Rectovaginal Fistulas Secondary to Obstetrical Injury

Abstract

Rectovaginal fistula (RVF), defined as any abnormal connection between the rectum and the vagina, is a complex and debilitating condition. RVF can occur for a variety of reasons, but frequently develops following obstetric injury. Patients with suspected RVF require thorough evaluation, including history and physical examination, imaging, and objective evaluation of the anal sphincter complex. Prior to attempting repair, sepsis must be controlled and the tract allowed to mature over a period of 3 to 6 months. All repair techniques involve reestablishing a healthy, well-vascularized rectovaginal septum, either through reconstruction with local tissue or tissue transfer via a pedicled flap. The selection of a specific repair technique is determined by the level of the fistula tract and the status of the anal sphincter. Despite best efforts, recurrence is common and should be discussed with patients prior to repair. As the ultimate goal of RVF repair is to minimize symptoms and maximize quality of life, patients should help to direct their own care based on the risks and benefits of available treatment options.

Rectovaginal fistula (RVF) is a debilitating condition for women and a vexing problem for colorectal surgeons. RVFs secondary to complications of labor and delivery, so called obstetric fistulas, represent the most common etiology worldwide. Treatment strategies vary based on the underlying etiology, anatomy, and severity of symptoms. All patients should undergo a complete workup including history and physical examination, imaging, and objective evaluation of anal sphincter function. Only a minority of RVFs resolve without an operation and, despite a variety of repair techniques, the need for both staged procedures and recurrent repairs are common. Surgeons must choose the repair technique that best matches the patient's fistula location, etiology, and anatomy, as well as their own training and experience.

Epidemiology

Estimates of the incidence of obstetric RVF range from 0.2 to 4 per 1,000 deliveries ^{1 2 3 4} ; prevalence estimates range from 0 to 81 per 1,000 women. ¹ Lower rates are observed in developed countries (0.2 per 1,000 deliveries in Norway) ² and higher rates in less developed, low-income areas of the world (1.7 per 1,000 deliveries in sub-Saharan Africa and Southeast Asia). ⁴ This disparity reflects an important distinction in the underlying mechanisms by which obstetric fistulas occur. The vast majority of international patients develop fistulas as a result of prolonged labor with pressure necrosis of the perineal body and lower rectovaginal septum from compression by the fetal head. Unfortunately, in many cases, the damage is not limited to the rectovaginal septum and patients present with either vesicovaginal or urethrovaginal fistulas in addition to their RVF. ⁵ Western patients, on the other hand, tend to develop fistulas as a result of tearing of the perineal soft tissue during delivery. In the United States, obstetric anal sphincter injury (OASIS), which includes both third- and fourth-degree lacerations, occurs in 4 to 7% of vaginal deliveries. ^{6 7 8 9 10} Primary repair is typically performed at the time of injury in a layered fashion. ¹¹ Among patients having primary repair, 10% will have wound dehiscence ¹⁰ and 1 to 3% of patients will develop a RVF. ¹²

Risk Factors

Risk factors for obstetric RVF development mirror the mechanism of injury. In the developing world, these are related to obstructed labor including a lack of access to emergency obstetric care, home delivery, stillbirth, nulliparity, short stature, teenage pregnancy, and lack of financial or decision-making capacity. ^{5 13 14 15 16} In developed countries, these are related to OASIS, including extremes of maternal age, episiotomy, the use of forceps and/or vacuum, macrosomia, nulliparity, and occipital posterior positioning. ^{6 7 8 9 17} Several large studies suggest that the use of either forceps or vacuum assistance during vaginal delivery increases the risk of OASIS, while others identified an increased risk only among patients delivered via forceps. ^{6 7 8 9} One study using routine postpartum endoanal ultrasound found that 80% of primiparous women delivered via forceps had sphincter defects compared with no women in the vacuum-assisted group. ¹⁸ Similarly, multiple studies have found that the use of midline episiotomy increases the risk of OASIS when compared with mediolateral episiotomy or no episiotomy. ^{8 9}

Classification

Fistulas are classified by their location, size, etiology, and extent of damage to surrounding tissues. Daniels ^a categorized RVF based on the height of the vaginal opening: “low” fistulas near the posterior vaginal fourchette, “mid” fistulas between the posterior fourchette and the cervix, and “high” fistulas near the posterior vaginal fornix. ¹⁹ Contemporary literature tends to condense these into “high” and “low” RVFs with the dividing line being the dentate or the anal sphincter complex. Organizing RVFs by location is useful both as a method of identifying the underlying disease process (e.g., obstetric fistulas tend to be low while fistulas following pelvic surgery tend to be higher) and in planning for an eventual repair (e.g., low fistulas may be more amenable to a local or perineal approach repair while high fistulas often require abdominal exploration).

