A total of 83 patients were managed from the time of their initial surgery at our institution between 1996 and 2019. Eighteen patients admitted between 1996 and 2000 had to be excluded as they were either lost to follow-up or had substantial missing data. Of the remaining 65 patients, there were 39 males and 26 females, giving a male–female ratio of 1.5:1. Median gestational age was 37 weeks (range 30–41), with 25 (38.5%) born prematurely prior to 37 weeks. Median birth-weight (BW) was 2.49 kg (range 1.05–3.42). The ethnicity was predominantly Chinese (n = 34, 52.3%).

According to Gross anatomic classification for the 65 patients: the predominant variant was EA with distal TEF (Type C) in 57 (87.7%), followed by isolated EA (Type A) in 5 (7.7%) (Table 1). Thirteen (20%) had long-gap EA (LGEA) of which 7 patients were of Type C and 5 were Type A variant. There was no significant difference in demographic characteristics between the LGEA and non-LGEA groups (Table 2).

EA/TEF was suspected or diagnosed prenatally in only 23 (35.4%) patients; with 100% detection of isolated esophageal atresia (Gross Type A) infants, but only 31.6% of infants with associated TEF (Gross B–E; p < 0.05). Diagnosis was based on ultrasonographic findings of an absent or small stomach bubble in association with polyhydramnios. Interestingly, the blind-ending upper esophageal pouch was not visualized in any of our patients’ prenatal ultrasounds. The most frequently encountered post-natal presentation was the classic pooling of secretions with coiling of the feeding tube in the upper esophageal pouch (n = 52, 82.5%). Majority (n = 36, 55.4%) of the patients were born in our institution; 26 (40.0%) were transferred from other local hospitals, and 3 (4.6%) were referred from overseas centers.

The incidence of associated congenital anomalies was high in our cohort, present in 37 out of 65 patients (57.0%), with 20 (30.8%) having more than one anomaly. The most common associated anomalies were cardiac defects (n = 29, 44.6%) of which 21 (32.3%) were defined as major cardiac anomalies. This was followed by ano-rectal anomalies (n = 9, 13.8%), vertebral (n = 9, 13.8%), and renal anomalies (n = 7, 10.8%).

Twenty (30.8%) patients met the criteria for VACTERL association (vertebral anomalies, anal atresia, cardiac malformations, TEF, renal, and limb malformations) [9]. There was no difference in the proportion of VACTERL association between the LGEA and non-LGEA groups. Two patients were diagnosed with CHARGE syndrome (coloboma, cardiac defects, choanal atresia, growth retardation, genital hypoplasia, and ear deformities).

Eighteen (27.7%) patients needed to be intubated pre-operatively in the Neonatal ICU. The median time to the initial surgery in our cohort was 24 h (Table 3). This excluded 1 patient with H-type TEF who had delayed presentation and was diagnosed at 19 months of age. Routine pre-operative bronchoscopy identified laryngeal cleft in 2 patients (3.1%) and tracheomalacia in 3 patients (4.6%). The decision for placement of an occluding Fogarty catheter into the distal fistula was made as per surgeon’s preference (n = 16, 25%), to optimize ventilation and identification of the fistula prior to ligation.

The median esophageal gap assessed intra-operatively was 4VB (range 1.3–8.0) or 5.9 cm (range 2.0–12.0) in the LGEA group, and 1VB (range 0.4–3.0) or 1.5 cm (range 0.6–4.5) in the non-LGEA group (p < 0.001). An extra-pleural dissection was preferred in our cohort (n = 48, 73.8%), to avoid potentially severe complications of empyema and mediastinitis associated with trans-pleural dissection [10]. All non-LGEA patients (n = 52) deemed clinically stable at surgery underwent primary anastomosis. Most of the LGEA group (n = 10/13, 76.9%) underwent staged repair with open gastrostomy creation at their initial surgery. Surgical techniques employed at eventual esophageal reconstruction were delayed primary anastomosis (DPA) in 5 patients (38.5%), gastric transposition in 2, and colonic interposition in 1 patient. Anastomotic tension was documented in 6 (46.2%) LGEA patients and 11 (21.2%) of non-LGEA patients. Of those who underwent definitive repair, surgery was performed via an open right thoracotomy in 57 (n = 87.7%). Minimally invasive surgery was attempted in 5 (7.7%) patients from year 2009, however 3 required conversion to open thoracotomy due to difficult mobilization. Our patient with H-type TEF underwent ligation via a cervical approach at 19 months of age.

