Transanal Irrigation for Neurogenic Bowel Disease, Low Anterior Resection Syndrome, Faecal Incontinence and Chronic Constipation: A Systematic Review

Transanal irrigation (TAI) has received increasing attention as a treatment option in patients with bowel dysfunction. This systematic review was conducted according to the PRISMA guidelines and evaluates the effect of TAI in neurogenic bowel dysfunction (NBD), low anterior resection syndrome (LARS), faecal incontinence (FI) and chronic constipation (CC). The primary outcome was the effect of TAI on bowel function. Secondary outcomes included details on TAI, quality of life (QoL), the discontinuation rate, adverse events, predictive factors for a successful outcome, and health economics. A systematic search for articles reporting original data on the effect of TAI on bowel function was performed, and 27 eligible studies including 1435 individuals were included. Three randomised controlled trials, one non-randomised trial, and 23 observational studies were included; 70% of the studies were assessed to be of excellent or good methodological quality. Results showed an improvement in bowel function among patients with NBD, LARS, FI, and CC with some studies showing improvement in QoL. However, discontinuation rates were high. Side effects were common, but equally prevalent among comparative treatments. No consistent predictive factors for a successful outcome were identified. Results from this review show that TAI improves bowel function and potentially QoL; however, evidence remains limited.


Introduction
Transanal irrigation (TAI) has received increasing attention as a treatment option in patients with bowel dysfunction as it has shown to improve faecal incontinence (FI) and chronic constipation (CC) [1,2]. With TAI, water is introduced into the bowel through the anus, facilitating emptying of the rectosigmoid and the left colon [3]. By performing regular irrigations, control of bowel function including time and place of bowel movements can be re-gained [4]. In patients with FI, efficient and controlled emptying of the bowel can be achieved with TAI. This can prevent episodes of incontinence in between irrigations for an average of two days. In patients with CC, regular evacuation of the rectosigmoid with TAI can prevent constipation [3].
TAI is introduced when conservative treatment fails. At present, TAI is the only minimally invasive treatment option for bowel dysfunction. This has positioned TAI as an important treatment modality before introducing more invasive methods such as sacral nerve stimulation, antegrade colonic irrigation or stoma formation [5].
FU ≥ 12 months, and mixed if patients with both short-term and long-term FU were included.

Search Strategy and Data Extraction
On October 15, 2020, the electronic databases PubMed, Embase, and Cochrane Library were systematically searched for relevant studies. The search strategy was developed by all authors in collaboration with a librarian with expertise in systematic reviews. The search was performed using relevant MeSH-or Emtree terms and text words. The search strategy is presented in Figure 1. Covidence was used for the removal of duplicate publications, article screening and data extraction [16], and Web of Science was used to screen references and citing articles of all included studies.  Two authors (H.Ø.K. and M.M.) independently extracted information on author, study design, study population and outcomes of interest using an electronic spreadsheet in Covidence. Any disagreements during the screening or data extraction process were solved by consensus discussions between H.Ø.K. and M.M. or by a third party (T.J., K.K. or P.C.).

Risk of Bias and Quality Assessment
The risk of bias was assessed using a modified version of the Downs and Black checklist [17]. The checklist is validated for both RCTs and non-randomised studies [17]. It comprises 27 items covering reporting, external and internal validity, and statistical power. In the present version, item 27 addressing statistical power was modified so that a study was given one point if a power calculation was conducted and zero if it was not. For each question, one point was awarded if the study fulfilled the question (item 5 ranges from 0-2 points). Hence, the maximum score for randomised trials was 28 and non-randomised studies 25. Studies were classified as being excellent (26)(27)(28), good (20)(21)(22)(23)(24)(25), fair (15)(16)(17)(18)(19) or poor (≤14) [18]. The assessment was independently performed by two reviewers (H.Ø.K. and M.M.). Disagreements were solved by consensus discussion between the two authors or by a third party (T.J.).

