International Journal of Cardiology 177 (2014) 1031–1035
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Developing a risk prediction model for survival to discharge in cardiac arrest patients who undergo extracorporeal membrane oxygenation Sung Bum Park a,b, Jeong Hoon Yang a,c,⁎, Taek Kyu Park c, Yang Hyun Cho d, Kiick Sung d, Chi Ryang Chung a, Chi Min Park a, Kyeongman Jeon a,e, Young Bin Song c, Joo-Yong Hahn c, Jin-Ho Choi c, Seung-Hyuk Choi c, Hyeon-Cheol Gwon c, Gee Young Suh a,e a
Department of Critical Care Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, South Korea Department of Medicine, Korean Armed Forces Capital Hospital, Seongnam, South Korea c Division of Cardiology, Department of Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, South Korea d Department of Thoracic and Cardiovascular Surgery, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, South Korea e Division of Pulmonary and Critical Care Medicine, Department of Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, South Korea b
a r t i c l e
i n f o
Article history: Received 16 June 2014 Accepted 23 September 2014 Available online 2 October 2014 Keywords: Cardiopulmonary resuscitation Extracorporeal membrane oxygenation Predictor
a b s t r a c t Background: Limited data are available on a risk model for survival to discharge after extracorporeal membrane oxygenation (ECMO)-assisted cardiopulmonary resuscitation (ECPR). We aimed to develop a risk prediction model for survival to discharge in cardiac arrest patients who undergo ECMO. Methods: Between January 2004 and December 2012, 505 patients supported by ECMO were enrolled in a retrospective, observational registry. Among those, we studied 152 adult patients with in-hospital cardiac arrest. The primary outcome was survival to discharge. A new predictive scoring system, named the ECPR score, was developed to monitor survival to discharge using the β coefficients of prognostic factors from the logistic model, which were internally validated. Results: In-hospital death occurred in 104 patients (68.4%). In multivariate logistic regression, age ≤ 66, shockable arrest rhythm, CPR to ECMO pump-on time ≤ 38 min, post-ECMO arterial pulse pressure N 24 mm Hg, and post-ECMO Sequential Organ Failure Assessment score ≤ 14 were independent predictors for survival to discharge. Survival to discharge was predicted by the ECPR score with a c-statistics of 0.8595 (95% confidence interval [CI], 0.80–0.92; p b 0.001) which was similar to the c-statistics obtained from internal validation (training vs. test set; c-statistics, 0.86 vs. 0.86005; 95% CI, 0.80–0.92 vs. 0.77–0.94). The sensitivity and specificity for prediction of survival to discharge were 89.6% and 75.0%, respectively, when the ECPR score was N 10. Conclusions: The new risk prediction model might be helpful for decisions about ECPR management and could provide better information regarding early prognosis. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Cardiopulmonary resuscitation (CPR) has traditionally been the treatment of choice for cardiopulmonary collapse and it has been developed and modified to achieve better survival. Unfortunately, previous studies revealed a low survival to discharge rate that ranged from 7 to 26% [1–4], which declines rapidly if the duration of CPR exceeds 10 min [5–7]. Recently, extracorporeal cardiopulmonary resuscitation (ECPR) with a portable cardiopulmonary bypass system has been increasingly utilized to supply oxygenated blood in the absence of spontaneous cardiac pumping [8,9]. Several observation studies have shown
an improved survival rate compared to conventional CPR in patients with in-hospital cardiac arrest [10,11]. However, to date, there have been no guidelines for the application of ECMO in the setting of CPR and the prognostic factors under ECPR have not been fully established in patients with in-hospital cardiac arrest. Therefore, we investigated the predictive risk factors for in-hospital death and developed a risk prediction model to better inform physicians of suitable candidates and their expected survival rate in cardiac arrest patients undergoing ECPR. 2. Methods 2.1. Study population
⁎ Corresponding author at: Department of Critical Care Medicine, Division of Cardiology and Department of Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, 81 Irwon-ro, Gangnam-gu, Seoul 135-710, South Korea. Tel.: +82 2 3410 1768; fax: +82 2 2148 7088. E-mail address: [email protected] (J.H. Yang).
http://dx.doi.org/10.1016/j.ijcard.2014.09.124 0167-5273/© 2014 Elsevier Ireland Ltd. All rights reserved.
This was a retrospective, single-center, observational study of consecutive adult patients with in-hospital cardiac arrest who had ECPR at Samsung Medical Center between January 2004 and December 2012. This study received the Institutional Review Board approval and informed consent was waived. Clinical, laboratory, and outcome
1032
S.B. Park et al. / International Journal of Cardiology 177 (2014) 1031–1035
data were collected by a trained study coordinator using a standardized case report form. Additional information was documented by reviewing hospital records and telephone interviews when necessary. We included 505 consecutive patients who underwent extracorporeal membrane oxygenation (ECMO) life support during the study period. Of these patients, we excluded patients unrelated to CPR, patients under 18 years of age, patients with malignancy whose expected life span was less than 1 year, patients who had veno-venous ECMO, patients who were inconsistent with the ECPR definition, patients with insufficient medical records and patients with out-of-hospital arrest. Finally, a total of 152 patients with witnessed cardiac arrest rescued by veno-arterial ECMO life support were eligible for this study (Fig. 1).
