Content uploaded by Susan L Bratton
Author content
All content in this area was uploaded by Susan L Bratton on Apr 02, 2018
Content may be subject to copyright.
Pediatric and Neonatal Extracorporeal Membrane Oxygenation;
Does Center Volume Impact Mortality?
Carrie L. Freeman, MD, MA*, Tellen D. Bennett, MD, MS*, T. Charles Casper, PhD*, Gitte Y.
Larsen, MD, MPH*, Ania Hubbard, MD*, Jacob Wilkes, BS#, and Susan L. Bratton, MD, MPH*
*University of Utah, School of Medicine, Department of Pediatrics, Primary Children’s Medical
Center
#Intermountain Health Care
Abstract
Objective—Extracorporeal membrane oxygenation, an accepted rescue therapy for refractory
cardiopulmonary failure, requires a complex multidisciplinary approach and advanced technology.
Little is known about the relationship between a center’s case volume and patient mortality. The
purpose of this study was to analyze the relationship between hospital extracorporeal membrane
oxygenation annual volume and in-hospital mortality and assess if a minimum hospital volume
could be recommended.
Design—Retrospective cohort study
Setting—A retrospective cohort admitted to children’s hospitals in the Pediatric Health
Information System database from 2004-2011 supported with extracorporeal membrane
oxygenation was identified. Indications were assigned based on patient age (neonatal vs.
pediatric), diagnosis, and procedure codes. Average hospital annual volume was defined as 0-19,
20-49, or ≥50 cases per year. Maximum likelihood estimates were used to assess minimum annual
case volume.
Patients—A total of 7322 pediatric patients aged 0-18 years of age were supported with
extracorporeal membrane oxygenation and had an indication assigned.
Interventions—None
Measurements and Main Results—Average hospital extracorporeal membrane oxygenation
volume ranged from 1-58 cases per year. Overall mortality was 43% but differed significantly by
indication. After adjustment for case-mix, complexity of cardiac surgery, and year of treatment,
patients treated at medium (OR 0.86, 95% CI 0.75-0.98) and high (OR 0.75, 95% CI 0.63-0.89)
volume centers had significantly lower odds of death compared to those treated at low volume
centers. The minimum annual case load most significantly associated with lower mortality was 22
(95% CI 22-28).
Corresponding Author Carrie Freeman, MD, Pediatric Critical Care Medicine, 295 Chipeta Way, P.O. Box 581289, Salt Lake City,
UT 84158, Phone 801-587-7572, Fax 801-581-8686, cnfreeman@umc.edu.
HHS Public Access
Author manuscript
Crit Care Med. Author manuscript; available in PMC 2015 August 25.
Published in final edited form as:
Crit Care Med. 2014 March ; 42(3): 512–519. doi:10.1097/01.ccm.0000435674.83682.96.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Conclusion—Pediatric centers with low extracorporeal membrane oxygenation average annual
case volume had significantly higher mortality and a minimum volume of 22 cases per year was
associated with improved mortality. We suggest this threshold be evaluated by additional study.
Keywords
Pediatrics; Extracorporeal Membrane Oxygenation; Hospitals; Low-Volume; Critical Care; Risk
Adjustment; Cardiopulmonary Resuscitation
Introduction
Extracorporeal membrane oxygenation (ECMO) provides prolonged partial
cardiopulmonary bypass and has been used for infants and children with severe
cardiopulmonary failure unresponsive to conventional therapy since 1975.(1-3) More
recently, this complex technology has been successfully used emergently to rescue “failing”
cardiopulmonary resuscitation (E-CPR).(4-6) Initial successful applications of ECMO were
almost exclusively among term neonates with pulmonary hypertension; however, ECMO
has increasingly been used to support older children and adults with both cardiorespiratory
failure and cardiac arrest.(6-8) Practice in the United Kingdom has focused on regional
ECMO referral centers while development in the US has not been centralized. (7, 9)
There are numerous reports regarding increasing surgical experience and center volume
demonstrating lower mortality in many high risk surgical procedures.(10-13) These
observations led to recommendations regarding minimum volume standards for some
surgical procedures.(12) The favorable relationship between increasing volume and
improved outcome also exists for infants and children with some complex conditions.(14,
15)
Given that pediatric and neonatal ECMO are highly complex medical-surgical endeavors, a
reasonable hypothesis is that center experience and volume may be associated with
mortality. There are no large multicenter reports addressing pediatric ECMO center volume
and survival. We utilized a large administrative pediatric database to determine if after
adjustment for case mix, center volume was associated with mortality. Our hypothesis was
that an inverse relationship existed between ECMO center volume and mortality. Because
applications of ECMO are expanding among both children and adults, study of this high cost
rescue therapy is increasingly important.
