eTable 1. International Classification of Disease codes
eTable 2. Specific Maternal Comorbidities and Prevalence of Any Congenital Heart Defect
eTable 3. Prevalence of Specific Heart Defects for Early and Late Onset Preeclampsia, Singletons
eTable 4. Preeclampsia Onset Time and Prevalence of Congenital Heart Defects After Excluding Preexisting Hypertension/Superimposed Preeclampsia and Multiparous Deliveries
eTable 5. Association Between Preeclampsia and Congenital Heart Defects According to Gestational Age at Delivery
Auger N, Fraser WD, Healy-Profitós J, Arbour L. Association Between Preeclampsia and Congenital Heart Defects. JAMA. 2015;314(15):1588–1598. doi:10.1001/jama.2015.12505
The risk of congenital heart defects in infants of women who had preeclampsia during pregnancy is poorly understood, despite shared angiogenic pathways in both conditions.
To determine the prevalence of congenital heart defects in offspring of women with preeclampsia.
Design, Setting, and Participants
Population-level analysis of live births before discharge, 1989-2012, was conducted for the entire province of Quebec, comprising a quarter of Canada’s population. All women who delivered an infant with or without heart defects in any Quebec hospital were included (N = 1 942 072 neonates).
Preeclampsia or eclampsia with onset before or after 34 weeks of gestation.
Main Outcomes and Measures
Presence of any critical or noncritical congenital heart defect detected in infants at birth, comparing prevalence in those exposed and not exposed to preeclampsia.
The absolute prevalence of congenital heart defects was higher for infants of women with preeclampsia (16.7 per 1000 [1219/72 782]) than without it (8.6 per 1000 [16 077/1 869 290]; prevalence ratio [PR], 1.57; 95% CI, 1.48 to 1.67). Infants of preeclamptic women had no increased prevalence of critical heart defects (123.7 vs 75.6 per 100 000 [90/72 782 vs 1414/1 869 290]; PR, 1.25; 95% CI, 1.00 to 1.57; prevalence difference [PD], 23.6 per 100 000; 95% CI, −1.0 to 48.2) but did have an increased prevalence of noncritical heart defects (1538.8 vs 789.2 per 100 000 [1120/72 782 vs 14 752/1 869 290]; PR, 1.56; 95% CI, 1.47 to 1.67; PD, 521.1 per 100 000; 95% CI, 431.1 to 611.0) compared with infants of nonpreeclamptic women. Among specific defects, prevalence was greatest for septal defects. When stratified by variant of preeclampsia, infants of women with early onset (<34 weeks) preeclampsia had greater prevalence of critical heart defects (364.4 per 100 000 [20/5488]; PR, 2.78; 95% CI, 1.71-4.50; PD, 249.6 per 100 000; 95% CI, 89.7-409.6) and noncritical heart defects (7306.9 per 100 000 [401/5488]; PR, 5.55; 95% CI, 4.98-6.19; PD, 6089.2 per 100 000; 95% CI, 5350.0-6828.3), whereas infants of women with late onset (≥34 weeks) did not.
Conclusions and Relevance
In this population-based study, preeclampsia was significantly associated with noncritical heart defects in offspring, and preeclampsia before 34 weeks was associated with critical heart defects. However, the absolute risk of congenital heart defects was low.
Congenital heart defects are the most common anomalies in infants, affecting every 8 births per 1000.1 Heart defects are a major cause of infant morbidity and mortality despite significant advancements in medical care.2,3 Increased survival has resulted in higher prevalence of heart defects in adults (5.78 per 1000)4 and greater health care costs.5,6 The causes and risk factors for congenital heart defects are mostly unknown. Only 15% are attributed to genetic anomalies; a smaller percentage, to maternal infections, drugs, or environmental exposures.7 A better understanding of the origins of congenital heart defects is needed for prevention and earlier detection.7
Recent studies have sought to identify biomarkers associated with heart defects, including imbalances in proangiogenic signaling proteins, such as vascular endothelial growth factor and placental growth factor, and antiangiogenic proteins, such as soluble endoglin and fms-like tyrosine kinase 1.8,9 These same biomarkers are abnormally altered in preeclampsia, a disorder characterized by excess soluble endoglin and fms-like tyrosine kinase 1 relative to placental growth factor and vascular endothelial growth factor.10 Imbalance in these biomarkers is detectable before symptoms of preeclampsia ever manifest,7,9,11 suggesting that angiogenic mechanisms may be part of a shared pathway in preeclampsia and congenital heart defects. Studies increasingly show that the pathology of preeclampsia begins early and possibly even at the start of pregnancy, around the time of fetal heart morphogenesis.12 Despite the plausible link, evidence that preeclampsia is associated with congenital heart defects is largely absent. We sought to determine the relationship between preeclampsia and prevalence of congenital heart defects in offspring.