to achieve a sustainable activated clotting time ranging from 180 to 220 s. In order to achieve an ideal cardiac index greater than 2.2 L/min/body surface area (m2), central mixed venous saturation over 70%, and MAP over 65 mm Hg, the initial revolutions per minute of the ECMO device were adjusted based upon the above criteria. Blood pressure was monitored continuously through an arterial catheter, and arterial blood gas analysis was measured in the artery of the right arm to estimate cerebral oxygenation. Following ECMO, necessary subsequent interventions for arrest causes were performed such as percutaneous coronary intervention, coronary artery bypass grafting, heart transplantation, non-coronary cardiopulmonary surgery, and non-cardiopulmonary surgery. 2.4. Statistical analysis
2.2. Definition and outcomes In previous studies, the definition of ECPR included both successful veno-arterial ECMO implantation and pump-on during cardiac massage [10,11]. However, various unexpected situations occurred in ongoing ECPR scenes. Actually, when a return of spontaneous circulation (ROSC) occurs during ECMO cannulation, the practitioner does not remove the already inserted cannula and does not stop the process of ECMO pumpon. We included such cases in our ECPR definition as intention-to-treat. Accordingly, ECPR was defined as an intention-to-treat with hemodynamic ECMO support during cardiac massage regardless of interim ROSC [11]. In this study, CPR to ECMO pump-on time was defined as the time from initiation of cardiac massage to that of ECMO pumpon. As for recurrent arrest cases, if the duration of ROSC was sustained for more than 20 min, we made the following arrest event a standard initial point of cardiac massage [10]. The post-ECMO initial arterial pulse pressure and initial mean arterial pressure (MAP) were measured at the moment of proper status of ECMO flow right after pumpon. The post-ECMO initial Sequential Organ Failure Assessment (SOFA) score was measured in all patients using the worst value of each scoring item within 24 h of the event. The primary outcome was survival to discharge and the secondary outcome was all-cause death during follow-up. 2.3. Procedure CPR was led by the CPR team of the hospital and all facts related to the CPR scene were recorded according to Utstein-style guidelines by bedside nurses [12]. The request call for ECPR was up to the CPR leader. The final decision to institute ECMO and the ECMO cannulation procedure during CPR was determined by ECMO specialists such as interventional cardiologists or cardiac surgeons. The CPR team and the ECMO specialists were available in the hospital at all times. The ECMO circuits were always prepared and primed by ECMO perfusionists in the intensive care unit. The ECMO device and various kinds of cannulae were transported in a wheeling cart to the CPR scene by ECMO perfusionists or the victims were transported to the catheterization laboratory room. Verbal or written informed consent for ECMO application was obtained promptly from relatives by the attending physician in charge of the victims. The CAPIOX® auto-priming bypass system (Terumo Inc., Tokyo, Japan) was used in all cases of ECPR. A percutaneous vascular approach was first tried in all cases using the Seldinger technique but surgical cut down exposure was performed secondarily in those patients in whom percutaneous cannulation failed. The usual sites of vascular approach were the femoral vessels. Arterial cannulae sized 14 to 21 French and venous cannulae sized 21 to 28 French were used. Anticoagulation was achieved by a bolus of intravenously injected unfractionated heparin prior to cannulation. After successful ECMO pump-on during CPR, cardiac compression stopped and continuous heparinization for systemic anticoagulation began intravenously
Comparisons for continuous variables were made using the t-test or Wilcoxon ranksum test when applicable. Categorical data were tested using the chi-squared test. The collinearity of the variables, the goodness-of-fit, discrimination and calibration of the model, and the predictive value and optimal cut-off point of each relevant continuous variable were assessed by the area under curve (AUC) of the receiver operating characteristic (ROC) curve analysis and the Hosmer–Lemeshow chi-square statistics, respectively. On the multivariate binary regression model, covariates included those with a p value b 0.2 and those that were clinically relevant. To predict the survival to discharge in ECPR patients, we developed a new predictive score with the independent variables selected by the multivariate analysis. The score-based prediction rule was developed from logistic regression equations by using a regression coefficient-based scoring method. To generate a simple integer-based point score for each predictive variable, we assigned scores by dividing the β coefficients by the absolute value of the smallest coefficient in the final model and rounding up to the nearest integer. The total score for each participant was calculated by adding each component together. Among the derivation cohort, the survival to discharge was measured for each total score. We then explored whether the score of the new prediction scoring system exactly indicated better survival to discharge, using ROC curve analysis, computed AUC, and its corresponding 95% confidence interval. We validated the prediction rule internally using the .632+ bootstrap method in the original data set by sampling with a replacement for 1000 iterations. Survival curves were constructed using Kaplan–Meier estimates and compared with the log-rank test. Statistical analyses were performed with STATA® 12 (STATA Corp, College Station, TX, USA). All tests were two-tailed and p b 0.05 was considered statistically significant.
3. Results 3.1. Baseline and procedural characteristics Among 152 adult cardiac arrest patients who underwent ECPR, successful ECMO weaning was achieved in 67 patients (44.1%) and survival to discharge was identified in 48 patients (31.6%). The demographic patient data are shown in Table 1. There were no significant differences between non-survival and survival groups except for the diabetes. The initial arrest and procedural findings are shown in Table 2. A cardiogenic origin was the most common cause of cardiopulmonary resuscitation (n = 125, 82.2%) in this study. Applications of ECMO were mostly performed at the bedside in the intensive care unit (n = 76, 50%). Compared
Table 1 Baseline patient characteristics.
Fig. 1. Scheme of Group Distribution in the Registry.
Variables
Non-survivors (n = 104)
Survivors (n = 48)
p value
Age, years Gender (male) Body mass index, kg/m2 Comorbidities Diabetes Hypertension Dyslipidemia Current smoker Chronic kidney disease Peripheral vascular disease Previous myocardial infarction Previous PCI Previous coronary bypass surgery Previous stroke Clinical presentation Ischemic cardiomyopathy Non-ischemic
61.5 ± 16.4 66 (63.5%) 23.2 ± 3.8
57.6 ± 15.4 27 (56.3%) 23.2 ± 3.3
0.16 0.40 0.93
48 (46.2%) 44 (42.3%) 13 (12.5%) 27 (26%) 16 (15.4%) 9 (8.7%) 13 (12.5%) 20 (19.2%) 6 (5.8%) 15 (14.4%)
31 (64.6%) 24 (50%) 7 (14.6%) 12 (25%) 6 (12.5%) 2 (4.2%) 5 (10.4%) 10 (20.8%) 3 (6.3%) 9 (18.8%)
0.04 0.38 0.72 0.90 0.64 0.50 0.71 0.82 0.99 0.50 0.31
44 (42.3%) 60 (57.7%)
25 (52.1%) 23 (47.9%)
PCI = percutaneous coronary intervention. Values are mean ± standard deviation or n (%).
S.B. Park et al. / International Journal of Cardiology 177 (2014) 1031–1035 Table 2 Arrest and procedural characteristics. Variables Arrest cause Cardiogenic shock Septic shock Hypovolemic shock Respiratory deterioration Neurogenic shock Unknown First monitored arrest rhythm Asystole Pulseless electrical activity VF/Pulseless VT ROSC before ECMO pump on CPR to ECMO pump on time (minutes) Post-ECMO initial arterial pulse pressure Post-ECMO initial mean arterial pressure Location of ECMO insertion Intensive care unit Catheterization laboratory room General ward Emergency room Operating room Subsequent intervention Percutaneous coronary intervention Coronary bypass graft surgery Heart transplantation Non-coronary heart surgery Non-cardiac surgery Continuous renal replacement therapy Biochemical findings after ECPR Platelet Total bilirubin Creatinine Initial SOFA score after ECPR ECMO duration, hours
Non-survivors (n = 104)
Survivors (n = 48)
p value
83 (79.8%) 5 (4.8%) 11 (10.6%) 2 (1.9%) 1 (1%) 2 (1.9%)
42 (87.5%) 1 (2.1%) 5 (10.4%) 0 (0%) 0 (0%) 0 (0%)
26 (25%) 54 (51.9%) 24 (23.1%) 40 (38.5%) 46 ± 26 19 ± 20 62 ± 35
3 (6.3%) 26 (54.2%) 19 (39.6%) 18 (37.5%) 31 ± 17 36 ± 24 76 ± 21
48 (46.2%) 28 (26.9%) 4 (3.8%) 15 (14.4%) 9 (8.7%)
28 (58.3%) 13 (27.1%) 1 (2.1%) 4 (8.3%) 2 (4.2%)
22 (21.2%) 3 (2.9%) 0 (0%) 10 (9.6%) 5 (4.8%) 46 (44.2%)
13 (27.1%) 7 (14.6%) 3 (6.3%) 11 (22.9%) 2 (4.2%) 12 (25%)
0.42 0.01 0.03 0.03 0.99 0.02
179 ± 99 1.6 ± 2.2 1.6 ± 1.1 13 ± 2 67.2 ± 103.2
165 ± 65 1.1 ± 0.8 1.2 ± 0.6 12 ± 1 67.5 ± 116
0.31 0.05 0.02 0.03 0.99
0.76
1033
arrest rhythm, CPR to ECMO pump-on time, initial pulse pressure, and initial SOFA score (Table 3). In addition, we performed a ROC curve analysis with a logistic regression in order to estimate the predictability of survival to discharge and to identify the optimal cut-off value of each continuous variable. The optimal cut-off points for survival to discharge were; age ≤ 66 years, CPR to ECMO pump-on time ≤ 38 min, initial pulse pressure N 24 mm Hg, and initial SOFA score ≤ 14 points. 3.3. New prognostic scoring system
0.01
0.91 b0.001 b0.001 0.002 0.62
PEA = pulseless electrical activity; VF = ventricular fibrillation; VT = ventricular tachycardia; ROSC = return of spontaneous circulation; ECMO = Extracorporeal membrane oxygenation; CPR = Cardiopulmonary resuscitation; ECPR = ECMO-assisted cardiopulmonary resuscitation, SOFA = Sequential Organ Failure Assessment. Values are mean ± standard deviation or n (%).
