SL Lee, MS Wong, CW Leung, NS Tsoi, CY Yeung

Abstract

Congenital diaphragmatic hernia of Bochdalek type remains a difficult respiratory management problem despite recent advances in medical technologies. We reviewed the features of our patients with this condition from 1988 to 1996. The overall survival rate was 71 % which was comparable with other reported series even though extracorporeal membranous oxygenation was not available. Pen-operative ventilation index less than 1000 and oxygenation index less than 20 were associated with significantly improved survival (p < 0.05 ). These respiratory status may be useful indices to predict the outcome of infants with congenital diaphragmatic hernia.

Congenital diaphragmatic hernia (CDH) of the Bochdalek type remains a difficult respiratory management problem. The mortality rate has remained high despite advances of medical technologies including surfactant replacement therapy (SRT), high frequency oscillatory ventilation (HFOV), inhaled nitric oxide therapy (iNO) and extra-corporeal membrane oxygenation (ECMO). We reviewed our series of infants with CDH from 1988 to 1996 and compared our experience with others to see if we could improve our outcome in the future.

Patients and Methods

Twenty one patients with congenital diaphragmatic hernia treated at our institution from 1988 through 1996 were reviewed. Pre-operative data included documentation of a prenatal diagnosis, gestational age at birth, sex, birth weight, presenting symptomatology, associated anomalies, pre-operative arterial blood gases, ventilatory parameters, time from birth to repair-operation, post-operative arterial blood gases and ventilatory parameters. The best oxygen index (O.I.=mean airway pressure x FiO₂ x 100 / PaO₂) and ventilation index (V.I.=respiratory rate x mean airway pressure) of these patients requiring ventilatory support were calculated. Comparison of data from survivors and non-survivors was analysed using Fisher's exact test.

Results

Our patients with congenital diaphragmatic hernia comprised of 18 newborns and 3 beyond the neonatal period. There were 13 male (62%) and 8 female (38%). Eleven (52%) were full-term and 10 (48%) were preterm. The gestational age and birth weight of the preterm infants were 35.5 ± 2 weeks and 2.4 ± 0.4 kg respectively. All patients had Bochdalek defects and none had Morgagni defect. Seventeen infants (81%) had left-sided defects and 4 (19%) had right-sided defects. Thirteen patients (62%) were transferred from other hospitals. In seven patients (33%) a prenatal diagnosis of congenital diaphragmatic hernia was made by ultrasound examination and all of them were diagnosed in the third trimester (mean gestational age of diagnosis: 32.4 ± 3 weeks). Of these, 4 patients were complicated by polyhydramnios and 6 presented with respiratory distress at birth. Three of the 7 prenatally diagnosed infants died and all of them were complicated by polyhydramnios.

Seven newborns (33%) without prenatal diagnosis presented with respiratory distress at birth or within the first day of birth. Four infants (19%) presented between the second and third day of life with respiratory distress and feeding difficulties. Three patients (14%) were diagnosed at 2 months, 8 months and 4 years of age respectively because of incidental finding in chest radiographs. Eight patients (38%) also had other significant anomalies including: Down syndrome in 2; Patau's syndrome in 2; Noonan syndrome in 1; ventricular septal defect in 3; atrial septal defect in 2; patent ductus arteriosus in 1; pulmonary atresia in 1; volvulus of stomach in 1; intestinal malrotation in 1 and Meckel's diverticulum in 1.

Eighteen patients (86%) underwent operative repair of their diaphragmatic hernia. Eight of them required endotracheal intubation and positive pressure ventilation at birth. Conventional mechanical ventilation was continued in all except in one patient who required high frequency ventilation because of persistent hypercapnia. Eight patients who presented at birth underwent repair operation between 10 to 24 hours of life. Ten patients underwent repair operation beyond the first day of life.

Eighteen patients were operated on and fifteen survived their pen-operative course, thus the overall pen-operative survival rate was 71% (15/21). One infant who had Down syndrome and ventricular septal defect was not included as pen-operative death as she died ten months after the operation because of aspiration pneumonia and cardiorespiratory failure. Three infants (14%) died without attempted repair; one had Patau's syndrome, one died of streptococcal septicaemia and one died of severe respiratory failure despite aggressive ventilatory support. Another three patients died within 3 days after the repair operation. Two of them were complicated by persistent pulmonary hypertension. Post-mortem examination revealed severe pulmonary hypoplasia in all the infants who succumbed during the pen-operative period.

