Predictors and outcomes of early post-operative veno-arterial extracorporeal membrane oxygenation following infant cardiac surgery

Background We aimed to determine predictors of, and outcomes after, veno-arterial extracorporeal membrane oxygenation instituted within 48 h after cardiac surgery (early ECMO) in young infants. Methods Patients ≤ 6 weeks old having cardiac surgery from 2003 to 2012 were enrolled prospectively. Patients cannulated pre-operatively, intra-operatively, or ≥ 48 h post-operatively were excluded. Variables at p ≤ 0.1 on univariate regression were entered into multiple logistic regression to predict early ECMO. Early-ECMO cases were matched 1:2 for six demographic variables, and death by age 2 years old (determined using conditional logistic regression; presented as odds ratio (OR), 95% confidence interval (CI)) and General Adaptive Composite scores at age 2 years (determined using Wilcoxon rank sum) were compared; p ≤ 0.05 was considered statistically significant. Results Of 565 eligible patients over the 10-year period, 20 had early ECMO instituted at a mean (standard deviation) of 12.4 (11.4) h post-operatively, 10 of whom had extracorporeal cardiopulmonary resuscitation. Of early-ECMO patients, 8 (40%) were found to have residual anatomic defects requiring intervention with catheterization (n = 1) and/or surgery (n = 7). On multiple regression, the post-operative day 1 highest vasoactive-inotrope score (OR 1.02; 95%CI 1.06,1.08; p < 0.001), highest lactate (OR 1.2; 95%CI 1.06,1.35; p = 0.003), and lowest base deficit (OR 0.82; 95%CI 0.71,0.94; p = 0.004), CPB time (OR 1.01; 95%CI 1.00,1.02; p = 0.002), and single-ventricle anatomy (OR 5.35; 95%CI 1.66,17.31; p = 0.005) were associated with early ECMO. Outcomes at 2 years old compared between early-ECMO and matched patients were mortality 11/20 (55%) vs 11/40 (28%) (OR 3.22, 95%CI 0.98,10.63; p = 0.054) and General Adaptive Composite median 65 [interquartile range (IQR) 58, 81.5] in 9 survivors vs 93 [IQR 86.5, 102.5] in 29 survivors (p = 0.02). Conclusions The identified risk factors for, and outcomes after, having early ECMO may aid decision making in the acute period and confirm that neurodevelopmental follow-up for these children is necessary. The hypothesis that earlier institution of ECMO may improve long-term outcomes requires further study.


Background
Extracorporeal membrane oxygenation (ECMO) has been used to support children with congenital heart disease (CHD) since the 1970s [1,2]. Since then, ECMO has become an accepted intervention for pediatric patients who have failed conventional medical therapy and in whom cardiac and/or respiratory insufficiency is potentially reversible [1,2]. In addition, ECMO has been instituted as rescue during cardiopulmonary resuscitation (E-CPR) and in selected patients as a bridge to transplantation [1][2][3]. In the post-operative period following repair of CHD, veno-arterial ECMO may be required for (i) failure to separate from cardiopulmonary bypass (CPB), with ECMO cannulation performed in the operating room; (ii) progressive hemodynamic or hypoxic deterioration with cardiogenic shock (i.e., low cardiac output syndrome-LCOS) in the pediatric cardiac intensive care unit (PCICU); or (iii) refractory cardiac arrest occurring in the PCICU [1][2][3]. The Extracorporeal Life Support Organization (ELSO) has collected registry data since the 1980s and indicates that ECMO has been used for cardiac support in > 55,000 neonates and children [4]. The registry documents the survival to hospital discharge among 310 centers as follows: in neonatal cardiac ECMO 2695/6475 (42%), neonatal E-CPR 547/1336 (41%), pediatric cardiac ECMO 4265/8374 (51%), and pediatric E-CPR 1232/2996 (41%) [4]. Despite significant advances in long-term ventricular assist devices, ECMO remains the most commonly used form of mechanical cardiopulmonary support in infants and young children in the immediate post-operative period [2,3].
