Article Text

Download PDFPDF

Timing of diagnosis affects mortality in critical congenital heart disease
  1. Luke Eckersley1,
  2. Lynn Sadler2,
  3. Emma Parry3,
  4. Kirsten Finucane1,
  5. Thomas L Gentles1
  1. 1Greenlane Paediatric and Congenital Cardiac Service, Starship Children's Hospital, Auckland, New Zealand
  2. 2Department of Obstetrics & Gynaecology, National Women's Hospital, Auckland, New Zealand
  3. 3New Zealand Maternal Fetal Medicine Network, National Women's Hospital, Auckland, New Zealand
  1. Correspondence to Dr Tom Gentles, Paediatric Cardiologist, Green Lane Paediatric and Congenital Cardiac Service, Starship Children's Hospital, Private Bag 92024, Auckland 1042, New Zealand; tomg{at}


Objective Screening for critical congenital heart disease (CHD) with prenatal ultrasound or postnatal pulse oximetry has the potential to improve outcome. To guide screening recommendations, this study aimed to identify the proportion and outcome of major CHD diagnosed before (early) or after (late) postnatal discharge prior to the introduction of postnatal oximetry screening.

Design A retrospective, population-based review of all major CHD in New Zealand from 2006 to 2010. The timing of diagnosis relative to discharge and to intervention in critical and non-critical cases with intention to treat was determined, as was the relationship of diagnostic timing to mortality at 1 year of age.

Results Late diagnosis occurred in 20% of critical and 51% of non-critical cases. Mortality occurred in 18% of critical vs 8% of non-critical cases. Mortality was lower with an early diagnosis of critical CHD (early diagnosis 16% vs late diagnosis 27%, p=0.04). Isolated critical CHD benefited most from early diagnosis (mortality, early diagnosis 12% vs late diagnosis 29%, p=0.002). Early diagnosis occurred in >90% critical complex CHD and hypoplastic left heart syndrome, 85% d-transposition of the great arteries (d-TGA) and 53% critical left ventricular outflow tract obstruction (LVOTO). Deaths in d-TGA and LVOTO primarily occurred prior to intervention and for d-TGA most often when birth was distant from the cardiac centre.

Conclusions Excess mortality occurs following late diagnosis of critical CHD, and for d-TGA even with early diagnosis if intervention is not immediately available. Antenatal detection retains an important role in reducing mortality related to critical CHD.

  • Mortality
  • Cardiac Surgery
  • congenital heart disease
  • delayed diagnosis
  • atrial septal defect

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

  • The timing of diagnosis of critical congenital heart disease (CHD) affects outcome.

  • This has driven adoption of universal pulse oximetry screening in many jurisdictions.

  • Fetal screening has received less emphasis and funding.

What this study adds

  • In New Zealand, there are deaths attributable to late diagnosis of critical CHD.

  • Many of these deaths occur in conditions where pulse oximetry detection rates are low (coarctation) or where the outcome is adversely affected by birth distant to the surgical centre (transposition).


Congenital heart disease (CHD) is the most common class of congenital anomaly and is responsible for 30–50% of congenital infant mortality.1–3 CHD is often categorised as requiring intervention or resulting in death within 1 year (major CHD∼3/1000 live births) or 4 weeks (critical CHD∼1.2/1000 live births).4–6

Major CHD may not be diagnosed before birth or prior to postnatal discharge. Factors contributing to a later diagnosis may include absence of a murmur, mild hypoxaemia, inadequate training of staff and early postnatal discharge. A delay in diagnosis, particularly of critical CHD, is associated with a significant increase in mortality and morbidity.2 ,7–9

Screening for CHD aims to avoid late diagnosis. Referral for cardiac evaluation has been guided by universal fetal ultrasound screening in some countries,9–11 and postnatal pulse oximetry screening in many others.2 ,6 ,12–15

New Zealand has a population of 4.2 million with a birth rate of ∼65 000 per year. There are significant challenges to infant cardiac outcome including the extent of engagement with prenatal ultrasound screening and potential for birth remote from definitive cardiac care. There is no pulse oximetry screening programme.

The aim of this retrospective population-based study is to guide delivery of screening services for CHD by estimating how many babies born with major CHD in New Zealand from 2006 to 2010 received a late diagnosis. We examined deaths associated with these late diagnoses to determine whether an improved rate of prenatal diagnosis or universal postnatal pulse oximetry screening might have improved outcome.


Study population

Included were live births, still births and terminations of pregnancy (ToP) in New Zealand from 20 weeks or 400 g (if gestation was unknown) with major CHD from 1 February 2006 to 31 December 2010. CHD was considered major if there was a late ToP or stillbirth, or where the infant underwent cardiac surgery, interventional catheterisation or died at <1 year of age. Timing of diagnosis and mortality were analysed for live-born infants where there was an intention to treat.