Eleven of the 13 patients (85%) who were referred from outside institutions survived their operation. Ten infants were not symptomatic before the first day of life and all of them survived the pen-operative course. Available data on the maturity at birth, associated anomalies, the side of the defect, the best arterial blood gases and ventilatory parameters were compiled for analysis. A comparison of these data between the survivors and the non-survivors was made. The results were shown in Tables I and II.

Table I shows the patients' characteristics and the survival results. Two features have been found to be significantly associated with improved survival. They are (1) maturity beyond 37 weeks and (2) absence of symptoms within the first day of birth. Other factors, including prenatal diagnosis of the condition, have not been found to be significantly affecting the survival outcome.

Table II shows the respiratory management and survival results. Eleven patients (52%) required ventilatory support before the repair operation. Seven of these 11 patients had an 0.1. less than 20 (9.0 ± 7.0) and six of them survived the pen-operative course; four patients who had an 0.1. more than 20 (64.8 ± 50.0) all died. Seven of the same 11 patients had a V.I. less than 1000 (441 ± 192) and 6 of them survived; four patients who had a V.I. more than 1000 (1730 ± 627) all died. The survival rates of patients with the best pre-operative 0.1. less than 20 or V.I. less than 1000 were significantly higher than those patients with 0.1. more than 20 or V.I. more than 1000 (p<0.05).

Of the 18 patients who underwent repair operation, the best post-operative 0.1. and V.I. of 16 patients were analysed. Two survivors were excluded for analysis because their post-operative 0.1. and V.I. were not available. All thirteen of the 16 patients who had an 0.1. less than 20 (4.6 ± 2.6) survived the pen-operative course. The same 13 infants had a V.I. less than 1000 (440 ± 251) and all of them survived the pen-operative course. Three of the other 16 patients had an 0.1. more than 20 (37.2 ± 17.7) and V.I. more than 1000 (2886 ± 1506) all of them died. The survival rates of patients with best postoperative 0.1. less than 20 or V.I. less than 1000 were significantly higher than those patients with 0.1. more than 20 or V.I. more than 1000 (p<0.05).

Discussion

Twenty one babies with CDH were managed in our hospital from 1988-1996. Male babies were more predominant in our series, the male to female ratio was 1.6. Eighty-one percent were left sided CDH and none of them had bilateral involvement. Our overall pen-operative survival rate was 71 % which is comparable to that in large centres of other advanced countries^1-4 (Table III). Survival was even better in outborn babies (84%). This could be due to the referral pattern by which only those babies who could be reasonably stabilised would reach our hospital.

In the past, CDH was immediately repaired upon diagnosis with a view to remove the bowel from the thorax and close the diaphragmatic defect so that gaseous exchange can be improved through expansion of the ipsilateral lung. Since the work of Cartlidge⁵ who suggested that delaying the operation to allow a period of physiological stabilisation might reduce the mortality, the concept of emergency surgical repair has been challenged. Sakai et al⁶ indeed demonstrated that the respiratory system compliance decreased after surgical repair of CDH. The authors postulated that the condition of the respiratory-unstable infant would be much worse immediately following repair. The infant's flexible thoracic cage is often deformed due to the negative pressure created in the thorax after the removal of the hernia content and the associated increase of intra-abdominal pressure. Moreover, the diaphragm is frequently distorted after repair especially if the defect is large. The ipsilateral lung is hypoplastic and slow to re-expand while the contralateral lung may be hyperinflated due to release of mediastinal shift and positive pressure ventilation. There are multiple clinical trials^7-9 supporting the favourable effect of preoperative stabilization. Others failed to show any advantage but neither did they reveal harmful effect.^10,11 We have adopted the preoperative stabilization approach during the study period. All patients underwent surgery after a period of stabilization for more than 10 hours during which relatively adequate oxygenation and stable haemodynamic status was achieved. However, we cannot make a comparison between early and delayed repair because of the small number of patients in the "pre-stabilisation" era.