In this study, we aim to (i) determine predictors of early (within 48 h) post-operative veno-arterial ECMO (early ECMO) following cardiac surgery with CPB (or without CPB, but with sternotomy for central shunt or modified Blalock-Taussig shunt (MBTS) or pulmonary artery banding (PAB)) for infants age ≤ 6 weeks and (ii) compare survival and 2-year functional outcomes of patients who had early ECMO to a matched cohort of patients who did not have early ECMO. We used 10 years of data from the interprovincial Western Canadian inception-cohort Complex Pediatric Therapies Follow-up Program (CPTFP) [5]. We hypothesized that markers of severity of post-operative illness predict early ECMO and that outcomes after early ECMO are worse than that in matched patients.

Methods
This study is part of an interprovincial inception cohort outcomes follow-up program conducted in Western Canada, the CPTFP. Infants were identified at the time of complex cardiac surgery with CPB (or without CBP, but with sternotomy for a central shunt or MBTS or pulmonary artery banding (PAB)) and were followed prospectively [5]. Demographic, pre-operative, intra-operative, and post-operative variables that were previously agreed upon were collected prospectively. Long-term follow-up was performed, with parental or guardian consent, during follow-up visits at the tertiary site of origin. The follow-up study and database have been approved by each respective institutional health research ethics board.
In this study, we included all infants ≤ 6 weeks of age having cardiac surgery with CPB (or without CPB, but with sternotomy for central shunt or MBTS or PAB) at the Stollery Children's Hospital from January 1, 2003, to December 31, 2012. Patients having ECMO pre-operatively, transitioned to ECMO in the operating room for failing to separate from CPB, or having ECMO instituted ≥ 48 h post-operatively were excluded. The 48-h post-operative cutoff time for defining early ECMO was chosen as we wanted to identify predictors of veno-arterial ECMO cannulation in those patients who deteriorate early after cardiac surgery and who therefore might have been considered for ECMO cannulation in the operating room (prior to admission post-operatively).
The CPTFP records data prospectively post-operatively on day 1 (defined as the day of surgery). In this study, the acute-care data on day 1 post-operatively did not include values in the 2 h immediately preceding ECMO cannulation or values after ECMO cannulation. The values within 2 h of cannulation are likely to occur when patients are in extremis and thus do not reflect those values that would lead to the decision to cannulate (i.e., are not useful to predict in advance which patients may need ECMO). Acute-care data for patients who had early ECMO were retrospectively confirmed by chart review to be those at least 2 h prior to ECMO cannulation. These variables included the following: highest lactate, highest vasoactive-inotrope score [6], highest epinephrine dose, lowest pH, lowest base deficit, lowest PaO2, highest creatinine, open sternum, steroids started for blood pressure, dialysis, and seizure. We also collected other operating room and post-operative day 1 variables for the early-ECMO patients for descriptive purposes, including pediatric risk of mortality (PRISM III) score and risk adjustment for congenital heart surgery (RACHS-1) category, intra-operative inotrope use and lactate, indication for ECMO, and investigations done peri-operatively (transesophageal echocardiogram (TEE) and cardiac catheterization).
Over the 10-year time period of this study, we have had 4 pediatric cardiac surgeons operating and 11 cardiac intensivists. Clinical practice of the surgeons and intensivists is at their discretion, and we do not have specific protocols for management. Post-operative management is up to the bedside clinicians, usually a collaboration between the cardiac intensivists, cardiac surgeons, and cardiologists, who round each morning as a group on the cardiac patients in the PCICU. The preferred initial vasoactive agent for low cardiac output is epinephrine, and steroids are considered in moderate to severe low cardiac output syndrome (LCOS). Over the years, practices may have changed with emerging research, although we did not find that year of surgery was a predictor of early ECMO ( Table 1). Some of the practice, including catecholamine and steroid use, are given in the "Results" section.
