A calculator for mortality following emergency general surgery based on the American College of Surgeons National Surgical Quality Improvement Program database. Issue 6 (June 2017)

• Record Type: Journal Article
• Authors: Haskins, Ivy N.; Maluso, Patrick J.; Schroeder, Mary E.; Amdur, Richard L.; Vaziri, Khashayar; Agarwal, Samir; Sarani, Babak
• Is Part Of: Journal of trauma and acute care surgery. Volume 82:Issue 6(2017)
• Publication Date: 2017-06
• Subjects: Emergency -- mortality -- risk -- surgery; Surgical intensive care -- Periodicals; Surgical emergencies -- Periodicals; Wounds and injuries -- Surgery -- Periodicals; 617.026
• DOI: 10.1097/TA.0000000000001451
• Languages: English
• ISSNs: 2163-0755

Abstract

Abstract : BACKGROUND: The complex nature of current morbidity and mortality predictor models do not lend themselves to clinical application at the bedside of patients undergoing emergency general surgery (EGS). Our aim was to develop a simplified risk calculator for prediction of early postoperative mortality after EGS. METHODS: EGS cases other than appendectomy and cholecystectomy were identified within the American College of Surgeons National Surgery Quality Improvement Program database from 2005 to 2014. Seventy-five percent of the cases were selected at random for model development, whereas 25% of the cases were used for model testing. Stepwise logistic regression was performed for creation of a 30-day mortality risk calculator. Model accuracy and reproducibility was investigated using the concordance index (c statistic) and Pearson correlations. RESULTS: A total of 79, 835 patients met inclusion criteria. Overall, 30-day mortality was 12.6%. A simplified risk model formula was derived from five readily available preoperative variables as follows: 0.034*age + 0.8*nonindependent status + 0.88*sepsis + 1.1 (if bun ≥ 29) or 0.57 (if bun ≥18 and < 29) + 1.16 (if albumin < 2.7), or 0.61 (if albumin ≥ 2.7 and < 3.4). The risk of 30-day mortality was stratified into deciles. The risk of 30-day mortality ranged from 2% for patients in the lowest risk level to 31% for patients in the highest risk level. The c statistic was 0.83 in both the derivation and testing samples. CONCLUSION: Five readily available preoperative variables can be used to predict the 30-day mortality risk for patients undergoing EGS. Further studies are needed to validate this risk calculator and to determine its bedside applicability. LEVEL OF EVIDENCE: Prognostic/epidemiological study, level III.