Article Text

Download PDFPDF

One-year outcomes of infants born with congenital diaphragmatic hernia: a national population cohort study
  1. Anna-May Long1,2,
  2. Kathryn J Bunch2,
  3. Marian Knight2,
  4. Jennifer J Kurinczuk2,
  5. Paul Damian Losty3
  6. on behalf of BAPS-CASS
    1. 1 Department of Paediatric Surgery, Cambridge University Hospitals NHS Foundation Trust, Cambridge, UK
    2. 2 National Perinatal Epidemiology Unit, University of Oxford, Oxford, UK
    3. 3 Institute of Child Health, University of Liverpool, Liverpool, UK
    1. Correspondence to Professor Paul Damian Losty, Institute of Child Health, University of Liverppol, Liverpool L69 3BX, UK; paul.losty{at}


    Objective To report outcomes to 1 year, in infants born with congenital diaphragmatic hernia (CDH), explore factors associated with infant mortality and examine the relationship between surgical techniques and postoperative morbidity.

    Design Prospective national population cohort study.

    Setting Paediatric surgical centres in the UK and Ireland.

    Method Data were collected to 1 year for infants with CDH live-born between 1 April 2009 to 30 September 2010. Factors associated with infant mortality are explored using logistic regression. Postoperative morbidity following patch versus primary closure, minimally invasive versus open surgery and biological versus synthetic patch material is described. Data are presented as n (%) and median (IQR).

    Results Overall known survival to 1 year was 75%, 95% CI 68% to 81% (138/184) and postoperative survival 93%, 95% CI 88% to 97% (138/148). Female sex, antenatal diagnosis, use of vasodilators or inotropes, being small for gestational age, patch repair and use of surfactant were all associated with infant death. Infants undergoing patch repair had a high incidence of postoperative chylothorax (11/54 vs 2/96 in infants undergoing primary closure) and a long length of hospital stay (41 days, IQR 24–68 vs 16 days, IQR 10–25 in primary closure group). Infants managed with synthetic patch material had a high incidence of chylothorax (11/34 vs 0/19 with biological patch).

    Conclusion The majority of infant deaths in babies born with CDH occur before surgical correction. Female sex, being born small for gestational age, surfactant use, patch repair and receipt of cardiovascular support were associated with a higher risk of death. The optimum surgical approach, timing of operation and choice of patch material to achieve lowest morbidity warrants further evaluation.

    • congenital diaphragmatic hernia
    • mortality
    • outcomes research
    • prognosis
    • congenital abnorm

    Statistics from

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    What is already known on this topic?

    • Congenital diaphragmatic hernia (CDH) is a congenital birth defect which carries a high mortality.

    • Infants with larger defects, requiring patch closure have poorer outcomes than those closed primarily.

    What this study adds?

    • Infants with CDH have an increased risk of death if they are female, born small for gestational age, given surfactant, have a patch repair or receive cardiovascular support.

    • Patch closure may increase the risk of postoperative chylothorax and the possible influence of patch material on chylothorax incidence requires further investigation.

    • These outcome data will enable more specific parental counselling and better shared decision-making.


    Congenital diaphragmatic hernia (CDH) is an abnormality requiring surgical correction with a relatively high mortality due to lung hypoplasia, aberrant pulmonary vascular biology and associated major anomalies. Short-term survival outcomes from institutions around the world have been extensively reported, with the majority focussing on mortality before initial discharge from hospital.1–4 Significant morbidity also accompanies CDH with medium-term complications including: hernia recurrence, chylothorax, gastro-oesophageal reflux disease (GORD) and prolonged hospital stay.5 6

    This study reports infant mortality and morbidity in the first year following operation in a national population cohort of children born alive with CDH. Variables related to infant mortality are assessed. Factors associated with death before surgery in this cohort have been reported previously and were found to include: female sex, prenatal diagnosis and the need for inotropes or pulmonary vasodilators.7 The aim of this analysis was to investigate factors related to all deaths within the cohort including those before surgery and those in the first postoperative year.

    In addition to mortality, hernia recurrence, incidence of chylothorax, GORD and length of stay in hospital in the first postoperative year were assessed and their incidence after the use of different surgical techniques are compared. These include comparisons between primary versus patch diaphragm closure, open versus minimally invasive surgery (MIS) and the use of biological versus synthetic patch material.