Material and Methods
Data Source
The Pediatric Health Information System (PHIS) database, a multi-center administrative
database with data from over 40 children’s hospitals in the United States was used.
Participating hospitals provide data on demographics, outcomes, diagnoses, procedures, and
charges using Clinical Transaction Classification™ (CTC) codes for billed services. (16, 17)
Data are de-identified centrally which qualified for exemption from human subjects review
by the University of Utah Institutional Review Board.
Freeman et al. Page 2
Crit Care Med. Author manuscript; available in PMC 2015 August 25.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Patients
Patients admitted between January 1, 2004-December 31, 2011, <18 years of age, with an
International Classification of Diseases, 9th revision, clinical modification (ICD-9-CM)
procedure code for ECMO (39.65) or CTC code for ECMO (521181) were evaluated for
inclusion.
Diagnosis Groups
Diagnostic categorization emulated categorizations used by the Extracorporeal Life Support
Organization (ELSO) ECMO indications (Figure 1). See details of the diagnostic
categorization in the data supplement and Appendix 1. Seven diagnostic categories were
defined: congenital diaphragmatic hernia (CDH), neonatal or pediatric respiratory failure,
neonatal or pediatric cardiac disease, and neonatal or pediatric cardiac arrest. Available data
could not distinguish a cardiac arrest prior to initiation of ECMO from an ongoing arrest
when starting ECMO (i.e. E-CPR). All patients with a cardiac arrest were classified as
neonatal or pediatric cardiac arrest regardless of other diagnosis codes except for congenital
diaphragmatic hernia (CDH) as cardiac arrest in this group is rarely the indication for
ECMO(4).
Study Variables
The primary outcome was in-hospital mortality and primary exposure was annual hospital
ECMO volume. Covariates included demographics, year of admission as well ECMO
indication. Risk Adjustment for Congenital Heart Surgery (RACHS-1) was used to adjust
for complexity of cardiac surgical repair, as mortality is increased for patients with single
ventricle physiology after both cardiac surgery and E-CPR.(18-20)
Hospital Variables
We created an average annual ECMO volume for each hospital over the study period using
quarterly data and averaged to cases per year. Empirically, hospital volume was categorized
as low, medium or high; 0-19, 20-49, and ≥50 average ECMO cases per year based on
clinical assessment. Average annual ECMO volume was also evaluated continuously.
Subgroup Analysis
Two additional diagnostic categories, respiratory syncytial virus (RSV) bronchiolitis and
Stage 1 palliation in hypoplastic left heart syndrome (HLHS) were created due to their
consistent coding to evaluate homogenous groups. (Appendix 1)
Statistical Analyses
Statistical analyses were performed using SPSS 18.0 (Chicago, IL) and the R Language and
Environment. (21) Categorical data were compared using the chi square test and continuous
data using the Wilcoxon Rank Sum test; p <0.05 was considered significant. Multivariable
logistic regression was used to evaluate ECMO volume and hospital mortality. Center case-
mix was adjusted for indication for ECMO which included age, ECMO support year, and
RACHS-1 scores for classified congenital cardiothoracic procedures.
Freeman et al. Page 3
Crit Care Med. Author manuscript; available in PMC 2015 August 25.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
We also sought to evaluate a potential “cut point” for minimal annual ECMO volume
associated with improved survival using a maximum likelihood approach. This approach is
based on the assumptions that such a cut point exists and patients at centers falling on the
same side of the cut point have the same chance of survival. A likelihood was calculated for
each possible cut point and the optimal point was chosen as the value providing the highest
likelihood. To assess the precision of the cut point estimate, a confidence interval was
calculated using a nonparametric bootstrap method. (22)
Results
7322 children meeting study criteria underwent ECMO support from 2004-2011. Children’s
hospitals within this cohort performed an average of 1 to 58 ECMO cases per year. Overall
in-hospital mortality was 43%. Comparing patients who survived to those who died (Table
1), there were significant differences related to patient age, indication for ECMO, year of
ECMO support, length of stay, as well as treating hospital ECMO volume. 15 hospitals were
categorized as low, 22 as medium, and 3 as high volume centers representing 16%, 69%,
and 15% of the patient cohort respectively.
