During the last decade, several studies have identified a significant decrease in perioperative mortality among children undergoing surgery for congenital cardiac malformations.Reference Grech and Elliott1–Reference Frid, Bjorkhem, Jonzon, Sunnegardh, Anneren and Lundell5 In many surgical centres, data is available only for the first 30 postoperative days, and this data is commonly used to document the quality of the care.Reference Stark, Gallivan and Lovegrove4, Reference Moller, Powell, Joransen and Borbas6–Reference Stark, Gallivan and Davis11 Mortality after the first 30 postoperative days, nonetheless, is still significant,Reference Eskedal, Hagemo, Eskild, Aamodt, Seiler and Thaulow3, Reference Gibbs, Monro, Cunningham and Rickards12–Reference Morris and Menashe14 and is related to other factors than those responsible for early death. Previous studies have focused on the cause of death in adults with congenital cardiac malformations,Reference Oechslin, Harrison, Connelly, Webb and Siu15 sudden death,Reference Silka, Hardy, Menashe and Morris16 and on cause of death in patients with specific cardiac conditions.Reference Nollert, Fischlein, Bouterwek, Bohmer, Klinner and Reichart17–Reference Kopf and Mello21
The aim of our present study was to investigate the seasonal distribution of early and late death, and the causes of late death, in 1,753 children born in the period from 1990 through 2002 who underwent surgery for congenital cardiac malformations at Rikshospitalet, Oslo, Norway. During this period, this centre served as a tertiary centre for three-quarters of all children in Norway with congenital cardiac malformations. Detailed clinical information is presented for children who survived the first 30 days after surgery, but subsequently died.
Material and methods
Design
The study is a population-based, retrospective cohort study based on a registry of congenital cardiac malformations.
Sources of data
The registry of congenital cardiac malformations at the Department of Paediatric Cardiology at Rikshospitalet, Oslo, Norway, was used to identify the children suitable for inclusion in the study. The registry contains data on diagnosis, interventions, surgery, and detailed clinical outcome extracted from hospital journals, autopsy reports, as well as information from other hospitals, out-patient clinics, and local doctors. Identification of patients is based on the unique personal identification number from the National Population Registry, which also provides information on date of death or emigration. These data were merged with data from the Norwegian Medical Birth Registry, which includes data on causes of death in children who died the first year of life.
Population
Surgeons at the Rikshospitalet performed 75% of all the cardiac surgical procedures in children in Norway during the period of study. The birth rate in the source population was approximately 45,000 per year over this period. The population identified for inclusion in the study consists of 1,753 liveborn children undergoing surgery for congenital cardiac malformations at Rikshospitalet between January 1, 1990, and September 1, 2002. Only children who underwent open or closed cardiac surgery related to palliation or anatomical correction of structural cardiac defects were included. Children who only underwent procedures as implantation of pacemakers, pericardial drainage, or extracorporeal membrane oxygenation were excluded, as well as children who were treated only with catheter-based interventions. Children with acquired heart disease, arrhythmias, cardiomyopathy, and positional anomalies were excluded from the study if they did not also undergo surgery for structural congenital cardiac defects. Also excluded were 33 children undergoing surgery abroad, and 108 premature infants who underwent surgical closure of persistently patent arterial ducts before the age of two months. Excluding re-operations within 30 days, a total of 2,246 cardiac operations were performed in the population studied during this period.
Data regarding death and emigration was 99% complete for the population on March 1, 2005. For the children who died in the period between Sept. 1, 2002 and March 1, 2005, data regarding surgery and cause of death was updated and included in the analysis.
Variables
Cardiac malformations
Classification of cardiac malformations was based on results from echocardiographic examinations, cardiac catheterisations, and in some instances from findings at surgery or autopsy. The conditions were coded both according to the classification system developed by Van MieropReference Van Mierop22, Reference Van Mierop23 and the 10th edition of the International Classification of Diseases. Based on previous epidemiologic classifications,Reference Pradat, Francannet, Harris and Robert24, Reference Harris, Francannet, Pradat and Robert25 the cardiac malformations were assigned into the following three groups:
• The functionally univentricular arrangement included children registered with hypoplastic left heart syndrome, functionally univentricular physiology, and tricuspid atresia, in whom Fontan palliation was regarded as the ultimate surgical option.
