Objective The life expectancy of patients with congenital heart disease (CHD) has significantly improved with advances in their paediatric medical care. Mortality patterns are changing as a result. Our study aims to describe survival and causes of death in a contemporary cohort of adult patients with CHD.
Methods We reviewed 3068 patients in our adult CHD database (age ≥16 years, seen at least once in our centre between 2000 and 2015), and documented the number and causes of death, via Australia’s National Death Index. Survival and mortality patterns were analysed by complexity of CHD and by underlying congenital diagnosis.
Results Our cohort comprised 3068 adult patients (53% male). The distribution of patients (per the Bethesda classification) was 47% simple, 34% moderate and 18% complex (1% not classifiable). Over a median follow-up of 6.2 years (IQR 3.5–10.4), 341 patients (11%) died with an incidence of 0.4 deaths/100 patient years (py). Survival was significantly worse with increasing complexity of CHD (p<0.001); mortality rate in the simple group was 0.3 deaths/100 py with a median age of death 70 years, and in the complex group was 1.0 death/100 py with a median age of death 34 years. Overall, non-cardiac causes of death outnumbered cardiac causes, at 54% and 46%, respectively. The leading single cause of death was heart failure (17%), followed by malignancy (13%). Simple adult CHD patients mostly died due to non-cardiac causes such as malignancy. Perioperative mortality only accounted for 5% of deaths.
Conclusions Premature death is common in adults with CHD. Although heart failure remains the most common cause of death, in the contemporary era in a specialist CHD centre, non-cardiac related deaths outnumber cardiac deaths, particularly in those with simple CHD lesions.
- heart failure
- cardiac arrest
Statistics from Altmetric.com
Over the last few decades, advances in paediatric cardiology and cardiac surgery have resulted in markedly improved outcomes in patients with congenital heart disease (CHD). Now, approximately 90% of CHD children survive to adulthood.1 The reduction in CHD mortality has been most marked in infants, whereas adult CHD (ACHD) mortality rates have declined in parallel with the general population.2 Today in developed nations, the ACHD population outnumbers the paediatric CHD population.1 3
Historically, the main causes of death in the ACHD population have been cardiac, including congestive heart failure, sudden cardiac death (SCD) and perioperative death.4–7 This is particularly the case in complex forms of disease such as Eisenmenger syndrome, transposition with atrial switch procedures and single ventricle hearts.4 6–8 Repaired tetralogy patients are at higher risk of SCD.9 Recently, it has been suggested that mortality patterns in ACHD are changing, with competing non-cardiac causes of death as these patients live longer into adulthood.7 Nevertheless mortality remains substantially higher than in the general population, particularly in more complex forms of CHD.7
This study aims to investigate survival and causes of death in a contemporary ACHD population, taken from a state-wide quaternary referral centre.
A retrospective single-centre review was performed at a quaternary ACHD referral centre in Sydney, Australia. This is the only ACHD referral centre state-wide with an estimated catchment of 6–7 million people. Consent was an opt-out scheme, as approved by our ethics committee. We included 3068 patients, seen at least once between January 2000 and December 2015. Patients with an isolated patent foramen ovale (PFO), small atrial septal defects (ASD) and dextrocardia with no associated congenital heart defects were excluded. A total of 4277 patients are registered in our ACHD database; of these, 1035 were excluded as they had not been seen between 2000 and 2015, 168 were excluded with isolated PFO or small ASD and six were excluded with isolated dextrocardia.
A comprehensive list of all deaths was obtained from the National Death Index Survey (NDIS) which records vital status nationwide. Causes of death were obtained from the NDIS in the majority of cases; in occasional cases, an experienced ACHD cardiologist reviewed the medical records to define the cause of death. Perioperative death was classified as death within 30 days of surgery. Cardiac death was defined as death due to heart failure, SCD, acute myocardial infarction (AMI), endocarditis, aortic complications and unspecified cardiac death. SCD was defined as death due to a cardiac cause within 1 hour of the onset of symptoms or unwitnessed death in the absence of a known non-cardiac condition as the proximate cause of death. Chronic ischaemic heart disease (IHD) was deemed the cause of death if listed as the primary cause of on the death certificate as per ICD 10 coding (I25). The definition of chronic IHD as per ICD 10 encompasses old myocardial infarction and ischaemic cardiomyopathy. Vascular death comprised cerebrovascular accidents (CVA), rupture of aneurysm and dissection.