Outcomes assessments were completed at 2 years of age. Pediatric psychologists obtained the results of the Adaptive Behavior Assessment System, 2nd Edition (ABA-S-II), in the patient's respective referral institutions [7]. The preschool version of ABAS-II is a parent/caregiver-completed questionnaire to provide comprehensive, norm-referenced assessment of adaptive skills for children 0-5 years of age. The General Adaptive Composite (GAC) score from the ABAS-II is used to assess adaptive function in children, with a normative population mean of 100 and standard deviation (SD) of 15; higher scores indicate better performance, and a score < 70 is 2 SD below the mean (expected in 2.27% of the normative population).

Statistics
Continuous variables are presented as mean (SD) and median [interquartile range, IQR] as appropriate, and categorical variables are presented as counts (percentages). The first objective was to determine predictors of early ECMO (defined as within 48 h post-operatively) following cardiac surgery at age 6 weeks or less. To screen for variables associated with this outcome, we used univariate regression models including the a priori specified variables given in Table 1. Multiple logistic regression models consisted of variables from Table 1 found approaching significance at p ≤ 0.10 in the univariate analysis and, after screening for multicollinearity, are presented as odds ratios (ORs) with 95% confidence intervals (CIs) and two-sided p values. We explored two multiple regression models: one using highest vasoactive-inotrope score and highest post-operative day 1 lactate as continuous variables, and one using these as dichotomous variables (pre-specified as highest vasoactive-inotrope score > 30 and highest lactate > 6 mmol/L). The second objective was to compare outcomes of patients who had early ECMO to a matched cohort of patients who did not have early ECMO. For each early-ECMO patient, there were two matched patients; matching was for sex, gestational age (± 2 completed weeks), year of surgery (± 2 years), chromosomal abnormality, single-ventricle physiology, and socioeconomic index (± 15 points on a scale from 17.8 to 101.7, with a mean 42.7 and SD 13) [8]. Given the six demographic variables used for matching, it was a priori decided to match 1:2 and not higher in order to avoid inability to adequately match patients. We compared the outcome of 2-year mortality using conditional logistic regression, and results are presented as OR with 95% CI and two-sided p values. We compared the outcome of 2-year GAC score in survivors using Wilcoxon rank sum test. Statistical analyses were performed using SAS version 9.3, and a p value ≤ 0.05 was considered statistically significant.

Description of the cohort
Over the 10-year period, 585 infants less than 6 weeks of age had cardiac surgery with sternotomy. Thirteen patients were excluded as they had ECMO cannulation pre-operatively or in the operating room, and another seven excluded as they had ECMO cannulation more than 48 h post-operatively. Of the 565 patient study cohort, 545 patients did not have ECMO, and 20 patients had veno-arterial early ECMO. The demographic and peri-operative variables for all patients are given in Table 1. Among the 20 patients who had early ECMO, the mean (SD) and median [IQR] time to cannulation for ECMO was 12.4 (11.4) and 8.0 [3.1, 21.5] h post-operatively. There were 10 (50%) cannulated during E-CPR; a description of these patients and comparison between those having E-CPR and non-E-CPR are given in Table 2. The non-E-CPR ECMO patients were cannulated for LCOS (n = 10), sometimes associated with cardiac tamponade (n = 2) or hypoxia (n = 1); two of these patients had a post-operative cardiac arrest requiring CPR and had return of spontaneous circulation prior to ECMO cannulation. The E-CPR patients had cardiac arrest attributed to pulmonary overcirculation (n = 1), progressive LCOS (n = 5), cardiac tamponade (n = 1), possible pulmonary aspiration (n = 1), and sudden unexplained arrest (n = 2). The E-CPR patients had lower inotrope score on post-operative day 1, compared to the non-E-CPR ECMO patients ( Table 2).