All paediatric cardiac procedures in New Zealand are provided at a single centre in Auckland. Surgical and cardiac catheterisation data were acquired from departmental databases. Fetal data were acquired from the fetal cardiology database, which collects all fetal diagnoses of cardiac disease in New Zealand referred for a fetal cardiology opinion.

To ensure complete ascertainment, mortality data were obtained from the New Zealand Perinatal and Maternal Mortality Review Committee (PMMRC) and the Child and Youth Mortality Review Committee (CYMRC). Notification of death to these review committees is a legal requirement in New Zealand from 20 weeks’ gestation until age 24 years. The PMMRC and CYMRC databases include postmortem reports when available and undertake linkage to numerous other governmental datasets. International Classification of Diseases, 10th revision coding is used to record the underlying cause of death, other contributory factors and comorbidities.16

Data from each database were combined, using New Zealand National Health Index (NHI) number as a unique identifier. An NHI is assigned for all fetuses from 20 weeks’ gestation. For each record, the most appropriate unifying congenital cardiac diagnosis and any significant non-cardiac diagnoses were identified. The timing of diagnosis was identified from the fetal cardiology database, at chart review, or from the PMMRC case file as ‘early’ (antenatal, predischarge (from birth admission)) or late (postdischarge or postmortem). Medical records and/or PMMRC case files were also reviewed to audit clinical coding and identify additional diagnoses.

Exclusion criteria

Those with birth locations outside New Zealand, isolated patent ductus arteriosus, secundum atrial septal defects, partial anomalous pulmonary venous drainage, cardiomyopathies, arrhythmia and cardiac disease related to Marfan syndrome were excluded. Also excluded were those where a comfort care pathway was followed (figure 1).

Figure 1

Major congenital heart disease (CHD)—number of cases and outcome in infants with date of birth February 2006–December 2010. ToP, terminations of pregnancy.


Critical CHD: CHD that was duct-dependent, required intervention or resulted in death at <28 days of age.

Isolated CHD: CHD not part of a syndrome, nor associated with a major non-cardiac congenital anomaly.

Complex CHD: Double inlet left ventricle, double outlet right ventricle, pulmonary atresia, tricuspid atresia, Ebsteins anomaly, L-TGA, heterotaxy.

Syndromic diagnosis: An established unifying diagnosis made using genetic or clinical criteria.

Non-cardiac major congenital anomaly (ncMCA): An anomaly requiring medical or surgical intervention, or having a significant impact on quality of life, excluding syndromic diagnoses.

Late ToP: From 20 weeks’ gestation.

Statistical methods

Data are presented as number of cases, percentage or case incident rate per 1000 total birth registrations (including fetal deaths from 20 weeks or at least 400 g if gestation unknown). CIs of mortality figures were calculated using the Wald method. Fisher’s exact test was used for comparison of categorical variables (GraphPad Prism, GraphPad, San Diego, California, USA). Kaplan–Meier analysis of survival was censored at 1 year of age. Log-rank analysis was generated for comparison of survival curves using SAS software (SAS Institute).


Of the 313 478 births in New Zealand from 1 February 2006 to 31 December 2010, there were 906 cases of major CHD (incidence 2.89/1000 births). These included 105 late TOPs, 33 stillbirths and 34 cases who received comfort care and died after birth. The remainder constituted live births with intention to treat and were the group subjected to further analysis (734 cases, rate 2.34/1000 births) (figure 1). Of these, 353 (1.12/1000 births) had critical CHD, 314 of whom had isolated critical CHD. Of the 34 with major CHD treated with comfort care, 22 had critical CHD, which was isolated in 12. Major comorbidities contributed to the decision not to pursue active treatment in 22 of the 34. Syndromic and ncMCA diagnoses were less common in the intention-to-treat group than in the ToP and comfort care groups (138/734 (19%) vs 65/139 (47%) p<0.001).

Of 734 in the intention-to-treat group, timing of diagnosis (antenatal, predischarge and postdischarge) could be divided approximately into thirds. Critical CHD was more likely to be diagnosed antenatally (163/353, 46%) than non-critical CHD (85/381, 22%) (p<0.0001). Postdischarge diagnosis occurred in 20% of critical CHD and in 51% of non-critical CHD. Isolated major CHD was more likely to be diagnosed after hospital discharge (235/596, 39%) than syndromic or ncMCA (30/138, 22%) (p<0.0001).

Total mortality in the first year of life with major CHD (table 1) was similar following antenatal (13.2%), predischarge (13.8%) and postdischarge diagnosis (12%). This remained the case when analysis was restricted to isolated major CHD. The mortality for critical CHD diagnosed after hospital discharge was higher than that for non-critical major CHD (19/70 (27%) vs 13/196 (6.6%), p<0.001).

