PATIENTS AND METHODS
- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
Patients who had had a V VF and UVF repaired between 2000 and 2006 were identified from our theatre database, and their case-notes were reviewed retrospectively. Data were collected on patient characteristics, fistula aetiology, presentation and diagnosis, fistula position and size, and the surgical intervention and outcome.
All patients had an examination under anaesthesia and a cysto-urethrogram performed (a cysto-urethrogram is not taken). The surgery was performed by one of two consultant surgeons with a special interest in reconstructive surgery. If accessible, fistula repair was attempted using a vaginal approach. The fistula tract was excised in all cases, and in all but two cases a Martius fat pad  was harvested from the labia to use as an interposition graft. In cases where the vaginal access was deemed insufficient, an abdominal repair was used, as originally described by O’Conor et al.. When possible, either omentum or peritoneum was brought to the pelvis as an interposition graft between the closure layers. Patients with vaginal repairs were discharged within 48 h, and abdominal repairs within 5 days. In both the vaginal and abdominal approach catheters were left in situ for 3 weeks. A cysto-urethrogram was performed to confirm healing.
Patients with objective and/or subjective reports of continued leakage were reassessed (including video-urodynamics). Patients were followed up in the clinic at 3- and 6-month intervals.
Continuous data were compared using t-tests, and categorical data with Fisher’s exact and chi-squared tests; in all tests, P < 0.05 was considered to indicate statistical significance.
- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
In all, 41 women were identified in the 6-year period, all tertiary referrals from other institutions; 32 had a V VF and nine a UVF. The median (range) age was 45 (24–67) years and the mean wait between initial presentation locally and repair at our centre was 9 (3–96) months. The cause of the fistulae and previous attempts at closure are shown in Table 1. Most V VF (56%) occurred after hysterectomy, and most UVF (56%) after placing a suburethral sling or tape. In the two women with obstetric fistulae, the fistula had resulted from protracted labour, in Africa. Diagnosis was confirmed in 90% of the women by examination/cystoscopy under anaesthesia alone. Four patients had been referred with long-standing voiding dysfunction and leakage after multiple anti-incontinence procedures. In these cases the diagnosis was initially unclear but was confirmed after a three-swab methylene blue test.
Table 1. The causes of fistula
|Aetiology||No.||Fistula>3 cm||Previous repairs attempted elsewhere|
| Total abdominal hysterectomy||13|| ||4|
| Emergency Caesarean section||2|| || |
| Total abdominal hysterectomy||3||3|| |
| Emergency Caesarean section||1||1||1|
| + hysterectomy for bleeding||2||2||1 (two attempts)|
|Anti-incontinence procedures†||4|| || |
| Cystoplasty||2|| || |
| Bladder neck closure||1|| || |
| ‘Take-down’ colposuspension||1|| || |
| Cervical cancer||3|| || |
| Endometrial cancer||2|| || |
| Infected urethral diverticulectomy||2|| || |
|Anti-incontinence procedures†||7|| ||1|
| Bladder neck (autologous) sling||3|| || |
| Transvaginal tape||2|| || |
| Urethropexy||2|| || |
Four patients were deemed unsalvageable; three had fixed rigid urethras and a ‘frozen pelvis’ secondary to radiation up to 10 years before presentation. One patient had a completely cleaved bladder neck and urethra after obstructed labour. These patients were treated by supravesical diversion according to patient choice; two with bladder neck closure and a Mitrofanoff channel, one with an ileocolic neobladder and Mitrofanoff channel, and one with an ileal conduit.
The 37 remaining patients had 47 procedures at our centre to repair their fistulae; 23 transvaginal and 24 transabdominal (nine procedures had been performed before referral). The success rates are shown in Table 2. The fistula was closed in 34 of 37 (92%) patients (26 of 28 V VF and eight of nine UVF). Twenty patients with a V VF (71%) required one operation, five (19%) required a second procedure (of whom four had a previous attempt elsewhere) and one (4%) required a third intervention. One patient with a large V VF secondary to radiotherapy was deemed unsalvageable after an initial failed attempt at repair, and subsequently had a conduit constructed. One remaining patient awaits a second attempt at repair (having had one previous failed attempt before referral). Seven patients with UVF had a successful repair in one procedure (one having had a previous failure elsewhere), and one required two procedures, the second using Martius interposition. One remaining patient with a complex urological history and after a urethral tape UVF had three failed attempts at closure and was eventually managed successfully with the formation of a Mitrofanoff channel.