Others choose to separate “simple” fistulas from more complex disease. Simple RVFs are traditionally low, small (<2.5 cm), and due to obstetric trauma or cryptoglandular infection as opposed to “complex” fistulas, which are high, large, recurrent, or associated with inflammatory bowel disease, malignancy, or radiation. ²⁰ Distinguishing RVFs by the presence or absence of an associated anal sphincter injury can also be extremely useful in terms of operative planning. The utility of separating RVFs based on etiology remains a matter of debate. Given that obstetric fistulas tend to be simple and often occur in otherwise healthy women of childbearing age, it stands to reason that rates of successful repair may be higher. In fact, women with obstetric fistulas were found to be almost four times more likely to have resolution of symptoms at 6 months than women with fistulas due to nonobstetric etiologies. ²¹ Other studies, however, have found no significant difference in successful repair rates based on RVF etiology. ^{22 23}

Evaluation

History

All patients with suspected RVF should undergo a complete history and physical examination, with a particular focus on any gastrointestinal or gynecologic disease and on symptoms of RVF such as vaginal drainage, perineal skin irritation, dyspareunia, urinary tract infections, and fecal incontinence (FI). A thorough review of patients' bowel function is performed, including any recent changes to stool caliber or frequency and the presence of any bloody or purulent anal discharge. Any history consistent with inflammatory bowel disease, especially perianal Crohn's disease, should be evaluated fully. FI, if present, can be quantified using a bowel journal or any one of a variety of validated scales.

A thorough obstetric history should be performed, including number of vaginal and operative vaginal deliveries, episodes of prolonged labor, and prior episiotomies or lacerations. If a laceration did occur, reviewing previous operative records may aid in surgical planning. Any prior surgical terminations, dilation and curettage procedures, or other obstetric procedures should also be documented. Breastfeeding status and future childbearing desires require particular attention. Breastfeeding women can have decrements in continence that improve after cessation of breastfeeding. Since the degree of FI can influence the selection of the surgical procedure to perform, patients may want to delay operative intervention until breastfeeding ceases to establish what their baseline level of continence is. Patients strongly desiring additional vaginal deliveries should delay operative intervention until after childbearing is complete, as often a Caesarean delivery is recommended after a fistula repair.

Examination

Surgeons should perform a complete examination of the vagina, perineum, and anus. Prior incisions, lacerations, external openings, or radiation changes and the width of the perineal body are noted. A digital rectal examination is performed to localize the tract, rule out malignancy, and evaluate the strength and function of the anal sphincter muscles and any obvious defects. A vaginal examination is performed to rule out malignancy, assess for ongoing inflammation, and identify the fistula tract. ²⁰ Examination in the operating room may be necessary when a fistula is suspected but cannot be identified on office examination or imaging. Instillation of dilute methylene blue into the rectum or vagina with placement of a tampon in the other cavity can help to identify an occult fistula tract. Endoscopy can be performed to rule out concomitant mucosal inflammation, anastomotic leak, or malignancy. Electromyography and anorectal manometry play lesser roles but can be used to quantify anal sphincter function.

Imaging

All patients should have evaluation of the anal sphincter muscles prior to surgery via endoanal ultrasound or magnetic resonance imaging (MRI), as up to 40% of women have evidence of occult sphincter injury after vaginal delivery. ¹⁸ The decision between endoanal ultrasound and MRI should be based on which modality could best provide a thorough evaluation of sphincter defects, fistula anatomy, and any other pelvic abnormalities based on the resources available at the surgeon's institution.

Endoanal ultrasound requires a trained provider and specialized equipment but provides the most detailed assessment of the location and degree of anal sphincter injuries as well as a three-dimensional representation of the fistula tract. In fact, one study found a 100% positive predictive value for endoanal ultrasound in determining the location of RVF fistula tracts. ²⁴ MRI can identify fistula tracts with up to a 92% positive predictive value. ²⁴ A 3.0 T MRI machine with a dedicated protocol, small field of view, and vaginal ultrasound gel as contrast is preferred. ²⁵ Unilateral or bilateral puborectal atrophy, especially in the setting of potential obstetric trauma, should raise the concern for anal sphincter injury or dysfunction. ²⁵ Computed tomography does not provide the same level of anatomic detail as MRI, it can be useful in suggesting the presence of an RVF, evaluating for undrained pelvic sepsis, and ruling out concomitant intestinal inflammation, especially diverticulitis. Fluoroscopy, via contrast enema or vaginography, can be used as a confirmatory test with sensitivities ranging from 79 to 100%. ²⁰