    Case identification

    Cases were identified and data collected using the British Association of Paediatric Surgeons Congenital Anomaly Surveillance System described previously.8 Initial standardised data collection forms were sent to reporting clinicians from all 28 specialist paediatric surgical centres in the UK and Ireland for each of the live-born infants within the population cohort. Data were double entered into a customised database and missing data were sought from reporting centres and clinicians. Further postoperative outcome data were obtained by sending an additional standardised data collection form 1 year after surgery to all surgical centres for information pertaining to all surviving infants.

    Outcome measures

    Data were collected and analysed on death, length of postoperative stay (LOS), chylothorax, GORD and hernia recurrence within one postoperative year. LOS was defined as date of primary operation, that is, CDH repair, to date of home discharge.

    Non-survivors were excluded from the analysis of LOS. Death was defined as either death before surgical correction or death in the first postoperative year. GORD was defined as the use of medication for gastro-oesophageal reflux. Incidence of postoperative pneumothorax(s), adhesive bowel obstruction and neurological sequelae were also documented. Neurological sequelae were defined as any known central nervous system dysfunction including: seizure activity, developmental delay or abnormalities detected on cerebral imaging.

    Statistical analyses

    Stata V.13 was used for all statistical analyses. Median and IQR values are presented for continuous data and differences analysed using Mann-Whitney U test. Due to less-than-complete follow-up for some infants, both minimum and maximum estimated incidences were calculated for some outcomes. These were defined as follows: in both calculations, the numerator was the reported incidence of the outcome at 28 days and, where data were returned, at 1 year. For the ‘minimum’ estimate, the denominator was the total number of children in the cohort. For the ‘maximum’ estimate, the denominator was the number of children who either returned follow-up forms, died or reported the outcome in question on the early data collection form. In the case of postoperative outcomes (table 1 and online supplementary table S1), denominators were restricted to those children who had undergone surgery.

    Table 1

    ‘Maximum’ incidence of given postoperative outcomes according to potential risk factors

    Risk factors assessed for their relationship with the postoperative outcomes described above included side of hernia, use of a patch, patch material and surgical approach (open surgery or MIS).

    Univariable and multivariable logistic regression analyses were used to assess factors associated with the risk of infant death. Possible risk factors were determined a priori from review of the literature and included: baseline demographic variables, antenatal diagnosis and presence of liver in the chest on antenatal ultrasound, markers of physiological instability for example, use of pulmonary vasodilators or inotropic support and organisational factors such as postnatal transfer and centre case-volume. Factors with a significant ORs (p<0.10) when fitted singly were included in a multivariable model in a forward stepwise method in order of their statistical significance. Those that improved the fit of the model on likelihood ratio testing were retained. Factors with a p value of <0.05 within the final model were considered to be statistically significant/associated with infant death.

    Small for gestational age (SGA) was investigated as a risk factor for mortality: this was defined as being less than the 10th percentile for birth weight for gestational age using British normative data from the LMS growth study.9 Birth weight is collinear with gestational age and related to SGA. This would therefore not be carried forward to the multivariable model alongside either of these factors. Factors that would only be relevant to certain groups of infants, for example, those who had an antenatal diagnosis or those who underwent surgery were also not carried forward into the multivariable model as to do so would have restricted the analysis to these small subsets of infants. Such factors included postnatal transfer and surgical repair type.


    From a live-born cohort of 219 babies with CDH, 182 underwent surgical repair (83%). Complete follow-up was obtained on 184 infants (85% of the original cohort of 219). Overall known survival to 1 year among the cohort was 138/184 (75%, 95% CI 68 to 81).

    Postoperative outcomes

    A total of 10 infants who underwent surgery died within the following year (7%, 95% CI 3 to 12) of the 148 infants who underwent surgery in whom complete follow-up was available. The timing of postoperative death is shown in figure 1. The causes of postoperative death were as follows: six had intractable pulmonary hypertension, two were unable to be weaned from ventilation (one of these had severe cerebral atrophy), one had sepsis and the cause of one death was unknown.

    Figure 1

    Kaplan-Meier graph showing timing of death from surgery in a 12-month period in infants with congenital diaphragmatic hernia.

    Overall LOS was 20 days (IQR 11–35 days) for those who survived. Table 1 shows the comparative analysis of the maximum incidence of key postoperative outcomes according to potential risk factors. The analysis of the minimum incidence results (shown in online supplementary table S1) are not markedly different and so all further results and discussion are based on the maximum estimates.

    The hernia recurrence rate was 12/150 (8%, 4.2%–13.6%). Postoperative chylothorax was observed in only 13/150 (9%, 4.7–14.4) infants.

    Nine of 52 CDH newborns who had patch repair died postoperatively (17%). The median length of hospital stay in infants having a patch reconstruction was 41 days (IQR 24–68); the median length of stay for babies undergoing primary CDH defect closure was 16 days (IQR 10–25).

