Since the original description by Bartlett et al.1 in 1976, indications for the use of extracorporeal membrane oxygenation (ECMO) have greatly expanded. One such indication has been the use of ECMO as a supportive therapy for manual cardiopulmonary resuscitation (CPR) in the presence of cardiac arrest in adults, a methodology also known as E-CPR. Much has been written about the use of ECMO within this context but little more than anecdotal case reports and brief clinical series are available in the scientific literature, and none recommend or discourage its deployment based on evidence-based criteria. Our review of the literature and meta-analysis (MA) had the primary objective of finding predictors of mortality for adult patients in these extreme conditions.
After a review and exemption by the institutional review board, we performed an electronic database search of PubMed from January 1990 to March 2007 with the following search terms: ECMO, extracorporeal life support (ECLS), heart arrest, resuscitation, cardiopulmonary resuscitation (CPR) and using the limit function for “humans.” One reviewer (A.Y.) read all 141 abstracts retrieved during the original search to identify articles appropriate for full text evaluation. When no abstract was available, the full article was reviewed. All selected articles, a total of 20 publications, that met the pre-established criteria were included.
Once an article was accepted for inclusion in the MA, individual patients were identified from each publication, and a unique identifier was assigned. Data on a number of variables (demographic, hospital course, and survival to discharge) were tracked and recorded for each individually identified patient.
All case reports and observational studies reporting on adults (older than 18 years) treated with ECMO after a witnessed cardiac arrest were considered and individually evaluated.
We rigorously defined cardiac arrest as the need for chest compressions or defibrillation being administered for a nonperfusing cardiac rhythm.
Publications in languages other than English were excluded. Studies reporting out-of-hospital cardiac arrest and/or uncertainty regarding onset time of the cardiac arrest were not included. Studies describing the use of ECMO as an elective therapy for patients with deteriorating circulation but not in cardiac arrest were also excluded. To achieve further accuracy in our estimations, individual authors of articles that fell short of our inclusion criteria were contacted to request a full set of data for each patient. If, after contacting the author, the individual patient data was still insufficient to validate our statistical analysis, then those studies were dismissed.
Information on the authors, institution, population, and dates were checked to identify duplicate publications. Duplicated patients in consecutive reports from the same institution or author were excluded. Patient data not reported in the original article but available in the duplicated article was included without duplicating patients. Our search yielded 141 titles, leading to 68 abstracts considered appropriate for full text evaluation, of which 20 articles were included in the MA (11 clinical series and nine case reports).2–21 The most frequent reasons for exclusion were failing the predetermined inclusion criteria, patients younger than 18 years, and inability to obtain detailed patient information. For a bibliographic search explanation see Figure 1.
Unadjusted odds ratios (OR) of dying before hospital discharge were calculated using logistic regression with single independent variables. Those independent variables associated with hospital mortality were included in the development of a multiple logistic regression models (SAS System Software package). When needed, Student's t test was used for comparing mean values among groups.
Rationale for Selection and Coding of Data
Because of the number of nonstandardized diagnoses reported in the included series and to facilitate analysis, we created a classification system with 15 diagnostic groups. Table 1 shows diagnoses and survival for all groups with n > 1 patient. Single case reports, although included in the MA, are not included in this table. Sex, ECMO relationship to a surgical procedure, ECMO deployment under 30′, ECMO modality (venous/venous or arterial/venous), presence of any complication, and survival to discharge were collected as dichotomous values. Age, weight, duration of CPR, ECMO timing, setup time, cannulation technique, cannulation site, setting of ECMO initiation, length on ECMO, and outcome were entered as continuous values. A list of all the variables collected and their data format is available in Table 2.
Variables for which information was lacking in a large proportion of the patients were not included in the MA. These included technique used for left heart decompression, reason for withdrawal of ECMO support, and ultimate cause of death.
Data from the 135 individually identified patients described in the included articles were abstracted into the MA (male:female = 1.6:1). All the included patients had in common treatment with E-CPR as a rescue modality after suffering a witnessed cardiac arrest.
Overall survival to hospital discharge for patients who underwent ECMO support after cardiac arrest was 40% (54 of 135). Survival was higher, although not enough to be statistically significant for females (46.5%).