with non-survivors, survivors had a higher prevalence of shockable rhythm, short CPR to ECMO time, high pulse pressure after ECMO pump-on, high MAP after ECMO pump-on, high sequential surgical interventions such as coronary bypass surgery, heart transplantation, and non-coronary surgery after insertion of ECMO, low level of serum creatinine, and low initial SOFA score. In addition, survivors were less likely to receive continuous renal replacement therapy. 3.2. Prognostic factors under ECPR In multivariate analysis with adjustment for prognostic covariates, significant predictors of survival to discharge were age, first monitored
The risk scoring system is shown in Table 3. Based on the logistic regression analysis, we created a new prognostic scoring system using the β coefficients and named it the ECPR score. The lowest β coefficient, 1.39, corresponded to approximately 2 points. As a result, points were assigned as follows; 2 points for age ≤ 66 years (β = 1.53), 3 points for arrest rhythm of PEA (β = 2.21), 4 points for arrest rhythm of VF/pulseless VT (β = 2.76), 3 points for CPR to ECMO pump-on time ≤ 38 min (β = 1.89), 2 points for initial pulse pressure N 24 mm Hg (β = 1.39), and 4 points for initial SOFA score ≤ 14 (β = 2.51), resulting in 15 possible points for the ECPR score. According to ROC curve analysis with logistic regression, the optimal cut-off point for the ECPR score for survival to discharge was N10 points. The sensitivity and specificity for prediction of survival to discharge were 89.6% and 75%, respectively, when the ECPR score was N10 points. Patients with an ECPR score N 10 points showed a 26-fold better survival to discharge rate (odds ratio [OR] 25.80, 95% confidence interval [CI] 9.24– 72.05, p b 0.001) than those with an ECPR score ≤ 10 points. Survival to discharge was predicted by an ECPR score with a c-statistics of 0.8595 (95% CI 0.80–0.92, p b 0.001, Fig. 2), which was similar to the c-statistics obtained from internal validation using the .632+ bootstrap method (training vs. test set: c-statistics 0.86 vs. 0.86005, 95% CI 0.80– 0.92 vs. 0.77–0.94, p b 0.001). The study population was divided into four groups according to ECPR score quartile as follows; quartile I, 0 to 7 points (n = 39, survival to discharge: 5.1%), quartile II, 8 to 10 points (n = 44, survival to discharge: 6.8%), quartile III, 11 to 12 points (n = 50, survival to discharge: 58%), and quartile IV, 13 to 15 points (n = 19, survival to discharge: 73.7%). Each quartile increment was associated with a 6.0-fold increase (OR 5.46, 95% CI 3.06–9.73, p b 0.001) for survival to discharge (Fig. 3A). In addition, the patients with an ECPR score N 10 points showed significantly better survival during followup duration in Kaplan–Meier estimates (p b 0.001) (Fig. 3B). 4. Discussion In our study, we investigated the risk factors predictive of survival and developed a new prognostic scoring system to predict survival to discharge for in-hospital cardiac arrest adult patients who underwent ECPR using a single-center registry for nine years. Old age, arrest rhythm of asystole, delayed ECMO pump-on, low initial pulse pressure, and poor initial SOFA score were independent predictors of in-hospital death. The
Table 3 Multivariate analysis and prognostic scoring system.
Pre-ECPR factors
Intra-ECPR factor Post-ECPR factors
Variables
β coefficient
Age ≤ 66 years Diabetes First monitored arrest rhythm Asystole Pulseless electrical activity VF/pulseless VT Ischemic cardiomyopathy CPR to ECMO time ≤ 38 min Initial pulse pressure N 24 mm Hg Initial MAP N 57 mm Hg Initial SOFA score ≤ 14
1.53
2.21 2.76 1.89 1.39 2.51
Odd ratio
95% confidence interval
p value
New ECPR score
4.61 2.47
1.63–13.03 0.97–6.33
0.004 0.06
2
9.15 15.76 0.72 6.62 4.00 2.33 12.29
1.91–43.89 2.91–85.36 0.27–1.90 2.08–21.10 1.50–10.64 0.71–7.61 2.71–55.74
0.01 0.001 0.50 0.001 0.001 0.16 0.001
3 4 3 2 4
ECPR = ECMO-assisted cardiopulmonary resuscitation; VF = ventricular fibrillation; VT = ventricular tachycardia; ECMO = extracorporeal membrane oxygenation; CPR = cardiopulmonary resuscitation; MAP = mean arterial pressure; SOFA = Sequential Organ Failure Assessment.
1034
S.B. Park et al. / International Journal of Cardiology 177 (2014) 1031–1035
Fig. 2. Predictability of a new ECPR score for survival to discharge.