• The group of severe cardiac defects included children with the following 9 conditions: atrioventricular septal defects, concordant atrioventricular and discordant ventriculo-arterial connections, or transposition, double outlet right ventricle, tetralogy of Fallot, totally anomalous pulmonary venous connection, pulmonary atresia with intact ventricular septum, interrupted aortic arch, common arterial trunk, and Ebstein's malformation.
• The groups of less severe cardiac defects included children with other cardiac defects than those placed in the first two groups.
In addition, each child who died was given a principal diagnosis based on the morphologically and/or hemodynamically most important lesions determined by the consensus of two paediatric cardiologists.
Extra-cardiac anomalies
Significant extra-cardiac congenital malformations and syndromes recorded in the registry of congenital cardiac malformations were included in the analysis. Minor anomalies, such as syndactyly of fingers or toes, isolated vascular anomalies of the skin, congenital pigmented naevuses, and congenital dislocation of the hip are examples of anomalies not entered into the registry. In 5 children who died, 3 in the perioperative period and 2 late, extra-cardiac anomalies were recorded in the data on causes of death from Norwegian Medical Birth Registry, but not in the registry of congenital cardiac malformations. These data were included in the analysis.
The children were assigned into three groups based on the presence or absence of extra-cardiac anomalies. In the first group there were children with no extra-cardiac anomalies. Allocated to the second group were children with Down's syndrome, provided that Down's syndrome was the only extra-cardiac anomaly. Patients with extra-cardiac anomalies other than Down's syndrome formed the third group.
Seasons
According to the Norwegian Meteorological Institute, the six coldest months in Norway are November through April. This period was defined as the winter season, and the months May through October as the summer season.
Cause of death
Perioperative death was defined as death within 30 days after cardiac surgery. Late death was defined as death later than 30 days after cardiac surgery. In addition, those children who died before discharge from hospital after cardiac surgery, but more than 30 days after the operation, were included as late deaths. Based on the available data, the cause of late deaths was determined by the consensus of two paediatric cardiologists (Eskedal and Thaulow). In children who died outside hospital without having an autopsy performed, information from primary health care centres and local doctors was used to determine the cause of death. Six categories of late death were defined:
• The group with respiratory infection included children diagnosed with infections in the upper or lower respiratory tract.
• The group with non-respiratory infection included children with the diagnoses of septicaemia, gastroenteritis, and unspecified viral infection (International Classification of Diseases, 9th edition: 079.9).
Children in these first two categories included those who died in hospital who were diagnosed with infection during the final hospital admission, those with diagnosed infection at autopsy, and those with infection diagnosed outside hospital the during the last 7 days of life.
• The category of sudden death included children with unexpected instantaneous death, or death within 24 hours after the onset of acute symptoms or signs (International Classification of Diseases, 9th edition: 798, International Classification of Diseases, 10th edition: R96).
• The group of procedure related deaths included children who died as a consequence of cardiac catheterisation or non-cardiac surgical procedures.
• The group of those with other non-cardiac deaths included children with trisomy 13, Werdnig- Hoffman disease, and children who died from an external cause of injury or from malignant disease.
• The final group of cardiovascular death included all those children not allocated to one of the other five categories, which also contained children who died due to progressive myocardial failure or nonsudden death secondary to arrhythmia.
Statistical analyses
Frequencies and percentages were used to summarize the data. Cross tabulation with p values from Pearson x2 tests plus odds ratios as a measure of effect size were used. Version 13.0 of the Statistical Product and Service Solutions was used in the calculations. A p value less than 0.05 was considered statistically significant.