The early-ECMO patients were investigated with TEE in 16 (80%) and cardiac catheterization in 9 (45%) post-operatively, with 8 (40%) found to have residual anatomic defects requiring intervention with catheterization (n = 1) and/or surgery (n = 7). Of the 8 with residual anatomic defects, 5 had an intra-operative TEE, 3 of which detected the residual lesion that later worsened post-operatively, and 2 of which missed the residual lesion. Catheter interventions included bilateral stenting of stenotic pulmonary arteries after ECMO decannulation; one other patient had dilation of a coarctation of the aorta 3 months after ECMO decannulation. Surgical interventions included clipping a central shunt due to overcirculation (followed by a right pulmonary artery band after decannulation), Sano revision due to coronary artery compression, left coronary re-implantation due to kinking, left coronary osteoplasty due to stenosis, MBTS and pulmonary valve replacement, left ventricle  Fig. 1. In Fig. 1, we show two thresholds: our pre-specified thresholds of highest lactate > 6 mmol/L and highest vasoactive-inotrope score > 30, and a post hoc threshold of highest lactate > 9 mmol/L and highest vasoactive-inotrope score > 27.

Outcomes of early (within 48 h) post-operative ECMO and matched comparison patients
Comparison of the early-ECMO patients (n = 20) to the matched patients (n = 40) is given in Table 4. As expected, on univariate analysis, early-ECMO patients had higher post-operative day 1 highest lactate and vasoactive-inotrope score, and lower post-operative day 1 base deficit (   PRISM III Score pediatric risk of mortality score, RACHS-1 risk adjustment for congenital heart surgery score, TAPVD total anomalous pulmonary venous drainage, TGA transposition of the great arteries, VSD ventricular septal defect a Surgery performed was as follows: Norwood-Sano for n = 6 HLHS, n = 2 HLHS variants, and n = 1 tricuspid atresia; Norwood-BT for n = 1 HLHS variant; TAPVD repair for n = 2 TAPVD; right ventricle to pulmonary artery conduit with closure of VSD for n = 1 pulmonary atresia; central shunt for n = 1 HLHS variant; pulmonary artery banding for n = 1 tricuspid atresia; arterial switch with repair of VSD, or atrial septal defect and LVOTO, or atrial septal defect, VSD, right ventricle muscle bundles, and LVOTO for n = 3 d-TGA; right ventricle outflow tract obstruction resection and pulmonary valve plasty for n = 1 poly-valvular disease with Noonan's syndrome; Ross-Konno procedure for n = 1 with complex LVOTO (sub-aortic stenosis, aortic valve stenosis, hypoplastic aortic arch with coarctation) post-operative day 1 highest vasoactive-inotrope score and lactate, and lowest base deficit. These findings may help predict the need for early ECMO before emergent deterioration. However, these variables may not be discriminating enough to predict for the individual patient, even in single-ventricle patients. For example, having a vasoactive-inotrope score > 30 occurred in 31% of single ventricle and 43% of biventricular patients having early ECMO but also in 5% and 4% of those not needing early ECMO. Third, compared to a matched cohort of patients, having early ECMO was associated with higher 2-year mortality (OR 3.22, 95% CI 0.98, 10.63; p = 0.054) and with statistically significantly worse functional outcome measured by GAC at 2 years of age. The 48-h post-operative time cutoff was chosen as we wanted to predict patients that deteriorate early after cardiac surgery. We are not aware of previous studies examining specifically prediction of, or outcomes of, early ECMO as defined here. A recent review found that studies of cardiac ECMO cohorts (E-CPR or not) have reported hospital survival in the range of 45-50%, with lower 2-year post-ECMO survival [3]. Consistent predictors of survival have included renal dysfunction, neurologic complication, and highest lactate prior to and during ECMO; the timing of initiation of ECMO was not examined as a predictor in most studies [3]. The few studies that did examine timing of initiation of ECMO post-operatively looked only at cannulation in the operating room vs. in the PCICU at any time post-operatively and had conflicting findings about association with survival [9][10][11][12][13][14]. The one study that examined early versus late post-operative ECMO cannulation defined these as at < 7 days vs. ≥ 7 days post-operatively and found no difference in survival [15]. Review of the ELSO database found that neonatal cardiac ECMO mortality was associated with longer time of mechanical ventilation before ECMO (< 10 h being optimal overall and < 15 h being optimal after stage 1 palliation for hypoplastic left heart syndrome); these findings suggest that early recognition and cannulation for ECMO post-operatively may be associated with improved survival [16,17].
