

15Data on ECMO are shown in Additional file 1: Table S6. Vascular access was percutaneous in most patients (98.0%), whereas cut-down was performed in 2.0% of patients. The most common locations for cannulation were emergency department (64.2%) and catheterization room (35.6%). The median (and IQR) size of the ECMO cannula for arteries and veins were 16.5 (15.0–16.5) Fr and 21.0 (19.5–21.0) Fr, respectively. The most common method of cannula removal was the surgical method (59.3%).

16This study presents data of 1644 patients with OHCA, of which 69.4% had initial shockable rhythm and 14.1% had favorable neurological outcome at hospital discharge. The proportion of favorable neurological outcome was 16.7%, 9.2%, and 3.9% in patients with shockable rhythm, PEA, and asystole, respectively. The survival rate at hospital discharge was 32.0%, 18.5%, and 10.8% for patients with shockable rhythm, PEA, and asystole, respectively.

17This study has several strengths. First, to the best of our knowledge, this study includes the largest cohort of patients with OHCA who received ECPR. Although several large cohort studies of (> 200) patients with OHCA who received ECPR have been reported in France, four other European countries, Korea, and Japan, this study included 1.5 times as much participants as the previous study with the largest number (916) of patients and about half the number of patients in a recently published systematic review of ECPR for OHCA. Second, unlike one previous study that included only patients with shockable rhythm, this study included patients with all types of cardiac rhythm. Third, this study included four institutions where more than 100 ECPRs were performed. Hence, the data used in this study can be considered a continuous dataset from institutions with low and high volumes of ECPR cases. Finally, this study described complications that were not considered in previous large studies.

18In previous studies that used detailed clinical data of more than 100 patients with OHCA who received ECPR, the proportion of favorable neurological outcome at discharge was 6–39%. This study included different types of patients, and favorable neurological outcome was observed in 16.7% and 9.2% of patients who had shockable rhythm and PEA, respectively. In this study, the rate of witnessed cardiac arrest was approximately 88%, and bystander CPR was performed in 66% of patients. Further, the median estimated low flow time was 55 min in patients with PEA. The fact that the scatterplot curves of favorable neurological outcome and survival rate are different in terms of estimated low flow time between patients with PEA (i.e., low proportion of favorable neurological outcome despite relatively high survival rate over a short estimated low flow time) and patients with PEA strongly suggests heterogeneity. It is important to carefully determine suitable candidates for ECPR in initial PEA patients. In patients with initial asystole, ECPR may also be considered because, in select populations, the survival rate was greater than 10%.

19Age is strongly associated with poor outcomes over time, and it is an independent prognostic factor, as shown with the large dataset used in this study. In previous studies, it was reported that the proportion of favorable neurological outcome in patients over 75 years of age is 1.7–2.9%. It may be necessary to determine cutoffs according to institution, region, and other factors. As with age, the outcomes of estimated low flow time worsened with time. Optimal limited low-flow time, defined as time from CPR initiation to ECPR, was reported by Otani et al. to be 58 min, and it was reported in several other studies to be within 60 min. It is difficult to determine optimal limited low-flow time; therefore, it is important to minimize time to ECMO.

20Regarding complications associated with ECPR, the most common complication was bleeding, with the rates of cannulation site bleeding and other types of hemorrhage at 16.4% and 8.5%, respectively, but it was reported to be 8–70% in previous studies. Thirty percent of ECPRs were performed in relatively high-volume ECPR centers. This may be because the proportion of obese patients in Japan is lower than those in Western countries and because of the high availability of skilled emergency physicians and acute care surgeons in the emergency departments of high-volume institutions. Since ECPR-related complications are associated with poor outcome, it is necessary to shorten the time to ECMO and to initiate ECPR without complications. Training in cannulation technique may be considered in future training systems at high-volume ECPR centers to prevent and control complications.

21Broad application of ECPR to patients with OHCA was observed in this study. This seems to be fundamentally associated with the national health insurance system in Japan. Since medical bills are mainly paid by the Japanese government, physicians in charge do not pay attention to unpaid medical bills during ECPR initiation. Moreover, in Japan, trained emergency physicians initiate ECPR with no involvement from cardiovascular surgeons; in contrast, in the USA, cardiovascular surgeons are usually involved throughout the entire ECMO process. Although a recently reported cost-effectiveness study on ECPR for patients with OHCA concluded that ECPR is a robust and economically acceptable resuscitative strategy after considering all parameters, it is difficult to establish uniform standard inclusion criteria. It may be possible to expand the indications for ECPR, but any expansion must be balanced by considering factors such as cost, complications, religious views, and local area-specific conditions. The large cohort dataset used in this study contains details of ICU management, CT data, cost, and socioeconomic status; therefore, a statistically confirmed study will provide robust conclusions and further hypotheses on ECPR.

22This study has several limitations. First, this was a retrospective study with variation of inclusion criteria in each participating institution; therefore, we reported the inclusion and exclusion criteria of all 36 SAVE-J II hospitals in another paper. Second, there was no control group (i.e., a patient group that did not receive ECPR). Third, data on long-term outcome was not obtained. Fourth, confounders of time course, such as number of ROSC and total ROSC time until ECMO establishment, were not obtained from this dataset. Due to lack of various dataset, estimated low flow time would be different actual low flow time. Fifth, complications were a composite outcome, and specific complications were not assessed.

23In this large cohort, data on the ECPR of 1644 patients with OHCA show that the proportion of favorable neurological outcomes at hospital discharge was 14.1% and survival rate at hospital discharge was 27.2%. Complications were observed during ECPR in 32.7% of patients, and the most common complication was bleeding, with the rates of cannulation site bleeding and other types of hemorrhage at 16.4% and 8.5%, respectively.