Esophageal Atresia and Respiratory Morbidity

Data from the French register were used. This population-based study was designed within the framework of the French national plan for rare diseases, with the aim of prospectively collecting data from all children born with EA in France, as described in previous publications.^15–18  The register was approved by the National Informatics and Privacy Committee (Commission Nationale de l’Informatique et des Libertés) and was qualified by the National Committee of Registers. All data were deidentified, with oral and written consent by the parents or caregivers. Data were collected prospectively by specialized physicians in each tertiary care center, during the initial hospitalization in the neonatal period and at 1-year follow-up, by using questionnaires validated by a multidisciplinary national committee of experts. A clinical research assistant contributed to data collection at each center, and all questionnaire responses were double checked by 2 professionals to ensure accuracy. The register was recorded in ClinicalTrials.gov (NCT02883725).

Inclusion criteria were children born with EA from January 1, 2008, to December 31, 2016, and alive at the age of 1 year. Children for whom 1-year follow-up data were missing were excluded. The EA type was defined according to Ladd classification.⁴  A long-gap was defined as an inability to perform end-to-end anastomosis within the first 15 days after birth for anatomic reasons, thus excluding other reasons for delaying surgery, such as extreme prematurity or severe cardiac malformation. The existence of chromosomal abnormality and/or associated malformations (eg, neurologic, urogenital system, cardiovascular, anorectal, costovertebral, laryngeal cleft, or others) and whether they were part of vertebral defects, anal atresia, cardiac defects, trachea-esophageal fistula, renal anomalies, and limb abnormalities (VACTERL)¹⁹  or another syndromic association (such as coloboma, heart defects, choanal atresia, growth retardation, genital abnormalities, and ear abnormalities [CHARGE])²⁰  was noted.

During neonatal hospitalization, the following data were collected: antenatal hydramnios; term (preterm birth was defined as birth before 37 weeks of amenorrhea [WA]); birth weight (BW); small for gestational age (defined as a BW <10th percentile for term); season of birth; EA surgery details (ie, thoracotomy or thoracoscopy, colonic or gastric replacement, esophago-cervicostomy); length of invasive ventilation or noninvasive ventilation and oxygen therapy; need for tube feeding (ie, gastrostomy or nasogastric tube) and/or inhaled treatment at initial discharge; and length of hospital stay.

At 1-year follow-up, the following data were collected: history of fundoplication, cardiac surgery, or aortopexy; GERD (defined as a positive pH test result and/or esophagitis on endoscopy); complications (eg, recurrent tracheoesophageal fistula and/or a history of esophageal dilatation); and nutritional status at 1 year (eg, oral or tube feeding, a need for gastrostomy, undernutrition [defined as a weight-for-height z score ≤−2.0 SDs]). A history of readmission after the initial discharge (except for patients with an initial stay of 1 year), the number and length of each event, and the main reason for readmission (eg, respiratory or other) were noted. Readmissions were defined as hospitalizations for respiratory diseases if a respiratory cause was retained as the main cause by the physician and further classified as respiratory infections (eg, bronchiolitis or pneumoniae), a brief unexplained resolved event (BRUE) related to tracheomalacia or unexplained (excluding events linked to GERD or anastomotic stricture), or a hospitalization for airway fibroscopy. Respiratory treatment at 1 year included the following: inhaled corticosteroid (ICS) treatments, inhaled anticholinergic bronchodilator, as-needed short-acting β-agonists, or chronic noninvasive mechanical ventilation.

Categorical variables are expressed as number (percentage), and quantitative variables are expressed as median (range). Normality was assessed by using histograms and the Shapiro-Wilk test. Between-group comparisons (ie, patients included versus deceased and patients included versus those with missing data at follow-up) and univariate analyses for the 2 outcomes (ie, readmission for respiratory cause and respiratory treatment at 1 year) were performed by using χ² or Fisher’s exact tests for categorical variables and Student’s t or Mann–Whitney U tests for quantitative variables. Factors with a P value <.10 in univariate analyses for the 2 outcomes were retained in logistic regression models. Full models were simplified with a backward selection procedure by using a removal criterion of 0.05. To avoid case deletion in multivariate analyses, missing data were imputed by multiple imputations by using the regression-switching approach (chained equations; m = 10 imputations).²¹  The imputation procedure was performed under the missing-at-random assumption by using all variables, with the predictive mean-matching method for continuous variables and logistic regression (binary, ordinal, or multinomial) models for categorical variables. Rubin’s rules were used to combine the estimates derived from multiple imputed data sets.²²  All statistical tests were performed at the 2-tailed α level of .05 by using SAS software, release 9.4 (SAS Institute, Inc, Cary, NC).

