Geriatrics & Gerontology International | 2019
Prediction of short‐term mortality after emergency surgery in octogenarians
Abstract
Dear Editor, We read the recent article by Bolger et al. assessing the outcomes of octogenarians after emergency surgery and the factors affecting postoperative mortality in the general hospital setting with great interest. By the multivariate analysis, they showed that American Society of Anesthesiologists (ASA) status classification and intensive care unit (ICU) utilization were the best predictors of 30-day postoperative mortality. Other than the limitations described in the Discussion section, however, we note that several methodological issues of this study seem important so as to avoid any optimistic interpretation or misinterpretation of results. First, we noted that their multivariate model for statistical adjustment only included the preoperative variables associated with postoperative mortality, but not the intraoperative risk factors and postoperative complications that can significantly affect postoperative mortality. Besides the preoperative health status and comorbidities of patients, the surgical burden and major postoperative complications are also important determinants of short-term mortality after surgery. Thus, only including preoperative factors in a multivariate model might not fully explain postoperative outcomes. For example, the Surgical Apgar Score based on the estimated blood loss, lowest heart rate and lowest mean artery blood pressure during surgery describes a combination of surgical complexity and the individual patient’s response to surgical stress, and has been proved as a simple assessment method of patients’ surgical burden. The available evidence shows that the Surgical Apgar Score is a powerful predictor of 30-day postoperative morbidity and mortality both in fit and frail older patients undergoing emergency abdominal surgery. Furthermore, common postoperative complications, such as adverse cardiovascular events, acute kidney injury and pneumonia, have been significantly associated with short-term mortality of older patients undergoing elective and emergency surgery. We argue that no inclusion of intraoperative risk factors and postoperative complications into the model would have tampered with the inferences of multivariate logistic regression analysis for risk factors of 30-day postoperative mortality and their odds ratios. It has been shown that compared with the surgical risk scores only based on preoperative risk factors, the models including intraoperative risk factors, such as the Surgical Apgar Score and Surgical Risk Score, can improve the predictive ability for postoperative complications and mortality. Second, this study only evaluated the associations of the ASA status classification and ICU utilization with 30-day postoperative mortality, but did not determine the performances of the two factors in predicting postoperative mortality. To determine predictive ability of two factors for postoperative mortality, after multivariate analysis, the authors should further construct the receiver operating characteristic curve and carry out sensitivity analysis to obtain the sensitivity, specificity, and positive and negative predictive values of the ASA status classification and ICU utilization for postoperative mortality in the validation and development sets. The area under the receiver operating characteristic curve can indicate the discrimination ability of the ASA status classification and ICU utilization in predicting postoperative mortality. Furthermore, by providing the predicted probabilities and observed frequencies for postoperative mortality based on the ASA status classification and ICU utilization, the readers can estimate whether there is a good overall agreement between predicted probabilities and observed frequencies in the development and the validation sets. Third, there have been many models or risk scores available for the prediction of short-term mortality after emergency surgery, including the Portsmouth Physiological and Operative Severity Score for the enumeration of Mortality and Morbidity (P-POSSUM) model, Biochemistry and Hematology Outcome Model (BHOM), Surgical Outcome Risk Tool (SORT), Charlson Comorbidity Index, Goldman score and Surgical Apgar Scores, National Emergency Laparotomy Audit risk model, and others. A limitation of this study design is no comparison for predictive values of the ASA status classification and ICU utilization for 30-day postoperative mortality with that of any established risk score or model. Thus, an important question that cannot be answered by this study is whether the predictive ability of the ASA status classification and ICU utilization for 30-day postoperative mortality in octogenarians undergoing emergency surgery equals or surpasses these established risk scores or models. In view of the aforementioned methodological limitations of this study, we do not agree with the authors’ conclusion that both ASA status classification and ICU utilization are proved as the best predictors for 30-day mortality of octogenarians undergoing emergency surgery.