Background

An estimated 780,000 people in the USA are living with Crohn’s disease (CD), and the incidence of the condition is increasing rapidly worldwide [1, 2]. In addition to the debilitating effects of intestinal inflammation, approximately 35% of patients with CD develop fistulas [3]—abnormal connections between the gastrointestinal tract and an epithelial-lined surface such as skin (external or cutaneous fistulas) or internal organs; the peritoneal space, retroperitoneal areas, or the thorax (internal fistulas) [4]. Fistulas involving the vagina, such as rectovaginal fistulas (RVFs) and anovaginal fistulas (AVFs), can be particularly upsetting and embarrassing for women, with the most common symptoms including passage of gas and/or stool via the vagina [5, 6].

Fistulizing CD is characterized by variable clinical presentations [3], but patients with RVF almost always have severe discomfort and pain, and may suffer psychological effects, including anxiety or poor body image due to malodorous drainage fluid [4, 6, 7]. For many, the natural history includes frequent recurrence after treatment and long episodes of actively draining fistulas [3]. Treatment for fistulas can be complex, including multiple pharmacological and surgical treatments, hospitalizations, and medical visits that result in high healthcare costs [6,7,8,9,10,11].

Despite RVF and AVF being debilitating conditions, there is limited observational literature on the epidemiology and disease burden of these conditions. We performed a systematic literature review (SLR) to summarize available information and highlight knowledge gaps regarding the disease burden for patients with CD-related RVF and AVF. The objective of this SLR was to identify evidence relating to incidence and prevalence, pharmacological and surgical interventions, pre-specified clinical outcomes (e.g., healing and response rates), patient-reported outcomes (PROs), and healthcare resource utilization (HCRU) including costs in observational studies of populations with RVF and AVF.

Methods

Search strategy and selection criteria

The SLR was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [12]. The protocol for this study was registered with the International Prospective Register of Systematic Reviews (PROSPERO; registration number CRD42020177732).

Eligibility criteria were organized around the elements of Population, Intervention, Comparison, Outcomes, Time, and Study Design (PICOTS) and are included in Additional file 1: Table S1. The electronic search was conducted on March 25, 2020, utilizing a search strategy incorporating terms for fistula type (RVF and AVF), study design, and outcomes of interest in PubMed and Embase (Additional file 1: Table S2 and Additional file 1: Table S3). Search results were filtered to include only English language (PubMed, Embase) and human studies (PubMed), and limit the publication time frame to 10 years prior to the search date. To ensure relevant studies were not missed in the electronic search, a manual search of key publications and references was conducted.

The titles and abstracts for studies identified through the search were independently screened by two reviewers trained in epidemiology and the conduct of SLRs in order to determine whether they met the PICOTS criteria. Studies meeting the PICOTS criteria were carried forward to the full-text review phase, where the full-text articles were independently assessed by each reviewer to determine final eligibility for data abstraction. Any discrepancies in either step were resolved by consensus, and if consensus could not be achieved, a third trained reviewer made the determination. Data from each eligible study were independently abstracted by two reviewers using a standardized data abstraction form. Both reviewers then jointly examined the abstraction spreadsheets in order to synthesize the abstracted data into one master spreadsheet. Data were extracted for a range of variables, including study type and design, population, outcomes, and limitations.

Studies included in this article met the following criteria: (1) reported on RVF or AVF, (2) used an observational study design (i.e., case–control, cohort/registry, or cross-sectional study), (3) measured one of the outcomes of interest—incidence/prevalence, treatment patterns, clinical outcomes (healing/failure/recurrence rates, post-operative infection), PROs (Crohn’s Disease Activity Index [CDAI], Inflammatory Bowel Disease Questionnaire [IBDQ], Perianal Disease Activity Index [PDAI], 5-dimensional EuroQoL questionnaire [EQ-5D], fecal incontinence, pain, discharge/soiling, pad use, alterations in or dissatisfaction with sexual intercourse/activity), and HCRU/costs, and (4) were original research (Additional file 1: Table S1). Case series were designated as cohort studies only if they met all of the following pre-specified criteria: > 10 patients per fistula type, patients sampled based on their exposure only (not outcome), outcome assessed over a pre-specified follow-up period or mean/median follow-up reported, information available to calculate the absolute/relative risk. In addition, sampling had to be labeled as ‘consecutive’ or text had to indicate that all eligible patients were included to avoid selection of unique cases. Further, AVF and RVF were to be reported separately from other types of fistula.