We conducted a population-based study using hospital discharge abstracts compiled in the Maintenance and Use of Data for the Study of Hospital Clientele database that covers the entire province of Quebec, Canada.13 Nearly a quarter of Canada’s population lives in Quebec, and 99% of women deliver in the hospital.13 We extracted discharge abstracts for women paired with infants delivered at 20 weeks of gestation or more between 1989 and 2012. We excluded stillbirths because pediatric discharge abstracts were not available and excluded chromosomal anomalies because the pathology is genetic rather than angiogenic. We obtained a waiver for ethical review from the institutional review board of the University of Montreal Hospital Centre.
In this study, congenital heart defects were defined as structural abnormalities of the heart or intrathoracic great vessels formed before birth.14 Defects related to systemic circulation, arrhythmia, or cardiomyopathy were not evaluated. We considered the following critical defects: tetralogy of Fallot; transposition of the great vessels, including double-outlet right and left ventricle; truncus arteriosus; hypoplastic left heart; common ventricle; coarctation of the aorta, including interrupted aortic arch; and other critical defects (total anomalous pulmonary venous return, Ebstein anomaly, and tricuspid and pulmonary atresia). In addition, we considered noncritical defects of the endocardial cushion, ventricular septum, atrial septum, valves, aorta, and pulmonary artery; heterotaxy; and all remaining noncritical defects. Patent ductus arteriosus may not be pathologic before term and was evaluated only in infants born at 37 weeks or more. Infants receiving a diagnosis of congenital heart defects were identified with the International Classification of Diseases, Ninth Edition (ICD-9) (1989-2005) and ICD-10 (2006-2012) over the ranges 745 to 747.4, Q20 to Q26.4, and Q26.8 to Q26.9 (eTable 1 in the Supplement). To increase ascertainment, we also identified surgical procedures to correct tetralogy of Fallot, common truncus, total anomalous pulmonary venous return, and defects of the endocardial cushion, atrial septum, ventricular septum, or valves. Procedures were documented with the Canadian Classification of Diagnostic, Therapeutic, and Surgical Procedures (1989-2005) and the Canadian Classification of Health Interventions (2006-2012).
In addition, we evaluated infants with any site-specific defect of the aorta or pulmonary artery, valves, and septum, regardless of whether defects were critical. Finally, we considered the total number of defects, including single and multiple ones (2 or more). Only defects diagnosed before discharge could be evaluated in this study.
Preeclampsia is defined as hypertension and proteinuria developing after 20 weeks of gestation in women who were previously normotensive.15 In Canada, diagnostic criteria for preeclampsia include blood pressure (>140 mm Hg systolic or 90 mm Hg diastolic) and proteinuria.15,16 Women who received a diagnosis of preeclampsia were identified through ICD codes. We retained any severity of preeclampsia, including mild (642.3, 642.4, and O13), severe (642.5, 642.6, O14, and O15), and superimposed (642.7 and O11). Superimposed preeclampsia consists of new-onset proteinuria among women with preexisting hypertension. Eclampsia, an advanced form of preeclampsia characterized by convulsions, was rare (n = 949) and included with severe preeclampsia. Mild preeclampsia was defined as insignificant proteinuria (<300 mg/24 h or <30 mg/mmol creatinine in a random urine sample) and severe preeclampsia as significant proteinuria (≥300 mg/24 hours or ≥30 mg/mmol creatinine).16 Gestational hypertension may be in the spectrum of preeclampsia17 and was included in the definition of mild preeclampsia. In addition, the ICD-10 places gestational hypertension in the category of mild preeclampsia.