Whether earlier cannulation for ECMO post-operatively or avoiding deterioration to E-CPR can result in better neurologic outcomes is not known. After cardiac ECMO, neurologic outcomes with comprehensive follow-up suggest mental delay in 38% of survivors and no clear difference with E-CPR, but this is based on very small numbers [3]. Recently, the CPTFP reported a review of 10 years of cardiac ECMO in 98 patients and found 65% survival to hospital discharge, 51% survival to 5 years, full-scale intelligence quotient of survivors 79.7 (SD 16.6) with 25% below 2 SD of the population mean, and GAC of 79.2 (SD 19.4); E-CPR and location of cannulation (operating room vs. PCICU) were not associated with survival or neurocognitive outcomes [18]. Of interest, survival improved in the more recent era, and this was associated with lower lactate and inotrope scores prior to ECMO cannulation [18]. In 51 cardiac patients having E-CPR, the CPTFP reported 49% survival to hospital discharge, 43% survival to 5 years, full-scale intelligence quotient of survivors 76.5 (SD 15.9), Data are given as odds ratio (OR) and 95% confidence interval (95% CI) a Two multiple regressions are shown: with inotrope score and lactate as continuous measures (model 1) and with inotrope score and lactate as dichotomous measures (model 2). Dialysis is not used in multiple regressions, as too few patients had dialysis GAC of 74.5 (SD 20.3) with 47% having a score below 2 SD of the population mean, and 24% having both full-scale intelligence quotient and GAC below 2 SD of the population mean, similar findings to those of all cardiac ECMO cases in the 10-year review [19]. When the CPTFP examined health-related quality of life, this was lower in cardiac ECMO survivors than in other children with CHD having surgery in early infancy; whether cannulation was from the operating room or done in the PCICU was not associated with this outcome [20]. Finally, of 502 CPTFP patients having cardiac surgery early in infancy, ECMO was an independent predictor of mortality at 4.5 years (hazard ratio 1.93, 95% CI 1.17, 3.18, p = 0.010) and of neurocognitive outcome, with an effect size for full-scale intelligence quotient of − 13.6 (95% CI − 21.3, − 5.9, p = 0.001) and for GAC of − 12.7 (95% CI − 20.3, − 5.0, p = 0.001) [21]. Although ECMO is associated with adverse outcomes, the timing of ECMO cannulation was not explored [21]. The present results are compatible with these findings from the CPTFP (including no statistically significant difference in outcomes between E-CPR and non-E-CPR early-ECMO patients) and others, and extend these findings by examining predictors and outcomes specifically of patients having early ECMO initiated in the PCICU within 48 h of surgery.
There are some important implications of the findings from this study. First, the finding that 40% of early-ECMO patients had residual anatomic lesions that were corrected by intervention using catheterization or surgery highlights the importance of patient assessment with early TEE and/or cardiac catheterization if patients are not progressing as expected. The safety and importance of TEE and cardiac catheterization to identify and manage residual lesions, particularly in patients on ECMO, has been noted by others [22][23][24][25][26][27][28]. Kato et al. found that patients who received catheterization within 48 h after ECMO cannulation had fewer respiratory complications and better 30-day survival than those who had later catheterization [27]. Agarwal et al. found that earlier detection of residual lesions during the first 3 days of ECMO (compared with later detection) was associated with a higher rate of decannulation and survival to hospital discharge [23]. Howard et al. found that the time to diagnosis or correction of residual lesions was significantly shorter in neonatal cardiac ECMO survivors (1 vs 2 days, p = 0.02) [28]. Second, the variables we found as independently predictive of early ECMO may not be easily modifiable. Vasoactive-inotrope score and base deficit reflect severity of progressive LCOS, single-ventricle anatomy is not modifiable, and CPB time is difficult to change. However, these variables may be used to help predict patients who are at high risk to need post-operative ECMO, should be on an ECMO alert, and should have an early TEE and consideration of a b Fig. 1 Association of day 1 post-operative highest lactate and vasoactive-inotrope score with early ECMO. Receiver operating curves for day 1 post-operative highest a lactate and b inotrope score association with early