Sixty-seven patients (49.6%) were successfully weaned from ECMO support, but only 54 of them were discharged alive from the hospital, with an attrition rate of 19.4% (13 patients) between ECMO discontinuation and discharge. Patients who were not discharged home after ECMO withdrawal included two patients on whom ECMO was used as a bridge to transplant and both were discharged alive. Four cases, where ECMO was used as a bridge to a ventricular assist device, had a 50% discharge home. The rest of the patients in the group either died while on mechanical support or had ECMO withdrawn because of major complications or family request.
Manual CPR in the 102 patients for whom data were available had a mean duration of 40 minutes (range 1′–180′). Some patients were placed on ECMO after a short period of CPR because of low cardiac output, despite recovery of cardiac rhythm. Although not statistically significant, there was a trend toward better survival for those who had CPR for <30′ before ECMO was instituted (OR 1.9; 95% CL, 0.9–4.2). In 21 patients, ECMO was initiated while full manual CPR was still in place, and 10 patients survived to hospital discharge (47.6%).
Median ECMO run was 54 hours (range 0–3881), with females having a significantly shorter run (p = 0.04). For the purpose of this analysis, duration of ECMO support was measured in days, and patients were divided into four equivalent groups. When evaluated in this fashion, the group sustained by ECMO between 0.875 and 2.3 days displayed a trend toward higher survival (61%) with lower OR for mortality (OR 0.2; 95% CL, 0.07–0.6) compared with the rest (Table 3).
Median age was 56 years (range 18–83). Age group was also a mortality risk. Patients for whom age was available (n = 133) were divided into four similar groups. Compared with group 1 (age 17–40), OR for mortality was significantly higher for group 2 (41–56 years) (OR 2.9; 95% CL, 1.6–8.2) and group 4 (older than 67 years) (OR 3.4; 95% CL, 1.2–9.7) and higher, although not significantly, for group 3 (Table 4).
The overall occurrence of complications was not well described in the articles included in our MA. Prevalence of major clinical complications was specifically mentioned in 15 of 135 patients, and absence of any complications was specifically addressed in only six patients. The existence of complications of any type was not addressed at all in the rest of the included patients (n = 114), seriously limiting our ability to draw any conclusions on that topic. According to data from studies other than those included in our MA, the occurrence of complications as a result of the use of E-CPR is a relatively common phenomenon. Frequent blood and blood products transfusions (47%), multiorgan system failure (30.4%), and an incidence of neurologic events of variable severity ranging between 5.6% and 17.5% seem to be the most commonly described complications.22,23
Information on the hospital setting at which E-CPR was initiated was available only for 82 patients. In most cases (50 of 82), E-CPR was started in the intensive care unit with only 12 survivors to discharge (24%). A second group of patients (n = 23) suffering from myocardial infarction or pulmonary embolization required emergency E-CPR to be instituted in the catheterization laboratory, and 10 of those patients survived to discharge (43.4%). A small group (n = 4) had ECMO implemented in the operating room with no survivors. Finally, a second small group (n = 5) needed E-CPR in the emergency department setting to treat deep hypothermia, acute myocardial infarction, and left internal mammary graft avulsion. Three patients (60%) survived to hospital discharge in this group. No significant conclusion could be drawn from the information on hospital setting because of small size of the samples and the variety of the initial diagnosis.
The information on the site of cannulation for ECMO and its relationship to survival was very limited in the majority of articles reviewed for this MA, and therefore no firm conclusion could be drawn. The only piece of solid information was that survival to discharge for 45 patients connected to ECMO through femoral cannulation was 55.5% (25 patients). For the purpose of comparison, we may speculate that all patients treated with E-CPR after postcardiotomy arrest (50% survival) were indeed cannulated through the sternotomy. If that was the case, the relationship between survival and cannulation site might be irrelevant.
The relationship between mortality and the diagnosis that originated the cardiac arrest was difficult to establish because of the large number of diagnoses and the low number of patients for many of those diagnoses. In the three largest groups, acute myocardial infarction (n = 46), pulmonary embolus (n = 21), and postcardiotomy arrest (n = 24), the rates of survival to discharge were 36.9%, 57%, and 50%, respectively.
Lower hospital survival (25% and 33%, respectively) was observed in a group of patients with cardiac arrest due to left ventricular rupture (n = 4) and a heterogeneous group of patients (n = 6) that included arrest due to nontraumatic pulmonary causes (adult respiratory distress syndrome, Werner granulomatosis, and Hanta virus). Except for cardiac arrest after a traumatic injury to the lungs (n = 6) with a survival of 50%, the results of E-CPR were rather dismal for other small groups. Viral cardiomyopathy (n = 7), acute heart transplant failure (n = 5), and arrest due to refractory arrhythmias (n = 3) had no survivors.
