Estimating risk of adverse cardiac event after vascular surgery using currently available online calculators.

BACKGROUND: The decision to proceed with vascular surgical interventions requires evaluation of cardiac risk. Recently, several online risk calculators were created to predict outcomes and to lead to a more informed conversation between surgeons and patients. The objective of this study was to compare and further validate these online calculators with actual adverse cardiac outcomes at a single institution. METHODS: All patients from January 2011 through December 2015 undergoing carotid endarterectomy (CEA), infrainguinal lower extremity bypass, open abdominal aortic aneurysm (AAA) repair, and endovascular aneurysm repair (EVAR) on the vascular surgical service were included using the Society for Vascular Surgery Vascular Quality Initiative database at our health system. Additional information was collected through retrospective chart review. Each patient was entered through three online risk calculators: (1) the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) estimates the risk of cardiac arrest and myocardial infarction (MI); (2) the Revised Cardiac Risk Index (RCRI) estimates risk of MI, pulmonary edema, ventricular fibrillation, primary cardiac arrest, and complete heart block; and (3) the Vascular Study Group of New England (VSGNE) Cardiac Risk Index estimates risk of postoperative MI only. Observed adverse cardiac events (ACEs) were compared with expected values for each calculator using a χ2 goodness-of-fit test. Institutional Review Board exemption was obtained. RESULTS: A total of 856 cases were included: 350 CEAs, 210 infrainguinal bypasses, 77 open AAA repairs, and 219 EVARs. For CEA, no risk calculator showed statistically significant variation from the observed values (NSQIP, P = .45; RCRI, P = .17; VSGNE, P = .24). For infrainguinal bypass, NSQIP slightly underpredicted adverse events (P = .054), RCRI strongly underpredicted (P = .002), and VSGNE showed no difference (P = .42). For open AAA repair, NSQIP (P = .51) and VSGNE (P = .98) were adequate predictors, but RCRI strongly underpredicted the adverse events (P ≤ .0001). Finally, EVAR cardiac outcomes showed greater adverse events than predicted by all three calculators (NSQIP, P = .02; RCRI, P = .0002; and VSGNE, P = .025). Pooled data for the entire group documented that the VSGNE proved an accurate tool for prediction (P = .34), whereas ACEs were underpredicted by NSQIP (P = .0055) and RCRI (P ≤ .001). CONCLUSIONS: Although online cardiac risk calculators of adverse surgical events are easy to use and to reference in broad surgical decision-making, there is significant variability in their predictability at the procedure and institutional level. Our data suggest that ACEs often occur at a higher rate than expected on the basis of calculated risks profiles, thus creating a platform for future discussion about preoperative evaluation and postoperative care decision-making models.

Vascular Graft Infection: Incidence and Potential Risk Factors.

PURPOSE: The purpose of this study was to identify factors that increase the risk of vascular graft infections (VGI) in patients following abdominal or lower extremity revascularization surgery. DESIGN: Retrospective, descriptive study. METHODS: We reviewed the electronic health records of 223 patients who had undergone abdominal or lower extremity revascularization procedures from July 2012 to November 2014, looking for factors associated with VGI. We reviewed 28 preoperative, intraoperative, and post-operative factors. Descriptive statistics (mean, range, and standard deviation) were used to describe the sample; χ was used to determine correlations between the risk factors and subsequent VGIs. The level of significance was determined at P = .05, with a confidence level of 95%. RESULTS: We identified 33 cases of VGIs for the 223 charts reviewed, yielding an incidence rate of 15%. Seventeen of the 33 patients with VGI (51.5%) were male. The average age of patients who experienced VGI was 60.9 years (standard deviation, 12.2 years, range, 29-81 years). Preoperative factors that were shown to show statistical significance for the development of VGI were sequential procedures (P = .003), diabetes mellitus (P = .002), hemoglobin A1c more than 7.0 (P = .0002), blood glucose more than 180 mg/dL (P = .0006), and lack of mobility (0.0097). Intraoperative factors associated with VGI were hemostatic agents applied to the surgical field intraoperatively (P = .003) and perioperative hypoxemia (P = .027). Postoperative factors associated with VGI were discharge from the hospital to skilled nursing facility or acute rehabilitation facility (P = .005) and unscheduled clinic visits (P = .008). CONCLUSION: We measured a 15% incidence of VGI and identified multiple pre-, intra-, and postoperative associated factors. Vigilance is required to prevent VGI and knowledge of specific risk factors is important.