    Postoperative chylothorax occurred in 20% of children who underwent patch closure and 2% of those closed primarily. All chylothoraces following patch closure developed after the use of synthetic patch material, for example, Gore-tex, Polypropylene or Polyester (11/34 (32%) and none occurred in the 19 infants closed with biological patch material. These were bovine or porcine collagen grafts, for example, Permacol, Tutopatch or Surgisis.

    The number of infants receiving medical therapy(s) for GOR was 40/150 (27%). Over a third of infants closed with a patch were treated with medication for GORD (20/53, 38%); a fifth of those who had CDH primary repair had GOR medication (20/97, 21% infants). Eight of 182 of infants who underwent surgery for CDH had a fundoplication in the first postoperative year (4%). Five of these had a Nissen, one was of a Thal type and the other two antireflux surgery procedures were of undefined type. The timing of fundoplication was clearly documented in six infants (range 34–247 days) following CDH primary operation (median 153 days).

    Univariable analysis identified 10 factors eligible for inclusion in the multivariable model (table 2). After likelihood ratio testing (significance level 0.05), gestational age was the only variable excluded from the final model (table 3).

    Table 2

    Characteristics of infants who died before 1 year of age and those who survived

    Table 3

    Factors associated with death before 1 year

    Other morbidity

    Complete data on each of the postoperative morbidities were not obtained for all infants. Data presented here therefore represent those for infants with complete data on each given defined outcome. Twenty infants were reported as having had a pneumothorax after surgery 20/182 (11%), two of these also had a pneumothorax prior to surgery. Nine of 140 infants (6%) were known to have experienced adhesive bowel obstruction in the year following CDH repair. Adhesive bowel obstruction occurred in 3/45 (7%) of those closed with a patch and 6/95 (6%) who had a primary closure. This occurred in 0/10 who had MIS compared with 9/130 (7%) in the open group.

    Twelve infants with CDH were known to have been discharged home on oxygen therapy 12/150 (8%). Seven of the 42 infants who underwent patch repair went home on oxygen (17%) compared with 5/108 (5%) undergoing primary repair. Three children were reported as having early evidence of a pectus chest wall deformity. Fourteen children from the cohort were reported as having associated neurological sequelae. Three of these developed seizure activity and four were described as having developmental delay.


    This UK and Ireland population-based study has shown that the overall survival of live-born infants with CDH was 75% at 1 year. It is evident that the major burden of mortality in this population born with CDH occurs before operation with 78% of all deaths occurring before surgery and 93% of stable infants with CDH who undergo surgery surviving to 1 year. In addition to the risk factors for early mortality identified in our previous report, antenatal diagnosis, female sex, the need for inotropes and pulmonary vasodilators, this study further identified that being SGA is a risk factor for infant mortality in those born with CDH (adjusted OR 6.6). Forty-one per cent of all SGA infants with CDH died before 1 year. This is the first study, to the best of our knowledge, to have identified SGA as a risk factor for mortality in those born with CDH and we recommend it should be included as a covariate in all future analyses of CDH mortality in live-born infants. Pulmonary hypoplasia and lung immaturity are likely to be more marked in these infants and affect their major requirement(s) for cardiopulmonary support.

    Postoperative survival was also strongly linked to diaphragm defect size and 9 of 10 children who died after surgery had undergone diaphragm closure with a patch. These findings share common themes with the Canadian CAPSNET CDH population-based studies and the International CDH Study Group Registry both of whom identified a strong relationship between defect size and survival.2 4

    This study did not demonstrate a relationship between CDH mortality and side of hernia or hospital case volume, factors cited in international publications that have suggested a significant impact on CDH prognosis.10–14 The lack of a demonstrable ‘volume-outcome relationship’ among this CDH cohort herein reported likely relates to the relatively small number of CDH babies cared for in ‘low volume’ UK and Ireland centres 40/217 (18%), precluding reliable statistical analysis. No conclusions can, therefore, be drawn from this study about case volume and outcome.

