RESUMO
BACKGROUND: Anastomotic stricture is the most common postoperative complication in infants undergoing repair of esophageal atresia with or without tracheoesophageal fistula (EA/TEF). Stricture indices (SIs) are used to predict infants at risk for stricture requiring dilation. We sought to determine the most accurate SI and optimal timing for predicting anastomotic dilation. MATERIALS AND METHODS: A retrospective study of infants undergoing repair of EA/TEF between 2008 and 2013 was performed. Esophagrams were used to calculate four SIs (upper pouch esophageal anastomotic stricture index [U-EASI], lower pouch esophageal anastomotic stricture index [L-EASI], lateral SI, and anterior/posterior SI). The best performing SI was identified. Logistic regression analysis was performed to determine if a first or second esophagram SI threshold was associated with dilation. A receiver operating characteristic curve measured the accuracy of the model using SIs to predict dilation. RESULTS: Of 45 EA/TEF infants included, 20 (44%) had postoperative strictures requiring dilation. As the best performing SI, logistic regression analysis showed that U-EASI as a continuous variable was predictive of dilation (P = 0.03) but was not significant at U-EASI ≤ 0.37. However, U-EASI ≤ 0.37 was associated with needing earlier dilation. On second esophagram (median, 38 days), U-EASI of ≤0.39 was significantly associated with dilation (OR: 7.8, 95% CI: 1.05-57.58, P = 0.04). The area under the receiver operating characteristic curve of the U-EASI model controlling for days to esophagram demonstrated improved predictive ability from first (AUC 0.73) to second esophagram (AUC 0.81). CONCLUSIONS: Calculation of the SI utilizing a U-EASI ≤ 0.39 on the delayed esophagram is associated with future anastomotic dilation. A multi-institutional study is necessary to confirm the predictive ability of the U-EASI.