Risk of bias assessment

Two independent reviewers evaluated risk of bias in each article using the Risk Of Bias In Non-randomized Studies of Interventions (ROBINS-I) tool for observational studies [13]. The ROBINS-I tool allows investigators to rate the risk of bias in non-randomized studies in a systematic way through the evaluation of seven pre-specified domains of bias (confounding, selection of participants, classification of interventions, deviations from intended interventions, missing data, measurement of outcomes, and selection of reported result). Any disagreements were settled by consensus. If consensus could not be achieved, a third trained reviewer made the final determination.

Results

Systematic literature review

Figure 1 displays the PRISMA diagram. The electronic search returned 514 articles; an additional 68 were identified from a manual search of other sources; 121 duplicates were deleted. Of the 461 records screened based on their titles and abstracts, 316 were included in the full-text assessment. Of those, 149 were excluded based on the inclusion/exclusion criteria (Fig. 1). A total of 16 studies were included in the qualitative synthesis for RVF and AVF: 14 addressing RVF only, 1 addressing AVF only, and 1 addressing both RVF and AVF (combined) (Table 1).

Fig. 1
figure 1

PRISMA flow diagram. AVF anovaginal fistula, PRISMA Preferred Reporting Items for Systematic Reviews and Meta-Analyses, RVF rectovaginal fistula

Full size image
Table 1 Characteristics of papers included in the SLR (n = 16 studies)
Full size table

Most studies (n = 14) were retrospective cohort studies or case series that met the review definition for cohort studies [9, 10, 14,15,16,17,18,19,20,21,22,23,24,25]. The SLR also included one retrospective, unmatched case–control study [26] and one prospective cohort/registry study [27]. Two of the retrospective studies [14, 19] included prospective follow-up data collection (e.g., via telephone). All but two of the included reports [25, 27] came from either single- or multi-site, clinic-based studies, most often from surgical centers.

Two of the papers were identified as having a low risk of bias, and 10 papers had a moderate risk of bias. Serious or critical risk of bias was identified in three papers [19, 20, 26]. Risk of bias could not be determined in one source owing to lack of information in a published poster abstract [18] (Table 1).

Incidence and prevalence

Three of the 16 included papers provided population-based estimates of RVF incidence or prevalence (Table 2). A study of a population-based inflammatory bowel disease cohort in the well-defined South Limburg area of the Netherlands from 1991‒2014 reported that 17 of 728 female patients with CD (2.3%) had RVF [27]. A large, claims-based study conducted in the Truven Health MarketScan database by Schwartz et al. (2019) identified cases of CD with codes for fistulizing disease and estimated that > 6000 women were affected by RVF in the USA [25]. It should be noted that this study had a limited follow-up period and used International Classification of Diseases codes that may have poor validity in this area.

Table 2 Incidence and prevalence of Crohn’s-related RVFs and AVFs: key findings and commentary (n = 3 studies)
Full size table

Schwartz et al. (2019) conducted their database analysis to support findings from an SLR. Their SLR reported only one study which the authors used to estimate the prevalence of patients with CD who had 1 (65.4%), 2 (19.2%), 3 (8.2%), 4 (4.5%), and 5 (2.7%) episodes of Crohn’s-related RVF. The median duration of fistula episodes for patients with 1, 2, 3, 4, and 5 episodes was 2.2, 7.1, 12.1, 17.1, and 22.0 years, respectively. The weighted average of medians of the duration of fistulizing CD was 5.1 years [25].

A study using data from the Rochester Epidemiology Project found that 3.1% of patients (13/414) diagnosed with CD between 1970 and 2010 had at least one RVF or AVF episode [21]. The South Limburg study estimated the overall cumulative probability of developing an RVF among female patients as 0.7% after 1 year, 1.7% after 5 years, and 3.1% after 10 years from CD diagnosis. The cumulative 10-year probability of developing an RVF among female patients with CD was 1.7% for patients diagnosed with CD between 1999 and 2011 (down from 5.7% for diagnosis between 1991 and 1998, which is prior to the introduction of anti-tumor necrosis factor agents [anti-TNFs]) [27].