We considered the possibility that congenital heart defects could be more strongly associated with some variants of preeclampsia. There is evidence that early-onset preeclampsia may have different etiology than late-onset preeclampsia.10 Therefore, in addition to severity (mild, severe, and superimposed), we identified early- and late-onset preeclampsia, defined as less than 34 weeks of gestation at admission vs greater than or equal to 34 weeks, respectively. We computed week of gestation at admission from the mother’s admission date and gestational age at delivery18 and excluded 1365 infants missing data on gestational age from this part of the analysis. As an alternate possibility, we determined whether preeclampsia was complicated by intrauterine growth restriction (ICD codes 764 and P05) because studies suggest that it may characterize complex variants of preeclampsia.17
We included covariates that were potential confounders of the relationship between preeclampsia and congenital heart defects: age (<25, 25-34, and ≥35 years), parity (unknown, 0, 1, and ≥2 previous deliveries), maternal comorbidity (preexisting hypertension or diabetes, obesity, anemia or other blood disorders, thyroid disorders, epilepsy or mood disorders, connective tissue diseases, tobacco or substance abuse, and heart disease caused by atherosclerotic, chronic rheumatic, or congenital disorders) (eTable 2 in the Supplement),7 multiple birth, socioeconomic deprivation,19 and calendar period (1989-1996, 1997-2004, and 2005-2012). Data on parity and socioeconomic deprivation were missing for 1.1% and 6.4% of women, respectively. These women were included in the analysis in separate categories. Missing data on parity or socioeconomic deprivation were not statistically related to either preeclampsia or congenital heart defects.
We calculated the absolute prevalence of congenital heart defects per 1000 infants for women with and without preeclampsia. For rare defects, we calculated the incidence per 100 000 infants. We used generalized estimating equations to calculate prevalence ratios (PRs), prevalence differences (PDs), and 95% CIs for each defect, comparing preeclampsia with no preeclampsia. We used robust error estimators to account for clustering of multiple births and multiple infants born to the same woman over time.20,21 We used log-binomial models with a Poisson distribution to achieve convergence because several heart defects were rare and this method reliably estimates associations under such settings.21 We adjusted all models for maternal age, parity, comorbidity, multiple birth, socioeconomic deprivation, and calendar period, and assessed model fit with deviance residuals.
In sensitivity analyses, we examined models in which we excluded women with preexisting hypertension or superimposed preeclampsia because we did not have information on medication use. Chronic hypertension may be associated with use of blood pressure–lowering medications that have teratogenic potential, some of which may cause heart defects.11 In addition, we restricted analyses to singleton pregnancies and to nulliparous women because multiple birth and multiparity can affect severity of preeclampsia and possibly risk of heart defects. Finally, we examined the association between preeclampsia and congenital heart defects according to gestational age at delivery (<34 vs ≥34 weeks) because preeclampsia onset time may be misclassified for a proportion of women.
We carried out analyses in SAS version 9.3 and used 2-sided tests with statistical significance set at P = .05.
There were 1 942 072 infants delivered at 20 weeks of gestation or more in this study, including 17 296 with heart defects. The sample included 50 840 multiple births but excluded 7609 stillbirths and 2797 infants with chromosomal anomalies.
The overall prevalence of heart defects was 8.9 per 1000 infants (Table 1). Prevalence was higher for infants of women with preeclampsia than without preeclampsia (16.7 vs 8.6 per 1000). Critical, noncritical, site-specific, and multiple defects were more prevalent in infants exposed to preeclampsia than not exposed (Table 2). Among critical defects, infants of preeclamptic women had higher prevalence of tetralogy of Fallot (41.2 vs 18.4 per 100 000), hypoplastic left heart (16.5 vs 12.0), and coarctation of the aorta (33.0 vs 16.8). Among noncritical defects, prevalence was higher for defects of the endocardial cushion (38.5 vs 13.4), ventricular septum (405.3 vs 279.2), atrial septum (755.7 vs 280.2), valve (92.1 vs 33.1), and pulmonary artery (208.8 vs 72.5). The prevalence of patent ductus arteriosus at term was also higher in infants of women with preeclampsia than without preeclampsia (224.9 vs 123.6). All 3 site-specific defects, including aorta or pulmonary artery, valve, and septum, were more common in infants of women with preeclampsia than without preeclampsia.