Data Synthesis
Results are presented separately for NBD, LARS, and FI and CC of heterogeneous origin. If data regarding NBD or LARS were separately presented in articles reporting data on FI and CC of heterogeneous origin, results were presented along with NBD or LARS results. Study and patient characteristics, details on TAI, primary and secondary outcomes, and quality assessment of each study are presented in tables and summarised descriptively. Due to the heterogeneity of outcomes and study designs, a meta-analysis was not conducted.

Results
In total, 1698 studies were identified through the database search. Another two studies were identified through the screening of references from the included studies. After the removal of 383 duplicates, the remaining 1317 studies were screened by title and abstract independently by two authors (H.M.L. and M.M.). As a result, 1151 studies were excluded, leaving 166 studies for full-text screening. Full-text screening was completed independently by two authors (H.M.L. and M.M.). Twenty-seven studies met the inclusion criteria. A flowchart of the screening process is presented in Figure 2.
Eight studies measuring pre-and posttreatment scores including patients with SCI, MS or SB showed a significant improvement in bowel function [7,[20][21][22][23][26][27][28]. One cross-sectional study reported a prevalence of severe NBD among TAI users of 41% and a proportion of 17% as being dissatisfied or very dissatisfied with TAI [25]. A retrospective study found a mean NBD score of 6.25 and a mean CCIS of 0.50 among current TAI users [24]. One study showed a successful outcome in all patients [19].
Five studies reported QoL data. Three studies used validated PROMs [7,26,27] and two studies non-validated PROMs [21,23]. Two studies measuring pre-and posttreatment scores including patients with MS measured generic QoL [26,27]. One study showed no significant difference in the Short Form (36) Health Survey (SF-36) scale scores [26,36] and the other no difference in the European Quality of Life-5 Dimension (EQ-5D) score [37], but a significant improvement in the European Quality of Life Visual Analogue Scale (EQ-VAS) score [27]. One study including patients with SCI measured disease-specific QoL using the American Society of Colon and Rectal Surgeons Faecal Incontinence Score (FIQLS) [7,32]. The study showed a significant difference in the coping/ehavior and embarrassment scales, but not in the lifestyle or depression/self-perception scales between patients treated with TAI and conservative treatment [7].
The discontinuation rate ranged between 3 and 66% [7,20,21,23,24,26,27]. Reported reasons for discontinuation were expulsions of the catheter, bursting of rectal balloons, time consumption, heavy administration, dislike of treatment, adverse events and inefficacy. Two studies systematically reported the frequency of side effects with a range between 29 and 36% of patients experiencing side effects [7,23], the most frequent of which were abdominal pain, sweating/hot flushes, general discomfort, headache and perianal/anorectal pain. No studies reported health-economic results; however, two studies showed a reduction in urinary tract infections requiring treatment and reduction in contacts with health care professionals [7,27].
Using a multivariable analysis, one study identified several factors associated with a positive outcome of individual bowel scores; however, no consistent factors were identified [20]. To identify predictive factors for a positive outcome, four studies compared the compliant group with the non-compliant group; one study showed a higher proportion of patients with tetraplegia and patients depending on help in the non-compliant group [23]; one showed a higher baseline CCIS, SF-36 score and maximum tolerated volume to rectal balloon distension in the compliant group; one showed that impaired anal electrosensitivity was predictive for a successful outcome [27]; and one found no significant difference between the groups [24].