No studies of AVF incidence or prevalence were identified. However, it should be noted that the terminology around fistula classification is not completely standardized. Of specific relevance to the current SLR is the interchangeable use of the terms AVF and ‘low RVF’ [28]. Studies do not always specify whether the RVFs described in their results include ‘low RVF’ (i.e., AVF).

Treatment patterns

Treatment patterns broadly refers to surgical procedures, medications, and conservative treatment including dietary modifications. It is known that RVF and AVF require substantial treatment; however, our systematic review did not identify any population-based studies addressing treatment patterns in these conditions. The current SLR includes 12 (non-population-based) studies that address treatment patterns by fistula type.

The sheer number of preceding interventions and repeat interventions per patient described in the included studies, along with ineffective treatments leading patients to try alternative treatments and procedures, are indicative of a high burden of disease. It is important to note that many of the studies identified through this SLR are clinic or hospital based and provide a snapshot of treatment patterns at that institution at the time of study. Similarly, clinic and hospital studies generally have limited numbers of patients and aim to compare one intervention versus another intervention, rather than providing a broad picture of treatment patterns for all patients with the condition of interest. While the focus of this article is the patient burden of RVF/AVF, the underlying burden of CD management including medications and surgeries is substantial.

Seven of 12 treatment pattern studies reported that patients had or required additional procedures before and/or after the intervention of interest, demonstrating a substantial treatment burden. It was frequently reported that patients had prior surgeries, such as anorectal surgery and fistula repair. For example, in a hospital-based study of 51 consecutive patients with CD who were undergoing treatment for RVF during 1998–2005, 40% of patients had previous anorectal surgery for CD. The median number of previous RVF surgical repairs in the group was 2–3, depending on the intervention group [15].

Six studies reported patients having other prior surgical interventions by fistula type, although some studies did not indicate whether the procedures were for the treatment of fistulas or the underlying CD. Procedures included seton drainage, diverting stoma creation, fistula plug, flap repair, fistulotomy, fistulectomy, fibrin glue, and sphincteroplasty [14,15,16, 19, 23, 26]. It should be noted that there is no single standard surgery for patients with AVF/RVF. Surgery of choice is dependent on location of fistula, severity, prior surgeries, degree of incontinence, and the surgeon’s clinical assessment and views on specific techniques (see Table 3).

Table 3 Studies providing information on treatment patterns (n = 12 studies)
Full size table

Another important aspect of treatment burden is the need for additional surgeries following the interventions of interest in the published studies. For example, 14 of the 51 patients (27%) described in the study above [15] eventually required proctectomies. In another study, seven of 21 patients with AVF who underwent transanal endorectal advancement flap repair received a second flap, three received a third flap, and one was diverted [16].

In addition to the surgical and procedural burden, this SLR indicated that patients with RVF/AVF report a heavy medication burden for both CD and fistula. For example, 94% of patients (48/51) who underwent treatment for RVF during 1998–2005 had received previous medication therapy, though it is unclear whether this was therapy for CD or fistula, specifically [15]. Seven of the 12 studies reporting on treatment patterns provided details on prior use of medications to manage CD and/or fistula. In a hospital-based study of 65 women who underwent surgery to close a RVF, 40% (26/65) had taken immunomodulators and 30.8% (20/65) had taken steroids within 3 months prior to surgery [14]. Reported medications include anti-TNF biologics, corticosteroids, azathioprine, methotrexate, 6 mercaptopurine, and antibiotics [9, 14,15,16,17, 19, 26, 27] (Table 3).

In addition to these surgical and medical treatments, other treatments for CD and RVF/AVF may include conservative management techniques, such as local wound debridement, a low residue diet, and sitz baths, although the data for these approaches are not enumerated in the literature [20].

Clinical outcomes

Eleven studies included data on clinical outcomes for treatments of RVF/AVF (Table 4). The variability in treatments, study design, and description of outcomes further demonstrates the complexity of the clinical situation. As with the classification of the fistula overall (e.g., AVF vs ‘low RVF’), investigators use varying terminology, with or without clear definitions, to describe outcomes of interest (e.g., healing, closure, response).