Repair

General Considerations

No single repair technique consistently outperforms others in achieving successful fistula closure. In a recent systematic review, Göttgens et al were unable to recommend any specific repair technique due to small sample sizes, weak methodology, and high degrees of variation within the literature. ²⁶ The literature largely consists of patient series in which different techniques are used for different groups of patients according to surgeon preference. ^{19 27 28} Smoking, ²² obesity, ²³ and exposure to pelvic radiation ²⁹ have been associated with significantly higher rates of recurrence.

Despite a paucity of high-quality comparison studies, important principles of RVF repair can be drawn from the literature and from prior experience with RVF repair. First, source control must be achieved via drainage of fluid collections, seton placement, or diversion prior to repair. Waiting at least 3 to 6 months is necessary to ensure that infection and inflammation have resolved. Second, the rectovaginal septum needs to be reestablished using healthy tissue, either locally, if the adjacent tissue is of sufficient quality, or via tissue transfer with a pedicled flap. Finally, the choice of repair must take into account FI, both current and after attempted repair.

Conservative Management

A minority of women will heal their RVF without surgical intervention. In a recent national survey of urogynecology practices, 17% of patients presenting with RVF were initially managed expectantly, and after a median follow-up of 8 weeks, 66% of patients reported resolution of symptoms. Patients managed expectantly were significantly more likely to have small (<1.5 cm) superficial fistula tracts compared with patients managed surgically. ³⁰ Lo et al found that 64% of RVF patients treated nonoperatively had improvement in their symptoms at 12 months, but this was less than the 89% who had improvement at 12 months after surgery. ³¹

In addition to medical management, there are options for limited repair. Fibrin glue injection has definitive closure rates in the 10 to 15% range, and thus is typically reserved for patients who cannot tolerate more invasive procedures. ²⁷ A pilot study of autologous fat graft injection to the fistula tract achieved tissue healing in 77% of women after an average of 1.8 procedures, although roughly one-third of their small sample was protected with a diverting stoma. ³² Bioprosthetic plugs have limited data to support their efficacy. ³³ Gajsek et al reported a 44% success rate at 11 months using a porcine small intestine xenograft. ³³ Other studies report closure rates of 11 to 86% and suffer from small sample sizes and limited follow-up. ^{27 34 35} The moderate success of isolated mesenchymal stem cell injection in the treatment of RVF has led to the development of stem cell infused fistula plugs. In their pilot study, Lightner et al found partial improvement in RVF symptoms among all five patients. However, all had residual defects on imaging and physical examination; none opted for ileostomy takedown at their 6-month follow-up. ³⁶

Operative Repair

Patients with “simple” fistulas are divided into four categories based on the location of the defect relative to the anal sphincter complex and the presence or absence of a sphincter defect ( Fig. 1 ). Patients with defects at the level of the sphincter complex typically do not require additional bulking procedures while those with defects above the sphincters benefit from transposing healthy tissue into the septum. Similarly, patients with defects in the anal sphincter muscles and symptoms of FI benefit from a concomitant sphincter repair as a part of their RVF operation. It is less clear what to do with patients who have a sphincter defect on examination or imaging but no symptoms of incontinence. We lean toward performing sphincter repair in this situation given the evidence that concomitant sphincteroplasty may improve RVF closure rates. ³⁷

Fig. 1.

Treatment algorithm for obstetric rectovaginal fistula. ERAF, endorectal advancement flap; EVAF, endovaginal advancement flap; LIFT, ligation of intersphincteric fistula tract; RVF, rectovaginal fistula.

Defect at the Level of the Sphincter Complex, Intact Sphincter Muscles

Endorectal Advancement Flap

Since its original description by Laird ^b in 1948 and its modification and popularization by Rothenberger et al in 1982, the endorectal advancement flap (ERAF) has become the most common operation performed for initial RVF repair. ERAF brings healthy tissue into the anal canal to buttress septal repair on its high-pressure side. ^{38 39} The procedure is performed with the patient in the prone position, typically under general anesthesia. A mechanical bowel prep can be given the day prior and a urinary catheter is placed at the time of surgery. The anal canal is exposed and the fistula opening is identified. A flap is raised starting just distal to the opening and continuing at least 4-cm proximal to the defect ( Fig. 2 ). The flap is trapezoidal, twice as wide at the base as at the apex, and includes mucosa, submucosa, and at least some of the inner circular muscle. ³⁹ The defect is cored out or curetted and the exposed sphincter muscle is plicated. Finally, the apex of the flap is trimmed to remove the fistula opening and the flap is secured to the anoderm and adjacent rectal mucosa using a series of interrupted absorbable sutures. The vaginal defect is intentionally left open to drain any fluid.

