Predictors of outcome in patients with congenital diaphragmatic hernia requiring extracorporeal membrane oxygenation

Journal of Pediatric Surgery (2007) 42, 1345 – 1350

www.elsevier.com/locate/jpedsurg

Predictors of outcome in patients with congenital diaphragmatic hernia requiring extracorporeal membrane oxygenation Ravindranath Tiruvoipatia,b,*, Yana Vinogradovac, Gail Faulknera,b, Andrzej W. Sosnowskia,b, Richard K. Firmina,b, Giles J. Peeka,b a

Department of ECMO, Glenfield Hospital, LE3 9QP Leicester, United Kingdom Department of Cardiac Surgery, Glenfield Hospital, LE3 9QP Leicester, United Kingdom c Division of Primary Care, University of Nottingham, University Park, NG7 2RD Nottingham, United Kingdom b

Index words: Diaphragmatic hernia; ECMO

Abstract Background: The role of extracorporeal membrane oxygenation (ECMO) in patients with congenital diaphragmatic hernia is still evolving. The use of ECMO is invasive with potential complications during instrumentation for cannulation and heparinization. There are no reliable predictors of outcome in patients requiring ECMO. We aimed to identify (a) the factors that could predict outcome and (b) the incidence and relation of complications during ECMO to outcome. Methods: bPreQ ECMO (age, sex, birth weight, blood gasses, and ventilator settings) and bonQ ECMO variables (mode of ECMO, use of nitric oxide, surfactant, liquid ventilation, inotropes, timing of repair, and complications on ECMO) were analyzed to identify predictors of outcome. Results: Fifty-two patients were included. The overall survival was 58%. Mean duration of ECMO (181 F 120 vs 317 F 156 hours, P = .001), use of nitric oxide (6 vs 10, P = .049), and renal complications (4 vs 14; P b .001) differed between survivors and nonsurvivors. The survival of patients requiring ECMO support for more than 2 weeks is significantly lower than that of patients requiring ECMO support for less than 2 weeks (18% vs 68%, P = .005). Multiple logistic regression revealed ECMO duration of 2 weeks or more and renal complications to be associated with mortality. Conclusion: No pre-ECMO variable could be identified as predictor of mortality. Prolonged duration of ECMO and renal complications on ECMO were independently associated with mortality. D 2007 Elsevier Inc. All rights reserved.

The management of congenital diaphragmatic hernia (CDH) remains a challenging problem in spite of advances

* Corresponding author. Department of ECMO, Glenfield Hospital, LE3 9QP Leicester, United Kingdom. Tel.: +44 7789 716 818; fax: +44 1162 502 374. E-mail address: [email protected] (R. Tiruvoipati). 0022-3468/$ – see front matter D 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.jpedsurg.2007.03.031

in antenatal and neonatal intensive care. In addition to mechanical ventilation, use of other treatments such as antenatal steroids, fetal surgery, nitric oxide (NO), surfactant, liquid ventilation, high-frequency oscillatory ventilation (HFOV), and extracorporeal membrane oxygenation (ECMO) have all been tried with variable results. Although the ideal management strategy in patients with CDH is unclear, current management strategy usually

1346 involves low pressure (b25 cm H2O), low rate of ventilation with tolerance for high Paco2, and judicious use of ECMO. Using this strategy, a survival rate as high as 93% has been reported [1]. The reported survival in patients with the use of ECMO ranges from 44% to 86% [1,2]. Although the efficacy of ECMO as compared with other forms of ventilation is not established by randomized controlled evaluation, it is used in situations where the patients fail to respond to medical therapy [3]. The primary role of ECMO is to provide gas exchange and prevent barotrauma, volutrauma, biotrauma, and oxygen toxicity associated with mechanical ventilation. However, the use of ECMO is invasive and has potential complications of cannulation and heparinization. Studies reporting good survival rate by using non-ECMO treatments argue that the use of ECMO is invasive and may be associated with increased morbidity and costs [4,5]. Several attempts were made to identify prognostic factors that could aid in selection of patients who could benefit most with ECMO treatment. Prognostic factors such as antenatal diagnosis, low birth weight, size of the diaphragmatic defect, herniation of the liver into the thorax, ventilation scores, lung compliance measurements, occurrence of pneumothorax, ability of reaching Po2 higher than 100 mm Hg, level of pulmonary hypertension have been reported in various reports. However, there is no prognostic factor that was found to be consistent and reliable for patients requiring ECMO [5,6]. This may be a consequence of the variability in patient selection criteria, the practice of treatments such as ECMO, and the extreme clinical instability of these newborns [1,7,8]. In addition, the uncommon occurrence of CDH makes it unlikely to have a large series from single centers with standard entry criteria and management. This series represents the 13-year experience of a single tertiary care ECMO center where patient selection and management have been standard over the study period. Extracorporeal membrane oxygenation was used in all patients referred after the failure of conventional treatments they received. The primary aim of this study was to identify bpreQ ECMO and bonQ ECMO variables that could predict mortality in patients with CDH requiring ECMO. In addition, we also aimed to determine the incidence and relation of complications during the course of ECMO to the survival of patients.