    One-year outcome data generated from this nationwide study are valuable for benchmarking, parental counselling and defining survival metrics and compare favourably with a recent large study published from France.15 Here, it is noteworthy that although preoperative mortality was similar in the UK and Ireland versus France (17% UK vs 18% France), the largest health outcome differences were observed in the mortality rate(s) of babies after CDH operation, 10/148 (7%) UK versus 21/114 France (18%).15 This difference is statistically significantly different (p=0.004 [Χ2 test]). These potential background differences are important to note particularly if international randomised controlled trials for interventions for CDH are planned in the future and/or for the interpretation of those already in progress, for example, the sildenafil versus inhaled nitric oxide for CDH (CoDiNOS trial).16

    When comparing outcomes of different surgical techniques, this study noted the fact that median postoperative length of stay appeared different in those primarily repaired with versus without a patch (40.5 vs 16 days). Our data suggest that the incidence of postoperative chylothorax may be higher in those repaired with a patch compared with those repaired primarily. Other authors have also found this to be the case,17–19 whereas others have not.20 The risk of early hernia recurrence in all CDH infants was 6% after primary closure vs 11% in those having a patch. It is noteworthy that few newborns in this series underwent minimally invasive repair (n=10 with complete follow-up). A robust comparison of clinical outcomes in these infants compared with those repaired with conventional open surgery was therefore not possible.

    This study is the first such report to suggest that there may be a relationship between the deployment of a synthetic material patch, for example, Gore-tex or Polypropylene and postoperative chylothorax. There are, as yet no clear data to suggest the superiority of one patch material over another and a lack of superiority in preventing recurrence was not demonstrated here due to the relatively small number of infants treated in this way.21 Large-scale clinical trials seeking to evaluate patch materials together with high-quality observational studies with meta-analyses may provide future answers.

    Regarding other postoperative outcomes, in the first year, we found that 27% of infants were prescribed medications for GORD and yet the rate of early fundoplication in the first postoperative year was low (4%). These metrics are significantly lower than several international published studies5 22 reflecting contemporary practice in the UK and Ireland.23 Adhesive bowel obstruction was an uncommon early event in this series, affecting 6% infants in the first year after repair. Risk factors for this remain unknown but it is an important to counsel parents about this possible complication prior to discharge.5 23 24

    Early neurological sequelae were likely under-reported here due to the relatively short duration of follow-up (1 year). Despite this, some infants were reported by collaborating centres to have developmental impairment(s) and/or seizure activity. Future well-designed studies incorporating large patient numbers in registries/networks may uncover the true incidence of neurological and developmental sequelae including neurosensory impairment, epilepsy and autism.25–29 Parent groups, supported by medical professionals have long called for a national registry of infants with CHD to prospectively collect data on infant characteristics and outcomes (CDH UK Parent Group – Personal Communication, 2018).

    This population study reports morbidity and mortality in the first year after birth in live-born infants with CDH. Postoperative survival of babies at 1 year following primary operation was 93%. Infants born with CDH who are SGA were identified as an ‘at-risk' group for mortality.

    The associated well-defined relationship(s) between large CDH defect size as indicated by the requirement for patch repair and adverse clinical outcome is notable. It is clear that technical aspects of operation for CDH, for example, open versus MIS or choice of prosthetic patch material warrant further evaluation to define patient selection criteria and to inform ‘best practice’.


    The authors would like to acknowledge the support, dedication and hard work of all those who contributed to this study. Sean Marven, Elizabeth Draper, Paul Johnson, Peter Brocklehurst, David Howe and Judith Rankin were involved in the original set-up and design of the study.



    • Contributors PDL, MK and JJK designed and coordinated the study along with the BAPS-CASS collaboration. A-ML and KJB analysed the data and A-ML wrote the manuscript which was reviewed and edited by all other authors.

    • Funding This study was funded by Action Medical Research (SP4407, GN1739).

    • Competing interests None declared.

    • Ethics approval Research ethics committee approval for this study was received from the London Research Ethics Committee: Ref 09/H0718/10.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Collaborators The British Association of Paediatric Surgeons Congenital Anomaly Surveillance System (BAPS-CASS) Collaborators: Marven S, Draper E, Johnson P, Brocklehurst P, Howe D, Rankin J, Aslam A, Jawaid W, Wilkinson D, Robb A, Lander A, Burki T, Coates L, Cusick E, Fishman J, Madden N, Adikibi B, Mackinlay G, Sabharwal A, Larcombe C, Curry J, Besarovic S, Ashour K, Johnson P, Carnaghan H, Davenport M, Sugarman I, Nour S, Tsang T, Paul A, Davies B, Mclaughlin D, Puri P, Abel R, Driver C, Mahomed A, Ervine E, Mccallion W, Phelps S, Craigie R, Morabito A, De La Hunt M, Hosie G, Lloyd K, Taylor R, Drewett M, Okoye B, Richards C, King J, Nellihala LP, de La Hunt M, Huddart S.

    • Patient consent for publication Not required.

    Linked Articles

    • Fantoms
      Ben J Stenson