Diagnosis of prosthetic vascular graft infection using the management aortic graft infection collaboration (MAGIC) criteria

Background: Vascular graft infection (VGI) is a devastating complication that necessitates immediate diagnosis to reduce morbidity and mortality. In 2016, The Management of Aortic Graft Infection Collaboration (MAGIC) proposed a diagnostic algorithm for identifying aortic graft infection which uses clinical, radiological, and laboratory findings to determine risk. The purpose of this study was to retrospectively validate the MAGIC criteria for VGI diagnosis and potentially broaden its application. Methods: An IRB approved retrospective cohort study was performed with 28 suspected VGI patients from our institution. A blinded chart review categorized patients based on MAGIC criteria into “infected,” “suspected infection,” or “insufficient evidence” groups. Due to the MAGIC criteria’s placement of a third outcome, “suspected infection,” the results were dichotomized into two systems whereby “suspected infection” patients were considered as infected (S = I) or uninfected (S = NI). Both dichotomized systems from the MAGIC criteria were then compared to the clinical diagnosis (with surgical evidence when possible) from which sensitivity and specificity were calculated. Results: A total of 15 patients (54%) presented with definitive VGI, two patients (7%) presented with suspicion for VGI, and 11 patients (39%) presented with insufficient evidence to suspect VGI; placement in these categories were determined by the MAGIC criteria at entry to the study. When using clinical diagnoses, 17 out of 28 had VGI (61%). When using the categorization system S = I, sensitivity was 88.2% and specificity of 100%, for S = NI, sensitivity dropped to 76.4%, but specificity remained at 100%. These results were recomputed using only 25 of the 28 patients who were good surgical candidates. Using this new surgical threshold and smaller cohort to compare MAGIC against, sensitivity was calculated to be 93.3% with a specificity of 100% when S = I. When S = NI, the sensitivity was calculated to be 86.7% with specificity remaining at 100%. Conclusions: The MAGIC criteria provide excellent specificity and reasonable sensitivity for diagnosis of VGI for peripheral and aortic revascularization.