Table 2. The overall VVF and UVF cure rates
| ||Overall cure rates, n/N (%) |
|After 1st repair||20/28 (71)||7/9 (78)|
|After 2nd repair*||5/6*||1/2|
|After 3rd repair**||1/1*||0|
|Total attempts at repair||35||12|
|Total success||26/28 (94)||8/9 (89)|
Complications were confined to those having V VF repair. One patient required antibiotic therapy for a wound infection after transabdominal repair, two were treated for UTIs and one developed (mild) stress urinary incontinence, which did not require surgical intervention.
Patient age (P = 0.21), menopausal status (P = 0.11), body mass index (P = 0.45), American Society of Anesthesiology grade (P = 0.51), and duration between fistula onset and repair (P = 0.28) were not significant factors in the outcome of V VF and UVF. Fistula aetiology and complexity of urological history (after a single procedure or multiple/complex interventions) were similar between those patients having a successful repair and those having persistent fistulae (P > 0.05). The results according to surgical approach are shown in Table 3. Similar surgical steps were described in successful and unsuccessful procedures. Whilst the cure rates after vaginal V VF repair (all 11 cured) were higher than those after abdominal V VF repair (18 of 24 cured) this was not statistically significant (P = 0.13). The route of previous failed repair did not influence the method of subsequent repair, unsuccessful transabdominal repair being just as likely to be repaired successfully via a vaginal or abdominal approach. All of the primary and secondary V VF repair failures were for fistulae of >3 cm, whilst only two of the cured patients had fistulae of similar size (P = 0.02). In five of the six failed cases, omentum was not available for interposition. In two of these cases peritoneum was used as an alternative interposition tissue, but in the remaining three cases no tissue was available, and the defect was closed primarily (in three layers) with no interposition. Peritoneum was the only available tissue for interposition in five patients, of whom three had a successful outcome. The one patient in whom failure occurred in the presence of omental interposition was a large (4 cm) fistula after radiotherapy, with significant ischaemic bladder changes, who was subsequently felt to be unsalveagable and diverted. Thus, 15 of 16 (94%) V VF repairs using omental interposition were successful, but the success rate decreased to 37.5% if peritoneal or no interposition was possible (P = 0.002).
Table 3. Success rates by surgical approach, as n/N (%) or n/N
| omental interposition||15/16 (94)||–|
| peritoneal interposition||3/5||–|
| no tissue interposition||0/3||–|
| Total||18/24 (75)||–|
| Martius fat interposition||10/10||6/8*|
| rectus fascia||–||0/1|
| no tissue interposition||1/1||2/3|
- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
The V VF was closed in 94% and the UVF closed in eight of nine patients. There was no significant difference in outcome according to the route of V VF repair, although vaginal V VF repair were uniformly successful, whilst abdominal V VF repair failed in a quarter of the patients. Failure of V VF repair was associated with a large fistula (>3 cm) and absence of omental interposition. UVF was more successful with Martius fat pad interposition (six of eight) than with no interposition (two of three), although this was not statistically significant.
Most fistulae in Westernised practice result from gynaecological intervention, particularly abdominal hysterectomy, with estimated rates of 0.3–2%, but was the cause for only 39% of the present patients. The other aetiologies are rare and reflect the tertiary nature of these referrals and the full range of causes; emergency Caesarean section in 12%, radiotherapy for pelvic malignancy in 12%, multiple complex anti-incontinence interventions in 27% and infection in 5%. By contrast, in a large series from Sloan-Kettering and the Mayo Clinic, fistulae were caused by gynaecological intervention in 91% and 82%, obstetric procedures in 8% (Mayo), radiation in 4% and 6%, and other causes in 5% and 4%, respectively [6,7]. The two remaining cases in the present series resulted from obstructed labour in Africa, a substantial issue for third-world healthcare (occurring in three of every 1000 deliveries in rural communities), but rarely seen in modern obstetric practice .
The ideal timing of intervention is still debated . For cases in which tissue is otherwise healthy, early vaginal repair within 2–3 weeks of injury is possible without increased morbidity or failure rates [7,10]. Others advocate a wait of 8–12 weeks before repair [11,12]. All of the present patients were referred >6 weeks after injury and early repair was not an option; we would be prepared to do so for simple fistulae.