Table 4 Interventions and success and failure rates in published studies (n = 11 studies)
Full size table

Nevertheless, most studies include some assessment of success of the surgical procedure. For example, Haennig et al. (2015) identified the median ‘interval to fistula closure’ after seton drainage and infliximab treatment in 12 patients with RVF as 30.6 months [10]. Milito et al. (2019) measured median time to ‘complete healing’ in 43 patients with RVF as 6 months (range: 2–11 months) [18]. Other studies measured the rate of closure or healing, some by surgical type and some across surgical types. Of the nine studies that reported healing/success/closure across multiple surgical types, rates varied from 14.4 to 81% [9, 18], with seven ranging between 50 and 75% [14,15,16,17,18,19, 23]. Some of the variation may be explained by differences in study design, population characteristics, and surgical types included.

‘Recurrence rates’ were specifically reported in two studies and ranged from 13 (complicated recurrence due to development of multiple perianal fistulas with severe sepsis) to 55.8% across multiple procedures [22, 24]. Five studies reported post-operative infection rates; however, two [16, 22] did not report rates by fistula type. In the three studies that did report RVF/AVF-specific rates, 1–13% of patients experienced a post-operative infection, including one abscess [17], one urinary tract infection [19], and two cases of severe sepsis [24]. Further complicating interpretation of these results is the variance in median or mean follow-up duration which ranged from 13 months [24] to 7 years [16].

Patient-reported outcomes

One of the 16 included studies offered findings collected through PRO instruments. El-Gazzaz et al. (2010) analyzed quality of life (QoL) data from the 12-item Short-Form Health Survey, Fecal Incontinence Quality of Life (FIQL), and Female Sexual Function Index (FSFI) questionnaires administered at surgical follow-up visits. The authors report ‘modest’ scores in the PRO instruments, with no significant difference between healed and unhealed women. For example, patients’ mean scores on the FIQL ranged from 2.5 to 3.1 (out of 5, with lower scores indicating lower QoL) in each of the scored domains (lifestyle, coping, depression, and embarrassment). FSFI total scores averaged 17.3 ± 6.7 and 17.9 ± 9.4 (from a possible total of 36, with lower scores indicating worse functioning) in healed and unhealed women, respectively. The surveys showed no statistically significant differences in QoL or sexual function among healed versus unhealed patients, and the authors suggest this may be due to the underlying effects of CD regardless of its complications [14]. This study is limited by the questionnaire completion rate among patients (45%) and reporting bias(es) regarding potential reticence of patients to discuss the sensitive nature of sexual health topics. Although a logical component of the disease burden in this population and one that may be captured in studies of CD overall, little is known about sexual interest and satisfaction among women with RVF/AVF. Similarly, more studies are needed to determine overall QoL and other insights into the patient experience that could be captured uniquely through PRO instruments.

Healthcare resource utilization

None of the 16 studies reported HCRU among patients by fistula type. More studies are needed to determine the direct and indirect costs of RVF and AVF, particularly as they relate to healthcare visits, copays, prescriptions, ancillary care such as psychological support, missed days at work and/or school, and productivity loss.

Discussion

CD-related RVF and AVF are rare and devastating complications of a life-altering and debilitating disease. The current systematic literature review highlights knowledge gaps regarding the disease burden for patients with CD-related RVF and AVF. Many women experience painful and embarrassing symptoms and must endure numerous medical and procedural interventions with limited hope for efficacy. Despite the immense disease burden experienced by these women, the literature of observational nature offers only a limited view of the treatment patterns, clinical outcomes, and PROs of CD-related RVF and AVF. After a systematic search, we found 16 studies that met the a priori criteria for inclusion and were qualitatively synthesized to characterize and quantify the global epidemiological burden of non-perianal CD-related RVF, and only one described AVF. Ten of the 16 studies were carried out in the USA; the six remaining studies were conducted in Europe.

We found that very few population-based epidemiology studies have been published; however, one Netherlands-based study estimates the prevalence of RVF to be 2.3% among female patients with CD. A notable gap in the literature is the lack of precise definitions and standardized use of terminology throughout studies. The addition of more robust, population-based studies and registry studies would contribute to the current knowledge gap around the epidemiology of RVF and AVF globally and could be a potential topic for further research.

