Author + information
- Received October 14, 2016
- Revision received February 19, 2017
- Accepted March 20, 2017
- Published online May 22, 2017.
- Hanna Carr, BSa,∗ (, )
- Sven Cnattingius, MD, PhDa,
- Fredrik Granath, PhDa,
- Jonas F. Ludvigsson, MD, PhDb,c and
- Anna-Karin Edstedt Bonamy, MD, PhDa,d
- aClinical Epidemiology Unit, Department of Medicine Solna, Karolinska Institutet, Stockholm, Sweden
- bDepartment of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
- cÖrebro University Hospital, Örebro, Sweden
- dDepartment of Women’s and Children’s Health, Karolinska Institutet, Stockholm, Sweden
- ↵∗Address for correspondence:
Dr. Hanna Carr, Clinical Epidemiology Unit, T2, Karolinska University Hospital, Solna, 171 76 Stockholm, Sweden.
Background In small clinical studies, preterm birth was associated with altered cardiac structure and increased cardiovascular mortality in the young.
Objectives The goal of this study was to determine the association between preterm birth and risk of incident heart failure (HF) in children and young adults.
Methods This register-based cohort study included 2,665,542 individuals born in Sweden from 1987 to 2012 who were followed up from 1 year of age to December 31, 2013. The main study outcome was diagnosis of HF in the National Patient Register or the Cause of Death Register. The association between preterm birth and risk of incident HF was analyzed by using a Poisson regression model. Estimates were adjusted for maternal and pregnancy characteristics, socioeconomic status, and maternal and paternal cardiovascular disease.
Results During 34.8 million person-years of follow-up (median 13.1 years), there were 501 cases of HF. After exclusion of 52,512 individuals with malformations (n = 196 cases), 305 cases of HF remained (0.88 per 100,000 person-years). Gestational age was inversely associated with the risk of HF. Compared with individuals born at term (≥37 weeks’ gestation), adjusted incidence relative risks for HF were 17.0 (95% confidence interval [CI]: 7.96 to 36.3) after extremely preterm birth (<28 weeks) and 3.58 (95% CI: 1.57 to 8.14) after very preterm birth (28 to 31 weeks). There was no risk increase after moderately preterm birth (32 to 36 weeks) (relative risk: 1.36; 95% CI: 0.87 to 2.13).
Conclusions There was a strong association between preterm birth before 32 weeks of gestation and HF in childhood and young adulthood. Although the absolute risk of HF is low in young age, our findings indicate that preterm birth may be a previously unknown risk factor for HF.
Between 5% and 13% of all live births occur before term (<37 weeks of gestation) (1,2). Although prematurity is still the main cause of neonatal death globally, high-income countries have experienced dramatic increases in survival rates in preterm infants over the past few decades (2,3). Knowledge about how the burdens of prematurity may be carried into later life in these steadily growing generations of new survivors is important for improving neonatal care, for meeting the future medical needs of these subjects, and for developing appropriate preventive measures.
Results from previous studies suggest that survivors of preterm birth are at increased risk of hypertension, stroke, and cardiovascular mortality but not ischemic heart disease (4–8). To the best of our knowledge, the association between preterm birth and risk of heart failure (HF) has not previously been explored. HF in children and young adults is an unusual but dangerous condition with high mortality rates (9,10). Congenital heart disease and cardiomyopathies, particularly idiopathic dilated cardiomyopathy, are the main causes of HF at young age (11–13). Incidence data for pediatric HF are scarce but were estimated to be 0.87 per 100,000 person-years in a study in the United Kingdom and Ireland of HF caused by cardiac muscle disease (14). Between 1987 and 2006, the incidence of HF among young adults in Sweden increased by 50%, and the proportion of cardiomyopathies as an underlying cause of HF increased from 15% to 25% (15).
Preterm birth entails exposure of the immature infant heart to extrauterine conditions. Evidence from animal models and small studies of preterm infants shows that preterm birth interferes with normal cardiac development in the neonatal period (16–19). In a cardiac imaging study of adults born preterm, ventricular mass in adulthood was seen to increase with lower gestational age at birth. Preterm birth was also associated with further alterations in cardiac structure and function (20).