All except one of our infants were put on conventional mode of ventilation (CMV) using high rate and low peak inspiratory pressure to minimize barotrauma. HFOV was employed in one baby who failed to be stabilized with CMV. There was a short term decrease in PaCO₂ but the infant finally died, due probably to the underlying severe pulmonary hypoplasia. HFOV is not only successful in management of acute respiratory failure in neonates, it is particularly useful for homogenously diffuse alveolar atelectasis. Bohn et al¹² showed that CDH babies with low risk could be hyperventilated with CMV. For high risk patients with critical pulmonary hypoplasia that failed to respond to CMV, HFOV was able to bring down the PaCO₂ and temporarily improve oxygenation- a phenomenon which was well studied in animals with pulmonary hypoplasia. Miguet et al⁴ has demonstrated the efficacy of preoperative stabilisation using HFOV while Paranka et al¹³ found that CDH babies with lung hypoplasia was least likely to respond to HFOV during various conditions of acute respiratory failure. It is not surprising to see contradicting results as different groups included patients with different degree of pulmonary hypoplasia which is probably the major determining factor for success. Early use of HFOV versus CMV still await further prospective randomised study. Clinical judgement and individualization of choices for mode of ventilatory management are of paramount importance. At present, we are still using CMV to start with and in those who fail to respond, HFOV would be considered.

None of our patients were given exogenous surfactant therapy. However, recent laboratory work has shown surfactant system dysfunction in neonates with CDH. Treatment with prophylactic exogenous surfactant has significantly improved the lung mechanics with increased total lung capacity, improved alveolar stability and increased compliance resulting in improvement in oxygenation and ventilation.¹⁴ It has also been demonstrated that prophylactic surfactant replacement therapy can also reduce pulmonary vascular resistance, increases pulmonary blood flow and causes a reduction in the right to left shunt.¹⁵ This would imply that surfactant deficiency may be partly responsible for the development of persistent pulmonary hypertension (PPHN) in neonates with CDH. It has been speculated that in many neonates with CDH, there is actually adequate pulmonary parenchyma but the dysfunction of surfactant system prevents the efficient use of available lung tissue and increase the incidence of PPHN. However, dramatic improvement in oxygenation, ventilation and pulmonary haemodynamics cannot be reproduced when surfactant is administered as rescue therapy.¹⁶ This is explained by some authors that surfactant is not delivered adequately, or the lungs have already incurred significant barotrauma and the surfactant is inactivated by alveolar proteins. Larger, multi-centred controlled trials are necessary to determine the efficacy, optimal dosage and timing and frequency of administration of surfactant replacement therapy in neonates with CDH. Partial liquid ventilation (PLV) may also be considered as an alternative in the future. Perfluorocarbons has a high oxygen carrying capacity with low surface tension. It is able to recruit collapsed alveoli because of its relatively high density. More important is the lack of deleterious effect on endogenous surfactant production. Two animal studies and a phase 1 human study¹⁷ have shown marked improvement in both the pulmonary compliance and gaseous exchange after initiation of PLV.

PPHN is a prominent pathophysiologic feature of CDH. Postmortem analysis and animal studies demonstrated a decrease in the total size of pulmonary vascular bed, a decrease in the number of vessels per unit area of lung and increased muscularization of arteries. Recently, it has been shown that nitric oxide synthase immunoreactivity is present in the main pulmonary trunks of CDH lambs.¹⁸ However, whether the quantity of endogenous nitric oxide produced is sufficient to produce vasodilatation or whether pulmonary vascular smooth muscle cells in CDH is relatively insensitive to the effect of nitric oxide was not investigated. In fact, the effect of inhaled nitric oxide therapy in the treatment of PPHN in CDH varied.¹⁹ Inhaled nitric oxide is only effective when it is combined with prior administration of surfactant in fetal lamb model²⁰ or when it is used in conjunction with HFOV in CDH neonates.²¹ Wilcox et at²² also showed that PLV could improve gas exchange and deliver nitric oxide successfully in fetal lamb model. All these studies suggested that for nitric oxide to be effective, it must be efficiently delivered to the alveolar-capillary interface. Nitric oxide therapy was only available in our institution since 1995. However, none of our infants developed severe PPHN that warranted this mode of treatment. The efficacy of nitric oxide therapy thus remains speculative.