In regression models, preeclampsia was associated with several congenital heart defects on adjustment for maternal characteristics (Table 3). Relative to no preeclampsia, infants exposed to preeclampsia had higher prevalence of any heart defect (PR, 1.57; 95% CI, 1.48-1.67) and noncritical defects (PR, 1.56; 95% CI, 1.47-1.67). Prevalence was elevated for all site-specific defects (septum, valve, and aorta/pulmonary artery) and for multiple defects. Among specific defects, prevalence was greatest for tetralogy of Fallot (PR, 1.67; 95% CI, 1.12-2.50), common ventricle (PR, 2.41; 95% CI, 1.09-5.33), endocardial cushion (PR, 2.23; 95% CI, 1.44-3.45), ventricular septum (PR, 1.24; 95% CI, 1.10-1.40), atrial septum (PR, 1.91; 95% CI, 1.73-2.10), noncritical valve (PR, 1.75; 95% CI, 1.33-2.29), and noncritical pulmonary artery defects (PR, 2.00; 95% CI, 1.66-2.41), as well as patent ductus arteriosus at term (PR, 1.55; 95% CI, 1.28-1.88). Preeclampsia was not associated with transposition of the great vessels, truncus arteriosus, hypoplastic left heart, coarctation of the aorta, or other critical defects. Among critical defects, infants of women with preeclampsia had an excess prevalence of tetralogy of Fallot (PD, 16.6 per 100 000; 95% CI, 2.0-31.1), and among noncritical defects the excess was highest for defects of the atrial septum (PD, 327.5 per 100 000; 95% CI, 265.4-389.7), ventricular septum (PD, 79.1 per 100 000; 95% CI, 32.2-126.0), and pulmonary artery (PD, 96.2 per 100 000; 95% CI, 64.3-128.2).
Gestational age at onset of preeclampsia appeared to be the most important risk factor for heart defects (Table 4). Women with early-onset preeclampsia had increased risk of having infants with heart defects relative to those without preeclampsia (PR, 5.53; 95% CI, 4.98-6.15). Early-onset preeclampsia was associated with critical (PR, 2.78; 95% CI, 1.71-4.50), noncritical (PR, 5.55; 95% CI, 4.98-6.19), septal (PR, 6.21; 95% CI, 5.47-7.05), valve (PR, 3.69; 95% CI, 2.34-5.80), aorta/pulmonary artery (PR, 6.33; 95% CI, 4.87-8.23), and multiple defects (PR, 5.31; 95% CI, 4.11-6.86) relative to no preeclampsia. Infants of women with early-onset preeclampsia had an excess of both critical heart defects (PD, 249.6 per 100 000; 95% CI, 89.7-409.6) and noncritical ones (PD, 6089.2 per 100 000; 95% CI, 5350.0-6828.3). In contrast, late-onset preeclampsia was unassociated or only weakly associated with heart defects. Preeclampsia with growth restriction was associated with all defects, but not as strongly as early-onset preeclampsia.
Early-onset preeclampsia was a significant risk factor for several critical and noncritical heart defects, including common ventricle, endocardial cushion, ventricular septum, atrial septum, and noncritical valve defects compared with no preeclampsia (Table 5). Late-onset preeclampsia was weakly associated with noncritical defects of the atrial septum, but the association was much weaker than for early-onset preeclampsia. Late-onset preeclampsia was associated with tetralogy of Fallot relative to no preeclampsia, but early onset was not.