Table 2 describes patient characteristics by ECMO volume category. Overall mortality was
significantly higher at low volume centers (47%) compared to medium and high volume
centers (42 and 41%) (p = 0.01). However, indication for ECMO also differed significantly
by center volume categories with more cardiac disease patients in the low ECMO volume
group, more cardiac arrest cases treated at high volume centers while neonatal respiratory
cases were more common at low and medium volume centers. ECMO indications by patient
age groups (neonatal vs. older children) are highlighted in Table 2.
After adjusting for these potential confounders, significantly higher mortality persisted at
low volume centers compared to medium [OR 0.86; (95% confidence interval 0.75-0.98)]
and high volume centers [OR 0.75; (95% CI 0.63-0.89)] (Table 3). Age was included within
indication for ECMO as neonatal versus older patients. A second logistic regression model
including average center volume as a continuous variable found that for each additional 10
patients per year, the odds of mortality decreased 5 percent (OR 0.95, 95% CI 0.92-0.98).
Subgroup Analysis
Due to concern regarding potential misclassification of ECMO indication, a subset of
patients who had consistent ICD-9-CM procedure and diagnosis coding was evaluated. CDH
(n=1016), HLHS with stage 1 palliation surgery (n=522), and patients with RSV
bronchiolitis (n=217) were identified using the ICD-9-CM procedure and diagnosis codes
described in Appendix 1. Table 4 shows this subset and compares mortality by center
volume. In-hospital mortality differed by indication and was 54% for CDH, 31% for RSV,
and 62% for HLHS undergoing stage 1 palliation. A similar multivariable analysis of this
subset adjusting for primary diagnosis as well as presence of a cardiac arrest and year of
treatment found that patients treated at both medium and high ECMO volume centers had
significantly lower odds of mortality [OR 0.74 (95% CI, 0.56-0.98) and 0.59 (95% CI,
0.42-0.83) respectively] compared to low volume centers.
Freeman et al. Page 4
Crit Care Med. Author manuscript; available in PMC 2015 August 25.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Finally, we evaluated center average annual ECMO volume and unadjusted mortality.
(Figure 2) Evaluating death and the annual ECMO volume at each center, the maximum
likelihood estimate of the optimal cutoff for volume was a minimum of 22 ECMO cases per
year. An identical result was found when risk factors were included in logistic regression
models. This was also the cutoff that produced the most significant difference between high
and low volume centers (p=0.00001). We found no evidence that the model assumptions
were violated. The 95% bootstrap confidence interval, from both univariate and multivariate
models, was (22-28 average annual cases).
Discussion
In this large retrospective multicenter database, we found that ECMO centers caring for
fewer than 20 ECMO cases annually had significantly higher case-mix adjusted mortality
than centers with larger ECMO volume. Centers had wide variation in application of ECMO
by indication as well as length of stay. However, when defining indications in a manner
similar to ELSO and utilizing a subgroup analysis, we continued to find a survival benefit
for infants and children treated at medium to large ECMO volume centers compared to those
treated at smaller centers. There was no significant difference in mortality between the
medium and high volume centers. ECMO requires complex coordination of multiple
providers to deliver care. Logically such care would appear sensitive to case volume;
however, this is the first large evaluation of case mix adjusted pediatric ECMO volume and
mortality.
Numerous studies have suggested an inverse relationship between surgical volume and
mortality.(12, 23) Bucher et al describes the positive impact of volume on in-hospital
mortality in infants with congenital diaphragmatic hernia also utilizing the PHIS database.
(14) Several reports found an association between small surgical volume and increased
mortality.(15, 24, 25) In addition, recent reports have found an increasingly complex
relationship with decreased mortality overall and the greatest difference in survival shown in
the most complex conditions.(26, 27) Extracorporeal Life Support Organization (ELSO)
does suggest that ECMO centers perform a minimum of 6 ECMO cases annually; (28)
however, this is based on expert opinion.