Table 1

Outcome by timing of diagnosis of major CHD with intention to treat actively

Kaplan–Meier analysis demonstrated similar survival for critical CHD with antenatal and predischarge diagnosis (figure 2A). There was no difference in the antenatal and predischarge proportion with syndromes or ncMCA (17/165, 10% and 17/118, 14%, respectively). The antenatal and predischarge group were therefore combined to represent early diagnosis (figure 2B). Early diagnosis was associated with a significant survival advantage compared with late diagnosis (figure 2B; p=0.04). The survival advantage was more marked in those with isolated critical CHD with 12% mortality in the early diagnosis group vs 29% in those diagnosed late (p=0.002) (table 1). The difference in mortality following late diagnosis of isolated critical CHD (65/314, 21%) and critical CHD with coexisting syndromic or ncMCA diagnosis (5/39, 13%) did not reach statistical significance.

Figure 2

(A) Survival with critical congenital heart disease (CHD) and intention to treat actively following (A) antenatal, predischarge and postdischarge diagnosis; (B) early (antenatal or predischarge from hospital) or late (postdischarge from hospital) diagnosis.

Antenatal diagnosis occurred in the majority of live-born hypoplastic left heart syndrome (HLHS) (23/29, 79%) and complex CHD (60/75, 80%) but less commonly in transposition of the great arteries (D-TGA) (38/93, 41%). Only 25% (15/62) of critical left ventricular outflow tract obstruction (LVOTO), primarily coarctation of aorta, was diagnosed before birth. Diagnosis before hospital discharge occurred in 90% of HLHS, 92% of complex CHD, 85% of D-TGA and 53% of LVOTO. Overall mortality <1 year of age was 38% for HLHS, 19% for critical LVOTO, 17% for complex CHD and 10% for d-TGA (table 2).

Table 2

Timing of diagnosis and outcome in selected critical CHD with intention to treat actively

There was no significant difference between early and late diagnosis for any type of critical CHD when each group was analysed separately; nor was there a significant difference in mortality between antenatal (2/38, 5%) and predischarge (6/41, 15%) diagnosis for d-TGA (p=0.27). Despite an early diagnosis, death was more likely to occur before intervention with LVOTO (3/4, 75%) and d-TGA (7/8, 88%) compared with HLHS (1/9, 11%) and complex CHD (2/12, 17%) (p<0.001).

Two of nine deaths with d-TGA had an antenatal diagnosis. One died as a result of an obstetric complication, while the other was a premature infant with pulmonary hypertension and necrotising enterocolitis. Of six diagnosed after birth and before discharge, four were born >100 km from the cardiac centre, the fifth was born prematurely at a nearby tertiary hospital with multiple comorbidities and the sixth died 4 months after operation of an unrelated cause. The only infant dying with d-TGA without an early diagnosis remained undiagnosed until postmortem.

There were eight deaths associated with a late diagnosis of LVOTO; four without any intervention, one after surgery and three in whom the diagnosis was made at postmortem. In total, 10 of the 12 deaths in the LVOTO group occurred without intervention. Critical CHD was diagnosed in four other infants after death; the infant with d-TGA, and one each with HLHS, atrioventricular septal defect and an anomalous left coronary artery from pulmonary artery. The incidence of CHD diagnosed at postmortem was 0.02/1000 live births (∼2% of critical CHD).


Detection of critical CHD is the primary target of antenatal and newborn cardiac screening programmes. We have shown that 20% of cases of critical CHD in New Zealand were diagnosed after postnatal discharge from hospital, comparable to the 22–30% found in other recent population-based studies.8 ,17–19

Late diagnosis of isolated critical CHD carried a significantly higher risk of mortality (29%) than early diagnosis (12%). Complex CHD/HLHS was diagnosed early and received intervention in most cases, but despite this had high mortality. The improved overall mortality conveyed by early diagnosis was mainly due to the low postintervention mortality of simple critical CHD (LVOTO/d-TGA).

New Zealand has the unusual circumstance of a single interventional centre for CHD, favouring high case ascertainment. As reporting of perinatal and infant deaths is mandated, we were able to identify cases diagnosed postmortem. The number of infants dying prior to diagnosis was higher than that reported in the last six years of the North of England study17 or in the USA.19 This difference may be due to the widely dispersed population in New Zealand or a reflection of more robust case ascertainment. The incidence of major CHD and critical CHD is similar to other population-based studies.2 ,6

Diagnosis and treatment of CHD is undergoing rapid change, and outcomes are generally improving.3 A short and recent period of data acquisition is therefore advantageous, and was achieved in this study due to the size of the population covered (65 000 births per year).