Multiple pregnancies was the only factor that influenced the association between preeclampsia and heart defects in sensitivity analyses (eTable 3 in Supplement). In particular, restricting analyses to singletons led to much stronger associations between early-onset preeclampsia and heart defects. In singleton pregnancies, early-onset preeclampsia was associated with greater prevalence of any heart defect relative to no preeclampsia (PR, 7.37; 95% CI, 6.61-8.21). The association was stronger for noncritical defects (PR, 7.43; 95% CI, 6.64-8.32) than critical ones (PR, 3.64; 95% CI, 2.17-6.10). For noncritical defects, early-onset preeclampsia in singleton pregnancies was most strongly associated with endocardial cushion (PR, 14.21; 95% CI, 6.97-28.98), atrial septum (PR, 11.99; 95% CI, 10.36-13.88), and pulmonary artery defects (PR, 14.19; 95% CI, 10.60-19.00). In contrast, analyses in which we excluded women with preexisting hypertension and superimposed preeclampsia (eTable 4 in the Supplement), restricted to nulliparous women, or used gestational age at delivery rather than preeclampsia onset time (eTable 5 in the Supplement) did not produce significantly different results.
We found an elevated prevalence of heart defects among infants of women with preeclampsia compared with no preeclampsia. Risk was elevated for defects affecting all general structures of the heart, including the aorta, pulmonary artery, valves, ventricles, and septa. Among the different variants of preeclampsia, early onset appeared to be the most important factor. Women with early-onset preeclampsia had significantly greater prevalence of infants with heart defects, both critical and noncritical, compared with those with no preeclampsia, whereas women with late onset had only marginally greater prevalence. Recent studies have linked preexisting hypertension with congenital heart defects7,11 but have not investigated preeclampsia despite basic research demonstrating that preeclampsia10 and heart defects9 share common biomarkers. Therefore, this study provides novel evidence of a relationship between preeclampsia and congenital heart defects, powered by data for a large population of pregnant women.
Lack of attention to preeclampsia and congenital heart defects in past research is understandable. Preeclampsia is by definition diagnosed only after 20 weeks of gestation, long after the major structures of the fetal heart have formed.14 Researchers have proposed that pathophysiologic changes in preeclampsia begin well before 20 weeks12 but only recently have detected imbalances in angiogenic biomarkers as early as 10 weeks in women who later developed preeclampsia.22 Overexpression of the antiangiogenic biomarkers soluble endoglin and fms-like tyrosine kinase 1 relative to angiogenic placental growth factor and vascular endothelial growth factor is thought to inhibit spiral artery remodeling at the placental interface,22,23 a process initiated at the start of pregnancy.24,25 Similar imbalances in these same biomarkers were recently observed in 31 fetuses with heart defects8 and 138 children before corrective surgery for heart defects.26
Furthermore, studies suggest that angiogenic factor imbalance is predominantly found in early- but not late-onset preeclampsia,10 supporting the notion that preeclampsia is heterogeneous, with early and late variants having different etiologies yet both manifesting with symptoms of hypertension and proteinuria.27 Heart defects were more strongly associated with early- than late-onset preeclampsia in this study, reinforcing the possibility that these variants represent different diseases. However, this hypothesis requires testing because angiogenic factor imbalance is also found in women without preeclampsia who later develop intrauterine growth restriction.10
No variant of preeclampsia yielded associations with heart defects that were as strong as early onset. The associations were much weaker for late-onset preeclampsia. Women with severe and superimposed preeclampsia had increased prevalence, but this probably reflects cases of early onset present in these categories. Furthermore, exclusion of multiple births from our data led to even stronger associations between early-onset preeclampsia and heart defects, suggesting that the pathology is unique to singleton pregnancies. This finding aligns with research suggesting that preeclampsia is a heterogeneous disorder in multiple pregnancies,15,25,28 in which elevated fms-like tyrosine kinase 1 levels are due to increased placental mass rather than pathologic overexpression25 and may not pose as great a risk for heart defects as preeclampsia in singleton pregnancies.
A related question is whether the initial stages of preeclampsia coincide with the period of fetal heart formation. Differentiation of cardiac tissues begins in the third week of gestation, and by week 8, the major structures of the heart (chambers, outflow tract, and endocardial cushions) have formed.14,29 Maturation continues after week 8, including the final stages of atrial and ventricular septation, division of the aorta and pulmonary artery, and valve development.14,29 In our study, preeclampsia, particularly early-onset disease, was associated with defects in nearly all cardiac substructures. Late-onset preeclampsia was associated with elevated, yet attenuated, prevalence of septal defects, which is plausible because this structure has a longer maturation period14 and is a potentially vulnerable to placental malfunction later in pregnancy.