For our study, ECMO indications were based on diagnosis codes and age as PHIS does not
have data regarding specific indication for ECMO. Centers differed both in annual case
volume and case mix. Survival with ECMO support differs by indication for
cardiorespiratory failure with the lowest mortality among neonates with respiratory failure
and substantially higher mortality for patients with cardiac failure after surgery for
congenital heart disease, cardiac arrest, and E-CPR.(29-35) The recent 2012 ELSO
international report of infants with CDH treated from 2004-2011 had an average annual
survival of 46% mirroring our results.(36) Likewise, patients with pediatric respiratory
failure requiring ECMO had the same average annual survival, 56%, echoing the in-hospital
mortality of our cohort with similar diagnoses.(29) Sherwin and colleagues found a 69%
mortality after stage one palliation in patients with HLHS supported by ECMO, similar to
our subgroup analysis mortality (62%) that included cardiac arrest patients who may be
classified as E-CPR cases in ELSO.(19, 32) Our neonatal cardiac arrest survival was 43%
Freeman et al. Page 5
Crit Care Med. Author manuscript; available in PMC 2015 August 25.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
and pediatric cardiac arrest survival was 52% which are similar to recent survival reported
with E-CPR (44-47%).(33-35)
The post hoc analysis for an annual volume threshold of 22 cases is substantially greater
than the ELSO recommendation of 6 cases per year. The PHIS hospitals are predominately
large free standing US Children’s Hospitals and likely are not representative of all ECMO
centers. Furthermore, information regarding the ECMO program structure at each hospital is
not available. Some institutions have a centralized unit and medical supervision for patients
on ECMO whereas others offer ECMO in several different locations and medical
supervision ranges from a core group to inclusion of all critical care physicians.
Unfortunately, evaluation of whether survival is affected by only hospital volume versus
provider volume and or specific intensive care unit (i.e. neonatal vs. pediatric vs. cardiac)
volume was not possible. The consistent association of higher mortality at small volume
centers should be validated by additional study. However, our findings lend support to the
regionalized approach used in the United Kingdom, although there is potential risk in
transporting these critically ill ECMO patients.
Our study is limited by the retrospective and observational nature of the data, and that many
ECMO specific data were not prospectively collected. When using ICD-9-CM codes, many
patients have overlapping codes and we chose to devise rules for diagnostic indications that
mirrored definitions used in the ELSO registry to enable comparison. Our method certainly
misclassifies some cases; however, our data regarding survival by indication are generally
similar to other reports. For instance, neonates sometimes had respiratory codes appropriate
for older ages; therefore assumptions were necessary regarding age. However, analysis of a
subset of pediatric patients with CDH, RSV and HLHS (more clearly defined diagnoses),
despite a smaller number in the cohort, found consistently higher case-mix adjusted
mortality at the low volume centers with an even stronger association.
Another limitation is in patients with cardiac arrest. This important diagnosis in the ECMO
patient population could not be accurately placed in time relative to ECMO cannulation:
prior to cannulation, during cannulation (E-CPR), or after going on ECMO. We were unable
to more finely adjust for severity of illness with respect to progressive organ failure not
captured by diagnosis coding. Finally, ECMO features such as mode of support, duration
and ECMO related complications cannot be ascertained reliably with this data source.
Severity of illness could not be fully adjusted for in this study and therefore our adjusted
mortality evaluation has limitations. In addition, centers may have different thresholds for
the application of ECMO support and inclusion and exclusion criteria likely vary. For
example, some larger volume centers may have a lower threshold for institution of ECMO
support due to experience and comfort with this advanced support. Additionally, larger
centers may care for a higher complexity of patients and the estimated mortality benefit seen
could have been underestimated. Clearly, these center differences could affect mortality, but
we are unable to test for these potential differences with the data available.
Additionally, there are also reports that demonstrate patient specific differences when
comparing administrative databases and clinical databases. (37) However, this imprecision
is unlikely to substantially affect our primary analysis which classified patients simply as
Freeman et al. Page 6
Crit Care Med. Author manuscript; available in PMC 2015 August 25.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
having heart disease or cardiac surgery as an indication for ECMO rather than a specific
anatomic diagnosis. The subset analysis did identify patients with HLHS palliated with a
Norwood procedure. We used diagnosis and procedure codes which Pasquali et al have
shown to differ when comparing registry and administrative data for the Norwood
procedure. When the Norwood procedure was identified in two databases, they found a 7%
difference in patient number and an absolute mortality difference of 1.7%. We used the
same codes to identify all patients in the PHIS database. When comparing mortality
difference in Norwood patients supported by ECMO, we found a 9% mortality difference
between high (58%) and low (67%) volume centers, and expect that this relatively large
mortality difference would likely persist even if some patients were misclassified by
diagnosis/procedure because the mortality for all neonates with cardiac disease was 47%.