Fig. 2.

Endorectal advancement flap. ( A ) Fistula. ( B ) Flap raised, with fistula opening visible. ( C ) Fistula opening closed. ( D ) Flap advanced, and portion of flap with fistula opening excised. ( E ) Flap sutured in place, with fistula opening underneath flap. ( F ) Incorrect flap construction, with inadequate coverage of fistula opening. Reprinted with permission from Abcarian H, ed. Anal Fistula: Principles and Management. New York: Springer; 2014:98.

Closure rates for ERAF range from 29 to 93%. ^{20 40} When selecting only obstetric RVF from the largest series of ERAF repairs, the range of closure rates tightens slightly ( Table 1 ). Sonoda et al report that older patients and patients with higher body surface area had higher closure rates while patients with a longer duration of symptoms or Crohn's disease had lower closure rates. ⁴¹ Ozuner et al found no difference in closure rates based on the etiology of the fistula, the use of perioperative antibiotics, or the use of antimotility agents but did find lower rates of closure among patients who had had a previous attempt at repair. ⁴²

Table 1. Subset of patients with obstetric RVF from series of patients undergoing ERAF.

Study (year)	Patients	Location	Size (cm)	Duration of symptoms (mo)	Bowel prep	Antibiotics	Flap	Recurrent (%)	Sphincteroplasty (%)	Diversion (%)	Follow-up period (mo)	Closure rate (%)	Post-op incontinence (%)
Rothenberger et al 39 (1982)	24	Low/mid	<2.5	–	Mechanical only	Intra-op	Full	–	29	3		86	–
Jones 71 (1987)	6	Low		–	Mechanical only	Intra-op	Partial	–	–	23	25	67	–
Lowry et al 37 (1988)	60	Low/mid	<2.5	–	Mechanical only	Intra-op	Full	40	31	–	25	75	–
Wise et al 44 (1991)	25	Low		2	Mechanical only		Partial	13	38	–	3	100	18
Kodner et al 38 (1993)	48	Low	<2.5	–	Mechanical and oral antibiotics	Intra-op	Partial	–	–	0	–	88	13
Khanduja et al 28 (1994)	27	–	–	13	Mechanical and oral antibiotics	24 h	–	–	52	0	16	93	7
Mazier et al 19 (1995)	19	Low/mid	<2.5	< 3	Mechanical and oral antibiotics	Intra-op	Full	31	0	–	49	94	5
Ozuner et al 42 (1996)	13	–	<3	12	Mechanical only	Intra-op	Partial	16	4	17	31	77	–
Tsang et al 45 (1998)	27	–	–	18	–			15	0		15	41	33
Hyman 72 (1999)	12	Low	–	–	–	72 h	Partial	8	0	0	39	92	0
Zimmerman et al 55 (2002)	9	Low	–	–	Mechanical only	Intra-op	Partial	14	33	24	–	44	–

Abbreviations: ERAF, endorectal advancement flap; Intra-op, intraoperative; Post-op, postoperative; RVF, rectovaginal fistula.

The relationship between ERAF and FI is complex and incompletely understood. Several studies report rates of FI as high as 20 to 30% after ERAF for RVF, although most studies do not report rates of either pre- or postoperative incontinence. Two separate mechanisms may be at work. First, creation of the flap may alter the function of the anal sphincter complex or interfere with secondary continence mechanisms such as the hemorrhoidal plexus. However, at least one study in which patients who underwent manometric evaluation did not show any difference before and after ERAF. ⁴³ Alternatively, RVF repair may unmask symptoms of FI that were previously classified as vaginal drainage or not apparent, while the fistula tract decompressed the rectal vault. ⁴⁴ Especially in the setting of obstetric RVF, where up to 75% of women have some level of sphincter defect, ²³ it is important to remember that ERAF does not address injuries to the sphincter complex and therefore may not be ideal for patients with a documented injury or clear symptoms of FI. For example, Tsang et al found a 45% improvement FI among RVF patients treated by sphincteroplasty compared with none among those treated by ERAF. ⁴⁵