ECPR score, which was developed using β coefficients of five prognostic factors, was identified as a good indicator of survival to discharge. Furthermore, patients with scores N 10 points showed significantly better outcomes during follow-up. In the setting of ECPR, shorter ECPR duration, postcardiotomy arrest, organ dysfunction indicators, arrest of cardiac origin, and no pulmonary hemorrhage were shown to be predictors of ECMO weaning and survival to discharge in previous studies. However, previous studies only focused on prognostic factors applied to a small, limited group and the results were not consistent [5,13–15]. Therefore, we not only investigated prognostic factors, but also developed a prognostic model to help physicians tailor aggressive therapies that can improve survival in patients who undergo ECPR. In a previous study on conventional CPR, a first monitored arrest rhythm of VF/pulseless VT was shown to be a predictor of survival to discharge, but not ECPR [16]. However, in the present study, the first monitored arrest rhythm was associated with a higher survival to discharge under ECPR. In line with a previous large-scale study, an important decision for implementation of ECMO in patients with in-hospital cardiac arrest might be the separation of nonshockable rhythm from VF/pulseless VT [17]. Furthermore, old age and the initial severity of organ failure were independent risk factors for in-hospital death. These findings suggest that decisions regarding
the implementation of ECMO during CPR should be made carefully while considering pre-ECPR factors such as correctable arrhythmic causes of cardiac arrest, uncorrectable age, and severity of organ dysfunction. As for intra-ECPR factors, the only clinically modifiable predictor of survival to discharge was CPR to ECMO pump-on time. The probability of survival to discharge was inferred by logistic regression [8]. In our study, the probability of survival to discharge decreased to 0.45, 0.37, 0.30, 0.24, and 0.18 as the ECMO pump-on time was delayed to 20, 30, 40, 50, and 60 min, respectively. According to our findings, it seems clear that we can achieve improved survival when a patient has ECMO pump-on as soon as possible. Thus, to achieve better survival in ECPR, we might have to set up a hospital-specific strategy for ECMO pump-on within 45 min from initiation of CPR to achieve the maximum survival rate of conventional CPR of 0.26 [1–4]. In particular, a new post-ECPR factor in this study was that the initially high pulse pressure after properly adjusted ECMO pump flow was an independent predictor of survival to discharge. A high pulse pressure right after ECMO pump-on indicates that the initial systolic cardiac capacity is enough to overcome the reverse flow pressure of arterial ECMO pumping and thus can be regarded as an important predictor of survival to discharge. To date, a robust risk model applicable to ECPR has not been developed. Accordingly, using the unmodifiable condition of the patient and the modifiable variables of ECPR, the new prognostic scoring system we created, the ECPR score, could provide improved information on early and long term prognosis in patients undergoing ECPR in real world practice. In particular, when a patient has prolonged ECMO life support without recovery of autologous heart pumping, a physician may consider the termination of ECMO life support or heart transplantation in case the patient has several prognostic risk factors for inhospital death or has a low ECPR score below 10. Our study had several limitations. Our investigation had several limitations. First, the nonrandomized nature of the registry data could have resulted in selection bias. Second, this study has no data regarding the neurologic outcomes of patients, which may be related to the withdrawal of aggressive management. Third, our registry did not include hemodynamic parameters such as pulmonary capillary wedge pressure measured by a pulmonary arterial catheter or left ventricle size or E/e′ measured by echocardiography. Lastly, future studies are needed for external validation of the new ECPR score from another cohort in order to apply it to a wide range of ECPR scores because we performed only an internal validation of the ECPR score.
Fig. 3. Survival to discharge and all-cause death during follow-up according to ECPR score. (A) Survival to discharge in population quartiles of the ECPR score; (B) Kaplan–Meier curves for all-cause death in ECPR score N 10 (solid line) versus ECPR score ≤ 10 points (dashed line).
S.B. Park et al. / International Journal of Cardiology 177 (2014) 1031–1035
In conclusion, using age, shockable arrest rhythm, CPR to ECMO pump-on time, post-ECMO arterial pulse pressure, and post-ECMO SOFA score, we created a good risk prediction model for survival to discharge in ECPR. Our results suggest that this new scoring system might be helpful in ECPR management decision making and could provide better information on early prognosis. Funding source None. Conflict of interest The authors declare no conflicts of interest. The authors of this manuscript have certified that they comply with the Principles of Ethical Publishing in the International Journal of Cardiology. Acknowledgments We appreciate the excellent statistical support of Seonwoo Kim, PhD, and Joonghyun Ahn, MS at the Samsung Biomedical Research Institute. References [1] Kouwenhoven WB, Jude JR, Knickerbocker GG. Closed-chest cardiac massage. JAMA 1960;173:1064–7. [2] Nachlas MM, Miller DI. Closed-chest cardiac resuscitation in patients with acute myocardial infarction. Am Heart J 1965;69:448–59. [3] Ehlenbach WJ, Barnato AE, Curtis JR, Kreuter W, Koepsell TD, Deyo RA, et al. Epidemiologic study of in-hospital cardiopulmonary resuscitation in the elderly. N Engl J Med 2009;361:22–31. [4] Linko E, Koskinen PJ, Siitonen L, Ruosteenoja R. Resuscitation in cardiac arrest. An analysis of 100 successive medical cases. Acta Med Scand 1967;182:611–20.
1035
[5] Chen YS, Chao A, Yu HY, Ko WJ, Wu IH, Chen RJ, et al. Analysis and results of prolonged resuscitation in cardiac arrest patients rescued by extracorporeal membrane oxygenation. J Am Coll Cardiol 2003;41:197–203. [6] Schultz SC, Cullinane DC, Pasquale MD, Magnant C, Evans SR. Predicting in-hospital mortality during cardiopulmonary resuscitation. Resuscitation 1996;33:13–7. [7] Cummins RO, Eisenberg MS, Hallstrom AP, Litwin PE. Survival of out-of-hospital cardiac arrest with early initiation of cardiopulmonary resuscitation. Am J Emerg Med 1985;3:114–9. [8] Chen YS, Yu HY, Huang SC, Lin JW, Chi NH, Wang CH, et al. Extracorporeal membrane oxygenation support can extend the duration of cardiopulmonary resuscitation Crit Care Med 2008;36:2529–35. [9] Shin TG, Jo IJ, Sim MS, Song YB, Yang JH, Hahn JY, et al. Two-year survival and neurological outcome of in-hospital cardiac arrest patients rescued by extracorporeal cardiopulmonary resuscitation. Int J Cardiol 2013;168:3424–30. [10] Chen YS, Lin JW, Yu HY, Ko WJ, Jerng JS, Chang WT, et al. Cardiopulmonary resuscitation with assisted extracorporeal life-support versus conventional cardiopulmonary resuscitation in adults with in-hospital cardiac arrest: an observational study and propensity analysis Lancet 2008;372:554–61. [11] Shin TG, Choi JH, Jo IJ, Sim MS, Song HG, Jeong YK, et al. Extracorporeal cardiopulmonary resuscitation in patients with inhospital cardiac arrest: a comparison with conventional cardiopulmonary resuscitation Crit Care Med 2011;39:1–7. [12] Cummins RO, Chamberlain D, Hazinski MF, Nadkarni V, Kloeck W, Kramer E, et al. Recommended guidelines for reviewing, reporting, and conducting research on inhospital resuscitation: the in-hospital '‘Utstein style'’ American Heart Association. Circulation 1997;95:2213–39. [13] Thiagarajan RR, Laussen PC, Rycus PT, Bartlett RH, Bratton SL. Extracorporeal membrane oxygenation to aid cardiopulmonary resuscitation in infants and children. Circulation 2007;116:1693–700. [14] Prodhan P, Fiser RT, Dyamenahalli U, Gossett J, Imamura M, Jaquiss RD, et al. Outcomes after extracorporeal cardiopulmonary resuscitation (ECPR) following refractory pediatric cardiac arrest in the intensive care unit Resuscitation 2009;80:1124–9. [15] Alsoufi B, Al-Radi OO, Nazer RI, Gruenwald C, Foreman C, illiams WG. Survival outcomes after rescue extracorporeal cardiopulmonary resuscitation in pediatric patients with refractory cardiac arrest. J Thorac Cardiovasc Surg 2007;134:952–9 [e2]. [16] Sandroni C, Nolan J, Cavallaro F, Antonelli M. In-hospital cardiac arrest: incidence, prognosis and possible measures to improve survival. Intensive Care Med 2007; 33:237–45. [17] Nadkarni VM, Larkin GL, Peberdy MA, Carey SM, Kaye W, Mancini ME, et al. First documented rhythm and clinical outcome from in-hospital cardiac arrest among children and adults JAMA 2006;295:50–7.