Given these limitations, this is the first large multicenter report to describe this inverse
relationship and many of our data did closely mirror those of other ECMO literature as well
as the most recent ELSO July 2102 International Summary. (36)
Conclusions
These findings suggest a minimum ECMO volume may be required to maximize ECMO
program performance and achieve better survival. Regionalization of pediatric and neonatal
ECMO centers when geographically possible, may improve survival. Improved competence
may enable centers to focus improvements and successfully care for higher risk cases.
Additional investigation into a potential minimum volume for neonatal and pediatric ECMO
is needed, and a minimum threshold of 20-22 cases per year may provide the framework for
such continued evaluation. Merging information from complementary databases such as
ELSO and PHIS would likely provide useful information to improve knowledge related to
ECMO indications, complications and survival.
Supplementary Material
Refer to Web version on PubMed Central for supplementary material.
References
1. Bartlett RH, Gazzaniga AB, Jefferies MR, et al. Extracorporeal membrane oxygenation (ECMO)
cardiopulmonary support in infancy. Transactions - American Society for Artificial Internal Organs.
1976; 22:80–93. [PubMed: 951895]
2. Bartlett RH, Andrews AF, Toomasian JM, et al. Extracorporeal membrane oxygenation for newborn
respiratory failure: forty-five cases. Surgery. 1982; 92(2):425–433. [PubMed: 7101133]
3. Bartlett RH, Roloff DW, Cornell RG, et al. Extracorporeal circulation in neonatal respiratory
failure: a prospective randomized study. Pediatrics. 1985; 76(4):479–487. [PubMed: 3900904]
4. Thiagarajan RR, Laussen PC, Rycus PT, et al. Extracorporeal membrane oxygenation to aid
cardiopulmonary resuscitation in infants and children. Circulation. 2007; 116(15):1693–1700.
[PubMed: 17893278]
5. Shin TG, Choi JH, Jo IJ, et al. Extracorporeal cardiopulmonary resuscitation in patients with
inhospital cardiac arrest: A comparison with conventional cardiopulmonary resuscitation. Critical
care medicine. 2011; 39(1):1–7. [PubMed: 21057309]
6. Morris MC, Wernovsky G, Nadkarni VM. Survival outcomes after extracorporeal cardiopulmonary
resuscitation instituted during active chest compressions following refractory in-hospital pediatric
Freeman et al. Page 7
Crit Care Med. Author manuscript; available in PMC 2015 August 25.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
cardiac arrest. Pediatric critical care medicine : a journal of the Society of Critical Care Medicine
and the World Federation of Pediatric Intensive and Critical Care Societies. 2004; 5(5):440–446.
7. Peek GJ, Mugford M, Tiruvoipati R, et al. Efficacy and economic assessment of conventional
ventilatory support versus extracorporeal membrane oxygenation for severe adult respiratory failure
(CESAR): a multicentre randomised controlled trial. Lancet. 2009; 374(9698):1351–1363.
[PubMed: 19762075]
8. Unosawa S, Sezai A, Hata M, et al. Long-term outcomes of patients undergoing extracorporeal
membrane oxygenation for refractory postcardiotomy cardiogenic shock. Surgery today. 2012
9. Petrou S, Bischof M, Bennett C, et al. Cost-effectiveness of neonatal extracorporeal membrane
oxygenation based on 7-year results from the United Kingdom Collaborative ECMO Trial.
Pediatrics. 2006; 117(5):1640–1649. [PubMed: 16651318]
10. Luft HS, Bunker JP, Enthoven AC. Should operations be regionalized? The empirical relation
between surgical volume and mortality. The New England journal of medicine. 1979; 301(25):
1364–1369. [PubMed: 503167]
11. Houghton A. Variation in outcome of surgical procedures. The British journal of surgery. 1994;
81(5):653–660. [PubMed: 8044538]
12. Birkmeyer JD, Finlayson EV, Birkmeyer CM. Volume standards for high-risk surgical procedures:
potential benefits of the Leapfrog initiative. Surgery. 2001; 130(3):415–422. [PubMed: 11562662]
13. Birkmeyer JD, Siewers AE, Finlayson EV, et al. Hospital volume and surgical mortality in the
United States. The New England journal of medicine. 2002; 346(15):1128–1137. [PubMed:
11948273]
14. Bucher BT, Guth RM, Saito JM, et al. Impact of hospital volume on in-hospital mortality of infants
undergoing repair of congenital diaphragmatic hernia. Annals of surgery. 2010; 252(4):635–642.