Vaginal Advancement Flap

Vaginal or endovaginal advancement flaps (EVAF) are performed in an identical fashion to ERAF except that the incision is made in the posterior vaginal wall instead of on the rectal side of the septum. There are no head-to-head comparisons of EVAF and ERAF in the literature. Small studies of EVAF report similar success rates to ERAF (80–93%), although the vast majority of patients had both concomitant bulking procedure and diverting stoma performed at the time of surgery. ^{46 47 48 49} The decision to perform EVAF as opposed to ERAF depends on surgical training (i.e., urogynecology vs. colorectal surgery), but also on the quality of the rectal tissue. EVAF may be well suited to repair of RVF due to Crohn's disease since it allows a flap to be generated with minimal manipulation of the diseased rectum. ⁴⁹ Since all of the dissection is performed on the vaginal side, EVAF also significantly diminishes the risk of FI when compared with ERAF, although some reports suggest increased rates of dyspareunia due to vaginal scarring. ⁴⁸

Ligation of Intersphincteric Fistula Tract

Although initially developed for transsphincteric fistula-in-ano, there are limited reports of using ligation of intersphincteric fistula tract for RVF repair. An incision is made in the intersphincteric groove and carried into the rectovaginal septum, separating the external and internal sphincter muscles without entering the fistula tract. The tract is divided and ligated on each side with absorbable suture. The repair is tested from each side with either hydrogen peroxide or a fistula probe. Similar to ERAF, the vaginal opening is cored out and left open to drain. Success of the procedure is predicated having a healthy, intact muscle layer on either side of the fistula tract. ⁵⁰ However, unlike in a transsphincteric fistula-in-ano where the intersphincteric groove can be identified lateral to the tract in an unaffected plane, the perineal body in the setting of RVF is typically replaced by scar, making identification, and isolation of the intersphincteric space more difficult. Moreover, the sphincter muscles are often damaged by the obstetric injury that led to the RVF, making isolation of healthy muscle impossible.

Defect at the Level of the Sphincter Complex, Defect in Sphincter Muscles

Sphincteroplasty

Concomitant disruption to anal sphincter complex should be addressed at the time of RVF repair via sphincteroplasty. Compared with end-to-end repair, overlapping sphincteroplasty is associated with lower incontinence rates, lower levels of fecal urgency, and more durable results. ¹² For low RVFs confined to the sphincter complex, overlapping sphincteroplasty with debridement or simple suture ligation of the fistula openings is typically sufficient for RVF closure since the divided tract is separated by well-vascularized sphincter muscle. Overlapping sphincteroplasty is performed in the prone position. A curvilinear incision is made over the perineal body. Scar tissue is dissected free and the ends of the sphincter muscles are identified ( Fig. 3 ). The muscles are dissected to allow overlap with minimal tension. Allis clamps are placed on the ends of the sphincter and used to pull across the midline while a finger is placed in the anus to determine appropriate sphincter tone. The muscles are then sutured to each other with absorbable horizontal mattress sutures. The vaginal and rectal openings of the fistula tract are debrided and either left open or closed with absorbable suture. The skin incision is partially closed with interrupted absorbable sutures. Patients are typically admitted at least overnight for pain control and an aggressive bowel regimen to minimize straining.

Fig. 3.

Overlapping sphincteroplasty. ( A ) Incision over perineal body. ( B ) Division of scar. ( C ) Reapproximation of internal sphincter muscle, if separately identifiable. ( D ) Overlapping repair of external sphincter. ( E ) Partial closure of wound. Reprinted with permission from Abcarian H, ed. Anal Fistula: Principles and Management. New York, NY: Springer; 2014:153.

Small studies show a 100% closure rate for sphincteroplasty following obstetric trauma. ^{47 51} Tsang et al presented the largest series of sphincteroplasty for low RVF repair with 28 of 38 patients (80%), ultimately achieving closure. ⁴⁵ Patients with preoperative incontinence, a manometric deficit or a defect on endoanal ultrasound, all had above average rates of closure (84, 86, and 88%, respectively) and significantly higher rates of closure than similar patients who underwent ERAF (33, 33, and 33%, respectively). Sphincteroplasty resulted in a significant reduction in symptoms of FI (60% preoperatively vs. 32% postoperatively) while ERAF did not (33% both pre- and postoperatively).