Analysis was performed using SPSS V.22 (IBM). Data for categorical variables are reported as frequencies and percentages (%) and data for continuous variables are presented as medians with IQR. Kaplan-Meier curves for survival overall and by complexity of CHD were constructed. The time of entry for our survival analysis is age 16, when patients transfer from paediatric to adult services. There is left censoring for patients <16 years old (ie, some patients would have died prior to referral to ACHD centre). Patients were followed until either time of death, or last known vital status via the NDIS (which ensures complete data for survival). We did not adjust for other variables in our survival analysis. Death rates were compared using the Χ2 test and medians with the Mann-Whitney test. Statistical significance was inferred at a two-tailed p≤0.05.
Patient years (py) followed was calculated by multiplying the number of patients by the total years between their 16th birthday (age at referral to our ACHD centre) and last known vital status. Last known vital status was taken either as time of death, or if alive, the date of last NDIS verification (31st December 2015). Mortality incidence was then calculated by dividing the number of deaths by total patient years followed, then multiplying by 100 (expressed in deaths/100 py). This was calculated overall and by Bethesda classification of complexity for underlying CHD.10 Lesions were defined as simple, moderate and complex.10 If multiple lesions were present, patients were assigned to the lesion of greatest complexity.
Study population and mortality
The study population included 3068 patients (53% male) followed for 85 779 patient years total. By the Bethesda classification, 47% had simple CHD, 34% moderate and 18% complex. One per cent were not able to be classified and included diagnoses such as anomalous coronary arteries. In the study period, 341 patients (11%) died, with an incidence of 0.4 deaths/100 py. Median age at death was 54 years (IQR 34–72), with a male predominance in deceased patients (60%). Overall survival to age 40 was 95%, 92% to age 50 and 88% to age 60. There were 147 deaths (10%) in the simple group, 101 deaths (10%) in the moderate group and 91 deaths (17%) in the complex group. This corresponded to an incidence of 0.3, 0.4 and 1.0 deaths/100 py in the simple, moderate and complex groups, respectively. Patient demographics in the study population are seen in table 1. Risk of mortality was significantly increased with increasing complexity of CHD (p<0.001) (figure 1).
The leading cause of mortality in our cohort was heart failure (17%), followed by malignancy (13%) and SCD (10%). Cardiac death as a whole accounted for 46% of all deaths, as opposed to non-cardiac death in 54% (p=0.34). Median age of cardiac death was 46 years old and of non-cardiac death was 60 years old (p=0.01). The most common causes of non-cardiac death included malignancy (13%), vascular (8%) and pneumonia (8%). Most of the malignancies were gastrointestinal (n=13), followed by unspecified malignant site (n=10), lung (n=6) and haematological (n=4) malignancies. Suicide and overdose was the cause of death in six cases (2%), with the median age of death in this subgroup 34 years old (IQR 29–51). Table 2 displays causes of death by complexity of underlying CHD.
The leading causes of death among the 154 cardiac deaths were heart failure (38%), SCD (23%) and acute or chronic IHD (20%). Figure 2 illustrates that endocarditis and SCD inflicted earlier mortality compared with IHD and heart failure. The median age of death related to endocarditis and SCD was 40 and 35 years, respectively, which is lower than the overall cardiac median age of death.
The median age at death in the simple CHD group was 70 years (IQR 58–77), with the most common causes of death being malignancy (n=30), followed by heart failure (n=20) and IHD (n=18). The predominant lesions in the simple group included bicuspid aortic valve (BAV) in 710 (49%) and ASD in 395 (27%). Median age at death in the moderate and complex groups was 51 years old (IQR 35–69) and 34 years old (IQR 28–43), respectively. The most common causes of death in the complex group were heart failure (25%) followed by SCD (22%).
Perioperative mortality accounted for 5% (n=16) of deaths, with equal numbers of simple and moderate/complex patients, respectively (n=8). The majority of these deaths occurred following elective cardiac surgery (n=15). Despite the procedures being elective, they were often high-risk such as redo operations or extraction of infected pacemaker leads.
