Developing a risk calculator for mortality following emergency general surgery

    loading  Checking for direct PDF access through Ovid


To the Editor:
The recent article by Haskins and colleagues1 developing a risk calculator for mortality following emergency general surgery was of great interest. By multivariate logistic regression analysis, they showed that the five variables most strongly associated with 30-day mortality were age, nonindependent functional status, preoperative sepsis, blood urea nitrogen level, and serum albumin level. Based on the strength of association of five variables with postoperative mortality, a mortality risk calculator was established. Furthermore, the authors described that this calculator had the good discriminative powers for mortality risk in both the derivation (C statistic 0.83) and testing (C statistic 0.83) data sets. Although the valuable study has been actualized, some issues in methodology seem important to avoid any optimistic interpretation or misinterpretation of results.
First, to determine discrimination ability of this risk calculator for postoperative mortality risk, providing only the C statistic value is not enough, especially for low-risk patients. The authors should perform the sensitivity analysis and construct the receiver operating characteristic curve to obtain the sensitivity, specificity, and positive and negative predictive values of this risk calculator for postoperative mortality. Furthermore, it is better to provide a cutoff value of this risk calculator for postoperative mortality, rather than a decile risk stratification of postoperative mortality.
Second, most important, the mortality risk calculator included only the preoperative variables associated with postoperative mortality, but not the intraoperative and postoperative risk factors that can significantly affect mortality of emergency surgery patients. The available evidence indicates that surgery type, operative time, intraoperative large blood loss, and blood transfusion are associated with increased risks of mortality after emergency surgery.2,3 In fact, the surgical Apgar score based on the estimated blood loss, lowest heart rate, and lowest mean arterial pressure during surgery has been shown as a good predictor of mortality after emergency high-risk intra-abdominal surgery.4 Moreover, postoperative complications, such as myocardial infarction, acute kidney injury, pulmonary complications, and stroke, have the strongest associations with mortality after emergency surgery. Especially, postoperative acute kidney injury and pulmonary complications are common among patients with emergency surgery. Even mild postoperative acute kidney injury and pulmonary complications are associated with increased early postoperative mortality. Actually, postoperative complications have been regarded as important targets for emergency surgical quality improvement initiatives.5 Thus, we argue that discriminating ability and predictive value of the risk calculator designed by authors would have further been improved, if the equation had included intraoperative and postoperative factors affecting postoperative mortality of emergency surgery patients.
Finally, much effort has been made to find an optimal scoring system that can accurately predict the risk of mortality following emergency surgery. Hitherto, there have been many established risk-prediction models with different discriminatory powers for postoperative mortality of emergency surgery patients, such as the American Society of Anesthesiologists physical status classification, Surgical Risk Scale, surgical Apgar score, CORES (calculation of postoperative risk in emergency surgery), E-PASS (Estimation of Physiologic Ability and Surgical Stress), American College of Surgeons National Surgical Quality Improvement Program Surgical Risk Calculator, Simple Prognostic Index, Acute Physiology and Chronic Health Evaluation II, Simplified Acute Physiology Score II expanded, POSSUM (Physiological and Operative Severity for the Enumeration of Mortality and Morbidity), Portsmouth POSSUM, Manheim peritonitis, Charlson Comorbidity Index, and so on. However, no single scoring system has extensively been validated across multiple centers and geographical locations. A limitation of this study design is no comparison for predictive value of this risk calculator for postoperative mortality with that of any established risk score or model.
    loading  Loading Related Articles