Figure 1 shows a flowchart of the study population. Collectively, 1460 children were born with EA in France between 2008 and 2016. After excluding of 97 (7%) patients who were deceased at 1-year follow-up and 76 (5%) with missing data at 1 year, 1287 patients (59% male) were included in the main analysis. Their main characteristics are represented in Table 1. The children were born preterm in 38%, diagnosed as small for gestational age in 27%, and had a history of prenatal hydramnios in 48%. EA was classified as type III or IV in 89%, and the diagnosis of long-gap EA was retained in 170 (14%). At least 1 associated malformation was observed in 52%, including cardiovascular anomalies (26%) and VACTERL associations (18%). The median age at end-to-end esophageal anastomosis was 1 day (0; 368), and the median length of initial hospitalization was 28 days (6; 754). At discharge, 20% of children needed tube feeding, and 7% were prescribed an inhaled treatment.

Comparing the 97 deceased patients’ characteristics with the study population (Table 1), they had higher rates of preterm birth (P < 10⁻³), hydramnios (P = .008), and associated malformations (P < 10⁻³); circumstances of death were reported in 80 (83%) patients and classified as follows: respiratory cause in 21 (26%), cardiovascular cause in 18 (23%), neurologic cause in 15 (19%), gastrointestinal cause in 7 (9%), severe breath-holding event leading to death in 3 (4%), other causes in 16 (20%), and limitations of life-sustaining treatments in 24 (30%).

A total of 76 (5%) patients had missing follow-up data at 1 year and were excluded. Overall, their neonatal characteristics were comparable with those of the study population, except for a shorter duration of invasive mechanical ventilation (P = .03) (Table 1).

At 1-year follow-up, 36 (3%) patients had recurrence of tracheoesophageal fistula, 28 (2%) had undergone aortopexy, 141 (11%) had undergone antireflux surgery, and 454 (36%) had undergone esophageal dilatation (Table 2). A total of 95 (12%) had GERD despite proton pump inhibitor treatment at initial discharge among 1129 of 1235 (91%), 176 (16%) were diagnosed with undernutrition, 148 (16%) needed tube feeding, and 287 (23%) had a history of gastrostomy tube in the first year.

Collectively, 776 (61%) patients were readmitted after initial discharge in the first year, 399 (31%) for a respiratory cause, corresponding to 51% of all readmission causes. The main causes of respiratory readmissions were as follows: infections in 249 (64%), BRUE in 48 (12% [among whom 6 (12%) underwent aortopexy]), and other causes in 91 (23%; Fig 2). Respiratory treatments were ongoing in 389 (35%) patients, maintenance inhaled treatment in 91%, ICS in 91%, inhaled anticholinergic bronchodilators in 9%, and chronic noninvasive ventilation in 7 (3%).

A total of 140 (11%) patients had a history of >2 readmissions for respiratory causes, including respiratory infections in 103 (59%), airway fibroscopy in 27 (13%), BRUE in 25 (12%), and other causes in 20 (11%). Ninety-four (73%) of these patients had ongoing respiratory treatments at 1 year. The other characteristics of this population are described in Table 2.

Factors significantly associated with readmission for respiratory causes in bivariate analyses are represented in Table 3. Notably, there was no influence of season of birth, small for gestational age, type of EA, long-gap EA, or associated malformation, except for laryngeal cleft. In the multivariate logistic regression model, factors significantly associated with readmission for respiratory causes were recurrence of tracheoesophageal fistula in the first year, aortopexy, antireflux surgery, and tube feeding at 1 year. Respiratory readmissions were also associated with inhaled treatment at 1 year. Although GERD was associated with readmission for respiratory causes in univariate analysis, it was not retained in the multivariate analysis.

Factors significantly associated with respiratory treatments at 1 year in bivariate analyses are represented in Table 4. There was no influence of the type of EA or long-gap EA. In the multivariate logistic regression model, factors associated with respiratory treatment at 1 year were male sex and laryngeal cleft.

