Maternal and Neonatal Factors Associated with Delayed Closure of Ductus Arteriosus in Term-Born Neonates: Insights from the Copenhagen Baby Heart Cohort Study

The ductus arteriosus (DA) allows blood to shunt from the pulmonary artery to the aorta in the fetal circulatory system. In most full-term neonates, closure of DA occurs within 48 h of birth [1]. However, we have shown that in 0.6% of neonates, the DA closure is delayed beyond the first week of life [2]. This is defined as a persisting ductus arteriosus (PDA). The incidence of PDA among neonates born at term is estimated to be approximately 1 in 2,000 births, and PDA constitutes up to 10 percent of all congenital heart defects (CHD) [3]. The PDA is often asymptomatic and may only be detected through echocardiography, resulting in uncertainty about the incidence [4].

After birth, the pressure decreases in the pulmonary vascular bed, and an open DA will shunt blood from the aorta back into the pulmonary circulation. This may contribute to pulmonary hyperperfusion and represent a left ventricle volume overload. The characteristics of the shunt, i.e., size and shape and the presence of comorbidities, will contribute to determining the range and severity of symptoms and complications associated with the condition, ranging from asymptomatic to severe conditions, such as pulmonary hypertension. Determining if a PDA needs intervention is done based on clinical presentation, auscultation, and echocardiographic measurements. The notion of an increased risk of endocarditis in smaller lesions is often no longer the basis for systematic intervention [3].

The etiology of PDA in healthy-term neonates remains poorly understood. Previous research has identified maternal obesity, diabetes, and female sex as potential contributors to delayed ductal closure due to their influence on fetal development and vascular function [5‒8].

Identifying neonates at risk for PDA can be challenging due to the often asymptomatic nature of the condition in its early stages. Leveraging data from the Copenhagen Baby Heart Study (CBHS) cohort, we aimed to determine factors linked to delayed ductal closure.

This retrospective cohort study utilizes data from the CBHS, a prospective cohort study conducted between April 1, 2016, and October 31, 2018. All pregnant women from three hospitals in the Capital Region of Denmark (Rigshospitalet, Hvidovre Hospital, and Herlev Hospital) were invited to participate [9]. Participants were offered a heart ultrasound based on a validated standardized neonatal echocardiographic protocol. To ensure consistent and uniform echocardiographic assessments, all sonographers received training and certification under the supervision of a pediatric cardiologist. The aim was to include all neonates born in the inclusion period. Just over 50% of the birth population was successfully included in the study. The CBHS obtained written consent from parents or guardians and has been approved by the Regional Ethics Committee of the Capital Region of Denmark (H-16001518) [10].

The CBHS database includes 28,503 newborns. We excluded neonates born preterm (gestational age <37+0 weeks) (n = 1,579), neonates with congenital heart defects (CHD) other than secundum atrial septal defects or patent foramen ovale (n = 995), participants if their echocardiographic examinations were compromised by movement, poor visualization, or inadequate imaging of the pulmonary artery (n = 3,645). To focus on the neonatal period, we excluded neonates with echocardiographic examinations performed more than 28 days after birth, and to minimize the influence of physiological open DA, we excluded cases scanned before day three (n = 2,718) [2]. Demographics are presented in Table 1.

The research team retrospectively reviewed all echocardiographic examinations. A retrograde flow from the aorta into the pulmonary artery was considered an open DA. A board-certified pediatric cardiologist confirmed all positive findings. The primary outcome was an open DA in the CBHS baseline echocardiographic examination.

We included variables based on findings provided in previous research. To evaluate the association of the outcome with neonatal factors, we included the following: Neonatal characteristics included sex, age at echocardiography, gestational age, twin birth, and proportions of neonates with abnormal length or weight at birth. We calculated gestational age-specific 10th and 90th centiles for birth weight, birth length, and placental weight in the control group. We then classified the cases as below, within, or above the defined reference range.

To evaluate the association with perinatal factors, we included the Apgar score. To evaluate the association with maternal factors, we included maternal ancestry, maternal overweight (BMI 30–34.9 kg/m²), maternal obesity (BMI ≥35 kg/m²), smoking during pregnancy, maternal diabetes (gestational and type one), maternal thyroid disease during pregnancy, size of the placenta, and late pregnancy (>40 years).

We categorized continuous variables and coded them as binary dummy variables. We generated a correlation matrix using R to assess potential correlations among the predictor variables. We did not identify any significant or concerning patterns. Using the logbin function in Rstudio utilizing a log-binomial general linear model, we calculated the association between exposures and the outcome by creating a model for each predictor. Rows with missing values were excluded separately for each model to maximize the use of available data for the relevant variables. We analyzed exposures individually and adjusted for potential confounding factors (aRR). Confounding factors include diabetes type 1, maternal BMI, maternal age, neonatal proportions at birth, and maternal smoking (presented in online supplementary Table 1; for all online suppl. material, see https://doi.org/10.1159/000543915). We calculated mean, median, and corresponding interquartile ranges (IQR) or standard deviation for demographics.