1. Methods This study is a retrospective review of case records, microfilms, and our database between September 1991 and December 2004. All patients with CDH managed in our unit were included. Our criteria for accepting neonates for ECMO treatment are an oxygenation index of more than 40, gestational age of more than 35 weeks, and birth weight of more than 2.0 kg, with no major chromosomal defects and no significant intraventricular bleeding. Patients were

R. Tiruvoipati et al. only accepted for ECMO therapy if the referring pediatric intensivist, the ECMO consultant, and the pediatric intensivist of our hospital agreed that conventional intensive care (the use of protective ventilation and permissive hypercapnia, use of surfactant, inhaled NO, and a trail of HFOV) had failed. Pre-ECMO and on-ECMO variables including gestational age, sex, birth weight, age at the time of ECMO cannulation, acid-base status, blood gasses and ventilator settings before commencing ECMO, mode of ECMO (venoarterial [VA] vs veno-venous [VV]), use of NO, surfactant, liquid ventilation, vasoactive agents, blood and blood products, timing of repair of CDH, and complications (mechanical and patient related) of ECMO were analyzed to identify the predictors of mortality. Complications that developed during the course of ECMO were collected according to the Extracorporeal Life Support Organization registry forms. Complications were classified into mechanical (related to ECMO circuit) and patient complications. Mechanical complications (requiring change of equipment or intervention) included oxygenator failure, pump malfunction, cracks in connectors, air in circuit, ECMO cannula problems, and clots in circuit. Patient complications included gastrointestinal hemorrhage (requiring blood transfusions or other intervention), surgical site bleeding (requiring blood transfusions or other intervention), seizures, cerebral infarction (diagnosed by cranial computed tomography or ultrasound), intracranial hemorrhage, renal complications (requirement of continuous venovenous hemofiltration), pulmonary complications (pneumothorax or pulmonary hemorrhage), requirement of inotropes, requirement of cardiopulmonary resuscitation, cardiac arrhythmias requiring antiarrythmic treatment, hypertension, and culture-proven infections.

1.1. Extracorporeal membrane oxygenation technique The ECMO technique used in our unit was described in detail elsewhere [9]. Briefly, cannulation for VV ECMO was performed with a Jostra or OriGen 12 or 15 doublelumen cannula (Jostra7, Herringen, Germany; OriGen Biomedical, Austin, Tex) inserted in the right internal jugular vein. VA ECMO was performed by inserting a single-lumen cannula in the right internal jugular vein for drainage and in the right common carotid artery for return. The oxygenators used were of silicone membrane (Avecor/Medtronic 0800, Minneapolis, MN) (before September 2001) or polymethyl pentene (Medos 0800-1200, Medizintechnik AG, Stolberg, German) (from September 2001); and the maximum activated clotting time (ACT) was between 160 and 200 seconds. Patients without signs of improvement in lung function after a period of 10 to 14 days on ECMO support were treated with other modes such as liquid ventilation, surfactant, NO, and HFOV at the discretion of the ECMO

Predictors of outcome in patients with congenital diaphragmatic hernia Table 1

1347

Comparison of variables before institution of ECMO between survivors and nonsurvivorsa

Variable Gestational age (wk) Prenatal diagnosis Age at referral Birth weight (kg) Male-female pH Pao2 Pco2 Sao2 HCO3 PIP PEEP Apgar score at 1 min Apgar score at 5 min

Survivors (n = 30) Valid n

Mean F SD

22

38.6 F 2.4 20 (67%) 24 hb 2.95 F 0.42 18:12 7.23 F 0.16 37.50b 60.5 F 24.4 62.1 F 27.4 20.1 F 3.3 34.5 F 5.6 4b 4.62 F 2.82 6.33 F 2.65