Mortality according to congenital diagnosis
The highest mortality rates were seen in Eisenmenger syndrome, complex (excluding Eisenmenger, transposition of the great arteries (TGA) and Fontans), TGA, and Fontans; 48%, 19%, 14% and 11%, respectively. The youngest and oldest patients at the time of death were 16 and 94 years old. Online supplementary table 1 shows detailed causes of death by underlying congenital heart diagnosis. Heart failure (n=5, 42%) was the most common cause of death in Fontans, followed by SCD (n=2, 17%). In patients with Eisenmenger syndrome, two-thirds of deaths were cardiac in nature, with SCD (n=7, 25%) being the most common cause, followed by heart failure (n=5, 18%). Additionally, in tetralogy of Fallot (TOF), heart failure and SCD were the main causes of death, at 20% and 17%, respectively. Patients with aortic coarctation are classified as moderate severity by Bethesda, but represent an entity which has a spectrum of treatments, from balloon angioplasty to more complex surgical intervention. The majority of these patients had aortic coarctation with BAV (54%). Vascular causes (aortic complications and CVA) and SCD were the most common causes of death in the coarctation group; 36% and 15%, respectively. Patients with Ebstein’s anomaly died equally as often from heart failure as from perioperative causes (n=2, 29% each).
Ventricular septal defects comprised a mixture of simple to moderate complexities (excluding Eisenmenger). The median age of death was 64 years (IQR 36–75) and non-cardiac causes of death were in the slight majority (n=14, 54%). The most common cardiac cause of death was attributed to AMI and IHD; six cases combined (23%). Simple complexity lesions consisted predominantly of ASD and BAV. Patients with ASD and BAV had near normal median age of deaths, 72 and 71 years, respectively, compared with the general Australian population (median age of death 78.9 years old).11 In those few patients who died under the age of 50 years (n=20), cardiac and non-cardiac deaths were similar in frequency (eight cardiac deaths and 12 non-cardiac deaths). Interestingly, in the BAV cohort, endocarditis was the only reason for cardiac death (n=4) below the age of 50 years.
This study provides insights into changing patterns of mortality in a large contemporary ACHD cohort. The overall mortality rate of 0.4 deaths/100 py is significantly less than that found by Engelings et al in a cohort of predominantly moderate–complex CHD patients (1.7 deaths/100 py).6 Our death rate was comparable with Diller et al (0.7 deaths/100 py) in an ACHD population with a similar distribution of complexity.6 7 The median age of death in our simple group was 70 years, only 9 years less than the Australian population average age at death of 79 years12. Median age of death in our moderate and complex CHD groups (51 and 34 years), in stark contrast, remains well below predicted population expectations.
In the context of an ACHD cohort with nearly half of our subjects having simple lesions, non-cardiac causes of death outweighed cardiac causes (54% vs 46%). This finding has not previously been reported in other ACHD mortality studies. Cohorts with predominantly moderate–severe CHD have reproducibly reported cardiac causes of death, in particular heart failure, as the major aetiology.4 6 7 13 Diller et al, in their study of nearly 7000 patients with ACHD, included a similar distribution of CHD complexity to our cohort.7 Although they observed a rising proportion of non-cardiac death, heart failure deaths remained substantially more common in our cohort (47% vs 17%). We observed a significant increase in non-cardiac death and malignancy in particular; our malignancy rates were up to three times that reported in prior studies. A similar distribution of types of malignancies causing death was seen, as compared with the general population.6 7 14
Reassuringly perioperative morality was low in our study, at 5%, comparing favourably with previously reported modern-day cohorts (7%–16%).6 15 This likely reflects improved operative techniques, better patient selection and specialist cardiac anaesthesia. Over 50% of our perioperative deaths were due to high-risk redo operations or extraction of device leads, also recognised to be high risk.
In the simple CHD group, mortality below the age of 50 years was uncommon, occurring only in 20 patients (14% of that group). Endocarditis was the single most common cause, and related to a BAV in 5 out of 7 endocarditis deaths. Beyond 50 years, non-cardiac causes of death were nearly double of cardiac deaths (84 vs 43), consistent with the observations of others.2 7 Malignancy (23%), vascular (8%) and pneumonia (7%) were the three most common non-cardiac causes identified. With improved survival relative to CHD and a longer lifespan, traditional risk factors for CAD accrue, and this was evident with 7 AMI and 11 IHD deaths. Heart failure (n=19, 15%), however, still remained the most common cardiac cause of death.
Survival in the moderate–complex CHD group unfortunately still remains significantly reduced. In keeping with prior literature, cardiac causes predominated, with heart failure most commonly identified, followed by SCD.6 7 13 This was particularly evident in those patients who died under the age of 50 years. In those who survived beyond 50 years, presumably with better ‘haemodynamics’ of their complex lesions, non-cardiac causes of death competed, with rising numbers of malignancies and vascular events.