There are other explanations for our findings. Women with preeclampsia have closer obstetric follow-up, with increased ultrasonographic imaging to assess fetal growth. Similarly, neonates of preeclamptic pregnancies have more morbidity and may be monitored more closely after birth. Opportunities to detect heart defects may therefore be more numerous compared with those for normotensive pregnancies. Detection would most likely be greater for noncritical defects because critical ones are more obvious and easier to capture, regardless of preeclampsia. However, it is extremely unlikely that detection bias explains all our findings; if it were responsible, we would expect more associations between late-onset preeclampsia and heart defects. Associations with late-onset preeclampsia were weak despite close medical follow-up.
Our results help advance the current understanding of the pathophysiology of preeclampsia and congenital heart defects. The relationship between them supports the notion that these disorders share common risk factors and etiology, beginning very early in pregnancy and involving a long cascade of events affecting the development of fetal heart structures throughout gestation. Prevention of both preeclampsia and heart defects may well depend on the ability to elucidate these pathways more clearly in future research. Until then, clinicians should be alert to the possibility that preeclampsia may increase the risk of heart defects in fetuses, although more research is needed in other settings to confirm our findings before modification of clinical practice.
We used data for a large population of live-born infants but did not have information on heart defects in stillbirths, miscarriages, or terminated pregnancies. Birth defect prevalence is underestimated when stillbirths and elective terminations are excluded,30,31 but mainly for nervous system and chromosomal anomalies rather than heart defects.30 We do not know whether critical heart defects disproportionately caused stillbirth in preeclamptic women, but there is no reason to suspect that pregnancy termination was more common. Underascertainment from these sources would attenuate the association between preeclampsia and heart defects. We also did not have information on heart defects diagnosed later in childhood or adulthood. Long-term follow-up studies assessing maternal history of preeclampsia and lifelong risk of heart defects may be merited.
We had limited data on risk factors for congenital heart defects, such as antihypertensive medications or noncompliance with folic acid supplementation.1,11 Sensitivity analyses in which we excluded women potentially exposed to teratogenic medications did not produce significantly different results. Smoking paradoxically decreases the risk of preeclampsia24 and is inconclusively linked with congenital heart defects1 and therefore cannot explain the findings. We adjusted for preexisting diabetes, a risk factor for both preeclampsia and heart defects. We adjusted all models for maternal comorbidity but cannot exclude the possibility of residual confounding. Other limits include possible misclassification of heart defects on hospital discharge abstracts and lack of information on reproductive history. Early-onset preeclampsia may have been misclassified as late onset if women were initially treated as outpatients for a few days,18 leading to overestimated associations between late-onset preeclampsia and heart defects. More rarely, late-onset preeclampsia may have been misclassified as early onset if women were admitted before 34 weeks but developed preeclampsia only afterward. Given the strength of our associations, however, it is unlikely that more accurate classification of onset time would change the study conclusion. Diagnostic criteria may also have changed during the study. We did not have data on ethnicity. Generalizability of the findings to other populations requires further research.
In this population-based study, preeclampsia was significantly associated with noncritical heart defects in offspring, and preeclampsia with onset before 34 weeks was associated with critical heart defects. However, the absolute risk of congenital heart defects was low.
Corresponding Author: Nathalie Auger, MD, MSc, FRCPC, Institut national de santé publique du Québec, and University of Montreal Hospital Research Centre, 190, Boulevard Crémazie Est, Montréal, Québec H2P 1E2, Canada (firstname.lastname@example.org).
Author Contributions: Dr Auger had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Auger, Fraser, Healy-Profitós.
Acquisition, analysis, or interpretation of data: Auger, Healy-Profitós, Arbour.
Drafting of the manuscript: Auger, Healy-Profitós.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Auger, Healy-Profitós.
Obtained funding: Auger, Fraser.
Administrative, technical, or material support: Auger, Fraser, Healy-Profitós.
Study supervision: Auger, Fraser.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
Funding/Support: This study was supported by the Canadian Institutes for Health Research (MOP-130452). Dr Auger was supported by a career award from the Fonds de recherche du Québec-Santé (grant 25128).
Role of the Funder/Sponsor: The funding organizations were not involved in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.