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Developing a risk prediction model for survival to discharge in cardiac arrest patients who undergo extracorporeal membrane oxygenation Sung Bum Park a,b, Jeong Hoon Yang a,c,⁎, Taek Kyu Park c, Yang Hyun Cho d, Kiick Sung d, Chi Ryang Chung a, Chi Min Park a, Kyeongman Jeon a,e, Young Bin Song c, Joo-Yong Hahn c, Jin-Ho Choi c, Seung-Hyuk Choi c, Hyeon-Cheol Gwon c, Gee Young Suh a,e a
Department of Critical Care Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, South Korea Department of Medicine, Korean Armed Forces Capital Hospital, Seongnam, South Korea c Division of Cardiology, Department of Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, South Korea d Department of Thoracic and Cardiovascular Surgery, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, South Korea e Division of Pulmonary and Critical Care Medicine, Department of Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, South Korea b
a r t i c l e
i n f o
Article history: Received 16 June 2014 Accepted 23 September 2014 Available online 2 October 2014 Keywords: Cardiopulmonary resuscitation Extracorporeal membrane oxygenation Predictor
a b s t r a c t Background: Limited data are available on a risk model for survival to discharge after extracorporeal membrane oxygenation (ECMO)-assisted cardiopulmonary resuscitation (ECPR). We aimed to develop a risk prediction model for survival to discharge in cardiac arrest patients who undergo ECMO. Methods: Between January 2004 and December 2012, 505 patients supported by ECMO were enrolled in a retrospective, observational registry. Among those, we studied 152 adult patients with in-hospital cardiac arrest. The primary outcome was survival to discharge. A new predictive scoring system, named the ECPR score, was developed to monitor survival to discharge using the β coefficients of prognostic factors from the logistic model, which were internally validated. Results: In-hospital death occurred in 104 patients (68.4%). In multivariate logistic regression, age ≤ 66, shockable arrest rhythm, CPR to ECMO pump-on time ≤ 38 min, post-ECMO arterial pulse pressure N 24 mm Hg, and post-ECMO Sequential Organ Failure Assessment score ≤ 14 were independent predictors for survival to discharge. Survival to discharge was predicted by the ECPR score with a c-statistics of 0.8595 (95% confidence interval [CI], 0.80–0.92; p b 0.001) which was similar to the c-statistics obtained from internal validation (training vs. test set; c-statistics, 0.86 vs. 0.86005; 95% CI, 0.80–0.92 vs. 0.77–0.94). The sensitivity and specificity for prediction of survival to discharge were 89.6% and 75.0%, respectively, when the ECPR score was N 10. Conclusions: The new risk prediction model might be helpful for decisions about ECPR management and could provide better information regarding early prognosis. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Cardiopulmonary resuscitation (CPR) has traditionally been the treatment of choice for cardiopulmonary collapse and it has been developed and modified to achieve better survival. Unfortunately, previous studies revealed a low survival to discharge rate that ranged from 7 to 26% [1–4], which declines rapidly if the duration of CPR exceeds 10 min [5–7]. Recently, extracorporeal cardiopulmonary resuscitation (ECPR) with a portable cardiopulmonary bypass system has been increasingly utilized to supply oxygenated blood in the absence of spontaneous cardiac pumping [8,9]. Several observation studies have shown
an improved survival rate compared to conventional CPR in patients with in-hospital cardiac arrest [10,11]. However, to date, there have been no guidelines for the application of ECMO in the setting of CPR and the prognostic factors under ECPR have not been fully established in patients with in-hospital cardiac arrest. Therefore, we investigated the predictive risk factors for in-hospital death and developed a risk prediction model to better inform physicians of suitable candidates and their expected survival rate in cardiac arrest patients undergoing ECPR. 2. Methods 2.1. Study population
⁎ Corresponding author at: Department of Critical Care Medicine, Division of Cardiology and Department of Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, 81 Irwon-ro, Gangnam-gu, Seoul 135-710, South Korea. Tel.: +82 2 3410 1768; fax: +82 2 2148 7088. E-mail address: [email protected] (J.H. Yang).
http://dx.doi.org/10.1016/j.ijcard.2014.09.124 0167-5273/© 2014 Elsevier Ireland Ltd. All rights reserved.
This was a retrospective, single-center, observational study of consecutive adult patients with in-hospital cardiac arrest who had ECPR at Samsung Medical Center between January 2004 and December 2012. This study received the Institutional Review Board approval and informed consent was waived. Clinical, laboratory, and outcome
1032
S.B. Park et al. / International Journal of Cardiology 177 (2014) 1031–1035
data were collected by a trained study coordinator using a standardized case report form. Additional information was documented by reviewing hospital records and telephone interviews when necessary. We included 505 consecutive patients who underwent extracorporeal membrane oxygenation (ECMO) life support during the study period. Of these patients, we excluded patients unrelated to CPR, patients under 18 years of age, patients with malignancy whose expected life span was less than 1 year, patients who had veno-venous ECMO, patients who were inconsistent with the ECPR definition, patients with insufficient medical records and patients with out-of-hospital arrest. Finally, a total of 152 patients with witnessed cardiac arrest rescued by veno-arterial ECMO life support were eligible for this study (Fig. 1).