29 29 30 29 30 29 25 29 23 22 26 24

Nonsurvivors (n = 22) Range

Valid n

Mean F SD

34-43

18

5 h-11 d 2.20-4.20

22 21 22 20 20 20 17 18 17 17 20 16

39.5 F 3.1 12 (55%) 19 hb 2.90 F 0.47 14:8 7.27 F 0.14 33.35b 59.3 F 33.4 63.9 F 22.0 21.3 F 5.1 34.3 F 3.8 5b 4.00 F 3.21 5.31 F 3.34

6.94-7.58 15-283 24-123.8 15-99 10.9-26.0 24-50 2-12 0-9 0-10

Range 34-42 6 h-18 d 2.0-3.80 6.95-7.6 10.2-142 29.3-179 25-99 13.1-14.2 28-40 2-10 0-9 0-9

Sao2 indicates oxygen saturation, arterial; PIP, peak inspiratory pressure; PEEP, positive end-expiratory pressure. a None of the differences were statistically significant. b Median.

consultant and intensivist to aid in weaning of ECMO. Repair of congenital diaphragmatic hernia was usually performed when the patients had stabilized on ECMO and had been diuresed or hemofiltered to their dry weight.

1.2. Statistical analysis The data were collected on an Excel spreadsheet (Microsoft Corp, Redmond, Wash) and exported to SPSS (version 12.0 for Windows; SPSS Inc, Chicago, Ill) for analysis. The v 2 test was performed on categorical variables, and the Student t test and Mann-Whitney nonparametric test were used for comparison of continuous variables. A P value of less than .05 was deemed to indicate statistical significance. Logistic regression analysis was performed to determine whether variables were independently associated with mortality and to estimate the odds ratios of survival associated with ECMO variables. The odds ratio were adjusted for birth weight (b2500, z2500 g), Apgar score at 1 minute, and prenatal diagnosis as these variables have been previously found to be associated with outcome in studies where ECMO was used in patients with CDH [10].

2. Results A total of 52 patients were managed using ECMO in our unit during the 13-year study period. Five of these patients had associated cardiac anomalies (pulmonary atresia in 2 patients, ventricular septal defect in 2 patients, and truncus arteriosus in 1 patient). The mean gestational age at birth was 38.9 (SD, 2.74) weeks and birth weight was 2.92 (SD, 0.44) kg. The Apgar scores were 4.35 (SD, 2.9) at 1 minute and 5.93 (SD, 2.9) at 5 minutes of birth. The mean age at the time of referral for ECMO was 2 (SD, 3.1) days. The mean Pao2/Fio2 ratio was 49.7 (SD, 49.8) mm Hg. The

overall survival to hospital discharge was 58%. As shown in Table 1, the pre-ECMO variables were not statistically different between survivors and nonsurvivors. The comparison of on-ECMO variables between survivors and nonsurvivors is shown in Table 2. The use of NO and mean duration of ECMO differed significantly between survivors and nonsurvivors. There was no significant difference between the use of blood and blood products between survivors and nonsurvivors (Table 3). On univariate analysis, on-ECMO variables including mean duration of ECMO (181 [SD, 121; range, 60-672] hours vs 318 [SD, 157; range, 73-738] hours; P = .001), use of NO (6 vs 10, Table 2 Comparison of treatments during ECMO between survivors and nonsurvivors Variable

Survivors (30)

Nonsurvivors (22)

Mean duration (h) on ECMO Duration (z14 d) Multiple ECMO treatments Mode (VA/VV/VV to VA) Vasopressors Vasodilators HFOV NO Surfactant Liquid ventilation Repair pre-ECMO On-ECMO repair

181 (SD, 121)a

318 (SD, 157)a

2b 2 14:15:1 28 8 6 6c 5 0 6 14

9b 3 9:11:2 18 9 7 10c 2 1 4 15

HFOV indicates high-frequency oscillatory ventilation. a Difference with P = .001. b Difference with P = .005. c Difference with P = .049.

1348

R. Tiruvoipati et al.

Table 3

Requirement of blood and blood productsa

Variable

Survivors (n = 30)

Blood (U) Fresh frozen plasma (U) Platelets (U) Cryo precipitate (U) 20% Human albumin solution (100 mL units) a

Nonsurvivors (n = 22)

Valid n

Mean F SD

Range

Valid n

Mean F SD

Range

26 26 26 26 26

24.4 4.81 42.4 3.65 4.62

7-58 0-20 5-92 0-16 0-14

21 21 21 21 19

31.5 6.24 54.3 4.62 5.32

9-84 0-24 5-176 0-17 0-17

F F F F F

14.9 6.07 29.3 4.42 3.89

F F F F F

17.0 6.24 48.1 4.79 5.24

None of the differences were statistically significant.

were related to the survival of patients. Among the patientrelated complications, the occurrence of renal complications is significantly related to the hospital survival of patients. Adjusting for birth weight, prenatal diagnosis, and Apgar score at 1 minute did not significantly change the odds ratios.