A number of investigators have reported success in treating CDH with ECMO.^1,8 Others have not observed this improvement.² According to the ECMO registry in USA in 1994, the survival rate for CDH treated by ECMO is 58 %¹ which suggested that there was no significant impact on the survival of newborn with CDH. Similarly, a large UK collaborative randomised trial of neonatal ECMO failed to show significant improved survival in this group of patients.²³ Staak et al¹ was able to show that ECMO was very valuable in a group of cases with borderline pulmonary hypoplasia that cannot be stabilised with conventional treatment. Both HFOV and ECMO have been shown to be equal in providing an approximately 15% improvement of salvage while ECMO produces a 15-20% of additional morbidity.²⁴ In the present series, even without ECMO our survival rate in CDH was still comparable to most other ECMO centres. Hence we do not recommend to introduce ECMO in the management of CDH in our locality with regard to the cost effectiveness; moreover earlier observation has indicated a relatively low incidence of PPHN in our population.²⁵

Attempts have been made to determine factors that can predict the outcome in CDH. With advances in prenatal sonography, more cases of CDH can be diagnosed before birth. A variety of features has been proposed to predict the outcome. Polyhydramnios has been shown to correlate with poor outcome but it usually appears in the third trimester and its value has been disputed. Prematurity and association with multiple congenital abnormalities in particular cardiac abnormality is widely accepted and documented to signify poor outcome in other series. ^26,27 In our study, maturity beyond 37 weeks and absence of symptoms within the first day of life were found to be significantly associated with improved survival. Other fetal sonographic parameters like polyhydramnios,²⁸ detection of CDH before 25 weeks,²⁹ presence of intrathoracic stomach,²⁸ small lung thorax transverse area ratio,²⁹ herniation of liver, underdevelopment of left heart and contralateral lung volume at level of an axial four chamber view³⁰ have also been shown to affect the outcome. However, none of these parameters occurring singly has been accurately predictive of postnatal death and on its own therefore should not alter the fetal management. At present stage, the advantage of prenatal sonography is to allow the preparative management of a critically ill newborn but it is hoped that it may also guide fetal intervention in the future.

Both preoperative and postoperative chest radiographs to assess the pulmonary hypoplasia has been evaluated but none of these has precluded survival. Clinical parameters that are associated with poor prognosis include early presentation within 6 hours of life, severe hypoxaemia with alveolar arterial oxygen difference AaDO₂ > 500 mmHg with FiO₂ 1.0 before operation and severe acidosis with pH less than 7. Bohn et at¹² demonstrated that if hyperventilation could produce PaCO₂ levels of less than 5.5 kPa (42 mmHg) and ventilation index less than 1000, the survival rate was 86%. In contrast, there was 100 % mortality if PaCO₂ remained greater than 5.5 kPa and ventilation index in excess of 1000. Johnston et al³¹ showed that postoperative postductal PaO₂ of greater than 70 mmHg is a good prognostic factor. In our study, we have found that an O.I. less than 20 and ventilation index less than 1000 both preoperatively and post-operatively were associated with significantly lower mortality. Lung function measurements have also been found by others to be useful in the assessment of CDH. Tracy et at³² suggested that infants with initial respiratory compliance of greater than 0.25 ml/cmH₂O/kg and tidal volume greater than 3.5 ml/kg did not require ECMO. Dimitriou et at³³ found that a respiratory compliance of less than 0.18 ml/cmH₂O/kg was most accurate predictor of outcome while Antunes et al³⁴ concluded that preoperative functional residual capacity of less than 9ml/kg may indicate fatal pulmonary hypoplasia in most infants with CDH.

The most important factor that determines survival is probably the severity of pulmonary hypoplasia as confirmed by our study. Post mortem examination in all the 6 patients who died either before surgical repair or within 3 days after the operation revealed significant pulmonary hypoplasia. On the other hand, only 2 out of the 6 patients who died were complicated by PPHN. Similar phenomenon was observed in other overseas studies. It seems that despite advances in the management of CDH, there remains a group of patients with pulmonary hypoplasia unsalvageable by any means. To tackle this problem, lobar lung transplant has been suggested and there was a successful case reported recently.³⁵ Fetal intervention to allow hypoplastic lung to develop may be an alternative. Several experimental animal models of in utero repair of CDH offered promise but early attempts in human have met with limited success only. It is shown in a lamb model that tracheal ligation can increase intratracheal pressure which may accelerate fetal alveolar growth accompanied by accelerated growth of pulmonary vasculature in the experimental animal with CDH.^36,37 However, it does not correct the surfactant deficiency associated with CDH.³⁸ Though appears promising, tracheal ligation remains experimental and has not been tried in Hong Kong.