Moreover, the definition of critical CHD was based on management and outcome. Larger studies have generally relied upon centralised coding by non-specialist staff for diagnosis and have therefore not been able to identify critical cases in a robust way.3 ,20

Cases that were associated with ToP and stillbirth or a comfort care pathway were excluded from analysis. The severity and complexity of CHD and the prevalence of syndromic and ncMCA in this excluded group are greater than that in the remaining cohort. All of these cases had an early diagnosis, and most a prenatal diagnosis. The mortality rate would most likely have been higher had they been actively treated.

Universal screening with pulse oximetry has resulted in cost-effective, improved detection of critical CHD.20 Prenatal diagnosis of d-TGA reduces early neonatal mortality and preoperative acidiosis7 while late diagnosis has been associated with medium-term neurological deficit.21 ,22 Similarly, postnatal diagnosis was associated with increased mortality for d-TGA in this study. The impact of late diagnosis and the influence of new screening protocols on long-term morbidity and neurological sequelae remain undetermined.

Our goal was to identify the size of the problem of postdischarge diagnosis of CHD in New Zealand and guide screening recommendations. Although complex cardiac disease accounts for the majority of postintervention mortality, failure to reach intervention was the major reason for death in simple critical CHD such as LVOTO and d-TGA.

Pulse oximetry studies have shown a sensitivity of ∼0.8 overall for critical CHD, and ∼0.6 when conditions commonly diagnosed by antenatal ultrasound are excluded.2 ,6 ,15 ,23 However, pulse oximetry sensitivity for LVOTO is ∼0.36,24 and LVOTO accounted for half the late diagnoses in this study. Provision of improved training and feedback on performance of fetal cardiac screening may improve antenatal diagnosis of LVOTO.25

A major finding of this study is the high mortality rate in those with d-TGA (9/93 (9.7%)) with only one death occurring after definitive surgery, and that due to an unrelated cause. Of those identified before hospital discharge (but not prior to birth), four of the five deaths occurred in infants delivered at a distance from the surgical centre and the other with significant comorbidity. d-TGA with severe hypoxaemia usually presents within the first few hours of birth. An immediate anatomic diagnosis, timely balloon atrial septostomy, expert cardiac intensive care and sometimes early operation may be lifesaving. These interventions are delayed if delivery is distant to the cardiac centre. In New Zealand, those with antenatally diagnosed critical CHD are delivered at the cardiac centre except in the rare case of unanticipated premature delivery. Improved antenatal detection has been predicted to reduce the risk of adverse outcome including death for d-TGA by 17% in Sweden,26 while in Holland the mortality rate decreased from 6.5% to 0%.27

The false positive rate for pulse oximetry detection of critical CHD ranges from 0.04% to 0.4%.2 ,6 ,15 ,28 ,29 Half of these have other pathology, including non-critical CHD, respiratory infection or sepsis. Protocols where failed oximetry screening triggered a clinician review rather than echocardiography have shown little change in resource use.30 The psychological impact of false positive and negative results is an issue for oximetry screening that requires ongoing assessment.31


This study only examined mortality as an outcome measure. We were unable to demonstrate a mortality difference for subgroups of cardiac lesions by timing of diagnosis; most probably because of the small sample size and low mortality rate in groups with hypothesised differences (eg, LVOTO). Late diagnosis may also contribute to morbidity, including cerebral and other end-organ damage.32 Analysis of major morbidity in addition to mortality could provide important additional information and increase sample size, without the need for data collection over several decades.


Late diagnosis of critical CHD in New Zealand occurs in around 20% of cases, with excess mortality particularly in cases with an otherwise good prognosis, such as LVOTO and d-TGA. The benefit of earlier postnatal diagnosis for d-TGA may be limited, and improved antenatal detection is more likely to impact on mortality and morbidity by allowing delivery close to a cardiac centre. Further work is required to improve early detection of LVOTO.


We would like to thank Charlene Nell, Desktop Support Administrator, for formatting the manuscript according to the guidelines and for excellent secretarial assistance. We acknowledge the support and usage of the PMMRC and CYMRC databases, Sinead McGlacken-Byrne for data auditing, and Steve Withy and Noelle Balbas for database management.


View Abstract


  • Contributors LE and TLG were involved in the study conception and design; analysis and interpretation of data; drafted the article; and final approval of the manuscript. LE and TLG were involved in the acquisition and analysis of data; revising the article; and final approval of the manuscript. All authors were involved in the study conception and design; interpretation of data; drafting and revising the article; and final approval of the manuscript.

  • Competing interests None declared.

  • Ethics approval The Health and Disability Ethics Committees of New Zealand, Multi-region Ethics Committee (MEC/12/EXP/077).

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

Linked Articles