[PubMed: 20881770]
15. Checchia PA, McCollegan J, Daher N, et al. The effect of surgical case volume on outcome after
the Norwood procedure. The Journal of thoracic and cardiovascular surgery. 2005; 129(4):754–
759. [PubMed: 15821640]
16. Weiss PF, Klink AJ, Hexem K, et al. Variation in inpatient therapy and diagnostic evaluation of
children with Henoch Schonlein purpura. The Journal of pediatrics. 2009; 155(6):812–818 e811.
[PubMed: 19643437]
17. Conway PH, Keren R. Factors associated with variability in outcomes for children hospitalized
with urinary tract infection. The Journal of pediatrics. 2009; 154(6):789–796. [PubMed:
19324369]
18. Flick RP, Sprung J, Gleich SJ, et al. Intraoperative extracorporeal membrane oxygenation and
survival of pediatric patients undergoing repair of congenital heart disease. Paediatric anaesthesia.
2008; 18(8):757–766. [PubMed: 18544147]
19. Chan T, Thiagarajan RR, Frank D, et al. Survival after extracorporeal cardiopulmonary
resuscitation in infants and children with heart disease. The Journal of thoracic and cardiovascular
surgery. 2008; 136(4):984–992. [PubMed: 18954640]
20. Jenkins KJ. Risk adjustment for congenital heart surgery: the RACHS-1 method. Seminars in
thoracic and cardiovascular surgery Pediatric cardiac surgery annual. 2004; 7:180–184. [PubMed:
15283367]
21. R Core Team. R: A language and environment for statistical computing. R Foundation for
Statistical Computing. 2012. cited Available from: http://www.R-project.org/
22. Efron B. Bootstrap Methods: Another Look at the Jackknife. Annals of Statistics. 1979; 7(1):1–26.
23. Kelley-Quon LI, Tseng CH, Jen HC, et al. Hospital type predicts surgical complications for infants
with hypertrophic pyloric stenosis. The American surgeon. 2012; 78(10):1079–1082. [PubMed:
23025945]
24. Hornik CP, He X, Jacobs JP, et al. Relative impact of surgeon and center volume on early
mortality after the Norwood operation. The Annals of thoracic surgery. 2012; 93(6):1992–1997.
[PubMed: 22516833]
25. Pasquali SK, Li JS, Burstein DS, et al. Association of center volume with mortality and
complications in pediatric heart surgery. Pediatrics. 2012; 129(2):e370–376. [PubMed: 22232310]
Freeman et al. Page 8
Crit Care Med. Author manuscript; available in PMC 2015 August 25.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
26. Chan T, Pinto NM, Bratton SL. Racial and insurance disparities in hospital mortality for children
undergoing congenital heart surgery. Pediatric cardiology. 2012; 33(7):1026–1039. [PubMed:
22349675]
27. Welke KF, O’Brien SM, Peterson ED, et al. The complex relationship between pediatric cardiac
surgical case volumes and mortality rates in a national clinical database. The Journal of thoracic
and cardiovascular surgery. 2009; 137(5):1133–1140. [PubMed: 19379979]
28. Extracorporeal Life Support Organization. [July 2012] ELSO Guidelines for ECMO Centers. 2010.
Available from: http://www.elsonet.org/index.php/resources/guidelines.html
29. Zabrocki LA, Brogan TV, Statler KD, et al. Extracorporeal membrane oxygenation for pediatric
respiratory failure: Survival and predictors of mortality. Critical care medicine. 2011; 39(2):364–
370. [PubMed: 20959787]
30. Cengiz P, Seidel K, Rycus PT, et al. Central nervous system complications during pediatric
extracorporeal life support: incidence and risk factors. Critical care medicine. 2005; 33(12):2817–
2824. [PubMed: 16352965]
31. Rajagopal SK, Almond CS, Laussen PC, et al. Extracorporeal membrane oxygenation for the
support of infants, children, and young adults with acute myocarditis: a review of the
Extracorporeal Life Support Organization registry. Critical care medicine. 2010; 38(2):382–387.
[PubMed: 19789437]
32. Sherwin ED, Gauvreau K, Scheurer MA, et al. Extracorporeal membrane oxygenation after stage 1
palliation for hypoplastic left heart syndrome. The Journal of thoracic and cardiovascular surgery.
