Can the American College of Surgeons National Surgical Quality Improvement Program Risk Calculator Predict Outcomes for Urgent Colectomies?
The American Surgeon
BACKGROUND: The American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) Risk Calculator (RC) predicts postoperative outcomes using 19 risk factors, including operative acuity. Acuity is defined by the calculator as emergent or elective only. The objective of this study is to evaluate the RC's accuracy in urgent (nonelective/nonemergent) cases.
METHODS: This is a retrospective review of the NSQIP data for patients who underwent urgent colectomies at a single tertiary care center over a 4-year period. Each urgent case was entered into the RC as both elective and emergent, and predicted outcomes were compared to actual postoperative outcomes. Receiver operating characteristic (ROC) curves were used when sufficient statistical power was present and the area under the curve (AUC) was calculated.
RESULTS: A total of 301 urgent colectomy patients were evaluated, representing 19% of all colectomies performed at our institution during the study period. Of the 15 possible postoperative outcomes, the RC showed high predictive value only for mortality (AUC elective .8467; emergent .8451) and discharge to a nursing/rehabilitation facility (AUC elective .8089; emergent .8105). The RC showed no predictive value for 6 outcomes and the remainder lacked statistical power to draw conclusions.
DISCUSSION: While the calculator predicted mortality and discharge to a nursing/rehabilitation facility, it did not accurately predict complications for urgent colectomies. Future versions of the calculator should focus on improving the predictive value by including urgent cases as a separate category.
Ziegler MA, Bauman JC, Welsh RJ, Wasvary HJ. Can the American College of Surgeons National Surgical Quality Improvement Program Risk Calculator Predict Outcomes for Urgent Colectomies? Am Surg. 2020 Dec 21:3134820973392. doi: 10.1177/0003134820973392. Epub ahead of print. PMID: 33345578.