Comparison of different risk-adjustment models in assessing short-term surgical outcome after transthoracic esophagectomy in patients with esophageal cancer

Abstract

Background Different risk-prediction models have been developed, but none is generally accepted in selecting patients for esophagectomy. This study evaluated 5 most frequently used risk-prediction models, including the American Society of Anesthesiologists, Portsmouth-modified Physiological and Operative Severity Score for the enUmeration of Mortality and morbidity (P-POSSUM), and the adjusted version for Oesophagogastric surgery (O-POSSUM), Charlson and the Age adjusted Charlson score to assess postoperative mortality after transthoracic esophagectomy. Methods Data were obtained from 278 consecutive esophageal cancer patients between 1991 and 2007. Performance in predicting postoperative mortality (in-hospital and 90-day mortality) were analyzed regarding calibration (Hosmer and Lemeshow goodness-of-fit test) and discrimination (area under the receiver operator curve). Results The Hosmer and Lemeshow goodness-of-fit test was applied to each model and showed a significant outcome for only the P-POSSUM score ( P = .035). The receiver operator curve indicated discriminatory power for P-POSSUM (.766) and for O-POSSUM (.756), other models did not exceed the minimal surface of .7. Conclusions Postoperative mortality after esophagectomy was best predicted by O-POSSUM. However, it still overpredicted postoperative mortality.