Neurological outcomes and duration from cardiac arrest to the initiation of extracorporeal membrane oxygenation in patients with out-of-hospital cardiac arrest: a retrospective study

Extracorporeal cardiopulmonary resuscitation (ECPR) is a novel CPR technique that uses veno-arterial extracorporeal membrane oxygenation (ECMO). Some observational studies have shown that, compared to conventional CPR, ECPR has a beneficial effect on survival and neurological outcomes in patients with both in-hospital and out-of-hospital cardiac arrest (CA) [1‐4]. Based on the 2015 American Heart Association guideline, ECPR is given a class IIb recommendation if it is performed rapidly in a specialized center, and should be used in patients with a potentially reversible etiology. Thus, the use of ECPR is increasing rapidly in emergency departments for patients with CA as an alternative for resuscitation.

However, the definite indication for ECPR has not been established, especially in patients with out-of-hospital cardiac arrest (OHCA) [5]. It is obvious that prolonged cardiac arrest without the return of spontaneous circulation (ROSC) precludes the possibility of ECPR. However, the acceptable duration from CA to the initiation of ECMO to achieve good neurological outcomes has not been clarified.

We sought to investigate the relationship between neurological outcomes and duration from CA to the initiation of ECMO in patients with OHCA treated with ECPR, and to determine the ideal time at which ECPR should be performed in these patients.

During the time period in which this study was conducted, 3451 patients experienced OHCA. Of these, 80 patients were treated with ECPR at our institution. One patient was excluded because he was 17 years old; therefore, 79 patients were finally included in the study (Fig. 1).

Table 1 shows the baseline characteristics and post-CA care of all patients and both the good and poor neurological outcome groups. The median age of the patients was 59 years (IQR, 48–64); 66 patients (83%) were men. Fifty-nine patients had an initial shockable rhythm. Twenty-one patients (27%) recovered spontaneous circulation transiently prior to admission to the hospital, but went into cardiac arrest again before arrival at hospital. Seventeen patients (22%) survived until hospital discharge. Of these, 11 cases (14%) had good neurological outcomes (CPC score, 1 or 2). Sixty-two patients (79%) died in the hospital; therefore, 68 patients (86%) had poor neurological outcomes (CPC score, 3–5). The proportion of patients whose OHCA was witnessed by EMS personnel was significantly higher among patients with good neurological outcomes than those with poor outcomes (73% vs. 19%; p < 0.01). Additionally, the proportion of patients with bystander CPR was significantly higher among patients with good neurological outcomes (91% vs. 53%; p = 0.02).

The median CA-to-arrival time was significantly shorter among patients with good neurological outcomes than those with poor outcomes (17 min [IQR, 3–25] vs. 28 min [IQR, 22–33]; p = 0.03). The median arrival-to-ECMO time was comparable between the groups (19 min [IQR, 14–31] vs. 20 min [IQR, 14–25]; p = 0.03). The median collapse-to-ECMO time was significantly shorter among patients with good neurological outcomes (33 min [IQR, 27–50] vs. 46 min [IQR, 42–56]; p = 0.03) (Table 2).

There are twenty-one patients who did not have an initial shockable rhythm, and thus qualified for exclusion based on our criteria for ECPR. This group was limited to patients with good neurological outcomes, and their diseases tended to be specific such as acute pulmonary embolism, and hypoxia due to acute respiratory failure (acute heart failure, bronchiolar asthma attack, and others). These patients were given epinephrine intravenously by EMS personnel, and consequently had ROSC at least once.

After controlling for potential confounders, we found that the adjusted odds ratio of CA-to-ECMO time for good neurological outcomes was 0.92 (95% confidence interval: 0.87–0.98, p < 0.01). In this model, male sex and transient ROSC before hospital arrival were also significantly associated with good neurological outcomes (Table 3). The adjusted odds ratio of CA-to-arrival adjusted by the same confounders was 0.91 (95% confidence interval; 0.85–0.97, p < 0.01).

Figure 2 shows the dynamic probability of survival to hospital discharge with good neurological outcome by CA-to-ECMO time. The rate of survival decreased abruptly when the CA-to-ECMO time exceeded 40 min wherein the rate of survival reduced to approximately 14%. Drawing a receiver operating characteristic curve of CA-to-ECMO time for good neurological outcome patients, we found that the optimal cutoff was 40.00 (sensitivity: 0.73, specificity 0.81), and the area under the curve was 0.71.

Three patients had a good neurological outcome at hospital discharge, although their CA-to-ECMO time was longer than 40 min. These three patients had common characteristics: they had transient ROSC before hospital arrival and an initial shockable rhythm.

In this study, CA-to-ECMO time was significantly shorter among patients with good neurological outcomes, and significantly associated with good neurological outcomes at hospital discharge. In addition, the probability of survival with good neurological outcome decreased when the CA-to-ECMO time exceeded 40 min.