Ethical considerations
The study was approved by the Regional Committee for Medical Research, the Norwegian Data Inspectorate, and the Norwegian Directorate of Health and Social Services.
Results
Of the children included, 141 (8%) had a functionally univentricular arrangement, 641 (36.6%) had severe cardiac defects and 971 (55.4%) had less severe cardiac defects. Down's syndrome was registered in 209 children (11.9%), and extra-cardiac anomalies other than Down's syndrome in 197 (11.2%). Of the children with Downs's syndrome, 9 had additional extra-cardiac anomalies. Of the 1,753 children undergoing surgery, 204 (11.6%) died during follow-up, with 124 (7.1%) dying in the perioperative period, and 80 (4.5%) late deaths. Age at initial cardiac surgery was less than 1 month in 29% of the children, and less than 12 months in 70%. Mean follow-up from birth was 8.1 years, with a range from zero to 15.2 years.
Of the children with the functionally univentricular arrangement, 36 (26%) died in the perioperative period, and 14 (10%) died subsequently. Of children with severe cardiac defects, 60 (9%) died in the perioperative period, and 39 (6%) subsequently. In children with less severe cardiac defects, there were 28 perioperative (3%) deaths, and 27 (3%) late deaths. Death in the perioperative period occurred in 10 (5%) children with Down's syndrome, and late deaths in 14 (7%) children. Children with extra-cardiac anomalies other than Down's syndrome suffered 17 (7%) perioperative deaths, and 20 (10%) late deaths. Gender was evenly distributed in children who died in the perioperative period, with 63 males and 61 females.
Seasonal distribution of deaths
Of the total number of cardiac operations, excluding reoperations performed within 30 days, 1,097 (48.8%) were performed during the winter season. The month of birth was evenly distributed in the cohort, with 882 (50.4%) born in winter season.
In Figure 1, we show perioperative and late deaths occurring during the winter and summer season. There was no significant difference in seasonal distribution of perioperative deaths. There were, however, significantly more late deaths during the winter season (p< 0.001). Excluding perioperative deaths, the children remaining at risk of late death were equally distributed regarding season of birth, with 815 (50%) born in winter season.
Figure 1 Seasonal distribution of perioperative deaths and late deaths in children undergoing surgery for congenital cardiac malformations.
Causes of late death
Cause of death could be determined in 78 of the 80 children dying late. Autopsy was performed in 41 children with late deaths, with 9 children dying outside hospital not submitted to autopsy.
In Figure 2, we show the seasonal distribution of the causes of death in the 78 children dying late. The cause of death was more likely to be a respiratory infection if death occurred during winter season compared with death occurring during summer season, with an odds ratio of 17.3, and 95% confidence intervals from 2.2 to 137, giving a p value of less than 0.01. The odds ratio for the cause of death to be an infection, be it respiratory or non-respiratory, was 3.3 for children who died in the winter season compared with children who died in the summer season, with 95% confidence intervals from 1.2 to 9.0, and p equal to 0.02.
Figure 2 Seasonal distribution of causes of late death in children undergoing surgery for congenital cardiac malformations. * Cause of death unknown in one child who died in winter season and in one child who died in summer season.
There were 7 children who died suddenly in the winter, and only one in the summer season. Autopsy was performed in only 2 of these 8 children. Of 8 sudden deaths registered, 4 occurred before the age of one year. In 3 of the 8, there was a functional univentricular arrangement palliated with a central shunt, 2 had double outlet right ventricle, 1 had an atrioventricular septal defect, 1 had Ebstein's malformation, and 1 one had an atrial septal defect in the setting of Down's syndrome. Of the children who died suddenly in the winter season, 3 were recovering after a respiratory infection.