Defect above the Level of the Sphincter Complex, Intact Sphincter Muscles

Above the sphincter complex, the rectovaginal septum is thin, poorly vascularized, and unlikely to heal without additional tissue transfer. For that reason, we favor transperineal repair in combination with a bulking procedure, such as transposition of the labial fat pad or bulbocavernosus muscle (the Martius flap), transposition of the gracilis muscle (gracilis flap), or the use of biologic mesh. All tissue transfer procedures begin in a similar manner. The patient is placed in a modified lithotomy position after induction of general anesthesia. Perioperative antibiotics are generally given, although not required. A transverse perineal incision is made over the rectovaginal septum and this dissection is carried superiorly, through the fistula tract and any surrounding scarred tissue until healthy soft tissue above the tract is encountered. The transferred tissue is passed through a subcutaneous tunnel between the harvest site and the perineal incision and then secured in place using absorbable sutures. The skin is loosely closed using interrupted sutures; a drain may be placed, depending on surgeon preference. Preoperative fecal diversion is controversial, especially since many tissue transfer repairs are performed for recurrent disease.

The Martius Flap

As originally described in 1921, the Martius flap involves the transposition of the bulbocavernosus muscle toward the urethral meatus to buttress a vesicovaginal or urethrovaginal fistula repair. Elkins' modification to this technique in 1990 transposes only the labial fat pad, thereby reducing both morbidity and operative time. ⁵² An elliptical incision is made over the labia majora. The skin and bulbocavernosus muscles are then separated from the labial fat pad, taking care to preserve the perineal branch of the pudendal artery running in the posterolateral pedicle. The fat pad is tunneled under the perineal skin and sutured to the apex of the rectovaginal septum that has been exposed via the transverse perineal dissection.

There are few papers reporting results of the Martius flap. A systematic review by Göttgens et al found seven papers with closure rates ranging from 65 to 100%. ²⁶ Songne et al reported on 14 RVF treated via the Martius flap. Four patients had a previous attempt at repair; 13 (93%) had resolution of symptoms at 3 months; the final woman required a second Martius' flap from the other side but did achieve complete healing. ⁵³ All women were diverted at the time of their initial operation and 12 were able to have their stoma taken down. Pitel et al reported the largest series of the Martius flap for RVF. After a mean follow-up of 29 months, 13 of 20 women (65%) had resolution of symptoms. ⁵⁴ No significant predictors were identified for flap failure. Although typically performed in isolation, at least one report of concomitant ERAF and the Martius flap did not demonstrate improved healing rates over ERAF alone. ⁵⁵ There are limited data regarding changes to quality of life after the Martius flap. Two small studies suggest some improvement in satisfaction with sexual activity after repair, despite several reports of increased dyspareunia. ^{55 56}

Gracilis Muscle Transposition

A pedicled gracilis muscle flap is an alternative to the labial fat pad. With the patient in lithotomy position, a longitudinal incision is made over the medial thigh, posterior to the saphenous vein. The gracilis muscle is freed to the level of the proximal neurovascular bundle. A second incision is made just above the knee to divide the distal tendinous insertion. Taking care to preserve the proximal pedicle, the muscle is transposed through a subcutaneous tunnel to the transverse perineal incision and sutured approximately 2 to 3 cm above the level of the fistula tract.

Studies reporting on gracilis flap are mostly small, single-institution series; since gracilis flap is rarely performed as the initial operation for RVF repair, the vast majority of gracilis patients have complex, recurrent fistulas. Two systematic reviews evaluated gracilis flap for RVF, including Göttgens et al who found seven papers reporting closure rates ranging from 43 to 100%. ²⁶ Hotouras et al identified 17 studies of 106 patients. Closure rates ranged from 33 to 100% but were typically between 59 and 70% among larger and higher quality studies. ⁵⁷ Wexner et al reported on 15 patients undergoing gracilis flap for RVF repair, the majority of which were due to Crohn's disease. All patients had a diverting stoma at the time of repair. After 3 months, three of the nine women with Crohn's disease (33%) and six of the eight women without (75%) had no evidence of fistula on anoscopy. ⁵⁸ Lefèvre et al performed gracilis flaps for RVF in eight women, again due mostly to Crohn's disease. All patients had a diverting stoma and at least one prior repair attempt. Six patients achieved RVF closure (75%) and five went on to stoma takedown. ⁵⁹ Park et al treated 11 patients with RVF with gracilis flap and report a closure rate of 73%. ⁶⁰ Only one study reported on sexual function with 50% of patients reporting dyspareunia and/or loss of libido. ⁶⁰