to achieve a sustainable activated clotting time ranging from 180 to 220 s. In order to achieve an ideal cardiac index greater than 2.2 L/min/body surface area (m2), central mixed venous saturation over 70%, and MAP over 65 mm Hg, the initial revolutions per minute of the ECMO device were adjusted based upon the above criteria. Blood pressure was monitored continuously through an arterial catheter, and arterial blood gas analysis was measured in the artery of the right arm to estimate cerebral oxygenation. Following ECMO, necessary subsequent interventions for arrest causes were performed such as percutaneous coronary intervention, coronary artery bypass grafting, heart transplantation, non-coronary cardiopulmonary surgery, and non-cardiopulmonary surgery. 2.4. Statistical analysis
2.2. Definition and outcomes In previous studies, the definition of ECPR included both successful veno-arterial ECMO implantation and pump-on during cardiac massage [10,11]. However, various unexpected situations occurred in ongoing ECPR scenes. Actually, when a return of spontaneous circulation (ROSC) occurs during ECMO cannulation, the practitioner does not remove the already inserted cannula and does not stop the process of ECMO pumpon. We included such cases in our ECPR definition as intention-to-treat. Accordingly, ECPR was defined as an intention-to-treat with hemodynamic ECMO support during cardiac massage regardless of interim ROSC [11]. In this study, CPR to ECMO pump-on time was defined as the time from initiation of cardiac massage to that of ECMO pumpon. As for recurrent arrest cases, if the duration of ROSC was sustained for more than 20 min, we made the following arrest event a standard initial point of cardiac massage [10]. The post-ECMO initial arterial pulse pressure and initial mean arterial pressure (MAP) were measured at the moment of proper status of ECMO flow right after pumpon. The post-ECMO initial Sequential Organ Failure Assessment (SOFA) score was measured in all patients using the worst value of each scoring item within 24 h of the event. The primary outcome was survival to discharge and the secondary outcome was all-cause death during follow-up. 2.3. Procedure CPR was led by the CPR team of the hospital and all facts related to the CPR scene were recorded according to Utstein-style guidelines by bedside nurses [12]. The request call for ECPR was up to the CPR leader. The final decision to institute ECMO and the ECMO cannulation procedure during CPR was determined by ECMO specialists such as interventional cardiologists or cardiac surgeons. The CPR team and the ECMO specialists were available in the hospital at all times. The ECMO circuits were always prepared and primed by ECMO perfusionists in the intensive care unit. The ECMO device and various kinds of cannulae were transported in a wheeling cart to the CPR scene by ECMO perfusionists or the victims were transported to the catheterization laboratory room. Verbal or written informed consent for ECMO application was obtained promptly from relatives by the attending physician in charge of the victims. The CAPIOX® auto-priming bypass system (Terumo Inc., Tokyo, Japan) was used in all cases of ECPR. A percutaneous vascular approach was first tried in all cases using the Seldinger technique but surgical cut down exposure was performed secondarily in those patients in whom percutaneous cannulation failed. The usual sites of vascular approach were the femoral vessels. Arterial cannulae sized 14 to 21 French and venous cannulae sized 21 to 28 French were used. Anticoagulation was achieved by a bolus of intravenously injected unfractionated heparin prior to cannulation. After successful ECMO pump-on during CPR, cardiac compression stopped and continuous heparinization for systemic anticoagulation began intravenously
Comparisons for continuous variables were made using the t-test or Wilcoxon ranksum test when applicable. Categorical data were tested using the chi-squared test. The collinearity of the variables, the goodness-of-fit, discrimination and calibration of the model, and the predictive value and optimal cut-off point of each relevant continuous variable were assessed by the area under curve (AUC) of the receiver operating characteristic (ROC) curve analysis and the Hosmer–Lemeshow chi-square statistics, respectively. On the multivariate binary regression model, covariates included those with a p value b 0.2 and those that were clinically relevant. To predict the survival to discharge in ECPR patients, we developed a new predictive score with the independent variables selected by the multivariate analysis. The score-based prediction rule was developed from logistic regression equations by using a regression coefficient-based scoring method. To generate a simple integer-based point score for each predictive variable, we assigned scores by dividing the β coefficients by the absolute value of the smallest coefficient in the final model and rounding up to the nearest integer. The total score for each participant was calculated by adding each component together. Among the derivation cohort, the survival to discharge was measured for each total score. We then explored whether the score of the new prediction scoring system exactly indicated better survival to discharge, using ROC curve analysis, computed AUC, and its corresponding 95% confidence interval. We validated the prediction rule internally using the .632+ bootstrap method in the original data set by sampling with a replacement for 1000 iterations. Survival curves were constructed using Kaplan–Meier estimates and compared with the log-rank test. Statistical analyses were performed with STATA® 12 (STATA Corp, College Station, TX, USA). All tests were two-tailed and p b 0.05 was considered statistically significant.
3. Results 3.1. Baseline and procedural characteristics Among 152 adult cardiac arrest patients who underwent ECPR, successful ECMO weaning was achieved in 67 patients (44.1%) and survival to discharge was identified in 48 patients (31.6%). The demographic patient data are shown in Table 1. There were no significant differences between non-survival and survival groups except for the diabetes. The initial arrest and procedural findings are shown in Table 2. A cardiogenic origin was the most common cause of cardiopulmonary resuscitation (n = 125, 82.2%) in this study. Applications of ECMO were mostly performed at the bedside in the intensive care unit (n = 76, 50%). Compared
Table 1 Baseline patient characteristics.
Fig. 1. Scheme of Group Distribution in the Registry.
Variables
Non-survivors (n = 104)
Survivors (n = 48)
p value
Age, years Gender (male) Body mass index, kg/m2 Comorbidities Diabetes Hypertension Dyslipidemia Current smoker Chronic kidney disease Peripheral vascular disease Previous myocardial infarction Previous PCI Previous coronary bypass surgery Previous stroke Clinical presentation Ischemic cardiomyopathy Non-ischemic
61.5 ± 16.4 66 (63.5%) 23.2 ± 3.8
57.6 ± 15.4 27 (56.3%) 23.2 ± 3.3
0.16 0.40 0.93
48 (46.2%) 44 (42.3%) 13 (12.5%) 27 (26%) 16 (15.4%) 9 (8.7%) 13 (12.5%) 20 (19.2%) 6 (5.8%) 15 (14.4%)
31 (64.6%) 24 (50%) 7 (14.6%) 12 (25%) 6 (12.5%) 2 (4.2%) 5 (10.4%) 10 (20.8%) 3 (6.3%) 9 (18.8%)
0.04 0.38 0.72 0.90 0.64 0.50 0.71 0.82 0.99 0.50 0.31
44 (42.3%) 60 (57.7%)
25 (52.1%) 23 (47.9%)
PCI = percutaneous coronary intervention. Values are mean ± standard deviation or n (%).
S.B. Park et al. / International Journal of Cardiology 177 (2014) 1031–1035 Table 2 Arrest and procedural characteristics. Variables Arrest cause Cardiogenic shock Septic shock Hypovolemic shock Respiratory deterioration Neurogenic shock Unknown First monitored arrest rhythm Asystole Pulseless electrical activity VF/Pulseless VT ROSC before ECMO pump on CPR to ECMO pump on time (minutes) Post-ECMO initial arterial pulse pressure Post-ECMO initial mean arterial pressure Location of ECMO insertion Intensive care unit Catheterization laboratory room General ward Emergency room Operating room Subsequent intervention Percutaneous coronary intervention Coronary bypass graft surgery Heart transplantation Non-coronary heart surgery Non-cardiac surgery Continuous renal replacement therapy Biochemical findings after ECPR Platelet Total bilirubin Creatinine Initial SOFA score after ECPR ECMO duration, hours
Non-survivors (n = 104)
Survivors (n = 48)
p value
83 (79.8%) 5 (4.8%) 11 (10.6%) 2 (1.9%) 1 (1%) 2 (1.9%)
42 (87.5%) 1 (2.1%) 5 (10.4%) 0 (0%) 0 (0%) 0 (0%)
26 (25%) 54 (51.9%) 24 (23.1%) 40 (38.5%) 46 ± 26 19 ± 20 62 ± 35
3 (6.3%) 26 (54.2%) 19 (39.6%) 18 (37.5%) 31 ± 17 36 ± 24 76 ± 21
48 (46.2%) 28 (26.9%) 4 (3.8%) 15 (14.4%) 9 (8.7%)
28 (58.3%) 13 (27.1%) 1 (2.1%) 4 (8.3%) 2 (4.2%)
22 (21.2%) 3 (2.9%) 0 (0%) 10 (9.6%) 5 (4.8%) 46 (44.2%)
13 (27.1%) 7 (14.6%) 3 (6.3%) 11 (22.9%) 2 (4.2%) 12 (25%)
0.42 0.01 0.03 0.03 0.99 0.02
179 ± 99 1.6 ± 2.2 1.6 ± 1.1 13 ± 2 67.2 ± 103.2
165 ± 65 1.1 ± 0.8 1.2 ± 0.6 12 ± 1 67.5 ± 116
0.31 0.05 0.02 0.03 0.99
0.76
1033
arrest rhythm, CPR to ECMO pump-on time, initial pulse pressure, and initial SOFA score (Table 3). In addition, we performed a ROC curve analysis with a logistic regression in order to estimate the predictability of survival to discharge and to identify the optimal cut-off value of each continuous variable. The optimal cut-off points for survival to discharge were; age ≤ 66 years, CPR to ECMO pump-on time ≤ 38 min, initial pulse pressure N 24 mm Hg, and initial SOFA score ≤ 14 points. 3.3. New prognostic scoring system
0.01
0.91 b0.001 b0.001 0.002 0.62
PEA = pulseless electrical activity; VF = ventricular fibrillation; VT = ventricular tachycardia; ROSC = return of spontaneous circulation; ECMO = Extracorporeal membrane oxygenation; CPR = Cardiopulmonary resuscitation; ECPR = ECMO-assisted cardiopulmonary resuscitation, SOFA = Sequential Organ Failure Assessment. Values are mean ± standard deviation or n (%).