Transanal Irrigation as Treatment for LARS
One RCT and four prospective cohort studies investigated TAI as a treatment for patients diagnosed with LARS [38][39][40][41][42]. Two studies hadshort FU [41,42], one had long FU [40], one had mixed FU [39] and one did not report any FU [38]. In total, 96 patients using TAI were included, with between 10 and 33 patients in each study. Four studies reported reasons for LARS, and the primary reason for LARS was resection for rectal cancer (89%) [39][40][41][42]. One study reported the operation type. In this study, 78% of patients had a total mesorectal excision [41]. Three studies were assessed to be of good methodological quality [40][41][42], one to be of fair methodological quality [39] and one to be of poor methodological quality [38].
Bowel function was assessed by validated PROMs in five studies [40][41][42][43][44] and by a non-validated PROM in one study [39]. One study used the William's Incontinence score [39,45], one the CCIS [36,37,40], one used the LARS score [46][47][48] and the Memorial Sloan Kettering Cancer Centre Bowel Function Instrument (MSKCC BFI) [41,49], and one the LARS score, the FIGS score and the obstructed defaecation syndrome (ODS) score [29,42,50]. QoL was assessed using the SF-36 in two studies [32,40,41] and in one study using the European Organisation for Research and Treatment of Cancer (EORTC-QLQ-C30) questionnaire [42,51].     Comparing pre-and post-treatment scores, all studies showed a significant improvement of bowel function. One study showed a significant improvement of the mental component of the SF-36 and a non-significant improvement in the physical component [32,40]. Another study showed an improvement in four (mental health, social functioning, role emotional, and bodily pain) of eight SF-36 scales [41]. One study using EORTC-QLQ-C30 showed an improvement in VAS scores of the Global health status domain [42].
The discontinuation rate ranged between 0 and 23% [39][40][41]. Reported reasons for discontinuation were time consumption, dislike of treatment, cancer recurrence, proctitis and pain during TAI. Two studies reported side effects with a range between 29 and 62% experiencing side effects [39,41] including abdominal cramps, minor rectal bleeding, leakage after irrigation, nausea and pain at insertion.
One study investigated predictive factors for a decrease in LARS score, but found none [41].

Transanal Irrigation as a Prophylactic Treatment for LARS
TAI compared to best supportive care as a prophylactic treatment for LARS immediately after ileostomy closure was investigated in an RCT with three months of FU [43]. Eighteen patients were randomised to TAI. One-year FU results were published later [44]. Patients were included if a low anterior resection for rectal cancer was performed. The studies were assessed to be of good methodological quality.
The irrigation volume during the trial was 1000 mL, and at 1-year FU the median (range) volume was 600 (200-1000) mL. During the trial, the median (range) irrigation time was 45 (30-60) min and all patients irrigated daily. At 1-year FU, irrigation was performed daily by 50% of patients. All patients self-administered TAI and were trained in TAI.
Bowel function was assessed by the number of defaecation episodes during the day and night and by the LARS score and the CCIS. QoL was assessed by the mental and physical components of the SF-36.
At 3 months of FU, the studies showed a significant difference between the groups in LARS score and CCIS, and in the number of defaecation episodes during the day and night. At 12 months of FU, a significant difference in the number of defaecation episodes during the day and night was observed, but no significant difference in the LARS score or CCIS was seen. At 3-and 12-months of FU, no significant difference in QoL measured by the SF-36 in patients using TAI compared with patients using best supportive treatment was observed.
After 3 months, 6% of patients had discontinued TAI; at the 1-year FU, 47% had discontinued. Among patients discontinuing at one year, 89% had discontinued because TAI was too time-consuming, and 11% had discontinued due to pain during irrigation.
Three prospective studies including patients with FI and CC of heterogeneous origin showed a significant improvement in bowel function with validated PROMs [55][56][57]. One of the studies showed significant improvement in QoL using the SF-36 [55] and the other an improvement in QoL on a non-validated 11-point Likert scale [57]. The last study showed no significant improvement in the FIQLS [56].
In the studies using non-validated PROMs to measure bowel dysfunction, one study reported an overall satisfaction with TAI of 73% [54], and one study showed a successful response to TAI in 50% of patients [53]. Using compliance as a success criterion, one retrospective study showed that 43% still irrigated at the 1-year FU. The study reporting data on only patients with FI used a non-validated measure and reported a successful outcome in 38% of patients [52].
In patients with chronic idiopathic constipation, overall satisfaction was reported in 67% of patients [60]. In patients with FI following sphincter damage after birth, no difference was seen when comparing the baseline and termination score [59].
Using a multivariate analysis, one study showed a significant association between satisfactory progress of the first training and TAI compliance [58]. A cross-sectional study showed higher satisfaction among younger adults <40 years [54]. One study found no association between incontinence score and anorectal physiology and a successful effect of TAI [53]. Another study found no correlation between baseline measures and duration of TAI treatment [60].