Biologic Mesh

Despite significant growth in the use of biologic mesh in RVF repair, evaluation of its efficacy remains difficult due to small case series with variation in surgical technique and implant specifications. Schwandner et al used porcine acellular dermal matrix as a part of ERAF repair in 21 patients with RVF, the majority of which represented recurrent disease. ⁶¹ After a mean follow-up of 12 months, 71% achieved closure. Ellis used a similar dermal implant via a transperineal approach and reported a closure rate of 81%. ³⁴ Göttgens et al implanted crosslinked porcine acellular dermal matrix via either a transperineal or transvaginal incision in 15 patients with RVF, achieving a 1-year closure rate of 64%. ⁶² In contrast, Mege et al reported only a 20% success rate using a different porcine acellular dermal matrix product and a transperineal approach. Interestingly, of the eight failures, three women had a recurrence of their RVF, while four women developed a new rectoperineal fistula and one woman developed a new vaginoperineal fistula. ⁶³

Defect above the Level of the Sphincter Complex, Defect in Sphincter Muscles

ERAF Plus Sphincteroplasty

Patients with a mid-to-high fistula tract and a defect in the anal sphincter muscles cannot be addressed by ERAF or a sphincteroplasty alone since ERAF does not correct the underlying muscle injury and sphincteroplasty is below the level of the septal defect. Combining these techniques can address both issues. Rothenberger et al report on 10 patients who underwent sphincteroplasty at the time of ERAF for concomitant incontinence with all 10 achieving satisfactory continence after the procedure. ³⁹ The overall RVF closure rate was 86%, although the authors do not report the rate for patients with and without sphincteroplasty. Similarly, an early series from Wise et al includes 15 women treated with ERAF and concomitant sphincteroplasty, 2 due to FI and 13 due to sphincter dysfunction on preoperative evaluation. All 15 women achieved return of anal continence. ⁴⁴ Khanduja et al evaluated 20 women who underwent concurrent sphincteroplasty from a larger series of ERAF patients; all achieved RVF closure. Fourteen (70%) reported perfect continence after a mean follow-up of 40 months, while four had incontinence to liquid stool and two had incontinence to gas. ⁶⁴ Lowry et al reported a 88% closure rate among 25 patients undergoing ERAF with concomitant sphincteroplasty versus 78% among 56 patients undergoing ERAF alone, although this did not reach statistical significance. ³⁷

Perineoproctotomy with Reconstruction

More extensive septal and sphincter injuries can be converted into a fourth-degree laceration and then rebuilt in layers. There is significant risk of FI since the anal sphincter muscles have to be divided to access the tract. For that reason, we do not recommend performing perineoproctotomy (PP, also known as episioproctotomy) in the absence of FI or a sphincter defect. The operation is typically performed in the prone position under general anesthesia. Perioperative antibiotics are given and a Foley catheter was placed. A fistula probe is placed into the tract and the tissue divided down to the probe using electrocautery. The ends of the divided sphincter muscle are identified and mobilized. The rectal wall is reapproximated using interrupted horizontal mattress sutures. ^{65 66} An overlapping sphincteroplasty is performed, then the vaginal wall is reapproximated and the skin loosely closed with interrupted sutures, leaving a small space and/or drain in the midline to prevent seroma formation.

Limited case series of PP exist, with the majority coming from a single specialized center with substantial experience. Hull et al compared 50 women undergoing PP to 37 undergoing ERAF for mostly obstetric-related RVF and reported 78% closure for PP compared with 62% for ERAF. ⁶⁵ Patients undergoing PP were more likely to have a history of obstetric injury (72 vs. 49%), episiotomy (71 vs. 38%), and diverting stomas (75 vs. 54%). Rates of FI dropped significantly among PP patients (50% preoperatively vs. 8% postoperatively), but not among patients undergoing ERAF. In addition, PP patients reported significantly better postoperative sexual function despite similar rates of dyspareunia. A separate analysis of PP from the same institution reported a 74% closure rate and significant improvement in FI scores but was unable to identify any patient- or surgeon-level predictors of recurrence. ⁶⁵ Mazier et al presented an early series of 28 patients undergoing PP but no closure rate was reported. None reported incontinence to stool, liquid, or gas after the procedure. ¹⁹

Complex

Recurrence

Given the fragility of the rectovaginal septum after injury and the substantial rate of surgical failure, surgeons should be prepared to deal with recurrence. Repeat evaluation is warranted to confirm recurrence, establish the anatomy of the tract (i.e., through the same defect at the same level or in a separate location), and evaluate surrounding tissues. Fluid collections should be drained to control local sepsis. Seton placement or even diverting stoma may be necessary, depending on location and quality of the tissue. Workup should rule out concomitant conditions prior to proceeding with recurrent surgical repair.