2012; 144(6):1337–1343. [PubMed: 22503203]
33. Raymond TT, Cunnyngham CB, Thompson MT, et al. Outcomes among neonates, infants, and
children after extracorporeal cardiopulmonary resuscitation for refractory inhospital pediatric
cardiac arrest: a report from the National Registry of Cardiopulmonary Resuscitation. Pediatric
critical care medicine : a journal of the Society of Critical Care Medicine and the World
Federation of Pediatric Intensive and Critical Care Societies. 2010; 11(3):362–371.
34. Wolf MJ, Kanter KR, Kirshbom PM, et al. Extracorporeal cardiopulmonary resuscitation for
pediatric cardiac patients. The Annals of thoracic surgery. 2012; 94(3):874–879. discussion
879-880. [PubMed: 22698774]
35. Huang SC, Wu ET, Wang CC, et al. Eleven years of experience with extracorporeal
cardiopulmonary resuscitation for paediatric patients with in-hospital cardiac arrest. Resuscitation.
2012; 83(6):710–714. [PubMed: 22306256]
36. Extracorporeal Life Support Organization. [2012 October 18] ECLS Registry Report, International
Summary. Jul. Available from: http://www.elsonet.org/index.php/registry/statistics/reports.html
37. Pasquali SK, Peterson ED, Jacobs JP, et al. Differential case ascertainment in clinical registry
versus administrative data and impact on outcomes assessment for pediatric cardiac operations.
The Annals of thoracic surgery. 2013; 95(1):197–203. [PubMed: 23141907]
Freeman et al. Page 9
Crit Care Med. Author manuscript; available in PMC 2015 August 25.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Figure 1.
Flowchart of Cohort Inclusions, Exclusions and Diagnostic Categorization
Freeman et al. Page 10
Crit Care Med. Author manuscript; available in PMC 2015 August 25.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Figure 2.
Average annual ECMO volume and mortality by center. PHIS centers listed in order of
increasing average annual ECMO volume, shown by bars. The solid line represents overall
in-hospital mortality by center. The dashed line is a weighted least squares regression line
for the relationship between hospital ECMO volume ranking and in-hospital mortality. Each
individual patient is weighted equally so the slope of the line is not disproportionately
influenced by any single center. The downward slope shows decreasing mortality with
increasing center volume.
Freeman et al. Page 11
Crit Care Med. Author manuscript; available in PMC 2015 August 25.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Freeman et al. Page 12
Table 1
Patient Demographics and ECMO Center Volume Comparing Pediatric Survivors to Non-Survivors
Variable Survivors Non-survivors p-value
N=4191 N=3131
n (%) n (%)
Age <0.001
0-7 days 2342 (56) 1760 (56)
8-30 days 156 (4) 156 (5)
31-365 days 721 (17) 494 (16)
1-10 years 663 (16) 432 (14)
>10 years 309 (7) 289 (9) 0.83
Male 2341 (56) 1741 (56)
Race <0.001
Black 814 (19) 453 (15)
White 1988 (47) 1493 (43)
Hispanic 656 (16) 479 (15)
Asian 91 (2) 78 (3)
Other 518 (12) 442 (14)
Unknown 124 (3) 186 (6)
Insurance 0.15
Public 2122 (51) 1579 (50)
Private 1417 (34) 1120 (36)
No insurance 87 (2) 63 (2)
Other 411 (10) 259 (8)
Unknown 154 (4) 110 (4)
Length of Stay (days)† 38 (21, 66) 19 (8, 19) <0.001
Indication for ECMO <0.001
Neonatal Respiratory Failure 986 (24) 236 (8)
Congenital Diaphragmatic Hernia 475 (11) 549 (18)
Neonatal Cardiac Arrest 417 (10) 555 (18)
Neonatal Cardiac Disease 590 (14) 531 (17)
Pediatric Respiratory Failure 511 (12) 394 (13)
Pediatric Cardiac Arrest 636 (15) 582 (19)