with non-survivors, survivors had a higher prevalence of shockable rhythm, short CPR to ECMO time, high pulse pressure after ECMO pump-on, high MAP after ECMO pump-on, high sequential surgical interventions such as coronary bypass surgery, heart transplantation, and non-coronary surgery after insertion of ECMO, low level of serum creatinine, and low initial SOFA score. In addition, survivors were less likely to receive continuous renal replacement therapy. 3.2. Prognostic factors under ECPR In multivariate analysis with adjustment for prognostic covariates, significant predictors of survival to discharge were age, first monitored
The risk scoring system is shown in Table 3. Based on the logistic regression analysis, we created a new prognostic scoring system using the β coefficients and named it the ECPR score. The lowest β coefficient, 1.39, corresponded to approximately 2 points. As a result, points were assigned as follows; 2 points for age ≤ 66 years (β = 1.53), 3 points for arrest rhythm of PEA (β = 2.21), 4 points for arrest rhythm of VF/pulseless VT (β = 2.76), 3 points for CPR to ECMO pump-on time ≤ 38 min (β = 1.89), 2 points for initial pulse pressure N 24 mm Hg (β = 1.39), and 4 points for initial SOFA score ≤ 14 (β = 2.51), resulting in 15 possible points for the ECPR score. According to ROC curve analysis with logistic regression, the optimal cut-off point for the ECPR score for survival to discharge was N10 points. The sensitivity and specificity for prediction of survival to discharge were 89.6% and 75%, respectively, when the ECPR score was N10 points. Patients with an ECPR score N 10 points showed a 26-fold better survival to discharge rate (odds ratio [OR] 25.80, 95% confidence interval [CI] 9.24– 72.05, p b 0.001) than those with an ECPR score ≤ 10 points. Survival to discharge was predicted by an ECPR score with a c-statistics of 0.8595 (95% CI 0.80–0.92, p b 0.001, Fig. 2), which was similar to the c-statistics obtained from internal validation using the .632+ bootstrap method (training vs. test set: c-statistics 0.86 vs. 0.86005, 95% CI 0.80– 0.92 vs. 0.77–0.94, p b 0.001). The study population was divided into four groups according to ECPR score quartile as follows; quartile I, 0 to 7 points (n = 39, survival to discharge: 5.1%), quartile II, 8 to 10 points (n = 44, survival to discharge: 6.8%), quartile III, 11 to 12 points (n = 50, survival to discharge: 58%), and quartile IV, 13 to 15 points (n = 19, survival to discharge: 73.7%). Each quartile increment was associated with a 6.0-fold increase (OR 5.46, 95% CI 3.06–9.73, p b 0.001) for survival to discharge (Fig. 3A). In addition, the patients with an ECPR score N 10 points showed significantly better survival during followup duration in Kaplan–Meier estimates (p b 0.001) (Fig. 3B). 4. Discussion In our study, we investigated the risk factors predictive of survival and developed a new prognostic scoring system to predict survival to discharge for in-hospital cardiac arrest adult patients who underwent ECPR using a single-center registry for nine years. Old age, arrest rhythm of asystole, delayed ECMO pump-on, low initial pulse pressure, and poor initial SOFA score were independent predictors of in-hospital death. The
Table 3 Multivariate analysis and prognostic scoring system.
Pre-ECPR factors
Intra-ECPR factor Post-ECPR factors
Variables
β coefficient
Age ≤ 66 years Diabetes First monitored arrest rhythm Asystole Pulseless electrical activity VF/pulseless VT Ischemic cardiomyopathy CPR to ECMO time ≤ 38 min Initial pulse pressure N 24 mm Hg Initial MAP N 57 mm Hg Initial SOFA score ≤ 14
1.53
2.21 2.76 1.89 1.39 2.51
Odd ratio
95% confidence interval
p value
New ECPR score
4.61 2.47
1.63–13.03 0.97–6.33
0.004 0.06
2
9.15 15.76 0.72 6.62 4.00 2.33 12.29
1.91–43.89 2.91–85.36 0.27–1.90 2.08–21.10 1.50–10.64 0.71–7.61 2.71–55.74
0.01 0.001 0.50 0.001 0.001 0.16 0.001
3 4 3 2 4
ECPR = ECMO-assisted cardiopulmonary resuscitation; VF = ventricular fibrillation; VT = ventricular tachycardia; ECMO = extracorporeal membrane oxygenation; CPR = cardiopulmonary resuscitation; MAP = mean arterial pressure; SOFA = Sequential Organ Failure Assessment.
1034
S.B. Park et al. / International Journal of Cardiology 177 (2014) 1031–1035
Fig. 2. Predictability of a new ECPR score for survival to discharge.