The median duration of peri-operative intubation was 5 days (IQR 4–9). From the year 2006 onwards, a post-operative contrast study to document integrity of anastomosis was routinely performed in our institution, at a median of 8 days (IQR 7–9.8) after surgery. Upon verification of esophageal integrity, the median time to initiation of feeds, either orally or via naso-gastric tube, was 8 days (IQR 6–9) post-operatively. Patients in the non-LGEA group had a shorter median length of stay when compared to the LGEA group (n = 22.5 v 48 days), and were more likely to be weaned to oral feeds by the time of discharge (p < 0.02, Table 3).

Our overall post-operative surgical morbidity rate was 32.3%. The most common complication was esophageal strictures (n = 20, 30.8%) with 85% (n = 17) presenting within the first year of life. Fifteen (75%) required serial (3 or more) dilatations. The cumulative number of mechanical dilatations across our cohort was 122, with 67 balloon dilatations (54.9%) performed by interventional radiologists and 55 endoscopic dilatations (45.1%) performed by pediatric surgeons. There was no significant difference in stricture rate, or the median number of dilatations between the LGEA (4.5, IQR 2.8–8) and non-LGEA group (5.5, IQR 3.3–8). The highest number of dilatations for a single patient was 38, who was also the only patient in our cohort to receive adjunct therapy with intralesional triamcinolone and mitomycin injections for recalcitrant esophageal stricture. The rate of perforation post-dilatation (n = 3/122, 2.4%) was low; all were treated conservatively with a period of nasogastric decompression and bowel rest. None of our patients with esophageal stricture required surgical resection and re-anastomosis.

Four patients (6.2%) developed anastomotic leak confirmed on routine contrast study within 2 weeks post-surgery. All were managed conservatively: kept nil by mouth with total parenteral nutrition, nasogastric stenting and intravenous antibiotics. Anastomotic leak was significantly associated with LGEA (p < 0.05, Table 4). Recurrent TEF occurred in 3 (4.6%) patients, 2 of whom were LGEA patients and required redo ligation via open thoracotomy at 5- and 35-months old. No patients developed post-operative vocal cord paralysis or symptomatic tracheal diverticulum.

In the immediate post-operative period, 42 patients (66.7%) were started on anti-reflux medications. Majority (n = 41, 63.1%) suffered dysmotility-related symptoms, with GERD in 28 (43.1%) and dysphagia in 20 (30.8%), most of whom were diagnosed before their first year of age (n = 16/28, 82.1% and n = 16/20, 80.0% respectively). Within this sub-group of patients on anti-reflux medications, 21 remain symptomatic, requiring long-term medications (range 2–23 years). There was no significant difference in incidence of these complications between the LGEA and non-LGEA groups.

GERD in our patients was diagnosed on impedance/pH study and/or contrast imaging demonstrating reflux up to the proximal or mid-esophageal level [11]. Nine were co-managed with a gastroenterologist. Reflux was deemed severe enough to require Nissen’s fundoplication in 9 (13.8%) patients (age range 2–43 months). Two patients had radiologic evidence of persistent reflux post-fundoplication although none had wrap disruption or required re-operation.

Esophagogastroduodenoscopy (OGD) was performed in 18 patients from our cohort (28.1%), 4 of whom were diagnosed with esophagitis. One demonstrated features of eosinophilic esophagitis. None were found to have Barrett’s metaplasia or esophageal malignancy.

