O2182: Post-Operative Esophagram: A Risk-Based Approach for Detecting Anastomotic Leak in Pediatric Esophageal Surgery

Background: Esophageal anastomotic leak (AL) is a significant complication following pediatric esophageal repair, leading to increased morbidity and prolonged hospitalization. Routine post-operative esophagrams are commonly performed to detect AL, but their necessity remains debated. This study identifies independent risk factors for AL and proposes a risk-based approach to post-operative imaging.

Methods: We conducted a single-center retrospective cohort study of all children at risk for AL who underwent a post-operative esophagram following pediatric esophageal repairs from 2018 to 2023. Demographic, clinical, and operative characteristics were evaluated. Associations with AL were assessed using multivariable mixed-effects logistic regression. Backwards elimination stepwise modeling was implemented with p >0.1 for variable removal.

Results: A total of 285 esophageal repairs were performed on 256 children (median age: 6.3 months, interquartile range [IQR]: 2.5–18.5). The most common procedures were esophageal stricture resection (24%), primary esophageal atresia repair (19%), traction-induced (Foker) long-gap esophageal atresia repair (primary 18%, rescue 8%), and esophageal replacements (14%). Overall, 62% of cases were reoperative. Post-operative esophagrams were performed at a median of 9 post-operative days (IQR: 7-12, range 2 to 45), identifying 17 leaks (5.9%), of which 8 were contained (3 symptomatic), and 9 were extravasating (all symptomatic).

Univariate analysis identified seven factors significantly associated with AL: female gender, rescue Foker procedure, Indocyanine green segmental esophageal hypoperfusion, suture line hypoperfusion, presence of a new large pneumothorax, a new moderate/large lung consolidation, and fever within 48 hours of the esophagram. Multivariable analysis identified three independent risk factors: segmental hypoperfusion (adjusted odds ratio [aOR] 10.1, 95% confidence interval [CI]: 1.9, 54.7, p=0.007), presence of a new large pneumothorax (aOR 24.5, 95% CI: 1.6, 381, p=0.022), and presence of a new moderate/large lung consolidation (aOR 11.1, 95% CI: 1.1, 110, p=0.04). The predicted probability of AL was 3%, 14%, 74%, and 90% for 0, 1, 2, and all 3 risk factors present, respectively (c-statistic = 0.842, 95% confidence interval: 0.720-0.965, p < 0.001). Notably, 14% of esophagrams resulted in contrast aspiration into the airway.

Conclusion: In our cohort, intraoperative esophageal perfusion features and post-operative radiographic findings were found to accurately predict the likelihood of AL. Presence of any of these risk factors warrants a post-operative esophagram, while absence of these factors suggested a routine esophagram may be unnecessary, sparing the child from aspiration, reducing ionizing radiation exposure, and lessening patient, parental, and provider stress. A risk-based approach to post-operative imaging warrants further validation at other centers with different AL risk profiles.