Our Fontan patients died most commonly from heart failure, although it is important to note that the majority of these were atriopulmonary connections. With changing surgical practices and the adoption of extracardiac conduits as current first-line approach, late cardiac complications will hopefully reduce.16 Eisenmenger patients, in keeping with prior reports, were at high risk of SCD.4 6 7 SCD and heart failure remained the most common causes of death in repaired TGA (mostly atrial switch), but again reflects an older surgical approach of atrial switch procedures; the later use of arterial switch operations may substantially reduce cardiac death in this group.17 Interestingly, SCD and heart failure remained the major causes of death in our TOF cohort. Raissadati et al identified no SCD in their TOF operated between 2000 and 2009, but this differed in timing and technique of operative approach from our older TOF cohort, who were mostly repaired in the 1980s–1990s.17
The reduced survival and predominance of heart failure and SCD in the moderate–complex CHD group highlights the persisting difficulties in preventing cardiac complications in these patients. Treatment guidelines are based on large studies in ‘structurally normal hearts’ and extrapolation to complex CHD is largely via small series or expert consensus.18 19 Predicting SCD and choosing the right time for prophylactic intervention remains particularly difficult, as traditional risk assessment based on ventricular function alone substantially underestimates risk.20 While specific lesions such as TOF have been well studied, predictive models in other lesions are lacking.21
As an ‘expert referral’ centre for ACHD, there is a referral bias for moderate–complex CHD. We excluded small ASD and additionally simple CHD such as BAVs may have been under detected or managed in peripheral hospitals given the need for less specialised care. Thus, our cohort study is lower than the expected size. Our incidence calculations rely on detecting the majority of ACHD in our region, and we acknowledge that both selection bias and a certain degree of ‘attrition’ as paediatric patients move to an adult centre are inevitable. As the only major ACHD referral centre state-wide, however, these effects are likely minimised. Furthermore, the NDIS ensures near-complete follow-up for vital status and mortality in Australia. In some patients, detailed medical records and death certificates were not available and classification of cause of death relied on the accuracy of correct coding by the NDIS. We acknowledge that this could lead to a degree of potential misclassification; however, every effort was made to validate coding via other medical records, where possible. We also acknowledge that because of relatively small numbers of deaths in each diagnostic group, certain diagnoses contain somewhat heterogeneous specific underlying conditions. For example, the ‘Fontan’ group contains atriopulmonary connection and total cavopulmonary connection patients, and left ventricular-type and right ventricular-type ‘single ventricle’ hearts. Finally, limitations of a single-centre retrospective study apply.
Mortality in moderate to complex ACHD remains substantially elevated above that of the general population, with cardiac death predominating in younger life and non-cardiac causes becoming more common in those over the age of 50 years. Patients with simple ACHD live significantly longer and are more likely to die of non-cardiac causes, such as malignancy and infection.
What is already known about this subject?
Mortality in adult congenital heart disease remains substantially higher than in the general population, particularly in more complex forms of disease.
As childhood surgical interventions improve, however, mortality patterns are changing, with a rising proportion of non-cardiac causes of death.
What might this study add?
This study for the first time reports that non-cardiac causes of death were more common than cardiac causes in an adult congenital heart disease cohort.
Patients with simple defects had mildly reduced survival compared with the population average, and were most likely to die from malignancy.
Survival in complex disease was significantly reduced, with heart failure and sudden cardiac death predominating below the age of 50 years.
Over the age of 50 years, cardiac and non-cardiac causes of death were equally common.
How might this impact on clinical practice?
This study highlights target areas for improving mortality rates in a contemporary adult congenital heart disease population.
Adult congenital clinicians should be aware of rising non-cardiac causes of death, such as malignancy and infection, in both simple defects and in patients with more complex disease who survive into mid-life.
Contributors CY was responsible for study design, data collection, statistical analysis and drafting of the manuscript. BM was responsible for drafting and editing of the manuscript, and assisted with analysis. IK contributed to statistical analysis and generation of figures. RLC contributed to study design, data collection, supervision and editing of the manuscript. DSC was responsible for study design, implementation and editing of the manuscript. All authors gave final approval for the manuscript and accept accountability for its contents.
Funding This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent Not required.
Ethics approval The study protocol was approved by the Institutional Review Board (reference: HREC/11/RPAH/626) of Royal Prince Alfred Hospital, Sydney Local Health District.
Provenance and peer review Not commissioned; externally peer reviewed.
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.