We also found that treatment patterns vary across the literature and are not well documented. Many studies were conducted in single centers and are limited by small sample size, retrospective study designs, and other factors that affect generalizability to broader populations. However, it is clear from the studies identified in this SLR that patients with RVF and AVF undergo myriad procedures both operative and non-operative, and also carry a heavy medication burden. Importantly, women with RVF and AVF experience frequent recurrence and subsequently undergo repeat procedures with variable success. Similarly, the SLR did not report robust literature on the effectiveness of these treatments in alleviating fistula symptoms or on the use of conservative management strategies (e.g., sitz baths and dietary approaches). Overall, there is a lack of standardized definitions or criteria to assess remission status, thus making it difficult to compare and make generalizable claims.

The range of techniques and multidisciplinary approach to address RVF and AVF is a function of the conditions’ distinct etiologies and the progressive nature of the underlying [28]. The American Society of Colon and Rectal Surgeons (ASCRS) provides a set of guidelines for the care of patients with RVF (Table 5). The literature does not provide reliable information about the efficacy of surgical treatments, and there is no clear treatment pattern for women with these conditions. The need for more robust population-based research and studies evaluating clinical effectiveness and comparative effectiveness across multiple therapeutic approaches is highlighted by the lack of high-quality evidence to support ASCRS treatment guidelines.

Table 5 ASCRS treatment guidelines for RVF
Full size table

It is the study authors’ opinion that additional knowledge gaps could be filled via studies that include population-based data from the USA and other countries, evaluate optimal induction and maintenance therapy, and incorporate additional PROs relating to QoL (including instruments such as the Wexner Incontinence Score and the FSFI questionnaire) [29, 30]. A review of data published since the search date uncovered an additional study by Seifarth et al. [31], which found a healing rate of 31.3% in patients with Crohn’s-related RVF after advancement flap placement. Although this demonstrates that research is ongoing in this area, data remains limited. The literature would benefit from the long-term follow-up of a longitudinal cohort of women with Crohn’s-related RVF or AVF to provide insights into patient outcomes well beyond the studied interventions.

Additionally, beyond the surgeries discussed in this review, it is important to note that other treatment methods, such as stem cell transplantation, have shown some degree of efficacy. A recent meta-analysis of Crohn’s fistula patients treated with stem cell transplantation utilized data from 7 studies of RVF and found a healing rate of 27.2% [32].

This study has several notable strengths. This is the first systematic review to identify knowledge gaps regarding the disease burden for patients with CD-related RVF and AVF. To accomplish this, a complete and thorough literature search with explicit eligibility criteria was undertaken across several databases, with articles subsequently screened for inclusion in this review. Selection, data extraction, and adjudication of risk of bias were done by two independent reviewers. Additional strengths of the current SLR include its compliance with established guidelines for SLRs, including the use of a pre-specified protocol and search criteria. The protocol was registered with PROSPERO to promote transparency and allow for future replication or updates. There are also a few limitation that should be noted. The SLR is limited by the overall lack of published data on Crohn’s-related RVF and AVF. Although the current search was designed to capture a wide range of literature, it is limited to recent publications in the last 10 years, and publications in English. Consequently, the resulting SLR might not be representative of the full body of published literature. Many of the studies were clinic-based, with short follow-up periods, and/or had small sample sizes, all of which limit generalizability. Furthermore, information on this topic area is only available where studies are published, so there is limited generalizability to populations where no data have been reported. Also, publication bias may have impacted the pool of available studies. Lastly, this SLR was designed to assess real-world evidence and therefore does not include data or follow-up from clinical trials. Although clinical trials include additional information on clinical outcomes and PROs, they are typically limited by short follow-up durations, strict inclusion criteria, and the inability to assess real-world outcomes. Despite these limitations, this SLR provides a unique summary of available data and highlights evidence gaps that can be addressed with further research.

Conclusion

This systematic review reports a heavy patient burden among women with Crohn’s-related RVF and AVF in terms of both symptoms and medical and surgical treatment. Substantial gaps in knowledge surrounding Crohn’s-related RVF and AVF remain and more observational research is needed to support professional treatment guidelines with high-quality evidence.