EA is one of the most dangerous diseases in paediatric surgery. Improvements in surgical techniques and postoperative care have improved the survival rate for EA [7]. Despite esophageal anastomosis, infants often face severe infection and even death caused by anastomotic leak. In the past, most of these patients were successfully treated conservatively with observation, drainage and the use of antibiotics. In 1990, MacKinnon was the first to propose that EA anastomotic leak was directly related to anastomotic tension [8]. Anastomotic leak is a serious complication after EA, which can lead to serious chest infection and even the possibility of refractory stenosis [9].
Possible factors associated with its onset include a poor anastomotic technique, the use of woven sutures, a two-layer anastomosis, a long-gap EA, blood transfusion, anastomosis under tension and gastresophageal reflux [10]. Factors associated with mortality included a delayed diagnosis, premature birth, low birth weight and the presence of CHD. In the Spitz classification, patients with a birth weight of more than 1.5 kg are classified as a low-risk group; however, it is still possible to be classified as a group with high mortality and poor prognosis in the Waterston classification because of severe congenital malformations [11]. In our study, the incidence of anastomotic leak was 29.7%. We found that the birth weight of the anastomotic leak group was lower than that of the non-anastomotic leak group (2.61 ± 0.58 vs. 2.91 ± 0.47, P = 0.023). Nine of seventeen infants with CHD weighed less than 2.5 kg. Low birth weight is often associated with a history of premature delivery, foetal distress and heart malformation [12]. Low birth weight may be associated with poorer peripheral circulation and lower cardiac output, which may lead to a poor blood supply at the local anastomotic site. [13].
Survival is directly related to birth weight and CHD status. Infants weighing over 1500 g with no major cardiac problems had nearly 100% survival, whereas the presence of one risk factor reduced survival to 80%, and when both risk factors were present survival was further reduced to 30–50% [14]. Folaranmi indicated that the probability of esophageal anastomosis increased significantly with increasing body weight. His results showed that birth weight was a significant variable associated with the probability of primary esophageal anastomosis (OR = 1.009, P = 0.004) [15]. A study from Turkey also reported a significantly higher incidence of anastomotic leak in very-low-birth-weight infants compared with cases in the low- and normal-birth-weight groups [16]. Our study showed that CHD did not increase the risk of esophageal anastomotic leak (P = 0.721). This is consistent with the conclusions of Japanese scholars, who believe that CHD has nothing to do with anastomotic leak or stenosis [17]. Another study found no evidence that thoracoscopic repair of esophageal atresia impaired outcomes in children with congenital heart disease [18]. More interestingly, survival after treatment for EA was not influenced by the presence of, or the accuracy of, the diagnosis of CHD in this series. With only a few exceptions, associated CHD should not change the strategies of EA repair [19].
In clinical practice, we found that infants with anastomotic leak had lower perioperative albumin. Low albumin is an independent risk factor for postoperative anastomotic leak of the oesophagus [20]. A low patient albumin level preoperatively may result in postoperative anastomotic oedema, which also increases anastomotic tension. Preoperative albumin levels in the anastomotic leak group were noticeably lower than those in the non-anastomotic leak group. However, in multiple factor analysis, albumin levels were not included in the regression equation, which may be related to the fact that the number of cases with low albumin levels (albumin < 28 g/L) was lower. The sample size should be further expanded to find a more appropriate cut-off value to evaluate the relationship between albumin level and the incidence of anastomotic leak.
The treatment of long-gap EA is still difficult [21]. There is an increased risk of anastomotic leakage in long-gap EA. It usually requires extensive mobilization of the esophageal stump, which may impair the vascular supply to the oesophagus and consequently impair the healing ability of the anastomotic site [22]. Surgeons should carefully anastomose under low tension to prevent anastomotic complications during the initial repair of EA/TEF [23]. In this study, the average defect length was 1.52 ± 0.95 cm and 1.98 ± 1.08 cm in the anastomotic leak group and 1.33 ± 0.82 cm in the non-anastomotic leak group, and the difference was statistically significant. Serious anastomotic leak was found in 6 of the 8 patients with long-gap EA. A long gap is an independent risk factor for anastomotic leak. We believe that delayed anastomosis of a long segment defect may be a better choice, especially for infants with a gap length > 3 cm (intraoperative or preoperative), and the timing of anastomosis should be carefully determined.
Using the receiver operating curve, the AUC of this indicator was found to be 0.732, which had a good predictive value. We believe the index can be used as a good predictor of anastomotic leak during the perioperative period. For infants with a preoperative evaluation of the ratio of loss length to birth weight greater than 0.7, esophageal extension and gastrostomy should be considered. Anastomosis can be attempted after the natural extension of the oesophagus within the chest reaches a gap length of less than 3 cm, therefore, delayed repair may be a better choice [24].
As with any retrospective study, there is bias associated with data collection; this study was limited to one institution, and other institutions may have produced different results. A prospective study with a large group of patients and long-term follow-up is necessary.