We hypothesized that preterm birth is associated with an increased risk of later HF. In a nationwide Swedish cohort study including >2.6 million live births, we investigated the association between gestational age at birth and risk of incident HF in childhood and young adulthood.
Patients and Methods
Study design and population
This registry-based cohort study included 2,665,542 individuals born in Sweden and registered in the Medical Birth Register between 1987 and 2012 (Figure 1). Individuals were followed up from 1 year of age until death, emigration, first diagnosis of HF or ischemic heart disease, or end of study (December 31, 2013), whichever came first. Start of follow-up was set to 1 year of age to avoid measuring HF as an immediate complication during neonatal care.
A unique personal identity number given to all Swedish residents allows for comprehensive cross-linking with other national registries (21). The caregivers are required by law to contribute information to these registries. The Medical Birth Register was started in 1973 and covers >98% of all births in Sweden (22). Since 1982, it is based on copies of standardized clinical record forms used in all antenatal care clinics and delivery and neonatal wards in the country, and it contains data on both mother and infant. The National Patient Register contains data on patient diagnoses and medical and surgical procedures (23). The registry covers all hospitalizations in Sweden from 1987 onward, and information on hospital-based outpatient care is included from 2001. The Cause of Death Register provides information on causes and dates of death in Sweden from 1961 (24). The Multi-Generation Register was created in 2000, and it includes individual index-persons born after 1932 who were alive in 1961 and links them to their parents (25). Information on educational level was collected from the Swedish Register of Education (26). Date of emigration was retrieved from the Register of the Total Population (27).
Data on the main exposure (i.e., gestational age at birth) were retrieved from the Medical Birth Register and categorized into 22 to 27 weeks (extremely preterm), 28 to 31 weeks (very preterm), 32 to 36 weeks (moderately preterm), and ≥37 weeks (term). Since the early 1990s, all pregnant women in Sweden are offered a diagnostic ultrasound scan in the early second trimester, usually between weeks 17 and 20, and >96% accept (28). When no information on ultrasound dating of pregnancy was available, the last menstrual period was used for assessing gestational age.
Data on size at birth were calculated as deviation from the estimated weight for gestational age and sex, based on the Swedish reference curve for intrauterine growth (29). Individuals were categorized as very small (<−2 SD), small (2 SD to <1 SD), appropriate (−1 SD to 1 SD), large (>1 SD to 2 SD) or very large (>2 SD). These data were also used to statistically correct for the possibility that an association between preterm birth and later HF is confounded by low birth weight for gestational age, a proxy for poor fetal growth.
The primary outcome was a diagnosis of incident HF without a previous diagnosis of ischemic heart disease in the National Patient Register or the Cause of Death Register. The International Classification of Diseases (ICD)-9th revision (ICD-9; used between 1987 and 1996) and 10 (ICD-10; introduced in 1997) were used to define HF (ICD-9 code 428 and ICD-10 code I50) and ischemic heart disease (ICD-9 codes 410 to 414 and ICD-10 codes I20 to I25).
From the Medical Birth Register, we included information on maternal factors such as age at delivery, country of birth, singleton or multiple pregnancy, diagnosis of hypertension, preeclampsia, diabetes mellitus, or gestational diabetes. Data on maternal smoking in the Medical Birth Register were divided into 2 groups according to information collected at the first antenatal visit, usually in weeks 8 to 12 of pregnancy. This information has been routinely collected since 1983, but data are sometimes missing, particularly for mothers of preterm individuals. Information on highest attained level of maternal education was retrieved from the Education Register and was categorized as ≤9 years, 10 to 12years, or ≥13 years of education.