Tracheomalacia was diagnosed by visual assessment via bronchoscopy of reduction in the cross-sectional tracheal luminal area during the expiratory phase of quiet breathing. Severity is classified as mild (50–75% reduction), moderate (75–90% reduction), or severe (> 90% reduction) [12, 13]. Twenty-two (33.8%) patients were diagnosed with tracheomalacia (17 mild, 5 severe) in our cohort, of which only 3 were identified at initial surgery. The remaining 19 patients were diagnosed within the first year of life when they became symptomatic, commonly presenting with persistent barking cough, stridor or wheezing. Thirteen had follow-up with flexible bronchoscopy. Two patients with severe tracheomalacia necessitated tracheostomy though none required aortopexy.

Respiratory symptoms were frequently encountered in our cohort (n = 42, 64.6%), of which the most common manifestation was recurrent lower respiratory tract infections (n = 23, 35.3%), defined as 3 or more episodes requiring hospitalization in a year. This was followed by chronic cough with recurrent pneumonia (n = 19, 29.2%), and asthma/bronchial hyper-reactivity (n = 13, 20%). Pneumonia and tracheomalacia were associated with LGEA in our cohort (p < 0.05).

Eighteen (27.7%) patients were co-managed with paediatric respiratory physicians. Bronchodilator use was recorded in 15 (23.1%) patients, with 8 currently on active treatment (duration of treatment ranging 3.25–23 years). Fifteen patients (23.1%) underwent flexible bronchoscopy, 13 of whom were on follow-up of tracheomalacia. Bronchoscopy was performed in 2 children as an adjunct investigation for reflux esophagitis and symptomatic esophageal pseudodiverticulum. Eleven (16.9%) children had computed tomography (CT) thorax for follow-up assessment of airway disease. Bronchiectasis was diagnosed in 3 (4.6%) patients.

Among the surviving patients, 29 were 5 years or older at the time of the study, 21 were between 1 to 5 years old, and 8 were less than 1 year old. Median duration of follow-up for the overall cohort was 5 years (IQR 1.5–10), while 5.5 years (IQR 1.75–9.25) for those remaining on active follow-up (n = 39, 60%; Table 5). Majority (n = 53, 82.8%) required multiple readmissions for EA/TEF-related issues with a median of 3 admissions (IQR 1–6) across the cohort. Six patients (13.8%) have been discharged or transferred to overseas specialist care, while 10 (15.4%) have defaulted from all follow-up. Three of 9 adolescent patients have complied with transition to adult care. There was no significant difference in median years of follow-up and readmissions, between the LGEA and non-LGEA group.

Overall survival of our operated patients was 58/65 (89.2%). All 7 mortalities were of the Gross Type C anatomical subtype, with a median birth weight of 2460 g. Two of them acutely deteriorated and died at their initial surgery: Sudden lung re-expansion followed by severe cardiorespiratory collapse occurred in one during thoracoscopic dissection of the upper esophageal pouch; in the other, acute cardiorespiratory collapse occurred peri-bronchoscopy. One patient with LGEA, concomitant type 2 laryngeal cleft and broncho-esophageal fistula required intensive care unit care from birth for recurrent respiratory infections, and eventually demised at 14 months old. Four patients died of late respiratory failure secondary to pneumonia and GERD-related complications at a median age of 8 months (range 5–32 months).

Table 6 compares survival by Spitz classification: majority of our infants (n = 34, 52.3%) belonged to the low risk (Class I: birth weight > 1500 g with no major congenital cardiac defect) group, with 97% survival [14]. All our infants in the high-risk (Class III: BW < 1500 and major congenital cardiac defect) group survived, although this group was limited to 5 patients. Mortality in our patients was associated with presence of major congenital cardiac anomaly on univariate analysis (p = 0.028; OR 15.9; 95% CI).