Several studies have shown that ECPR in OHCA patients may be a good alternative to conventional CPR [3, 4, 6]. The reported survival rate with minimal disability ranged widely from 5%–30% [3, 4, 11‐14]. However, the human and economic cost of ECMO management cannot be negligible. ECPR for patients with OHCA could reduce the severe burden on the hospital’s emergency department; however, to our knowledge, no study has investigated the cost effectiveness of ECPR for patients with OHCA. Adequate patient selection with strict indication criteria is essential if ECPR becomes a part of the standard care of patients with OHCA.

The indication for ECPR for patients with OHCA should include several factors including age, the presence of witnesses, bystander CPR, previous medical status, arterial pH, and serum lactate concentration at hospital admission, shockable cardiac rhythm, and transitory ROSC [10, 15]. However, the duration of CPR before the initiation of ECMO is a key factor for good neurological outcomes in patients with OHCA treated with ECPR. Moreover, in our study, the CA-to-ECMO time was shown to be an independent prognostic factor for good neurological outcomes. Therefore, the upper limit of CA-to-ECMO time should be inevitably included in the indication for ECPR for patients with OHCA. In inpatients with -hospital cardiac arrest (IHCA), Chen et al. reported that prolonged resuscitation more than 60 min before the initiation of ECMO was associated with poor neurological outcomes [16]. The prognosis of patients with IHCA treated with ECPR tended to be better than that of patients with OHCA [11, 17]. This is due to situational differences between patients with IHCA and those with OHCA. In patients with IHCA, CPR is more likely to be started immediately; furthermore, the quality of CPR and advanced life support is more secure, and the duration before the achievement of ECMO flow is shorter. Therefore, the ideal time criteria for ECPR should be different between patients with OHCA and those with IHCA. However, studies that discuss the time criteria in patients with OHCA are scarce. Hase et al. reported that the optimal cutoff duration of CPR for a good outcome was 45 min in patients with OHCA treated with ECPR [18]. In the present study, there was a large difference in the rate of survival to hospital discharge with good neurological outcome between the patients with a CA-to-ECMO time within 40 min and those whose time was over 40 min. The survival rate with good neurological outcome was about 35% when the duration of CPR was within 40 min; however, it decreased to about 15% if CPR was performed for over 40 min. This is similar to the result of the study by Hase et al. The acceptable survival rate to make ECPR in patients with OHCA cost effective is not clear. However, we think it is reasonable to regard the point at which there is a large difference in the survival rate as the ideal time. Based on the present study, the time limit of the duration of CPR before the initiation of ECMO might be around 40 min. We should consider ECPR in patients with OHCA if they are relatively young, have a witness and no terminal disease, and the initiation of ECMO is presumed to be within this time period. In other words, we should make every effort in both pre-hospital and in-hospital care to establish ECMO within 40 min from the onset of cardiac arrest. In our study, arrival-to-ECMO time was a median of 19 min, and about 10 min in faster cases. The acceptable duration prior to hospital admission was therefore limited and at most 30 min.

However, it is also true that time alone is not an absolute criterion, and some patients have good neurological outcome regardless of prolonged CPR. It is very important to find prognostic factors that predict good neurological outcomes in patients with prolonged CPR. In our study, three patients had good neurological outcomes, although their CA-to-ECMO time exceeded 40 min, and they all recovered spontaneous circulation transiently before hospital arrival. In these patients, cerebral perfusion could be slightly maintained around the duration of ROSC, and a longer time might be permissible. Recently, some physiological parameters specific to the individual patient during CPR, such as regional saturation of oxygen in the brain, have been reported to be able to predict neurological outcome in patients with OHCA [19, 20]. In the future, such parameters could be used to indicate patients with the potential for neurological recovery, regardless of prolonged CPR duration.

There are several limitations to this study. First, it was a retrospective study with a small sample size. Second, this study lacks some information about post-cardiac arrest care, such as transfusion, ventilator setting, and therapies for infectious complications. Additionally, explanatory variables in the logistic regression analysis were limited because of the small sample size. Therefore, some potential prognostic factors could not be adjusted. Third, neurological outcomes were evaluated only at the time of hospital discharge. CPC scores and/or survival rate after hospital discharge at 3 months or 6 months, for example, were not evaluated. Finally, the outcome of ECPR for patients with OHCA is affected by the system of emergent medical services and pre-hospital treatment. It is well known than manual CPR is difficult to perform in a moving ambulance, and mechanical CPR in this situation could provide different results with a varying optimal CPR duration before ECPR initiation. Therefore, the relationship between the prognosis and the duration of CPR before the initiation of ECMO might be different in other settings than ours. A well-designed, prospective, multicenter study with a large sample size is needed to determine the standard indication for ECPR for patients with OHCA.

The survival rate with good neurological outcome decreased when the duration until the initiation of ECMO exceeded 40 min in patients with OHCA treated with ECPR. The indication for ECPR in patients with OHCA should include a time criterion. Based on the present study, this ideal time might be within 40 min from the onset of CA.