Role of computed tomography in prediction of gastrointestinal fistula in patients with acute pancreatitis

Aim : To identify computed tomography (CT) features that predict gastrointestinal fistula (GIF) in patients with acute pancreatitis (AP). Methods: This retrospective study comprised consecutive patients with AP and GIF from June 2017 to June 2018. The diagnosis of GIF was based on upper gastrointestinal endoscopy, colonoscopy or surgery. A cohort of 19 matched patients from a prospective database of AP served as control group. Measures of severity, and clinical outcome were evaluated. CT parameters were compared between the groups to assess the features that could predict the development of GIF. Results : There was no difference between the two groups in terms of disease etiology, severity, drainage, and mortality. On univariate analysis, the CT features that were found to be significantly different between the two groups were the presence of bowel wall thickening (P=0.005), maximum thickness of the bowel wall (P=0.007), presence of air foci in extra pancreatic necrosis/ collection (P=0.013), discontinuity of the bowel wall (P=0.046) and the displacement/ compression of bowel by fluid collection (P=0.014). On multivariate analysis, all the above-mentioned CT findings except discontinuity of bowel wall were found to be statistically significant. Conclusion : CT is helpful in predicting GIF in patients with AP. (Acta gastroenterol. belg., 2019, 82, 495-500).