Acknowledgement

We are grateful to paediatric surgical colleagues of our hospital for providing the surgical management.

References

1. vd Staak FHJM, de Haan AFJ, Given WB, et al. Improving survival for patients with high risk congenital diaphragmatic hernia by using extracorporeal membrane oxygenation. J Pediatr Surg 1995;30:1463-7.

2. Steimle CN, Meric F, Hirsch RB, et al. Effect of Extracorporeal life support on survival when applied to all patients with congenital diaphragmatic hernia. J Pediatr Surg 1994;29:997-1001.

3. Mallik K, Rodgers B, McGahren E, et al. Congenital diaphragmatic hernia: Experience in a single institution from 1978 through 1994. Ann Thorax Surg 1995;60:1331-6.

4. Miguet D, Claris 0, Lapillonne A, et al. Preoperative stabilisation using high frequency oscillatory ventilation in the management of congenital diaphragmatic hernia. Crit Care Med 1994;22:S77-82.

5. Cartlidge PUT, Mann NP, Kapila L. Preoperative stabilisation in congenital diaphragmatic hernia. Arch Dis Child 1986;61: 1226-8.

6. Sakai H, Tamura M, Hosokawa Y, et al. Effect of surgical repair on respiratory mechanics in congenital diaphragmatic hernia. J Pediatr 1987;111:432-8.

7. Nakayama DK, Motoyama EK, Tagge EM. Effect of preoperative stabilisation on respiratory system compliance and outcome in newborn infants with congenital diaphragmatic hernia. J Pediatr 1991; 18:793-9.

8. Breaux CW, Rouse TM, Cain WS, et at. Improvement in survival of patients with congenital diaphragmatic hernia utilizing a strategy of delayed repair after medical and/or extracorporeal membrane oxygenation stabilization. J Pediatr Surg 1991;26:333-8.

9. Sigalet DL, Tierney A, Adolph V, et al. Timing of repair of congenital diaphragmatic hernia requiring extra corporeal membrane oxygenation support. J Pediatr Surg 1995;30:1183-7.

10. Wilson JM, Lund DP, Lillehei CW, et al. Delayed repair and pre-operative ECMO does not improve survival in high rise congenital diaphragmatic hernia. J Pediatr Surg 1992;27:368-75.

11. Nio M, Haase G, Kennaugh J, et al. A prospective randomized trial of delayed versus immediate repair of congenital diaphragmatic hernia. J Pediatr Surg 1994;29:618-21.

12. Bohn D, Tamura M. Persin D, et al. Ventilatory predictors of pulmonary hypoplasia in congenital diaphragmatic hernia confirmed by morphologic assessment. J Pediatr 1987;111 :423-31.

13. Paranka MS, Clark RH, Yoder BA, et al. Predictors of Failure of high Frequency oscillatory ventilation in Term infants with respiratory Failure. Pediatrics 1995;95(3):400-4.

14. Wilcox D, Glick PL, Karamanoukian H, et al. Pathophysiology of congenital diaphragmatic hernia. V. Effect of exogenous surfactant therapy on gas exchange and lung mechanics in the lamb congenital diaphragmatic hernia model. J Pediatr 1994;124:289-93.

15. O'Toole SJ, Karamanoukian H, Glick PL, et al. Surfactant decreases pulmonary vascular resistance and increases pulmonary blood flow in the fetal lamb model of congenital diaphragmatic hernia. J Pediatr Surg 1996; 31:507-11.

16. O'Toole SJ, Karamanoukian H, Glick PL, et al. Surfactant Rescue in the Fetal lamb model of congenital diaphragmatic hernia. J Pediatr Surg 1996;31:1105-9.

17. Pranikoff T, Ganger P, Hirschl R. Partial liquid ventilation in newborns with congenital diaphragmatic hernia. J Pediatr Surg 1996;31:613-8.

18. Karamanoukian H, Glick PL, Wilcox DT, et al. Pathophysiology of congenital diaphragmatic hernia X - localisation of lambs with surgically created congenital diaphragmatic hernia. J Pediatr Surg 1995;30:5-9.