The 20 cardiovascular deaths were evenly distributed in regard to season. There were 4 children who died before discharge from hospital, 14 died in hospital after re-admission, and 2 died outside hospital, with 3 children with Down's syndrome, and 4 having extra-cardiac anomalies other than Down's syndrome. Persistent pulmonary hypertension was registered in 5 children, and arrhythmias in 4 children. There were 4 children with a functionally univentricular arrangement, 7 had severe cardiac defects, and 9 had less severe cardiac defects. The principal cardiac diagnoses for the 9 children with less severe cardiac defects were coarctation of the aorta and ventricular septal defect in 3, coarctation of the aorta in 2, congenitally corrected transposition, tetralogy of Fallot with pulmonary atresia, valvar aortic stenosis, and ventricular septal defect.
The 5 procedurally related deaths included 2 children who died during cardiac catheterisation, and 3 who died during non-cardiac surgical procedures. The 6 non-cardiac causes of death were due to trisomy 13 in 2, Werdnig Hoffmann disease, multiple non-cardiac anomalies, external accident, and one death due to malignancy. The principal cardiac diagnosis in the children who died from non-cardiac causes were atrial septal defect in 2, ventricular septal defect in 2, and one each with tetralogy of Fallot with pulmonary stenosis and with pulmonary atresia.
In Table 1, we list the clinical diagnoses and aetiological agents in children who died with an infection. Among 25 children who died of a respiratory infection, only one died during the summer season, this being diagnosed as an unspecified respiratory infection.
Table 1 Clinical diagnosis and etiologic agent in children who died related to infections.
1: Respiratory syncytial virus detected in nasopharyngeal secretions.
2: Isolated from lungs at autopsy.
3: Isolated in blood culture.
4: Detected in faeces.
There were 5 children who died in hospital subsequent to infection by the respiratory syncytial virus. All died before the age of two years, with 4 of the 5 dying before they were one year old. Of these, 3 had a severe cardiac defect, and one had a functionally univentricular arrangement. Another child had pulmonary hypertension, and 3 were cyanotic before acquiring the viral infection. None had received prophylaxis with palivizumab.
In the 14 children who died with a non-respiratory infection, 6 died in the summer season, 5 with a clinical diagnosis of septicaemia, and one with unspecified viral infection.
None of the children who died with an infection were coded with influenzal infection according to International Classification of Diseases, 9th or 10th edition.
In Table 2, we show the distribution of cardiac defects and associated non-cardiac anomalies in all children dying late, and in the children who died with a respiratory or non-respiratory infection. In children who died from a respiratory infection, one-third had Down's syndrome. We compared the 10 children with Down's syndrome who either died in the perioperative period, the 14 dying late, and the 185 who survived, finding no differences in distribution of gender, gestational age, birth weight or age at initial cardiac surgery.
Table 2 Cardiac malformations and associated extra-cardiac anomalies in all late deaths and in children who died related to infections.
AV = atrioventricular, VSD=ventricular septal defect.
In Table 3, we show gender, survival-time after last cardiac surgery, age at time of death and place of death for all children dying late, and for children who died with an associated infection. Gender and survival-time after last cardiac surgery were evenly distributed in the 3 categories of late death. The majority of all late deaths occurred before the children were two years old, and 44 (55%) children died before the age of one year. The group with respiratory infections had the highest proportion of deaths during the first two years of life, with 80% of deaths occurring before the age of 2 years. Survival after the last cardiac surgery was short, with a median survival in the range from 4 to 6 months in all 3 groups.
Table 3 Gender, survival time after surgery, age and place of death in all late deaths and in children who died related to infections.
Late deaths before discharge from hospital after cardiac surgery
There were 11 children who died in the range of 34 to 90 days after the last operation, and 9 of them died before 1 year of age, with 6 dying in the winter season. The causes of death in these 11 children were septicaemia in 2, gastroenteritis in 1, death during cardiac catheterisation for 2, sudden death in 1, Werdnig Hoffmann disease in 1, and the final 4 deaths due to progressive cardiac failure. Associated non-cardiac anomalies were found in 5 of the 11 children, with 2 having Down's syndrome. The cardiac diagnosis was a functionally univentricular arrangement in 3, atrioventricular septal defect in 3, aortic stenosis in 2, coarctation of the aorta in 2, and ventricular septal defect in the final child.