Using data from the French national register, we report herein the largest series of children with EA, with the aim of assessing respiratory morbidity at 1 year. Our study clearly shows that respiratory morbidity is high during the first year of life, with respiratory causes representing >50% of readmissions, and one-third of the EA population receiving a respiratory treatment. In the logistic regression model, factors associated with readmission for respiratory causes were the need for antireflux surgery, tube feeding at 1 year, a recurrence of tracheoesophageal fistula, and severe tracheomalacia requiring aortopexy; the factors associated with respiratory treatment at 1 year were male sex and laryngeal cleft.

The main criteria chosen to assess respiratory morbidity was history of readmission for respiratory cause, and, therefore, severe events. These analyses confirm our preliminary results from the same register, with assessment of 1-year outcomes in 301 children born between 2008 and 2009, among whom the readmission rate was 59%, with 52% linked to a digestive cause and 48% linked to a respiratory cause.¹⁸  In other studies, readmission rates after EA repair in the first 2 years have been 55%, with 75% for a digestive cause¹⁴  and 44% for respiratory problems in the first year.¹²  We also confirm that respiratory infections are the leading cause of respiratory readmissions in this population, which affected 64% of respiratory readmissions in our study. Although most previous studies in children with EA have been retrospective, they have shown an estimated 27% to 43% risk of pneumonia in the first 2 to 5 years of age,^{7,12,13,23,24}  78% risk of recurrent episodes of bronchitis,²⁵  46% risk of exacerbation,²⁴  and 36% risk of wheezing or asthma symptoms in the first 3 years of life.²⁴  In a prospective monocenter study, Gischler et al²⁶  described that 74% of children with EA had >5 lower respiratory tract infections before the age of 5 years. In young infants, imprecise clinical diagnosis criteria, inconstant radiologic patterns, and low rates of microbiologic identification may lead to difficulty distinguishing between pneumonia, bronchitis and/or bronchiolitis, and wheezing and/or asthmalike symptoms, all of which are favored in EA because of the narrow airway, upper airway hypersecretion and clearance abnormalities, tracheal diverticulum, and aspiration.^8,27,28  Although infections were the most common cause of readmission in our population, we did not observe the stratified risk of hospitalization for bronchiolitis by birth season that was previously described in the general pediatric population.^29,30  The second leading cause of respiratory readmissions was a history of BRUE, which has also been described as a leading cause of mortality in the first months after EA repair.³¹  These episodes are favored by laryngotracheomalacia, preterm birth, small airway, and nutritional issues.³² 

The significant population size herein allowed us to assess factors associated with respiratory morbidity. We observed an association between respiratory morbidity and antireflux surgery but not between respiratory morbidity and either GERD or esophageal stricture and need for dilatation. Previous studies have highlighted a higher risk in GERD^7,12,24  and esophageal stricture cases,⁷  favoring chronic aspiration and esophageal motor abnormalities, which were not confirmed herein. Although our observational study does not allow causal deduction about the relations between these factors, it could be hypothesized either that frequent respiratory symptoms associated with GERD favor the indications for antireflux surgery, implying a causal relationship between GERD and respiratory manifestations, or that antireflux surgery itself (via aggravated esophageal dysmotility) could be responsible for frequent aspirations and, therefore, increased respiratory morbidity.³³  Suspected respiratory manifestations of GERD, such as asthmalike symptoms and chronic cough, are commonly encountered in pediatrics, without evidence of a causal relation.³⁴  Tube feeding at 1 year was also retained as a main factor associated with respiratory morbidity herein. This was not due to undernutrition, suggesting that feeding disorders leading to both chronic aspiration and dependency on tube feeding may explain this association. Feeding disorders are common in this population; although often related to delayed oral feeding, preterm birth, and esophageal stenosis or dysphagia, they can also be of anatomic origin.³⁵  We did not have access to precise information about swallowing disorders, which are known to be associated with chronic aspiration.³⁶  Fistula recurrence and laryngeal cleft were also associated with readmission for respiratory cause in the first year, further supporting the hypothesis that chronic aspiration is involved in children with EA who have respiratory morbidity. Although recurrence of tracheoesophageal fistula is an infrequent complication after esophageal repair, it is a classic cause of respiratory morbidity in the first months after EA repair.^33,37  The typical warning signs include cough precipitated by feeding, gaseous distention of the gastrointestinal tract, and pneumonia.³⁸  These nonspecific, repeated manifestations can lead to hospitalizations and delayed diagnosis. Another factor associated with respiratory morbidity in the multivariate logistic regression model was aortopexy. Previous studies have highlighted the importance of preoperative airway fibroscopy to assess tracheomalacia and diagnose laryngeal malformations in patients with EA^5,6,39  because these malformations increase the risk of dysphagia and direct aspiration leading to respiratory infections. Because aortopexy is required for severe and complicated tracheomalacia,⁴⁰  its association with respiratory morbidity is unsurprising.