ECPR score, which was developed using β coefficients of five prognostic factors, was identified as a good indicator of survival to discharge. Furthermore, patients with scores N 10 points showed significantly better outcomes during follow-up. In the setting of ECPR, shorter ECPR duration, postcardiotomy arrest, organ dysfunction indicators, arrest of cardiac origin, and no pulmonary hemorrhage were shown to be predictors of ECMO weaning and survival to discharge in previous studies. However, previous studies only focused on prognostic factors applied to a small, limited group and the results were not consistent [5,13–15]. Therefore, we not only investigated prognostic factors, but also developed a prognostic model to help physicians tailor aggressive therapies that can improve survival in patients who undergo ECPR. In a previous study on conventional CPR, a first monitored arrest rhythm of VF/pulseless VT was shown to be a predictor of survival to discharge, but not ECPR [16]. However, in the present study, the first monitored arrest rhythm was associated with a higher survival to discharge under ECPR. In line with a previous large-scale study, an important decision for implementation of ECMO in patients with in-hospital cardiac arrest might be the separation of nonshockable rhythm from VF/pulseless VT [17]. Furthermore, old age and the initial severity of organ failure were independent risk factors for in-hospital death. These findings suggest that decisions regarding
the implementation of ECMO during CPR should be made carefully while considering pre-ECPR factors such as correctable arrhythmic causes of cardiac arrest, uncorrectable age, and severity of organ dysfunction. As for intra-ECPR factors, the only clinically modifiable predictor of survival to discharge was CPR to ECMO pump-on time. The probability of survival to discharge was inferred by logistic regression [8]. In our study, the probability of survival to discharge decreased to 0.45, 0.37, 0.30, 0.24, and 0.18 as the ECMO pump-on time was delayed to 20, 30, 40, 50, and 60 min, respectively. According to our findings, it seems clear that we can achieve improved survival when a patient has ECMO pump-on as soon as possible. Thus, to achieve better survival in ECPR, we might have to set up a hospital-specific strategy for ECMO pump-on within 45 min from initiation of CPR to achieve the maximum survival rate of conventional CPR of 0.26 [1–4]. In particular, a new post-ECPR factor in this study was that the initially high pulse pressure after properly adjusted ECMO pump flow was an independent predictor of survival to discharge. A high pulse pressure right after ECMO pump-on indicates that the initial systolic cardiac capacity is enough to overcome the reverse flow pressure of arterial ECMO pumping and thus can be regarded as an important predictor of survival to discharge. To date, a robust risk model applicable to ECPR has not been developed. Accordingly, using the unmodifiable condition of the patient and the modifiable variables of ECPR, the new prognostic scoring system we created, the ECPR score, could provide improved information on early and long term prognosis in patients undergoing ECPR in real world practice. In particular, when a patient has prolonged ECMO life support without recovery of autologous heart pumping, a physician may consider the termination of ECMO life support or heart transplantation in case the patient has several prognostic risk factors for inhospital death or has a low ECPR score below 10. Our study had several limitations. Our investigation had several limitations. First, the nonrandomized nature of the registry data could have resulted in selection bias. Second, this study has no data regarding the neurologic outcomes of patients, which may be related to the withdrawal of aggressive management. Third, our registry did not include hemodynamic parameters such as pulmonary capillary wedge pressure measured by a pulmonary arterial catheter or left ventricle size or E/e′ measured by echocardiography. Lastly, future studies are needed for external validation of the new ECPR score from another cohort in order to apply it to a wide range of ECPR scores because we performed only an internal validation of the ECPR score.
Fig. 3. Survival to discharge and all-cause death during follow-up according to ECPR score. (A) Survival to discharge in population quartiles of the ECPR score; (B) Kaplan–Meier curves for all-cause death in ECPR score N 10 (solid line) versus ECPR score ≤ 10 points (dashed line).
S.B. Park et al. / International Journal of Cardiology 177 (2014) 1031–1035
In conclusion, using age, shockable arrest rhythm, CPR to ECMO pump-on time, post-ECMO arterial pulse pressure, and post-ECMO SOFA score, we created a good risk prediction model for survival to discharge in ECPR. Our results suggest that this new scoring system might be helpful in ECPR management decision making and could provide better information on early prognosis. Funding source None. Conflict of interest The authors declare no conflicts of interest. The authors of this manuscript have certified that they comply with the Principles of Ethical Publishing in the International Journal of Cardiology. Acknowledgments We appreciate the excellent statistical support of Seonwoo Kim, PhD, and Joonghyun Ahn, MS at the Samsung Biomedical Research Institute. References [1] Kouwenhoven WB, Jude JR, Knickerbocker GG. Closed-chest cardiac massage. JAMA 1960;173:1064–7. [2] Nachlas MM, Miller DI. Closed-chest cardiac resuscitation in patients with acute myocardial infarction. Am Heart J 1965;69:448–59. [3] Ehlenbach WJ, Barnato AE, Curtis JR, Kreuter W, Koepsell TD, Deyo RA, et al. Epidemiologic study of in-hospital cardiopulmonary resuscitation in the elderly. N Engl J Med 2009;361:22–31. [4] Linko E, Koskinen PJ, Siitonen L, Ruosteenoja R. Resuscitation in cardiac arrest. An analysis of 100 successive medical cases. Acta Med Scand 1967;182:611–20.
1035
[5] Chen YS, Chao A, Yu HY, Ko WJ, Wu IH, Chen RJ, et al. Analysis and results of prolonged resuscitation in cardiac arrest patients rescued by extracorporeal membrane oxygenation. J Am Coll Cardiol 2003;41:197–203. [6] Schultz SC, Cullinane DC, Pasquale MD, Magnant C, Evans SR. Predicting in-hospital mortality during cardiopulmonary resuscitation. Resuscitation 1996;33:13–7. [7] Cummins RO, Eisenberg MS, Hallstrom AP, Litwin PE. Survival of out-of-hospital cardiac arrest with early initiation of cardiopulmonary resuscitation. Am J Emerg Med 1985;3:114–9. [8] Chen YS, Yu HY, Huang SC, Lin JW, Chi NH, Wang CH, et al. Extracorporeal membrane oxygenation support can extend the duration of cardiopulmonary resuscitation Crit Care Med 2008;36:2529–35. [9] Shin TG, Jo IJ, Sim MS, Song YB, Yang JH, Hahn JY, et al. Two-year survival and neurological outcome of in-hospital cardiac arrest patients rescued by extracorporeal cardiopulmonary resuscitation. Int J Cardiol 2013;168:3424–30. [10] Chen YS, Lin JW, Yu HY, Ko WJ, Jerng JS, Chang WT, et al. Cardiopulmonary resuscitation with assisted extracorporeal life-support versus conventional cardiopulmonary resuscitation in adults with in-hospital cardiac arrest: an observational study and propensity analysis Lancet 2008;372:554–61. [11] Shin TG, Choi JH, Jo IJ, Sim MS, Song HG, Jeong YK, et al. Extracorporeal cardiopulmonary resuscitation in patients with inhospital cardiac arrest: a comparison with conventional cardiopulmonary resuscitation Crit Care Med 2011;39:1–7. [12] Cummins RO, Chamberlain D, Hazinski MF, Nadkarni V, Kloeck W, Kramer E, et al. Recommended guidelines for reviewing, reporting, and conducting research on inhospital resuscitation: the in-hospital '‘Utstein style'’ American Heart Association. Circulation 1997;95:2213–39. [13] Thiagarajan RR, Laussen PC, Rycus PT, Bartlett RH, Bratton SL. Extracorporeal membrane oxygenation to aid cardiopulmonary resuscitation in infants and children. Circulation 2007;116:1693–700. [14] Prodhan P, Fiser RT, Dyamenahalli U, Gossett J, Imamura M, Jaquiss RD, et al. Outcomes after extracorporeal cardiopulmonary resuscitation (ECPR) following refractory pediatric cardiac arrest in the intensive care unit Resuscitation 2009;80:1124–9. [15] Alsoufi B, Al-Radi OO, Nazer RI, Gruenwald C, Foreman C, illiams WG. Survival outcomes after rescue extracorporeal cardiopulmonary resuscitation in pediatric patients with refractory cardiac arrest. J Thorac Cardiovasc Surg 2007;134:952–9 [e2]. [16] Sandroni C, Nolan J, Cavallaro F, Antonelli M. In-hospital cardiac arrest: incidence, prognosis and possible measures to improve survival. Intensive Care Med 2007; 33:237–45. [17] Nadkarni VM, Larkin GL, Peberdy MA, Carey SM, Kaye W, Mancini ME, et al. First documented rhythm and clinical outcome from in-hospital cardiac arrest among children and adults JAMA 2006;295:50–7.