Discussion
In this study, we present data on the deaths of children undergoing surgery for cardiac malformations as based on follow-up information we hold on children born in Norway between 1990 to 2002. Late deaths after surgery accounted for two-fifths of all deaths occurring in our cohort of patients. We found that such late death was more than two times more likely to occur during the winter season than in the summer season. Excluding perioperative deaths, the children left at risk of late death showed no seasonal variation in season of birth or season of cardiac surgery that could influence the seasonal distribution of late deaths. Of the 11 children who died before discharge from hospital, but later than 30 days after surgery, 6 died during the winter season. Reclassification of these children to the group of those dying in the perioperative period had no impact on our conclusions.
There was a clear seasonal difference in causes of late death, especially respiratory infections, which were significantly more common as a cause of death during the winter period. It is difficult to be completely sure, in retrospect, about the leading cause of death. We found a striking seasonal difference, nonetheless, in the occurrence of late deaths, and this finding seems to be associated with the higher proportion of deaths caused by respiratory infections deaths during the winter season. Our data does not contain detailed information of how many children who survived respiratory infections throughout the year, and the risk of death due to a respiratory infection cannot be estimated. We also have to take account that the climate in Norway, with low temperatures in the winter months, might be a risk factor of death in itself in children with limited cardiac reserves. If so, the impact of season on mortality would be greater in countries with cold winters as compared with countries with mild climate throughout the year.
The impact of respiratory infection has been described both in adultsReference Fleming, Cross and Pannell26, Reference Meyers27 and children with cardiac disease.Reference Moler, Khan, Meliones, Custer, Palmisano and Shope28, Reference MacDonald, Hall, Suffin, Alexson, Harris and Manning29 In agreement with the present results, it has been documented in studies from national health registries that infection with both the influenza and respiratory syncytial viruses are major contributors to deaths and morbidity in children in general.Reference Fleming, Pannell and Cross30–Reference Leader and Kohlhase32 It is also known that children with cardiac malformations are at particular risk of serious infection by the respiratory syncytial virus.Reference MacDonald, Hall, Suffin, Alexson, Harris and Manning29, Reference Leader and Kohlhase32–Reference Cabalka34 In our study, only 5 of the 25 children who died from a respiratory infection had a verified infection with the respiratory syncytial virus. Since our data is obtained retrospectively, this number is probably too low, especially when taking account of the high prevalence of infection by the respiratory syncytial virus in hospitalised children under the age of one year in general.Reference Leader and Kohlhase32 There are several recent publications on the safety, efficacy and economic implication of prophylaxis with palivizumab in children with cardiac malformations.Reference Saji, Nakazawa and Harada35–Reference Feltes, Cabalka and Meissner40 Palivizumab has been used for prophylaxis in a few children with cardiac malformations in Norway since 1999, but our period of study is mainly from the time before palivizumab was introduced. Immunisation with palivizumab has, in one trial, been shown to be effective in reducing hospitalisation caused by respiratory syncytial virus in children with cardiac malformations, but mortality was not significantly reduced.Reference Feltes, Cabalka and Meissner40 Also important is that this trial enrolled only children with unoperated or partially corrected cardiac defects in a stable condition, and with no associated non-cardiac anomalies. We observe that in our cohort of children undergoing cardiac surgery, more than half of the children who died with an associated respiratory infection after the immediate postoperative period had associated non-cardiac anomalies.
It proved possible to isolate Streptococcus pneumonia in 3 children who died. In Norway, vaccination against pneumococcal disease was not routinely performed during the period of study unless there was also an accompanying immune defect. Our study did not identify any late deaths with a terminal infection due to influenza virus. This might be because of the limitations of our retrospective data.