Meta-analyses based on observational studies are frequently unable to extract the patient's data without the inherent biases of the original studies, (i.e., the retrospective design in all the studies included in our MA).24 Another common problem affecting this form of MA is a phenomenon called ecological bias, a type of statistical error by which patients included in the MA may not be representative of the population subjected to similar treatment but not included in the MA.24 To assess our MA for ecological bias, we reviewed the results among 165 adult patients with cardiac arrest treated with E-CPR from studies not included in our MA and found an overall survival rate similar to ours (37.6%).22,25,26
Authors often reported on complications as a simple rate and without further specification or not at all. Studies included in our analysis had limited information on the specific complications, and therefore their occurrence should be looked as an underestimation.
Historically, survival for witnessed arrests has benefited from significantly better results than for nonwitnessed arrests (25% vs. 7%, respectively), likely due to a swift implementation of manual CPR.27 Overall survival to hospital discharge after cardiac arrest is significantly low (22.4%) even in series, showing a high initial CPR survival (48.3%).28 Because of these poor results, mechanical extracorporeal ventilation and support of the perfusion of end organs during cardiac arrest (E-CPR) have been advocated for improving hospital survival.
Early animal research indicated that the use of some form of extracorporeal circulation had the potential to become an effective intervention for resuscitation when conventional techniques of precordial compression and external defibrillation fail to reverse cardiac arrest.29
Similarly, the initial clinical results of the use of E-CPR in small groups of patients with reversible diseases were encouraging.30 Results of E-CPR were particularly hopeful in the pediatric arena, where the experience in the use of ECMO for rescue therapy started earlier.31
Although the use of E-CPR in adults has gained some momentum, its implementation has been based on personal experiences and the availability of vast technical and human resources, rather than evidence-based criteria.
In our review, duration of manual CPR before implementation of ECMO seems to behave as a limiting factor regarding the measured outcome. Our findings, although not statistically significant, show a negative trend on hospital survival as the length of manual CPR extends beyond the initial 30 minutes. A number of authors in the pediatric E-CPR literature have speculated that an expeditious institution of ECMO should shorten the duration of manual CPR consequently influencing clinical outcomes; however, none of those studies found a significant difference in the duration of manual CPR between survivors and nonsurvivors.32,33 Our findings regarding duration of manual CPR seem to be in accordance with those reported by Chen et al.25 and is not included in our MA. They reported a statistically significant difference in survival (p < 0.05) on adults treated with E-CPR depending on whether the manual portion of the CPR remained below or went above 60 minutes (survival rate of 48% and 11.5%, respectively).
Age stratification demonstrated a trend of lower mortality in the youngest adults. We might speculate that the relationship between age and mortality is likely the result of the relationship between the age of the patient and the specific diagnosis leading toward the cardiac arrest event, although we present no significant evidence on that regard. Although we verified a trend favoring hospital survival among a younger group of patients (mean 51.2 years), no statistically significant difference was evident (p = 0.09) when we compared them with those who did not survive to hospital discharge (mean 56 years). These findings are similar to those reported by Saito et al.26 who found no statistical difference (p = 0.86) in the survival rates of patients younger and older than 75 years (44.3% and 41.7%, respectively) treated with E-CPR. A larger sample size may be needed to observe the effects of age over mortality.
Mortality for those with cardiac arrest due to heart transplant failure (n = 5) was 100%. Although lacking statistical value because of the small number of cases, this outcome was surprisingly worse than we expected. Although the original studies did not expand on the subject, one potential explanation for the lack of recovery reserve seen in failed cardiac grafts could be the presence of pulmonary hypertension or the increasingly common use of marginal organs. Other authors have found E-CPR as a beneficial solution for failing cardiac grafts, although in the large majority of patients, it was instituted as a preventive rather than as an emergency measure.34
A possible critique to this type of study is the long timespan over which the included studies were published (1990–2007) and the ensuing changes in patient management and technology. To assess the impact of improvements in technology and critical care management, we divided the identifiable population from the included studies into two groups. Those treated with E-CPR between January 1990 and December 1998 (group I) with a total of 89 cases and those treated with E-CPR between January 1999 and March 2007 (group II), which included 41 cases. Survival to discharge was 30% for group I and 59% for group II (p < 0.001), confirming that recent technological and management changes have had a strong influence on hospital discharge.