Individuals born with malformations that could possibly correlate with risk of HF were identified by searching the Medical Birth Register and the National Patient Register for 1 of the following ICD-9 or ICD-10 diagnoses: malformations of the circulatory system (745.0 to 747.9 or Q20 to Q28), congenital malformation syndromes due to known exogenous causes not classified elsewhere (759.8 or Q86), other specified congenital malformation syndromes affecting multiple organ systems (743.0/755/756.0/756.7/756.8/757.1/758.6/759.8 or Q87), other congenital malformations not classified elsewhere (759.0 to 759.9 or Q89), or chromosomal abnormalities (758.0 to 758.9 or Q90 to Q99). Children with a diagnosis of patent ductus arteriosus (747A or Q25.0) were not excluded from our analysis because this condition is very common after preterm birth and is a possible mediator of HF.
By using the Multi-Generation Register, it was possible to trace the registered father for almost 99% of individuals in the cohort. This approach enabled a search for both maternal and paternal diagnoses of HF or ischemic heart disease or death from HF or ischemic heart disease in the National Patient Register and Cause of Death Register, using the aforementioned ICD-9 and ICD-10 codes plus the earlier ICD version 8 codes for HF (428) and ischemic heart disease (410 to 414).
The association between preterm birth and HF was assessed in a Poisson regression analysis. Relative risks (RRs) (adjusted incidence rate ratios) and corresponding 95% confidence intervals (CIs) for each gestational age category were modeled by using log (person-years at risk) as an offset. Covariates for the adjusted models were chosen on the basis of an association with our main outcome at a level of p ≤0.20. Birth year period (1987 to 1995, 1996 to 2003, and 2004 to 2012), attained age during follow-up (in 5-year intervals), maternal age and education, subject sex, and birth weight were included as covariates in the first adjusted model (or gestational age in analyses of the association between birth weight for gestational age and risk of HF). In the second model, we also adjusted for maternal or paternal HF or ischemic heart disease. Adjusted models 1 and 2 were applied both before and after excluding individuals with malformations. Because of the large number of missing data on maternal smoking among preterm individuals, we adjusted for maternal smoking habits in a separate model, including individuals with complete data on all covariates.
Incidence rates for HF were also calculated on the basis of gestational age at birth and attained age at time of diagnosis (Online Table 1). Using a Poisson regression model, we estimated unadjusted incidence rate ratios for the same age intervals, also presented in Online Table 1. All data were analyzed by using SAS version 9.4 software (SAS Institute, Inc., Cary, North Carolina).
Ethical permission was obtained from the Regional Ethical Vetting Board in Stockholm (Etikprövningsnämnden) with the registration number 2011/195-31/2.
Among 2,665,542 individuals included in the study, 156,879 (5.9%) were born preterm; 5.14% were moderately preterm, 0.56% very preterm, and 0.18% extremely preterm (Online Table 2). Preterm individuals were more often low birth weight for gestational age than individuals born at term. Mothers of preterm individuals were more likely to be younger (≤19 years of age) or older (≥35 years of age), to have lower levels of education, to be smokers, and of non-Nordic origin. Maternal pregnancy complications and multiple pregnancies were more common in women with preterm births.
During follow-up (beginning at 1 year of age), there were 501 cases of HF. Three of these cases were deaths caused by HF. Total time of follow-up was 34.8 million person-years, yielding an incidence of 1.4 per 100,000 person-years. After exclusion of 52,512 individuals born with malformations (as specified in the Patients and Methods section), there were 305 cases of HF in 34.2 million person-years of follow-up (incidence 0.89 per 100,000 person-years). The median individual time of follow-up for all subjects was 13.1 years (IQR: 6.1 to 20.1 years).
Cohort characteristics in relation to outcome are presented in Table 1. Individuals diagnosed with HF were more often male than the healthy population. Mothers of children later diagnosed with HF were more often smokers and had lower levels of education. Maternal and paternal HF or ischemic heart disease was also more frequent in subjects with HF.
Preterm birth and risk of HF
Numbers and incidence rates of HF in relation to gestational age at birth are presented in Table 2. Incidence rates of HF were inversely related to gestational age at birth. Preterm birth was associated with an increased risk of HF across all 3 categories of prematurity, and risks increased with decreasing gestational age.