Sudden death occurred in 8 of the 1,629 children who survived the first 30 days after cardiac surgery. This is lower than previously reported,Reference Silka, Hardy, Menashe and Morris16, Reference Pelech and Neish41 and may relate to the relatively short follow-up time in our study. An increased incidence of sudden death during the winter season has been reported in the general population,Reference Straus, Bleumink, Dieleman, van der, Stricker and Sturkenboom42–Reference Arntz, Willich, Schreiber, Bruggemann, Stern and Schultheiss44 which is in agreement with the present observations. It may be that sub-clinical infections are of importance for sudden death in patients with a low cardiovascular reserve, but our data cannot answer this suggestion.
The complexity of the cardiac malformations was higher in the group of children dying late, with two thirds having either severe cardiac defects or a functionally univentricular arrangement, as compared with 45% in the total population studied. Furthermore, there was a high prevalence of associated extra-cardiac anomalies, these being found in 43% of children dying late, compared to 23% in the total population. This contrasts with the observations in children who died in the perioperative period, where associated extra-cardiac anomalies were registered in only 22%. The available data on extra-cardiac anomalies was more extensive in children who died than in children who survived in our study. If excluding extra-cardiac anomalies exclusively reported in the Norwegian Medical Birth Registry, the prevalence of extra-cardiac anomalies in late deaths would still be 40%. Given the limitations of analysing data culled from a register, the definite prevalence of extra-cardiac anomalies is probably higher than we report.
We have previously shown that overall survival improved in children with congenital cardiac anomalies born from 1995 to 1999, as compared to children born from 1990 to 1995, but this did not include children with extra-cardiac anomalies other than Down's syndrome.Reference Eskedal, Hagemo, Eskild, Aamodt, Seiler and Thaulow45 The group of children with cardiac malformations and associated extra-cardiac anomalies other than Down's syndrome accounts for one-quarter of late deaths, in contrast to just over one-tenth in the total population, and one-sixth of the perioperative deaths. Down's syndrome was registered more often in those dying late, specifically in just under one-fifth, compared with just over one-tenth of the total population. Down's syndrome was present in onethird of the children dying late from respiratory infections. This observation is in agreement with previous reports, which have shown that respiratory infections are leading causes of death in children with cardiac malformations and Down's syndrome.Reference Yang, Rasmussen and Friedman46, Reference Mikkelsen, Poulsen and Nielsen47
Our population is made up of children born in a decade in which paediatric cardiac surgery has made great advances in improving perioperative survival, despite the fact that more children with complex cardiac defects are undergoing surgery at an ever younger age.Reference Grech and Elliott1, Reference Eskedal, Hagemo, Eskild, Aamodt, Seiler and Thaulow3 In our study, the majority of children dying late were aged less than 2 years, and over half of these children died before they reached the age of one year. Survival after the last cardiac operation was short, with a median time under 6 months.
In most studies, the judgement of the quality of care is based on observations obtained during the first 30 days after surgery.Reference Stark, Gallivan and Lovegrove4, Reference Moller, Powell, Joransen and Borbas6–Reference Stark, Gallivan and Davis11 Our study extends this data, and suggests that children undergoing surgery for complex cardiac malformations represent a group at high risk during the first two years of life, particularly when additional extra-cardiac anomalies are present. Programmes for immunisation, close follow-up after cardiac surgery, and early hospitalisation when symptoms are detected of respiratory infections, might reduce late mortality. Our retrospective data, nonetheless, does not allow definite conclusions. Only prospective studies can provide clear evidence of which measures will be effective.
Acknowledgements
We thank Sørlandet Hospital, Kristiansand, Sørlandet Kompetansefond, and Rikshospitalet-Radiumhospitalet Medical Centre for the economic support that made this study possible. We also thank the support provided in regard to classification of cause of death by Dr Christian Fredrik Lindboe, Section of Pathology, Sørlandet Hospital. This study was funded by Sørlandet Regional Hospital, Kristiansand, Sørlandet Kompetanse fond and by Rikshospitalet-Radiumhospitalet Medical Centre, Oslo.