Reoperative RVF repair has an even lower rate of successful closure than initial repair. MacRae et al operated on 28 patients with at least one prior RVF repair using a variety of techniques. Overall, 71% of simple and 40% of complex fistula healed. ⁶⁷ Halverson et al also reported on recurrent RVF repairs in which 27 of 35 patients (82%) healed after a median of two operations. ⁶⁸ Two large studies also found a significant reduction in the likelihood of healing with each additional repair, with predicted closure rates dropping from 68% for patients with no prior repairs to 53% for patients with one to two prior repairs and 50% for patients with three or more. ^{22 23}

There are no clear data regarding how the initial choice of RVF repair technique affects closure rates for subsequent RVF repairs. As most obstetric RVFs occur at the level of the anal sphincter, we will briefly focus on the management of recurrence after ERAF and sphincteroplasty in more detail. Some have suggested that patients with prior failed ERAF may benefit from a different technique at the time of recurrent repair due to poor blood supply in the surrounding rectal wall; however, others found no evidence to recommend against repeat ERAF repair. ^{67 68} In their early study on ERAF, Lowry et al found only a slight attenuation in success between the first and second ERAF repair (88–85%), with a larger drop occurring at the time of the third attempt at repair (55%). ³⁷ Based on these results, we believe that it is acceptable to attempt a second ERAF repair after identifying and optimizing any factors that may have contributed to failure of the first repair such as nutrition, source control, medical comorbidities, and tension on the flap or poor vascularity of the flap. Recurrent repairs do not need to address a potential sphincter defect if the initial repair did not, unless new symptoms have developed or an injury occurred during the prior repair. Given the steep reduction in efficacy after a second ERAF repair, other options should be considered rather than a third ERAF procedure. The Martius or gracilis flap may help to bring in viable tissue to help bulk the rectovaginal septum. Although limited, there are reports of repeat Martius' flaps in the event of additional recurrence. ^{53 54}

Recurrence after sphincteroplasty appears to be less common than recurrence after ERAF and there are some data to support repeat attempt at sphincter repair. Tsang et al found a nonsignificant smaller decrease in closure rates from first to second repair for sphincteroplasty patients than for ERAF patients (80–75 vs. 45–25%). A recurrent tract through the muscle should be considered a technical failure and repeat sphincteroplasty or tissue transfer is warranted. On the other hand, recurrence through the lower perineal body without disruption of the sphincter complex can be treated with a lay open fistulotomy without affecting fecal continence. Patients with FI due to failed sphincter repair but no recurrent fistula could be considered for sacral neuromodulation, especially since newer data suggest an intact sphincter complex is not necessary to achieve functional improvement. ⁶⁹

Diversion

The need for fecal diversion in the setting of RVF repair remains a source of controversy. There are no randomized trials comparing closure rates with and without diversion and nonrandomized series suffer from significant selection bias. A small portion of patients, up to 15% in some studies, ²⁷ will achieve RVF closure through diversion alone, making it hard to separate the impact of the surgical repair from that of the diversion. One small randomized trial of fecal diversion at the time of sphincteroplasty for FI found no functional advantage to diversion. Incontinence scores improved equally across both groups and there was no difference in overall complications. ⁷⁰ One large study combining a variety of RVF repairs did find a higher closure rate among patients with a stoma (88–45%), ²⁷ although the majority of the literature found no benefit to stoma formation. ^{22 23 41 67 68} The decision to perform a diverting stoma, therefore, remains surgeon specific. Anecdotally, tissue transfer procedures appear to be more likely to be performed with diversion, although it is unclear whether this simply reflects higher rates of recurrent disease. Our practice has been to divert patients primarily for symptom relief rather than as a method of improving closure rates.

Conclusion

RVF remains a challenging clinical problem for colorectal surgeons and a significant detriment to the health of women. Preoperative workup should focus on identifying the level of the fistula, the presence of a sphincter defect, and any concomitant inflammatory or malignant disease. Repair should not be attempted until local sepsis has been controlled. There is no gold standard for RVF repair and all techniques have substantial failure rates. In general, obstetric fistulas tend to be low, at the level of the anal sphincter complex, and amenable to either ERAF or sphincteroplasty, depending on the status of the muscle. Injuries that extend higher into the rectovaginal septum often require additional tissue transfer or reconstruction of the septum and perineal body. Recurrent defects should be evaluated with the goal of determining the source of failure. Several local techniques, such as ERAF and sphincteroplasty, can be attempted a second time but repeated failure warrants more complex repair.