Choice of Operation and Specific Set of Preoperative Comorbidities Can Predict Adverse Postoperative Outcomes after Extra-anatomic Suprainguinal Bypass Procedures
Mohammad H. Eslami, MD, MPH, Efthymios Avgerinos, MD, Rabih A. Chaer, MD MS, Michael J. Singh, MD, Michel S. Makaroun, MD.
Division of Vascular Surgery, University of Pittsburgh Medical School, Pittsburgh, PA, USA.
Objective: The purpose of this study was to develop a simple risk predictive model for patients undergoing EAB for aortoiliac occlusive disease (AIOD).
Methods: American College of Surgeons National Surgical Quality Program databases (ACS-NSQIP: 2011-2014) for AIOD were queried to identify all the patients who underwent elective EAB based on CPT-procedure codes. Choice of operation (fem-fem bypass (FFB) vs ax-fem (AFB)), as well 2011-2013 data regarding perioperative variables, were used to develop an internally validated parsimonious risk model for composite adverse outcomes after EAB. This model was then externally validated using patients who had their procedure in 2014 (validation sample). Goodness of fit of the model was evaluated using C-statistic. The probabilities predicted by this model were correlated with the values reported for same patients using ACS-NSQIP algorithm.
Results: We identified 902 cases that underwent elective EAB (709 FFB, 193 AFB: derivation sample= 594 cases, validation sample=308 cases). Although AFB group included older patients (70.6 +/- 9.3 vs. FFB: 66.3 +/- 10.8, p<.001) with higher ASA classifications (p<.001), comparison of other demographics and perioperative risk factors between the two surgical groups were not statistically significant. Rate of adverse outcomes were significantly higher for AFB group (43.5% vs. 18.3%, p<.001) and AFB significantly increased the risk of adverse postoperative outcomes (Odds Ratio: 2.6, 95% Confidence Interval: 1.66-4.08, p<0.001). Our parsimonious model (5 variables: choice of operation for EAB, age and functional status, CHF and, functional status, ASA classification) showed a reasonable predictability and goodness of fit (Hosmer-Lemeshow goodness of fit p=0.166) in predicting adverse outcomes in both derivation (C-statistic =0.736) and validation (C=0.745) cohorts (Figure 1). This simple model correlates very well with the ACS-NSQIP probability prediction calculation (ρ=0.711 for derivation sample and ρ=0.745 for validation sample, Figure2).
Conclusions: A simple 5-variable model shows promise in predicting adverse outcomes for patients undergoing EAB. Based on this model, AFB independently predicts adverse outcomes, and given the many advances in endovascular surgery should be performed only selectively.
Back to 2018 Program
