Outcomes of Rectovaginal Fistula Repair

OBJECTIVES Rectovaginal fistulae (RVF) often represent surgical challenges, and treatment must be individualized. We describe outcomes after primary surgical repair stratified by fistula etiology and surgical approach. METHODS This retrospective cohort study included women who underwent surgical management of RVF at a tertiary care center between July 1, 2001 and December 31, 2013. Cases were stratified according to the following etiology: cancer (RVF-C), inflammatory bowel disease or infectious (RVF-I), and other (RVF-O). Patients with prior surgical treatment of RVF were excluded. Surgical approaches included local (seton, plug), transvaginal or endorectal, abdominal, diversion alone, or definitive (completion proctocolectomy with permanent colostomy or pelvic exenteration). Recurrence-free survival was estimated using the Kaplan-Meier method, and comparisons between subgroups were evaluated based on fitting Cox proportional hazards models. Censoring occurred at last relevant clinical follow-up. Factors contributing to recurrence-free survival were evaluated including age, body mass index, smoking status, fistula etiology, ileostomy, and surgical approach. RESULTS During the study period, 107 women underwent surgical repair of RVF. The most common fistula etiology was RVF-I (54.2%), followed by RVF-O (23.4%), and RVF-C (22.4%). Ninety-four women underwent fistula repair by the local (29.9%), transvaginal/endorectal (25.2%), abdominal approach (19.6%), or diversion alone (13.1%), whereas 13 underwent definitive surgery (12.2%). Recurrence-free survival was significantly different depending on surgical approach (P < 0.001), but not etiology (P = 0.71). Recurrence-free survival (95% confidence interval) at 1 year after surgery was 35.2% (21.8%-56.9%) for the local approach, 55.6% (37.0%-83.3%) for the transvaginal or endorectal approach, 95% (85.9%-100%) for the abdominal approach, and 33.3% (15%-74.2%) for those with diversion only. CONCLUSIONS Recurrence rates after RVF repair are high and did not differ by fistula etiology. Abdominal repair of RVF had significantly fewer recurrences.