The use of E-CPR in the face of witnessed cardiopulmonary arrest in adults might increase the chances of leaving the hospital alive up to 40%. Survival seems to be improved when E-CPR is used in younger patients for shorter periods of time and after an expeditious implementation. Instances of cardiopulmonary arrest treated with E-CPR with the best chances of hospital survival were pulmonary emboli, trauma, and postcardiotomy arrest. Neurologic sequelae and other major complications, although likely high, are poorly described in the reviewed literature. Recommendations on how to improve the survival to hospital discharge for patients undergoing E-CPR should focus on shortening of ECMO deployment time and refining neuroprotective strategies. A dedicated team of experts available on a 24/7 basis and preprimed equipment on stand-by seem to answer the deployment issue.35 A rapid cooling protocol for patients undergoing manual CPR seems to be the best neuroprotective strategy available until ECMO can be initiated.36,37 Before embarking on the costly task of instituting an ECMO program for adults, healthcare systems should carefully evaluate the comparative effectiveness of E-CPR compared with manual CPR alone.
1. Bartlett RH, Gazzaniga AB, Jefferies MR, et al
: Extracorporeal membrane oxygenation (ECMO) cardiopulmonary support in infancy. Trans Am Soc Artif Intern Organs
22: 80–93, 1976.
2. Reyftmann L, Morau E, Dechaud H, et al
: Extracorporeal membrane oxygenation therapy for circulatory arrest due to postpartum hemorrhage. Obstet Gynecol
107: 511–514, 2006.
3. Krumnikl JJ, Toller WG, Prenner G, Metzler H: Beneficial outcome after prostaglandin-induced post-partum cardiac arrest using levosimendan and extracorporeal membrane oxygenation. Acta Anaesthesiol Scand
50: 768–770, 2006.
4. Tiruvolpati R, Balasubramanian SK, Khoshbin E, et al
: Successful use of venovenous extracorporeal membrane oxygenation in accidental hypothermic cardiac arrest. ASAIO J
51: 474–476, 2005.
5. Moser B, Voelckel W, Gardetto A, et al
: One night in a snowbank: A case report of severe hypothermia and cardiac arrest. Resuscitation
65: 365–368, 2005.
6. Wang SS, Ko WJ, Chen YS, et al
: Mechanical bridge with extracorporeal membrane oxygenation and ventricular assist device to heart transplantation. Artif Organs
25: 599–602, 2001.
7. Younger JG, Schreiner RJ, Swaniker F, et al
: Extracorporeal resuscitation for cardiac arrest. Acad Emerg Med
6: 677–678, 1999.
8. Crowley MR, Katz RW, Kessler R, et al
: Successful treatment of adults with severe Hantavirus pulmonary syndrome with extracorporeal membrane oxygenation. Crit Care Med
26: 409–414, 1998.
9. Mair P, Hoermann C, Moertl M, et al
: Percutaneous venoarterial extracorporeal membrane oxygenation for emergency mechanical circulatory support. Resuscitation
33: 29–34, 1996.
10. Kurose M, Okamoto K, Sato T, et al
: The determinant of severe cerebral dysfunction in patients undergoing emergency extracorporeal life support following cardiopulmonary resuscitation. Resuscitation
30: 15–20, 1995.
11. Kawahito K, Ino T, Adachi H, et al
: Heparin coated percutaneous cardiopulmonary support for the treatment of circulatory collapse after cardiac surgery. ASAIO J
40: 972–976, 1994.
12. Kurose M, Okamoto K, Sato T, et al
: Extracorporeal life support for patients undergoing prolonged external cardiac massage. Resuscitation
25: 35–40, 1993.
13. Maggio P, Hemmila M, Haft J, Bartlett R: Extracorporeal life support for massive pulmonary embolism. J Trauma
62: 570–576, 2007.
14. Willms DC, Atkins PJ, Dembitsky WP, et al
: Analysis of clinical trends in a program of emergent ECLS for cardiovascular collapse. ASAIO J
43: 65–68, 1997.
15. Tsai SK, Wang MJ, Ko WJ, Wang SJ: Emergent bedside transesophageal echocardiography in the resuscitation of sudden cardiac arrest after tricuspid inflow obstruction and pulmonary embolism. Anesth Analg
89: 1406–1408, 1999.
16. Kolla S, Lee WA, Hirschl RB, Bartlett RH: Extracorporeal life support for cardiovascular support in adults. ASAIO J
42: 809–819, 1996.