P = .049), and renal complications (4 vs 14, P b .001) were significantly different between the survivors and nonsurvivors; use of NO, prolonged duration of ECMO, and development of renal failure requiring continuous venovenous hemofiltration (CVVH) during the course of ECMO treatment are all associated with decreased survival. The survival of patients who required ECMO support for less than 14 days was significantly better than that of patients who required ECMO support for more than 14 days (68% [28/41] vs 18% [2/11], P = .005). Logistic regression analysis of the variables included in the analysis revealed the duration of ECMO of more than 14 days and renal complications on ECMO to be independently associated with mortality (odds ratio, 0.10; 95% confidence interval [CI], 0.02-0.66; odds ratio, 0.09; 95% CI 0.02-0.34). The incidence of complications on ECMO and the relation of the complication to the survival are presented in Table 4. As shown in Table 4, none of the mechanical complications

3. Discussion The antenatal and postnatal management of patients with CDH is still evolving, and none of the therapies used to treat CDH have been successful in all patients. Given the wide variations in the management of patients with CDH it is not surprising that there are no predictors that are reliable. The primary cause of mortality in patients with CDH is refractory pulmonary hypertension because of pulmonary hypoplasia with reduced alveolar surface area and surfactant deficiency. Moreover, patients with CDH are known to have

Table 4

Incidence and survival in relation to the complications during ECMO

Variable

Incidence (%) in all patients

Discharged alive (%) with the complication

Unadjusted odds ratio (95% CI)

10 (19) 2 (4) 7 (14) 6 (12) 10 (19) 36 (69)

5 (50) 0 4 (57) 4 (67) 7 (70) 21 (58)

0.68 – 0.97 1.54 1.93 1.09

Mechanical complications Oxygenator failure Pump malfunction Cracks in connectors Air in circuit ECMO cannula problems Clots in circuit Patient complications Gastrointestinal hemorrhage Surgical site bleeding Seizures Cerebral infarction Intracranial hemorrhage Requirement of CVVH Pulmonary complications Inotropes Requirement of CPR Cardiac arrhythmias Hypertension Culture-proven infections

3 (6) 6 (12) 8 (16) 3 (6) 5 (10) 18 (35) 11 (21) 13 (25) 2 (4) 3 (6) 6 (12) 2 (4)

CPR indicates cardiopulmonary resuscitation. a Adjusted for birth weight, Apgar score, and prenatal diagnosis. * P b .01.

2 4 3 0 1 4 5 6 1 0 3 1

(67) (67) (38) (20) (22) (45) (46) (50) (50) (50)

Adjusteda Odds ratio (95% CI)

(0.17-2.71)

0.46 (0.1-2.2)

(0.19-4.87) (0.26-9.26) (0.44-8.49) (0.33-3.58)

0.61 1.68 2.39 0.72

1.5 (0.13-17.67) 1.54 (0.26-9.26) 0.38 (0.08-1.79) – 0.16 (0.02-1.5) 0.09* (0.02-0.34) 0.53 (0.14-2.04) 0.54 (0.15-1.9) 0.72 (0.04-12.25) – 0.7 (0.13-3.87) 0.72 (0.04-12.25)

(0.1-3.66) (0.25-11.24) (0.49-11.65) (0.18-2.84)

– 1.25 (0.15-10.21) 0.53 (0.1-2.83) – 0.25 (0.02-2.82) 0.06* (0.01-0.29) 0.52 (0.13-2.16) 0.59 (0.15-2.31) 0.62 (0.03-11.49) – 0.55 (0.07-4.33) 0.72 (0.04-12.69)