After exclusion of individuals with major congenital malformations and adjustment for maternal characteristics, subject sex, birth period, and birth weight for gestational age, the risk of HF was 17 times higher in subjects born extremely preterm, and 3.6 times higher in subjects born very preterm, compared with subjects born at term (Table 2). Additional adjustment for parental cardiovascular disease only minimally attenuated relative risks for HF. Adjusting for maternal smoking habits did not alter the described associations (Online Table 3). There was no significant increase in risk of HF for subjects born moderately preterm in the adjusted models.
The median age at diagnosis of HF was 16.5 years (IQR: 5.2 to 19.7 years). Online Table 1 displays age-specific incidence rates for HF in relation to gestational age at birth. Incidence rates dropped after the first 5 years of life and then rose again after 16 years of age. Subjects born before 32 weeks of gestation had the highest incidence rates of HF across all age categories. The same pattern was seen after excluding subjects born with malformations.
Birth weight for gestational age and risk of HF
We also found an association between low birth weight for gestational age and increased risk of HF (Table 3). Compared with infants born with appropriate birth weight for gestational age, those born very small for gestational age (>2 SDs below the mean) had a 3-fold risk of subsequent HF in the unadjusted analysis. After excluding subjects with malformations and taking maternal characteristics and gestational age into account in the adjusted models, this association was weakened and no longer significant. There was no indication of confounding by parental cardiovascular disease.
This registry-based cohort study of >2.6 million children and young adults found that preterm birth was associated with an increased risk of incident HF, also after adjustment for birth weight for gestational age and potential confounders. The RR was inversely related to gestational age at birth (Central Illustration). Individuals born extremely preterm and very preterm faced a 17-fold and >3-fold increased risk of HF, respectively; corresponding risk was not significantly increased for those born moderately preterm. A very low birth weight for gestational age (>2 SDs below the mean) was also associated with an increased risk of HF, but this risk increase was not significant after adjustment for potential confounding factors.
The mechanisms by which preterm birth may influence subsequent risk of HF in childhood and young adulthood remain elusive. A review of existing evidence concluded that individuals born preterm have slightly higher resting systolic blood pressure in early adulthood, which may increase their risk of developing hypertension (4,5). Hypertension is, in turn, 1 of the most important risk factors for developing HF in adults, also in the absence of ischemic heart disease (30,31). In childhood, even mild, untreated elevation of blood pressure may cause organ damage such as left ventricular hypertrophy (31). Thus, it is plausible that an elevation of blood pressure may contribute to the increased risk of HF observed in preterm individuals in our study.
Ischemic heart disease is a major cause of HF in adults (32) but comprises a very small proportion of HF in children and adolescents (15,33). Thus far, there have been no data confirming a link between preterm birth and ischemic heart disease (6,34,35). In the present study, individuals with ischemic heart disease as first event were censored and no longer contributed risk time in the study. Thus, ischemic heart disease is an unlikely explanation for the observed association between preterm birth and HF. Instead, cardiomyopathies, including the diagnostically broad “idiopathic dilated,” are considered a principal cause of HF in the younger population (15,33). In individuals born preterm, such heart muscle disease could be the result of cardiac remodeling after preterm birth.
The current understanding of cardiac development is that cardiomyocytes proliferate until late gestation and switch to an adult hypertrophic growth mode shortly after birth (36–38). Animal models show that the immature cardiomyocytes of the preterm heart adapt to extrauterine conditions through structural remodeling, which may have an impact on future cardiac function (16,17). In a small echocardiographic study of preterm infants, there were signs of delay in maturation of the myocardium at 28 days of age (18). The same study found signs of left ventricular diastolic dysfunction and greater dependence on atrial contraction in preterm infants. In addition to the major circulatory transition that occurs at birth, the preterm heart is often exposed to conditions that increase cardiac workload (e.g., patent ductus arteriosus), leading to important left-to-right shunting and bronchopulmonary disease with the risk of pulmonary hypertension (39). A cardiac imaging study of preterm infants found that patent ductus arteriosus was associated with significantly increased end-diastolic volumes and increased left ventricular mass (19). However, it is unclear to what extent such changes are reversed after ductal closure.