Discussion
Results from this review show that TAI is a beneficial treatment for both NBD, LARS, and FI and CC of heterogeneous origin with some studies reporting improvement in disease-specific and generic QoL. With few exceptions, the studies in this review have used TAI as second-line treatment when conservative treatment has failed. Therefore, results from this review mainly evaluate effects on bowel function among patients not responding to conservative treatment, i.e., patients with potentially more severe bowel dysfunction.
Change in bowel function and QoL was primarily measured with PROMs. PROMs allow for the evaluation of patients' perspectives on functionality and QoL [62] and have gained acceptance within this research field. The use of validated instruments has previously been identified as a limitation in TAI research [12]. Overall, 67% of the included studies used at least one validated bowel-specific PROM. However, 82% of studies published within the last ten years used validated measures, showing that this limitation is no longer prominent. Nine different PROMs were used to evaluate bowel function, and this inconsistency of outcome measures compromises comparability. Numerous bowel function measures exist, which have been developed and validated differently. The NBD score and the LARS score have been developed and validated to evaluate bowel function based on a correlation with QoL, whereas the CCCS and FIGS are correlated to physiological or clinical assessment. Consensus regarding core outcome measures would ensure comparability in future research.
Half of the studies measured QoL by generic and/or disease-specific QoL measures. Three studies used a disease-specific QoL measure [7,40,56] and two of these showed improvement [7,40]. Although the NBD and LARS scores are not QoL measures, their items correlate with an impact on QoL. The reported improvement of these scores in many of the included studies could therefore suggest an improvement in disease-specific QoL. Some studies showed improvement in generic QoL measured with SF36, EQ-5D, or EORTC-QLQ-C30 [27,[40][41][42]55], while other studies showed no significant change [26,43,44,56]. Two of the studies showing no improvement in generic QoL used TAI as a prophylactic rather than a symptomatic treatment [43,44]. Four studies used non-validated questions to measure QoL; three studies showed significant improvement in QoL [21,23,57]. The wording or themes explored by generic QoL instruments might be insensitive to changes in QoL resulting from an improvement in bowel function. We encourage research into generic QoL instruments sensitive to changes in bowel function that allow for a subjective valuation of the aspects of QoL that are most important to the individual patient.
Results show a high discontinuation rate at the 1-year FU of 19 to 57%, and several studies have based effect analyses solely on patients still performing irrigation at FU. Irrigation is known to be time-consuming and may involve practical difficulties. In order to overcome these challenges, patients have to experience a beneficial effect to continue the use of TAI [12]. Therefore, many studies consider the continuation of TAI as a successful outcome, and the high discontinuation rates in the studies included in this review suggest that TAI is beneficial only for a selected group of patients.
To predict a successful outcome and target the introduction of TAI to patients most likely to benefit from treatment, predictors of discontinuation have been studied. The studies included in this review reported no consistent predictive factors for a successful outcome. Using a multivariate analysis, Bildstein et al. found the progress of the first training to be a predictive factor for a successful outcome [58]. Almost all included studies in the present review reported that patients received TAI training prior to initiation, stressing that training is considered as an important part of the process. However, it is not evident which parameters the training comprises. In our clinic, all patients are taught irrigation by a specialised nurse, and the first irrigation performed by the patient or a caregiver is carried out under supervision at the clinic. In our experience, adequate training and patient support are important factors for patient compliance. Findings in this review partially support this; however, this must be further explored in future studies. Typically, clinical factors or basic demographic variables have been studied, such as age and sex, level of injury in SCI, mobility, tumour characteristics, stoma details, anorectal physiology, baseline bowel function and QoL scores. However, a successful outcome of TAI may also depend on personal characteristics such as the psychological profile and compliance with other treatment and hospital FU [5]. Future research should be directed towards better phenotyping TAI candidates. Among possible predicting factors for a successful outcome, socio-economic factors or personality traits should also be included.
Three of the major reasons for discontinuation identified through this review were technical problems, inefficacy and TAI being too timeconsuming. The primary technical problems reported were expulsion of the catheter, bursting of rectal balloons, and leakage around the catheter. Interestingly, technical problems were not reported as a reason for discontinuation amongst patients with LARS. Possible explanations might be the absence of a hyperreflective rectum in patients with LARS, which is seen in patients with NBD and can complicate rectal installation [63], or that data on technical problems was not reported.
