Outcome after surgery for perianal fistula: predictive value of MR imaging.

Abstract

The purpose of this study was to determine if MR findings are predictive of long-term outcome in a cohort of patients whose initial surgery was performed without access to the findings of MR imaging. Forty patients with surgically proven perianal fistulas underwent preoperative dynamic contrast-enhanced MR imaging. The MR and surgical findings were independently recorded on an identical anatomic form. Three patients were subsequently lost to follow-up. The outcome for the remaining 37 patients was determined from surgical review, case notes, and questionnaires. Minimum follow-up period was 14 months (range, 14-39 months). Outcome was determined by one observer who was unaware of the initial MR grading and had not been present during surgery. Outcome was considered unsatisfactory if further surgery was required. MR imaging was better than surgical exploration in predicting outcome (for MR imaging: positive predictive value, 73%; negative predictive value, 87%; sensitivity, 89%; and specificity, 68%; for surgical exploration: positive predictive value, 57%; negative predictive value, 64%; sensitivity, 73%; and specificity, 47%). MR classification of fistulas was significantly associated with outcome (p = .0004), and surgical classification was not significantly associated with outcome (p = .22, chi-square test). Also MR grades differed significantly for patients with satisfactory and unsatisfactory outcomes (p < .001, Mann-Whitney U test). MR imaging is valuable in the management of patients with perianal fistulas. MR imaging accurately reveals surgical anatomy and can be used to make better predictions regarding patient outcome than surgical findings.