Predictors of outcome in patients with congenital diaphragmatic hernia reduced left ventricular mass and left ventricular dysfunction [11-13]. Among these factors, pulmonary hypoplasia is thought to be the most important in determining the outcome [2,3]. Although isolated pulmonary hypertension may be potentially reversible, pulmonary hypoplasia is unlikely to be affected by the existing treatments including the use of ECMO. Attempts made to predict the degree of hypoplasia both radiologically as well as physiologically were proved to be difficult and unsuccessful. Although there are a few small series reporting the predictors of severity of lung hypoplasia prenatally using lung-head ratio, magnetic resonance lung volumetry, and postnatal modified McGoon index, these require further evaluation in larger series [1416]. Physiological predictors such as alveolar arterial gradient, arterial oxygenation, and severity of hypercarbia have been evaluated and were found to be unreliable. The use of birth weight and Apgar scores was reported to be most predictable by the CDH study group after logistic regression analysis [10]. However, this was later found to predict survival lower than the actual survival [1]. Predictors of mortality in patients with CDH who were not treated with ECMO [5,17] may not serve as good predictors for patients requiring ECMO. This may be noted from the results of the study by Keshen et al [5] where the predictive equation based on the outcome of patients who were managed conventionally was shown to predict outcome in 94% of patients who were managed without ECMO but could only predict the outcome in 75% of the patients requiring ECMO. Some studies have identified pre-ECMO variables such as birth weight, Apgar scores, gestational age, and prenatal diagnosis [10,18-20] as predictors of mortality. However, most of the studies reporting predictors are multicentric and so may have variability in the selection criteria and ECMO practice [10,18-20]. In our study with relatively standard patient selection and ECMO practice in the management of CDH, birth weight, Apgar scores, or other pre-ECMO physiological variables were not predictive of outcome, and prolonged duration of ECMO and renal complications (requirement of CVVH) on ECMO were independently associated with mortality. In this series, only 2 of 11 patients who required ECMO support for more than 336 hours (14 days) survived. Although there are no reliable markers available to assess the adequacy of lung development at the time of birth or before institution of ECMO, it may be inferred from these results that neonates requiring ECMO for more than 14 days are less likely to have adequate lung development and pulmonary hypertension is probably not reversible, and hence, patients may have a poor prognosis. The use of ECMO in our series was not associated with significant complications that could affect the survival. In our study, none of the mechanical complications on ECMO were associated with mortality. We believe this is because mechanical complications are diagnosed early and treated safely by our highly trained ECMO specialist nurses and perfusionists who provide 24-7 supervision of the ECMO

1349

circuit. This suggests that although ECMO is an invasive procedure, its use as such would not increase the mortality. The development of renal complications (requirement of CVVH) while on ECMO was associated with mortality. Although the exact cause(s) of renal failure in these patients are unclear, we believe this could be multifactorial, including the severe hypoxia in these neonates before instituting ECMO. The mortality in our series was not affected by the mode of ECMO (VV vs VA). These findings are similar to those reported earlier comparing the use of VV vs VA ECMO [21]. Nitric oxide was used more frequently in the group of patients who died as compared with those who survived. This is possibly because of severe lung hypoplasia and pulmonary hypertension in the group of patients who died. Although the UK collaborative trial [22] proved ECMO to be clinically effective as well as cost-effective compared to conventional ventilation in the treatment of neonatal respiratory failure, it included only a small number of neonates with CDH (total of 35 neonates with 18 in the ECMO arm and 17 in the conventional arm). These numbers are clearly too small to support either ECMO or conventional care in patients with CDH and hence does not reflect the clinical effectiveness or cost-effectiveness of ECMO in treating patients with CDH. The use of ECMO requires transfer to a specialist center, instrumentation of neck vessels as well as heparinization, all of which are associated with a potential risk. In addition, some centers have reported good results with conventional care, suggesting no significant advantage in the use of ECMO [4,5]. Given this variability of results, a well-designed prospective randomized controlled trial would be a more certain way to assess the effectiveness of ECMO as well as conventional care in an unbiased manner. This may also identify variables that are reliable in predicting the outcome in neonates who might benefit from the use of ECMO and would further clarify the role of ECMO in management of CDH. To conclude, the results of our study suggest that patients requiring ECMO support for more than 2 weeks are likely to have significant pulmonary hypoplasia and have a poor prognosis. The association of renal complications while on ECMO carries higher mortality. None of the pre-ECMO variables could predict the outcome of patients. An adequately powered randomized controlled trial may further clarify the efficacy as well as the cost-effectiveness of ECMO in treating neonatal patients with congenital diaphragmatic hernia.

References [1] Downard CD, Jaksic T, Garza JJ, et al. Analysis of an improved survival rate for congenital diaphragmatic hernia. J Pediatr Surg 2003; 38:729 - 32. [2] Wilson JM, Lund DP, Lillehei CW, et al. Congenital diaphragmatic hernia—a tale of two cities: the Boston experience. J Pediatr Surg 1997;32:401 - 5.

