Established Scoring Systems Have Limited Utility In Predicting Postoperative Adverse Outcomes Among Vascular Surgery Patients
Muhammad Saad Hafeez, MBBS, Salim G. Habib, MD, Michael C. Madigan, MD, Natalie D. Sridharan, MD MS, Rabih A. Chaer, MD MS, Mohammad H. Eslami, MD.
University of Pittsburgh Medical Center, Pittsburgh, PA, USA.
OBJECTIVES:
General surgical studies suggest 14-item risk analysis index (RAI) score correlates with postoperative mortality, leading to mandatory recording by some institutions for new surgical patients. Others have shown that frailty strongly predicts postoperative outcomes, leading to generation of scores like the modified 5-item frailty score(mFI-5). We evaluated the association of RAI and mFI-5 with postoperative outcomes and compared their ability to predict adverse events among patients undergoing open vascular operations.
METHODS:
We retrospectively analyzed patients undergoing elective index open vascular operations with preoperative RAI recorded at a multi-hospital system (July 2016 to December 2020). We excluded non-elective and endovascular cases. We calculated mFI-5 scores and generated mFI-6 scores by adding creatine (creatinine>1.6 mg/dL) to the mFI-5. Primary endpoints were 30-day postoperative death (POD), major morbidity events (MME), and 1-year mortality. (MME: myocardial infarction, cardiac arrest, arrhythmia, pulmonary embolism, unplanned intubation, stroke, acute kidney injury, and sepsis.) Appropriate statistical analyses were performed.
RESULTS:
Among 804 cases with recorded RAI (infrainguinal bypass=47.2%, carotid=34.3%, and open aortic=18.5%), there was an overall POD, one-year mortality, and MME of 2.0%, 7.5%, and 10.8%, respectively. The only score associated with POD was mFI-6 (OR=1.78[1.06-3.00];p=0.029), while RAI (OR=1.04[0.99-1.09];p=0.089) and mFI-5 (OR=1.38[0.78-2.46];p=0.270) were not (Figure). Receiver operating curve analyses demonstrated limited predictive ability with mFI-6 (Area under curve (AUC)=0.6280).
For one-year mortality, only RAI demonstrated association with reasonable predictive ability (OR=1.08[1.06-1.11];p<0.001; AUC=0.7340). For MME, mFI-5 (OR=1.12,[0.86-1.44];p=0.400) and mFI-6 (OR=1.23[0.97-1.56];p=0.086) were not significantly associated, but RAI was (OR=1.03[1.01-1.05];p=0.002), albeit with low predictive ability (AUC=0.5926).
CONCLUSIONS:
Existing scoring systems have limited utility in predicting adverse events after vascular operations, calling for development of better risk stratification systems specific to vascular surgical patients. 
Back to 2023 Abstracts