Purpose: Esophagectomy with three-field lymphadenectomy (3FLD) is a potentially curative treatment option for resectable esophageal cancer (EC), which is associated with significant morbidity. Risk scores are useful for patient assessment; this study compares seven different scores and indexes to find the best model for predicting morbidity in patients undergoing 3FLD. Methods: Six years of data from January 2010 to January 2016 were reviewed, patients with EC who underwent 3FLD was retrospectively scored using 4 predictive scores and 3 predictive index models. Postoperative morbidity was assessed according to the Extended Clavien-Dindo classification. The outcomes were the presence and severity of morbidity. Validation was performed by calculating the area under the ROC curve and by the assessment of collinearity among the variables independently associated with morbidity in the overall model to determine the best predictive model. Results: Two hundred thirty patients were included in the final analysis. Complications after 3FLD occurred in 168 patients (73%) (minor complications, n=96 [41%]; major complications, n=72 [31%]). The AUC values (<0.7) indicated that all scores and indexes had poor discrimination power in predicting the presence and severity of morbidity. Overall, the Steyerberg score was associated with the lowest risk of misestimation in predicting morbidity (p=0.0330). Conclusion: No score or index could predict the presence or severity of morbidity after 3FLD with good discrimination power. Age (>68 years) was the most critical factor affecting morbidity. The Steyerberg score model, based on the addition and subtraction of risk values was the best model for predicting morbidity after 3FLD.

This content is only available as a PDF.