1350 [3] Khan AM, Lally KP. The role of extracorporeal membrane oxygenation in the management of infants with congenital diaphragmatic hernia. Semin Perinatol 2005;29:118 - 22. [4] Al Shanafey S, Giacomantonio M, Henteleff H. Congenital diaphragmatic hernia: experience without extracorporeal membrane oxygenation. Pediatr Surg Int 2002;18:28 - 31. [5] Keshen TH, Gursoy M, Shew SB, et al. Does extracorporeal membrane oxygenation benefit neonates with congenital diaphragmatic hernia? Application of a predictive equation. J Pediatr Surg 1997;32:818 - 22. [6] Harrington KP, Goldman AP. The role of extracorporeal membrane oxygenation in congenital diaphragmatic hernia. Semin Pediatr Surg 2005;14:72 - 6. [7] Langham Jr MR, Kays DW, Ledbetter DJ, et al. Congenital diaphragmatic hernia. Epidemiology and outcome. Clin Perinatol 1996;23:671 - 88. [8] dos Santos LR, Maksoud-Filho JG, Tannuri U, et al. Prognostic factors and survival in neonates with congenital diaphragmatic hernia. J Pediatr (Rio J) 2003;79:81 - 6. [9] Ichiba S, Killer HM, Firmin RK, et al. Pilot investigation of hypothermia in neonates receiving extracorporeal membrane oxygenation. Arch Dis Child Fetal Neonatal Ed 2003;88:F128-33. [10] Estimating disease severity of congenital diaphragmatic hernia in the first 5 minutes of life. The Congenital Diaphragmatic Hernia Study Group. J Pediatr Surg 2001;36:141 - 5. [11] Molenaar JC, Bos AP, Hazebroek FW, et al. Congenital diaphragmatic hernia what defect? J Pediatr Surg 1991;26:248 - 54. [12] Baumgart S, Paul JJ, Huhta JC, et al. Cardiac malposition, redistribution of fetal cardiac output, and left heart hypoplasia reduce survival in neonates with congenital diaphragmatic hernia requiring extracorporeal membrane oxygenation. J Pediatr 1998;133: 57 - 62.

R. Tiruvoipati et al. [13] Schwartz SM, Vermilion RP, Hirschl RB. Evaluation of left ventricular mass in children with left-sided congenital diaphragmatic hernia. J Pediatr 1994;125:447 - 51. [14] Lipshutz GS, Albanese CT, Feldstein VA, et al. Prospective analysis of lung-to-head ratio predicts survival for patients with prenatally diagnosed congenital diaphragmatic hernia. J Pediatr Surg 1997;32: 1634 - 6. [15] Paek BW, Coakley FV, Lu Y, et al. Congenital diaphragmatic hernia: prenatal evaluation with MR lung volumetry—preliminary experience. Radiology 2001;220:63 - 7. [16] Suda K, Bigras JL, Bohn D, et al. Echocardiographic predictors of outcome in newborns with congenital diaphragmatic hernia. Pediatrics 2000;105:1106 - 9. [17] Numanoglu A, Morrison C, Rode H. Prediction of outcome in congenital diaphragmatic hernia. Pediatr Surg Int 1998;13:564 - 8. [18] Heiss KF, Clark RH. Prediction of mortality in neonates with congenital diaphragmatic hernia treated with extracorporeal membrane oxygenation. Crit Care Med 1995;23:1915 - 9. [19] Stevens TP, Chess PR, McConnochie KM, et al. Survival in early- and late-term infants with congenital diaphragmatic hernia treated with extracorporeal membrane oxygenation. Pediatrics 2002;110(3):590 - 6. [20] Davis PJ, Firmin RK, Manktelow B, et al. Long-term outcome following extracorporeal membrane oxygenation for congenital diaphragmatic hernia: the UK experience. J Pediatr 2004;144(3): 309 - 15. [21] Kugelman A, Gangitano E, Pincros J, et al. Venovenous versus venoarterial extracorporeal membrane oxygenation in congenital diaphragmatic hernia. J Pediatr Surg 2003;38:1131 - 6. [22] Roberts TE. Economic evaluation and randomised controlled trial of extracorporeal membrane oxygenation: UK collaborative trial. The Extracorporeal Membrane Oxygenation Economics Working Group. BMJ 1998;317:911 - 5.