Side effects were systematically reported in eight studies [7,23,39,40,55,57,58,60]. For NBD, side effects were reported to be experienced by 29 to 36% of patients, while this ranged between 29 and 62% for LARS and 22 and 59% for FI and CC of heterogeneous origin. There was no difference in the type of side effects reported among the different conditions. The most frequent side effects were abdominal cramps/pain, anorectal pain, nausea, sweating/hot flushes, minor bleeding and leakage of irrigation fluid. Christensen et al. reported no significant difference in the proportion of patients experiencing side effects during or immediately after TAI when comparing patients treated with TAI and those treated with conservative treatment [7]. This suggests that the side effects are not related to TAI, but to NBD itself. In SCI, autonomic dysreflexia during and after defaecation is even less pronounced when using TAI than with the usual digital manoeuvres to facilitate bowel emptying [64]. However, this finding has not been investigated for the LARS, FI or CC of heterogeneous origin. Only one study reported three serious adverse events, with no serious outcome [7], implying that such events are rare with the use of TAI. Bowel perforation is a potential risk related to TAI, and the risk has been reported to be 1 per 50,000 irrigations [65]. None of the included studies reported bowel perforations.
There are limitations to the included studies. So far, no RCTs have been conducted supporting the treatment of TAI compared with optimal conservative treatment in patients suffering from LARS, MS, FI or CC of other origin, and the risk of confounding as well as publication bias is known to be higher in non-randomised studies. FU varied between the studies, with the majority of studies having short FU time. Furthermore, conclusions may be limited by the fact that only a few studies have made power calculations, and the sample sizes of the included studies are generally modest, which may introduce type 2 errors. Generally, external validation was assessed to be of good quality in most studies; however, the modest sample size might indicate selection bias in the recruitment of patients. Systematic inclusion methods in prospective studies in the future could strengthen the evidence.
Another limitation is that many of the studies only included patients in their analysis who were still irrigating at FU. Therefore, the results primarily reflect improvements in a selected cohort. Future studies should include both intention-to-treat and per-protocol analysis. This is not necessarily a limitation; however, it should be taking into consideration when introducing TAI to patients. Since no consistent predictors supporting which patients could benefit from TAI have been identified until now, this selection process is difficult for the clinician. Therefore, a trial-and-error strategy for the introduction of TAI with focus on an individualised course of treatment has been suggested [5]. TAI is often combined with conservative modalities to optimize treatment; however, the majority of studies do not report concomitant treatment. Reporting of concomitant conservative modalities could help clinicians to optimize treatment. Another limitation to the studies is the missing reporting of clinical significance, and future studies should report results in a manner allowing for this to be assessed.
Limitations to this systematic review include a potential risk of publication bias if studies investigating TAI that found no significant results were not published. Inclusion criteria were restricted to the English language, which could have excluded relevant articles. In some early studies, different terms have been used for TAI -for example, wash-out-which were not included in the search. This may be a limitation to our search. However, we consider our search using irrigation sufficient as recent literature has used the terms TAI and rectal irrigation, which would have been included in our search. Furthermore, the literature search was limited to three databases, and additional eligible studies might have been identified through other databases.

Conclusions
Results from this review show that TAI improves bowel function and potentially improves QoL among patients with NBD, LARS, and FI and CC of heterogeneous origin; however, the evidence remains limited. Until now, the highest evidence of TAI improving bowel function and QoL is from three RCTs showing superiority of TAI over best supportive care [7,43] and TAI as more efficient than tibial nerve stimulation [42] In NBD, the majority of the evidence is for patients with SCI, MS or SB. A high discontinuation rate calls for improved patient selection to TAI. However, no consistent predictive factors for a successful outcome have been identified. In order to identify patients benefiting from TAI, a trialand-error approach may be used to assess if patients benefit from treatment. To optimize the possibility of a successful outcome of TAI treatment, it is important to conduct a personalised treatment course with supervision from specialised health-care personnel and to monitor outcomes of TAI.