It was expected that respiratory readmissions would be associated with more frequent inhaled treatment, although their therapeutic effects remain unproven in EA.³³  Interestingly, we found that the presence of cardiovascular malformations, a previously identified risk factor for respiratory morbidity in children with EA, was significantly associated with respiratory treatment.^33,41  Our findings regarding the male sex are novel. Although previous studies have shown that EA is more frequent in male patients,⁴²  male sex has not been associated with an increased morbidity in patients with EA.⁴³  In the general population, young boys are known to be more at risk for bronchiolitis and wheezing episodes in the first years of life.⁴⁴  Herein, EA type was not retained among the main factors associated with respiratory morbidity in the first year. Although previous findings have shown that patients with long-gap EA develop major digestive complications in the first year,⁴⁵  we did not observe increased respiratory morbidity in this group. However, the limited size of this subpopulation may have limited statistical power. Previous studies have highlighted that preterm birth,⁴¹  smaller BW, and hydramnios¹²  impact respiratory outcomes in the first year of life. Our study did not confirm these early reports.

Our study’s main limitation was missing data, especially regarding bronchoscopy findings, causes of respiratory readmissions, and precise indications for inhaled treatment. Imprecise diagnostic criteria and additional chronic disorders, such as tracheomalacia, made it difficult to distinguish between pneumoniae, bronchitis, and asthmalike exacerbations. In current clinical practice, aspiration rates are not accurately estimated. Furthermore, because of its descriptive and midterm follow-up, our study provides information on respiratory morbidity using clinical outcomes, whereas data from lung imaging and function in subsequent years would allow a more complete evaluation of long-term respiratory consequences. Despite these limitations, our population is comparable to previous descriptions in terms of EA types,^18,46,47  associated malformations,^2,14,47  preterm birth,^2,14,46  and mortality rates.^2,14,46  Our study is, to our knowledge, the largest prospective real-life population-based description of children born with EA, a rare disease, using standardized and centralized national data collection.

Collectively, these findings show that respiratory impairment in patients with EA is complex and multifactorial and that respiratory and digestive complications are closely linked in this disorder of embryological developmental origins.⁸  Identifying patients with EA who are at high risk for increased respiratory morbidity (ie, those presenting with a history of tube feeding, aspiration, recurrence of tracheoesophageal fistula, severe tracheomalacia, and laryngeal cleft) could lead physicians to refer to pediatric pulmonologists for close respiratory follow-up and care.²⁷  Consistent with this, the recently published care recommendations for respiratory complications among patients with EA who have tracheoesophageal fistula states that regular long-term follow-up by a multidisciplinary team with a specialized pediatric pulmonologist is imperative.³³  Upcoming data from our register that reflect assessments of chronic respiratory symptoms and lung function at the age of 6 years will provide additional insights into long-term respiratory evolutions. Additional evidence from large cohorts, national and international collaborations, are urgently needed to provide consensus guidelines for the respiratory monitoring of children born with EA.

Research assistant Katialine Groff provided invaluable help in managing the register data. We also thank Dr Laurent Michaud^† for his devotion to his patients and his EA expertise. Groupama Foundation provided financial support.

BRUE

brief unexplained resolved event

BW

birth weight

EA

esophageal atresia

GERD

gastroesophageal reflux disease

ICS

inhaled corticosteroid

VACTERL

vertebral defects, anal atresia, cardiac defects, trachea-esophageal fistula, renal anomalies, and limb abnormalities

WA

weeks of amenorrhea