This was a retrospective cohort study using routine data abstracted from medical records from three fistula repair hospitals supported by the FC+ Project in DRC between January 1, 2017, and December 31, 2019.

The study population consisted of women who had undergone surgical repair of urogenital fistula in sites supported by the FC+ project in DRC between January 2017 and December 2019. We limited the analytic sample to only those women who achieved successful fistula closure during repair (n = 718; Figure 1).

Variables related to the study's objectives were extracted from patients' medical records at fistula repair sites, using questionnaires developed using KoBoCollect software.

The dependent variable was the occurrence of residual incontinence in women with closed fistula after repair surgery. The independent variables included sociodemographic characteristics (age at admission, marital status, level of education, occupation, and area of residence), clinical characteristics (number of incontinence operations, cause and treatment of incontinence and the operating technique used, among others), parity, and length of stay. Fistula repair was called successful when the woman had a fistula closed and continent at the time of the assessment.

The data collected from patient records were analyzed using Stata 16 software (Stata Corporation, College Station, TX, USA). We described patient characteristics through calculating frequencies (%) and means (with standard deviation). Bivariate analyses were performed to explore factors associated with residual incontinence after fistula repair. Student's t-test was used to test for differences in continuous variables (age, number of previous surgeries, parity and length of stay or duration of catheterization) and Chi-square test for the remaining, categorical variables. Multivariable logistic regression was used to calculate unadjusted and adjusted odds ratios. The variables which had a p ≤ 0.20 in bivariate analyses were included in the logistic regression model. Differences were considered statistically significant where p ≤ 0.05.

The Ethics Committee of the School of Public Health of Kinshasa approved the study protocol (ESP/CE/153/2020).

A total of 718 women had successful fistula closure at hospital discharge. Figure 1 shows the selection process and criteria for our sample selection. The sociodemographic, clinical, and gynecological-obstetric characteristics of women included within our analytic sample are presented in Table 2 Patients' mean age was 35 ± 0.5 years. The majority of the women were married or in union (58.7%), housewives or farmers (68.5%), and lived in rural areas (82.6%). More than one-third (35.5%) reported no formal education. The average parity among women was four children per woman (4 ± 3 children).

Of the 632 women for whom information was available on delivery mode for the delivery resulting in the fistula, 368 (58.2%) delivered by cesarean section. Most women (46.2%) developed fistula due to cesarean section and prolonged labor (30.1%), and 270 women (37.6%) had a fistula duration of more than 4 years prior to repair. Vesicovaginal fistula was the most common type of fistula (77.4%) and 44.6% of women had a fistula size between 1.5 and 3 cm. Among included women, 308 women (42.9%) had any previous fistula repair surgeries. In terms of case management, the vaginal route was the primary surgical approach for 515 women (71.7%), and average length of stay was 23 days in hospital (Table 2).

Overall, 31 of 718 women (4.3%) had residual incontinence (Table 3). Among these 31 women, information on surgical care was available for 26. 11 of 26 women had incontinence after the first fistula repair and 10 after at least two surgical repairs. The main causes of incontinence identified were urethral incontinence (six cases), short urethra (six cases), severe fibrosis (three cases) and micro-bladder (two cases). Information of incontinence cause was missing for nine women. As treatment, 10 of these 26 women underwent perineal muscle rehabilitation, nine underwent reconstructive surgery, and information was missing for seven (Table 3).

The frequency of residual incontinence was higher among women who underwent repair at the Heal Africa (18 cases; 8.6%) and St. Joseph's (8 cases; 3.7%) sites compared to the Panzi site (5 cases; 1.7%) (Figure 2).

In bivariate analyses, parity, site, previous surgical repair, mode of delivery, number of prior surgeries, surgical route and length of hospital stay were significantly associated with residual urine incontinence after fistula repair (Table 4). However, in the multivariate logistic regression (Table 4), after adjusting for other variables, only the fistula repair site, number of previous surgeries, and surgical route remained independently associated with residual incontinence. Compared to women undergoing care at Panzi Hospital, those operated on by Heal Africa faced an over forty-fold increased adjusted odds of residual incontinence (aOR: 43.57; 95% CI: 4.26–445.26).

