LAUSR

To the Editor, With great interest we read the recent article by Giugliano and colleagues using only the surgical Apgar score (SAS) to predict the perioperative morbidity, mortality, and length of hospital stay in patients undergoing esophagectomy. By the Cochran-Armitage trend test and competing risks proportional hazards regression, they show that the SAS is a significant predictor of postoperative complications, and a lower SAS is associated with a prolonged length of hospital stay and a greater postoperative mortality. Other than the limitations described by the authors in the discussion, however, we note that several issues of this study are not well addressed. First, the authors only evaluated associations of the SAS with postoperative adverse outcomes, but did not determine the performances of the SAS in predicting postoperative adverseoutcomes. In our opinion, to determine prediction accuracy of the SAS for postoperative morbidity andmortality, providing only its hazard ratios bymultivariable analysis is not sufficient. We believe that the authors should have further performed the sensitivity analysis and constructed the receiver operating characteristic curve to obtain the sensitivity, specificity, positive, and negative predictive values of the SAS for postoperative adverse outcomes in the validation anddevelopment sets, as performed in previous studies. By providing the predicted probabilities and observed frequencies for postoperative adverse outcomes basedon the SAS, the readers can estimatewhether there is a goodoverall agreement between predicted probabilities and observed frequencies in the development and the validation sets. Furthermore, the area under the receiver operating characteristic curve can indicate discrimination ability of the SAS in predicting postoperative adverse outcomes. In addition, calibration assessment should also be carried out by the Hosmer-Lemeshow goodness-of-fit test. Second, there are many risk-prediction models with different discriminatorypowers foradverseoutcomesafter esophagectomy, such as the American Society of Anesthesiologists (ASA) physical status classification, modified frailty index, prognostic nutritional index, Portsmouth-modified Physiological and Operative Severity Score for the enUmeration ofMortality andmorbidity (P-POSSUM), the adjusted version for Oesophagogastric surgery (O-POSSUM), Charlson and the Age adjusted Charlson score, etc. A limitation of this study design is that there is no comparison for predictive values of the SAS for postoperative adverse outcomes with those of any established risk score or model. Thus, an important question that remains unanswered by this study is whether predictive performance of the SAS to facilitate early identification of adverse outcomes after esophagectomy equals or surpasses these established risk scores or models. Third, other than the surgical burden, health status and comorbidities of patients are also important determinants for postoperative adverse outcomes. The recent evidence indicates that compared to the use of the SAS alone, the risk-prediction model built by combining patient characteristics and ASA score has a higher predictive ability for postoperative mortality and adverse outcomes. Thus, we argue that more prospective studies are needed to determine whether improved outcomes can be achieved with the use of the SAS alone for risk-stratified triage and postoperative care modification of patients undergoing esophagectomy.