ECMO. Area under the curve is 0.709 (95% CI 0.581-0.838) and 0.823 (95% CI 0.734-0.911), respectively. The star indicates highest lactate > 6 mmol/L, highest vasoactive-inotrope score > 30; the triangle indicates highest lactate > 9 mmol/L, highest vasoactive-inotrope score > 27  [29]. If we use the cutoffs from this study, highest lactate of > 9 mmol/L had a sensitivity of 40% and specificity of 93%, and highest vasoactive-inotrope score of > 27 a sensitivity of 45% and specificity of 94% for early ECMO. Third, those patients having early ECMO are a high-risk group for mortality and poor functional outcomes. We have previously described that ECMO is a risk factor for adverse neurocognitive and health-related quality of life outcomes [20,21]. These children warrant neurodevelopmental follow-up in order to identify adverse outcomes and provide early intervention in order to optimize each patient's achievement of his/her full potential. This information may also help in counseling family when making decisions about early ECMO. Limitations of this study should be recognized. The small number of patients having early ECMO at a single center limits the generalizability of the findings and the power of the study to determine predictors. The retrospective data collection for some variables is also a limitation; however, this was done only for objective variables recorded in the chart on the first post-operative day in early-ECMO patients. This is an observational study and therefore cannot prove causation and cannot rule out residual confounding. Finally, we must acknowledge adverse outcomes from ECMO. We hypothesize that earlier prediction of the need for, and therefore cannulation for, ECMO is desirable; however, this must be balanced against the adverse effects of unnecessary ECMO resulting from inaccurate prediction. For this reason, better prediction models may be required before our findings can direct decision making. Strengths of this study include that the majority of the data is prospectively collected by the CPTFP, with 2-year follow-up on all early-ECMO patients and 97% of non-ECMO survivors. In addition, we included a large cohort (n = 565) of patients having cardiac surgery in early infancy over a period of 10 years at a referral Western Canadian cardiac surgical center.

Conclusions
Risk factors for early ECMO after cardiac surgery in young infants were identified, including single-ventricle anatomy and early markers of severity of illness (lactate, vasoactive-inotrope score, and base deficit). Early ECMO often (in 40%) indicated a residual anatomic defect requiring intervention and was associated with adverse 2-year outcomes (mortality and GAC score). Further work is necessary to improve prediction of early ECMO in larger cohorts of patients. The hypothesis that earlier institution of ECMO may improve long-term outcomes requires further study. The Western Canadian Complex Pediatric Therapies Follow-up Program has been supported over the years by contributions from the following: Alberta Health, Stollery Children's Hospital, Women and Children's Health Research Institute, referral-site follow-up clinics (Saskatoon and Regina SK, Winnipeg MB, Calgary AB, Vancouver BC), and Glenrose Rehabilitation Hospital Research Trust, with ongoing funding by Glenrose Rehabilitation Hospital. These funding agencies had no role in design and conduct of the study; collection, analysis or interpretation of the data; preparation, writing, review, or approval of the manuscript; or the decision to submit the manuscript for publication.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Authors' contributions GAK, DG, LR, GGG, DM, GYB, CMTR, and ARJ made substantial contributions to conception and design of the work. GAK, DG, CMTR, and GYB made substantial contributions to acquisition of the data. CMTR, IAD, and FM made substantial contributions to analysis of the data. CMTR, GYB, ARJ, and IAD had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. All authors made substantial contributions to interpretation of the data. GAK and ARJ drafted the first version of the manuscript. All authors revised the manuscript critically for important intellectual content. All authors had final approval of the version to be published. All authors agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Ethics approval and consent to participate
The Health Research Ethics Board of the University of Alberta approved this study [Pro00001030]. All parents/guardians signed an informed consent form.

Consent for publication
Not applicable.