Evidence of modulation of cardiac structure and function has also been found in older survivors of preterm birth. One study included 102 individuals born preterm who underwent cardiovascular magnetic resonance imaging at 20 to 39 years of age; they were compared with control subjects born at term (20). Ventricular mass in young adulthood increased with decreasing gestational age at birth. Higher systolic blood pressure in preterm individuals could not alone explain this finding, as the increase in ventricular mass was disproportionate relative to any elevation in blood pressure. There were no observable differences in left ventricular ejection fraction between the 2 groups, but both stroke volume and end-diastolic volume were lower in individuals born preterm. Preterm-born individuals also had reduced diastolic myocardial relaxation. Moreover, their right ventricular function was compromised, and 6% had a right ventricular ejection fraction below clinical reference values (40). We speculate that such alterations of cardiac function in preterm individuals may be a consequence of the combination of interrupted normal cardiac development and postnatal exposure to circulatory challenges, and HF could ultimately be an expression of this.
Strengths of the current study include the very large cohort and the registry-based nondifferential follow-up. Sufficient statistical power allowed us to examine the risk of HF even among extremely preterm infants, although they only comprised 0.18% of our cohort. Dividing gestational age into 4 categories also enabled us to show a strong dose–response relationship between low gestational age at birth and later HF. We were able to control for many confounding factors, including maternal and pregnancy characteristics, heritability of heart disease, and birth weight for gestational age. We had information on congenital malformations, and including or excluding children with malformations yielded essentially the same results among individuals born extremely preterm and very preterm. The National Patient Register and Cause of Death Register were used to ascertain HF. After excluding subjects born with congenital malformations, we found an incidence of 0.88 per 100,000 person-years, which is almost identical to that of a British-Irish study (0.87 per 100,000 person-years) (14). The validity of the HF diagnosis is high in Swedish registers, with a positive predictive value of 82% and even 95% for those with a primary diagnosis of HF (41).
The limitations of our study are typical for registry-based research. Children born at low gestational age are generally subject to closer medical follow-up, especially during their first years of life. Thus, we cannot rule out that surveillance bias has influenced our results. To avoid measuring HF as a direct complication in the neonatal period, all individuals diagnosed with HF before 1 year of age were excluded. We were unable to investigate if and how a hemodynamically significant patent ductus arteriosus and its treatment relate to risk of later HF because of the low number of HF cases in the lowest gestational ages and the probable underreporting of patent ductus arteriosus diagnosis in the national registries, compared with other prospective national cohorts (42). The same limitation applies to the use of antenatal corticosteroids, diagnosis of bronchopulmonary dysplasia, and duration of mechanical ventilation, which may all be factors that are potential mediators of the association between preterm birth and later risk of HF.
HF can be difficult to diagnose in a young patient, and reduced cardiac function may remain silent and unreported for a long time (13,43). If this is the case, our findings may be a late reflection of a more direct effect of prematurity. We attempted to investigate this topic further by looking closer at age at diagnosis, and we found that the excess incidence of HF in those born very or extremely preterm was higher at 1 to 5 years of age compared with later. This outcome could also be explained by those extravulnerable to disease, having already developed HF by the time they reach the age of the expected increase in incidence (i.e., depletion of susceptible effect). Another possibility is that subjects born in earlier birth years, thus contributing longer follow-up, are not representative of more recent preterm births. In summary, given the age profile of our cohort, this study captured effects of prematurity in childhood and adolescence but not effects later in adulthood.
Our results show a strong association between gestational age and risk of HF. However, the total number of cases of HF in our material is small (501 cases in >2.6 million individuals), with only 11 cases among those born extremely preterm. Any minor change in the number of cases would thus influence effect size. The great risk increase for HF we have observed in the most preterm subjects may be decreased but not easily silenced by such alterations. Also, we have excluded all cases of HF before 1 year of age (n = 753), which could have influenced the association between preterm birth and HF. Longitudinal follow-up would bring us closer to the true nature of this relationship, but as most survivors of extremely preterm birth are still young, this scenario will not be possible for some time yet. Furthermore, the outpatient section of the National Patient Register was not established until 2001, which further limited our possibility of observing patients over time or investigating underlying causes of HF in this particular cohort (23).