Pediatric Cardiac Disease 576 (14) 284 (9)
Year of ECMO 0.02
2004-2007 1858 (44) 1473 (47)
2008-2011 2333 (56) 1658 (53)
Center Volume (Average ECMO cases/year) 0.01
Low (0-19) (15 hospitals) 619 (15) 539 (17)
Medium (20-49) (22 hospitals) 2909 (69) 2137 (68)
Crit Care Med. Author manuscript; available in PMC 2015 August 25.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Freeman et al. Page 13
Variable Survivors Non-survivors p-value
N=4191 N=3131
n (%) n (%)
High (≥50) (3 hospitals) 663 (16) 455 (15)
Crit Care Med. Author manuscript; available in PMC 2015 August 25.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Freeman et al. Page 14
Table 2
Patient Characteristics by Center ECMO Volume
Characteristic Low Medium High p-value
N=1158 N=5046 N=1118
Overall Mortality 539 (47) 2137 (42) 455 (41) 0.01
Neonatal ECMO (<31 days) N=682 (59) N=3050 (60) N=607 (54)
Indication for ECMO n (%) n (%) n (%) <0.001
Respiratory Failure 188 (28) 885 (29) 149 (25)
Congenital Diaphragmatic Hernia 153 (22) 724 (24) 147 (24)
Cardiac Disease 202 (30) 792 (26) 127 (21)
Cardiac Arrest 139 (20) 649 (21) 184 (30)
Neonatal Mortality 320 (47) 1292 (42) 259 (43) 0.09
ECMO Year 0.001
2004-2007 370 (54) 1459 (48) 268 (44)
2008-2011 312 (46) 1591 (52) 339 (56)
Length of Stay (days)† 32 (16, 62) 31 (17, 58) 31 (16, 60) 0.67
Pediatric ECMO N=476 (41) N=1996 (40) N=511 (46)
Indication for ECMO n (%) n (%) n (%) 0.108
Respiratory Failure 159 (33) 580 (29) 166 (33)
Cardiac Disease 120 (25) 603 (30) 137 (27)
Cardiac Arrest 197 (41) 813 (41) 208 (41)
Pediatric Mortality 219 (46) 845 (42) 196 (38) 0.05
ECMO Year 0.08
2004-2007 211 (44) 797 (40) 226 (44)
2008-2011 265 (56) 1199 (60) 285 (56)
Length of Stay (days)† 23 (9, 47) 28 (11, 56) 26 (13, 47) 0.04
Crit Care Med. Author manuscript; available in PMC 2015 August 25.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Freeman et al. Page 15
Table 3
Center Volume and Mortality Risk Model
Factor Odds Ratio (95% confidence Interval)
Center Volume
Low 1 Reference group
Medium 0.86 0.75-0.98
High 0.75 0.63-0.89
Indications for ECMO
Neonatal Respiratory Failure 1 Reference group
Congenital Diaphragmatic Hernia 4.94 4.09-5.96
Neonatal Cardiac Disease 4.09 3.32-5.03
Neonatal Cardiac Arrest 6.21 4.97-7.76
Pediatric Respiratory Failure 3.28 2.70-3.99
Pediatric Cardiac Disease 2.54 2.04-3.16
Pediatric Cardiac Arrest 4.50 3.71-5.46
Years treated
2004-07 1 Reference group
2008-10 0.86 0.78-0.95
Crit Care Med. Author manuscript; available in PMC 2015 August 25.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Freeman et al. Page 16
Table 4
Subgroup Analysis of Mortality and Center Volume
CDH RSV HLHS Stage 1 Palliation p-value
N=1016 N=217 N=522
n (%) n (%) n (%)
Age †(days) 0 (0,1) 30 (1,654) 0 (0,1) <0.001
Male 596 (59) 136 (63) 305 (58) 0.52
Year of ECMO 0.01
2004-2007 500 (49) 81 (37) 245 (47)
2008-2011 516 (51) 136 (63) 277 (53)
Cardiac Arrest 86 (9) 71 (31) 303 (58) <0.001
Mortality 544 (54) 90 (42) 321 (62) <0.001
In-Hospital Mortality by Center Volume p-value 0.26 p-value 0.002 p-value 0.528
Low N=152 N=34 N=67
Number of deaths, n (%) 87 (57) 23 (68) 45 (67)
Medium N=718 N=132 N=366
Number of deaths, n (%) 387 (54) 48 (39) 224 (61)
High N=146 N=60 N=89
Number of deaths, n (%) 70 (48) 19 (32) 52 (58)
Congenital diaphragmatic hernia (CDH), respiratory syncytial virus (RSV), hypoplastic left heart syndrome (HLHS)
Crit Care Med. Author manuscript; available in PMC 2015 August 25.