We found that the likelihood of developing residual incontinence was three times higher among women who had any previous fistula repair surgery than those who had not (AOR: 3.17, 95% CI: 1.10–9.14). Compared to women who had an abdominal surgical route of repair, those with a vaginal surgical route of repair were about seven times more likely to experience residual incontinence (AOR: 6.78, 95%CI: 1.02–45.21).

We found 4.3% of women experienced residual urinary incontinence, which is substantially lower than rates reported elsewhere in sub-Saharan Africa which range from 5 to 71% (18, 25, 35–38). This could be explained by the level of experience of surgical teams at these sites, as well as the FC+ project's support through technical assistance, particularly surgical training and equipment provision.

We found that the frequency of residual incontinence varied according to the fistula repair sites, with residual incontinence less frequent among women operated on at the Panzi site than at the other two sites, despite adjustment for several factors reflecting fistula severity. A contrary result was reported by Delamou et al. (18), who found that the surgical site was not associated with residual incontinence in women (18). In our context, the differing expertise and the number of surgeons across the different sites might explain such results.

In the literature, the extent of urethral injury, tissue loss of the bladder neck/posterior urethra, overlying pubo-cervical fascia, circumferential fistulae and severe fibrosis have been reported as anatomical causes related to the occurrence of residual incontinence in women after repair (39). In our study, residual incontinence occurred among women with anatomical risk factors including urethral gap, short urethra, and severe fibrosis. We also found that having any previous fistula repair surgery was associated with a three-fold increase in residual incontinence. Similar results were reported by Kimassoum et al. (19), who found that out of 84 patients, 36 women experienced residual incontinence after the second fistula repair and 25 women experienced it after three or more fistula repairs (19). The findings also showed a significant relationship between vaginal surgical route and residual incontinence after repair. Further investigation will be needed to document better this phenomenon and guide practitioners on appropriate intervention among incontinent women after surgical fistula repair.

We found no sociodemographic characteristics to be associated with residual incontinence after repair; this finding is consistent with what has already been reported in the literature (18, 24, 25, 29, 40).

Furthermore, parity and mode of fistula causative delivery were not independently associated with residual urinary incontinence after fistula closure. Other studies have reported an association between parity and residual incontinence in women after repair (35, 41, 42). Finally, we did not find an association between certain fistula characteristics, including fistula size and duration of fistula, with residual incontinence in women after fistula closure. Fistula size was not independently predictive of residual incontinence in women after fistula closure in the published literature (18, 25, 43, 44).

Our study is one of the first to examine the factors associated with residual incontinence after fistula closure in women treated in the three main repair sites supported by FC+ in DRC. All three sites are referral hospitals specialized in fistula treatment with experience in both routine care and outreach. The three sites are assisted by the same external development partner for provision of training, equipment and technical support. The nearly equal distribution of the sample across the three included facilities is also an asset to the study's sample, representing the geographic diversity of repair sites: the St. Joseph site in the west, while Panzi and Heal Africa are located in eastern DRC. Furthermore, outreach repair campaigns are conducted by all the sites, allowing for full national coverage, especially those areas where women are most vulnerable.

The findings of this study should be interpreted with caution given some limitations. Due to the retrospective nature of our study, we are missing data on certain women's characteristics. We were also limited in the selection of variables which could be included within our analysis, and it is possible that not all confounding factors were examined or controlled for. For instance, we could not account for the cause of the fistula even though Cesarean section (CS) was reported as the leading cause of fistula occurrence. Yet, we acknowledge the fact that some women who had a prolonged labor or obstructed labor underwent CS and subsequently had a fistula. The fact that this information was not recorded in the registers might have overestimated CS related causes. This represents a limitation of the study which relied on routine data and calls for the systematic inclusion and collection of such information in medical records.

Additionally, the small number of sites did not allow us to explore the influence of site-level variables; for example, the number of qualified surgeons was not spread evenly across the three repair sites, which may have been an important factor responsible for differences in outcomes across sites. The number of women with residual incontinence is also small, which limited the statistical power for certain comparisons. Finally, the diagnosis of residual incontinence was not possible in patients who were lost to follow-up, i.e., could not show up for their 3-months post-repair visit.