17. Szocik J, Rudich S, Csete M: ECMO resuscitation after massive pulmonary embolism during liver transplantation. Anesthesiology
97: 763–764, 2002.
18. Fujimoto K, Kawahito K, Yamaguchi A, et al
: Percutaneous extracorporeal life support for treatment of fatal mechanical complications associated with acute myocardial infarction. Artif Organs
25: 1000–1003, 2001.
19. Hsieh PC, Wang S, Ko W, et al
: Successful resuscitation of acute massive pulmonary embolism with extracorporeal membrane oxygenation and open embolectomy. Ann Thorac Surg
72: 266–267, 2001.
20. Kawahito K, Murata S, Adachi H, et al
: Resuscitation and circulatory support using extracorporeal membrane oxygenation for fulminant pulmonary embolism. Artif Organs
24: 427–430, 2000.
21. Mabuchi N, Takasu H, Ito S, et al
: Successful extracorporeal lung assist (ECLA) for a patient with severe asthma and cardiac arrest. Clin Intensive Care
2: 292–294, 1991.
22. Mégarbane B, Leprince P, Deye N: Emergency feasibility in medical intensive care unit of extracorporeal life support for refractory cardiac arrest. Intensive Care Med
33: 758–764, 2007.
23. Bowen FW, Carboni AF, O'Hara ML: Application of “double bridge mechanical” resuscitation for profound cardiogenic shock leading to cardiac transplantation. Ann Thorac Surg
72: 86–90, 2001.
24. Lau J, Ioannidis JP, Schmid CH: Summing up evidence: One answer is not always enough. Lancet
351: 123–127, 1998.
25. Chen YS, Chao A, Yu HY, et al
: Analysis and results of prolonged resuscitation in cardiac arrest patients rescued by extracorporeal membrane oxygenation. J Am Coll Cardiol
41: 197–203, 2003.
26. Saito S, Nakatani T, Kobayashi J, et al
: Is extracorporeal life support contraindicated in elderly patients? Ann Thorac Surg
83: 140–145, 2007.
27. Dumot JA, Burval DJ, Sprung J, et al
: Outcome of adult cardiopulmonary resuscitations at a tertiary referral center including results of “limited” resuscitations. Arch Intern Med
161: 1751–1758, 2001.
28. Brindley PG, Markland DM, Mayers I, Kutsogiannis DJ: Predictors of survival following in-hospital adult cardiopulmonary resuscitation. Can Med Assoc J
167: 343–348, 2002.
29. Gazmuri RJ, Weil MH, von Planta M, et al
: Cardiac resuscitation by extracorporeal circulation after failure of conventional CPR. J Lab Clin Med
118: 65–73, 1991.
30. Younger JG, Schreiner RJ, Swaniker F, et al
: Extracorporeal resuscitation of cardiac arrest. Acad Emerg Med
6: 700–707, 1999.
31. Zaritsky A, Nadkarni V, Hazinski MF, et al
: Recommended guidelines for uniform reporting of pediatric advanced life support: The Pediatric Utstein Style. A statement for healthcare professionals from a task force of the American Academy of Pediatrics, the American Heart Association, and the European Resuscitation Council. Resuscitation
30: 95–115, 1995.
32. Allan CK, Thiagarajan RR, Armsby LR, et al
: Emergent use of extracorporeal membrane oxygenation during pediatric cardiac catheterization. Pediatr Crit Care Med
7: 212–219, 2006.
33. Dalton HJ, Siewers RD, Fuhrman BP, et al
: Extracorporeal membrane oxygenation for cardiac rescue in children with severe myocardial dysfunction. Crit Care Med
21: 1020–1028, 1993.
34. Leprince P, Aubert S, Bonnet N, et al
: Peripheral extracorporeal membrane oxygenation (ECMO) in patients with post-transplant cardiac graft failure. Transplantation Proc
37: 2879–2880, 2005.
35. Duncan BW, Ibrahim AE, Hraska V, et al
: Use of rapid-deployment extracorporeal membrane oxygenation for the resuscitation of pediatric patients with heart disease after cardiac arrest. J Thorac Cardiovasc Surg
116: 305–309, 1998.
36. Barone FC, Feuerstein GZ, White RF: Brain cooling during transient focal ischemia provides complete neuroprotection. Neurosci Biobehav Rev
21: 31–44, 1997.
37. Wang H, Olivero W, Lanzino G, et al
: Rapid and selective cerebral hypothermia achieved using a cooling helmet. J Neurosurg
100: 272–277, 2004.