19. Shah N, Jacob T, Exler R, et al. Inhaled nitric oxide in congenital diaphragmatic hernia. J Pediatr Surg 1994;29:1010-5.

20. Karamanoukian H, Glick PL, Wilcox DT, et al. Pathophysiology of congenital diaphragmatic hernia VIII: Inhaled nitric oxide requires exogenous surfactant therapy in the lamb model of congenital diaphragmatic hernia. J Pediatr Surg 1995;30:1-4.

21. Kinsella JP, Neish SR, Shaffer E, et al. Low dose inhalational nitric oxide in persistent pulmonary hypertension of the newborn. Lancet 1992;340:819-20.

22. Wilcox DT, Glick PL, Karamanoukian H, et al. Pathophysiology of congenital diaphragmatic hernia XIII: Perflurocarbon associated gas exchange (Page) rescue improves pulmonary mechanics, oxygenation and allows nitric oxide delivery in the hypoplastic lung congenital diaphragmatic hernia lamb model. J Crit Care Med 1995;23:1858-63.

23. UK Collaborative ECMO Group: UK Collaborative randomised trial of neonatal extracorporeal membrane oxygenation. Lancet 1996;348:75-82.

24. Jholenaar JC. Congenital diaphragmatic hernia, a deficit beyond surgical repair. In:Tibboel D & Vander Voort E,editors. Update in intensive care and emergency medicine no.25 Intensive care in childhood - a challenge to future. Springer 1996;100-3.

25. Yeung CY Chinese neonates are different. In: Yu V, Tsang R, Feng ZK, Yeung CY, editors. Textbook of neonatal medicine: a Chinese perspective. Hong Kong University Press 1996:877-83.

26. Cannon C, Dildy GA, Ward R, et al. A population based study of congenital diaphragmatic hernia in Utah: 1988 - 1994 Obstet - Gynecol 1996;87(6):959-63.

27. Fauza DO, Wilson JM. Congenital diaphragmatic hernia and associated anomalies: their incidence, identification and impact on prognosis. J Pediatr Surg 1994;29:1113-7.

28. Dommergues M, Louis-Sylvestre C, Mandelbrot L. Congenital diaphragmatic hernia: can prenatal ultrasonograph predict outcome? Am J Obstet Gynecol 1996;174(4):1377-81.

29. Metkus AP, Filly RA, Stringer MD, et al. Sonographic predictors of survival in Fetal diaphragmatic hernia. J Pediatr Surg 1996;31:148-51.

30. Guiband L, Filiafrault D, Garel L, et al. Fetal congenital diaphragmatic hernia. Accuracy of sonography in the diagnosis and prediction of outcome after birth. Ajr Am J Roentgenol 1996;166: 1195-202.

31. Johnston PW, Liberman R, Gangitano E, et al. Ventilation parameters and arterial blood gases as a prediction of hernia. J Pediatr Surg 1990;25:496-9.

32. Tracy T, Bailey P, Sadig F, et al. Predictive capabilities of preoperative and postoperative pulmonary function tests in delayed repair of congenital diaphragmatic hernia. J Paediatric Surg 1994;29:265-70.

33. Dimitriou G, Greenough A, Chan V, et al. Prognostic indicators in congenital diaphragmatic hernia. J Paediatric Surg 1995;30:1694-7.

34. Antunes M, Greenspan J, Cullen A, et al. Prognosis with preoperative pulmonary function and lung volume assessment in infants with congenital diaphragmatic hernia. Paediatrics 1995;96:1117-22.

35. Van Meurs, Rhine W, Benitz W, et al. Lobar lung transplantation as a treatment for congenital diaphragmatic hernia. J Pediatr Surg 1994;29:1557-60.

36. Di Fiore J, Fauza D, Slavin R, et al. Experimental fetal tracheal ligation reverses the structural and physiological effects of pulmonary hypoplasia in congenital diaphragmatic hernia. J Pediatr Surg 1994;29:248-57.

37. Di Fiore J, Fauza D, Slavin R, et al. Experimental fetal tracheal ligation and congenital diaphragmatic hernia : A pulmonary vascular morphometric analysis. J Pediatr Surg 1995;30:917-24.

38. O'Toole SJ, Iharma A, Karamanoukian H, et al. Tracheal ligation does not correct the surfactant deficiency associated with congenital diaphragmatic hernia. J Pediatr Surg 1996;31:546-50.