The approach to repair was determined by patient and surgical factors. For fistulae secondary to abdominal hysterectomy, the injury is most commonly caused by tear or de-vascularization of the posterior bladder wall during mobilization of the vagina. The injury is supratrigonal and accessible via a vaginal approach in most cases [6,7,13]. Vaginal surgery avoids the morbidity of a further laparotomy and provides a more rapid recovery for the patient. In 71% of the present patients injured after hysterectomy, this proved to be possible and the repair was successful in all these patients via a vaginal approach. In the remaining cases a vaginal approach was not considered appropriate. The fistulae were either high and/or large (>3 cm), and/or close to the ureteric orifice, and/or concomitant hysterectomy or augmentation cystoplasty (three cases) were required. Consequently, an abdominal approach was used more often (51%) than reported in other series (10–20%) [5,6]. Recently, laparoscopic and robot-assisted laparoscopic V VF repair was reported [14,15]. As most simple fistulae can be repaired vaginally with minimal morbidity and high success rates, a laparoscopic approach would only be indicated for those V VF requiring abdominal repair. It remains to be seen whether the success rates of open abdominal repair can be replicated with laparoscopic techniques for these often complex fistulae.
Patient factors do not appear to significantly determine the outcome. Previous (vaginal or abdominal) operative intervention did not influence the success of further attempts, a similar finding to that of Ayed et al., but conflicting with the reports of others, who have found reducing success rates with increasing number of repairs . Although failure was confined to the abdominal approach, this is probably related to case selection, with larger fistulae approached from above. Other series have shown equivalent results from either approach, with success rates consistently reported to be >90% for both vaginal and abdominal repair [6,7,18].
The major determinants of success in the present series were fistula size (>3 cm) and tissue interposition. The relative importance of these two factors is difficult to ascertain. All failures had extensive defects and limited tissue to interpose. Two other fistulae with large defects were primarily closed with omental interposition, and three smaller fistulae were closed with peritoneal interposition. In other series, fistula size has been an important determinant of failure, with authors’ advocating fistula of >1–3 cm as an indication for tissue interposition [7,16,19,20]. Most authors emphasize the importance of quality tissue interposition [5,16,20,21], although good results from transvaginal repair of simple fistulae with no tissue interposition have been reported . The nature of the tissue interposed varies; whilst peritoneal interposition has been successfully described in transvaginal repairs , we had difficulty harvesting sufficient flaps (transabdominally) for five patients with larger defects. Omental or Martius interposition were successful in all but two cases (after radiotherapy, and the patient with multiple sequential failures), and it is our practice to use a vascularized fat pedicle in both vaginal and abdominal repairs, whenever possible. In complex and radiation-related fistulae, where tissues are poorly vascularized and omentum/peritoneum might not be readily available, alternatives need to be considered. Only one of the five radiation-related V VF was successfully repaired (a fistula of >3 cm using omental interposition) with three being diverted as primary procedures and the other after a single failed attempt. In retrospect this should have been the initial management for this fistula too, but the patient was keen to have one attempt at closure. In future, we plan to explore the use of muscle flaps (rectus from above, gracilis from below [20,22]), as well as bowel segments  and/or ovary, which we have used successfully in some cases of uro-rectal fistulae repair.
This study has several limitations. By reviewing the outcomes of a consecutive group of ‘all-comers’, this series represents a heterogeneous population of patients, with a significant proportion of complex fistulae. Nevertheless, we consider that this group represents a true depiction of ‘real life’ experience from a tertiary unit in current practice. We analysed the factors important to success, and confirmed that they are consistent across fistulae from widely different aetiologies, and similar to those reported from more selected cohorts. Retrospective analysis is often limited by data that were not collected at the time. No information was available on quality of life, urinary and sexual function before and after fistula repair. We intend to address these specific issues in future prospective reports.
In conclusion, this report comprised a heterogeneous population of patients referred to a tertiary centre with V VF or UVF. Despite the varied causes, the repair was still successful in 94% of V VF and eight of nine UVF in which repair was considered possible, but more than one procedure was required in 29% and two of nine, respectively. Failure was significantly more likely in large (>3 cm) V VF in which omental interposition was not possible. No significant factors for UVF repair were identified. The importance of good quality tissue interposition, especially for those with more extensive defects, was apparent.