This study found a strong association between preterm birth and risk of incident HF in children and young adults. The increase in risk was inversely related to gestational age at birth, although absolute risks were low. Considering the rising number of individuals surviving preterm birth and the potential consequences of early onset of reduced cardiac function, the problem may grow with time. There may be a need for closer follow-up and assessment of cardiac health in survivors of extremely and very preterm birth.
COMPETENCY IN MEDICAL KNOWLEDGE: Survivors of preterm birth are at higher risk of developing hypertension and cardiovascular mortality in young adulthood than those born at term. The relationship between gestational age also applies to the risk of developing HF.
TRANSLATIONAL OUTLOOK: Further research is needed to elucidate the mechanisms compromising cardiac function after preterm birth and evaluate interventions to improve long-term cardiovascular health in these individuals.
This study was funded by the Swedish Research Council for Health, Working Life and Welfare (Dr. Bonamy, 2010-0643), Swedish Society for Medical Research (Dr. Bonamy), Stockholm County Council (Dr. Bonamy, clinical research appointment), the Swedish Heart and Lung Foundation (Dr. Bonamy, 20160578), and the Karolinska Institutet Distinguished Professor Award (Dr. Cnattingius). All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- confidence interval
- heart failure
- International Classification of Diseases
- relative risk
- Received October 14, 2016.
- Revision received February 19, 2017.
- Accepted March 20, 2017.
- 2017 American College of Cardiology Foundation
- Johansson S.,
- Iliadou A.,
- Bergvall N.,
- Tuvemo T.,
- Norman M.,
- Cnattingius S.
- Koupil I.,
- Leon D.A.,
- Lithell H.O.
- Kaufman B.,
- Lin K.,
- Patel A.,
- Naim M.,
- Shah M.,
- Shaddy R.
- Wong C.M.,
- Hawkins N.M.,
- Jhund P.S.,
- et al.
- Andrews R.E.,
- Fenton M.J.,
- Ridout D.A.,
- Burch M.
- Hirose A.,
- Khoo N.S.,
- Aziz K.,
- et al.
- Broadhouse K.M.,
- Finnemore A.E.,
- Price A.N.,
- et al.
- Lewandowski A.J.,
- Augustine D.,
- Lamata P.,
- et al.
- ↵National Board of Health and Welfare. Swedish Medical Birth Register. A summary of content and quality. 2003. Available at: http://www.socialstyrelsen.se/publikationer2003/2003-112-3. Accessed February 15, 2017.
- ↵National Board of Health and Welfare. Method report: cause of death statistics (in Swedish). Available at: http://www.socialstyrelsen.se/publikationer2010/2010-4-33. Accessed February 15, 2017.
- ↵Statistics Sweden. Evaluation of the Swedish Register of Education. Available at: http://www.scb.se/en_/Finding-statistics/Publishing-calendar/Show-detailed-information/?publobjid=3299. Accessed February 15, 2017.
- ↵Statistics Sweden. Tables on the population in Sweden 2009. Available at: http://www.scb.se/statistik/_publikationer/be0101_2009a01_br_be0110tab.pdf. Accessed February 15, 2017.
- ↵Swedish Agency for Health Technology Assessment and Assessment of Social Services. Routine ultrasound examination during pregnancy. Available at: http://www.sbu.se/en/publications/sbu-assesses/routine-ultrasound-examination-during-pregnancy/. Accessed February 15, 2017.
- Lloyd-Jones D.M.,
- Larson M.G.,
- Leip E.P.,
- et al.
- Haydock P.M.,
- Cowie M.R.
- Hsu D.T.,
- Pearson G.D.
- Zoller B.,
- Sundquist J.,
- Sundquist K.,
- Crump C.
- Kaijser M.,
- Bonamy A.K.,
- Akre O.,
- et al.
- Behrman R.E.,
- Butler A.S.
- Lewandowski A.J.,
- Bradlow W.M.,
- Augustine D.,
- et al.