We included 19,566 neonates in the present study (shown in online suppl. Fig. 1). A total of 190 (1%) had an open DA after day three. Characteristics of the cohort are presented in Table 1. We found the following factors to be associated with delayed ductal closure: maternal obesity aRR = 1.97 (95% CI: 1.01–3.82), maternal thyroid disorders aRR = 2.02 (95% CI: 1.2–3.41), low APGAR score at minute one aRR = 2.63 (95% CI: 1.16–6), high weight at birth aRR = 1.81 (95% CI: 1.23–2.64), and length at birth aRR = 1.67 (95% CI: 1.07–2.61) (presented in Table 2). Further analysis independent of age at the examination is presented in the online supplementary Table 2.

In this large prospective cohort study, we investigated the relationship between maternal and neonatal characteristics and delayed ductal closure in term-born neonates. Among the various factors identified, a pregnancy BMI over 35, maternal hypothyroid disorder, low Apgar at 1 min, and neonatal birth proportions showed an association. These findings enhance our understanding of the factors influencing delayed DA closure in term-born infants.

A registry-based cohort study conducted in Sweden, including more than 2,000,000 infants, revealed a correlation between maternal obesity and registered cases of PDA in term-born children. They reported a two-fold higher risk of PDA in children born term to mothers with BMI over 35, which is identical to the risk we report with delayed ductal closure [11, 12]. It is hypothesized that metabolic disturbances and abnormal placental nutrient supply play vital roles. Hyperglycemia, insulin resistance, and chronic inflammation related to maternal obesity could theoretically impair fetal heart development and impact the cardiac transition to postnatal circulation [5].

We observed associations between delayed ductal closure and maternal hypothyroidism during pregnancy. A hospital-based cohort study (n = 52,047) found an association between other CHD and hypothyroidism, not including the presence of an open DA as a CHD [13]. To our knowledge, no other studies have investigated the association between delayed ductal closure and hypothyroidism.

We observed an association between larger neonatal proportions and delayed ductal closure. Notably, this association persisted even after we adjusted for correlated factors such as maternal diabetes. Studies have associated low birth weight with the development of CHD. We found no studies reporting an association with isolated high birth weight [14].

We did not observe the hypothesized relationship between diabetes during pregnancy and delayed ductal closure [5]. However, the low number of cases observed did result in imprecise risk estimates with wide confidence intervals. Future studies targeting this specific subgroup may offer valuable insights [5, 8].

This study comprehensively examines risk factors associated with delayed ductal closure in full-term newborns in a population-based cohort. We assessed all findings under the supervision of a board-certified pediatric cardiologist to minimize the potential errors from data based on echocardiograms conducted by various ultrasound technicians at multiple locations. It is essential to acknowledge that just over 50% of children born during the study period in the three centers were included, introducing the possibility of selection bias in our results. The high likelihood of observing physiologically open DA due to early scanning presents a potential bias, particularly if early scanning is associated with other variables, such as hospitalization status. We excluded cases found before day three from our primary analysis to mitigate as many physiological open DAs as possible.

In this prospective cohort study, we identified risk factors for delayed ductal closure in a neonatal echocardiography. The maternal factors included obesity and thyroid disorder. Neonatal factors included neonatal proportion, e.g., weight and length. These findings contribute to identifying at-risk groups early to improve neonatal care and outcomes.

We want to thank the Copenhagen Baby Heart study for enabling the study and providing access to the comprehensive data pool, as well as the Danish Cardiovascular Academy for funding this research project. Furthermore, we thank the cardiologic research department at Herlev Hospital for their support and guidance.

Parents or guardians provided written consent to participate in the Copenhagen Baby Heart Study and its sub-studies. The Regional Ethics Committee of the Capital Region of Denmark has approved the CBHS (H-16001518) [9].

H.B. received lecture fees from Amgen, BMS, Sanofi, and MSD. Apart from this, no authors declare any conflict of interest.

The study was supported by the Danish Cardiovascular Academy, which is funded by the Novo Nordisk Foundation, Grant No. NNF20SA0067242. The Danish Heart Foundation, the Danish Children’s Heart Foundation, Candy’s Foundation, the Toyota Foundation, the Paediatric Department Hvidovre, and the Herlev-Gentofte Hospital Research Foundation supported this work. The funders did not participate in the study’s design, data collection, analysis, or interpretation.

Anton Friis Mariager is the first and corresponding author. Alberte Hammeken contributed to data collection by participating in the review of 25,000 echocardiograms. Mikkel Malham contributed to data collection by reviewing 25,000 echocardiograms, provided essential supervision of the process, and proofread the manuscript. Anna Sellmer proofread the manuscript and provided statistical support. Raheel Altaf Raja and Johan Navne contributed to data collection by reviewing 25,000 echocardiograms and proofreading the manuscript. Henning Bundgaard and Kasper Iversen provided supervision of the Copenhagen Baby Heart Cohort, essential research supervision, and manuscript proofreading. Dorthe Lisbeth Jeppesen, primary supervisor, contributed to data collection by reviewing echocardiograms, provided supervision of the Copenhagen Baby Heart Cohort, and proofread the manuscript. All the authors have seen and approved the submitted manuscript, have contributed significantly to the work, attest to the validity and legitimacy of the data and its interpretation, and agree to its submission. The authors attest that the article is the author’s original work, has not received prior publication, and is not under consideration for publication elsewhere.

We performed all statistical analyses using R Studio version 1.4.171. This article and its online supplementary material include all data generated or analyzed during the study. For further inquiries, the corresponding author could be contacted.