Fecal incontinence (FI) is a common condition affecting up to 15% of community-dwellers.1 It is socially disabling, and although clinical severity correlates with impaired QoL,2 it has recently been illustrated that the most important independent factor impacting QoL is gender.3
FI occurs when one or more structures and functions involved in continence are disrupted, and when the remaining mechanisms are unable to compensate. Anal sphincter weakness––due to direct muscular damage or neuropathic injury, rectal sensitivity impairment, and loose fecal consistency4 are some of the most common factors involved, and these frequently overlap. In a recent study we confirmed that, besides mechanical external anal sphincter (EAS) and internal anal sphincter dysfunctions due mainly to obstetric factors, previously unexplored impaired and delayed conduction through sensory anorectal pathways and secondary reduced cortical activation in response to electrical stimulation are very prevalent pathophysiological factors associated with FI in women.5 In effect the proportion of patients with FI with an afferent/sensory dysfunction was much higher than in the case of those with peripheral pudendal motor conduction impairments, which is a well-known pathophysiological mechanism for FI.5
FI treatment remains a clinical challenge. Surgical procedures have quite unpredictable long-term results in the long-term, and therefore conservative rehabilitation treatments are recommended as a first line treatment. These include Kegel exercises (K), biofeedback (BF), endoanal or endovaginal electrostimulation (ES), or tibial nerve neuromodulation (NM) through percutaneous or transcutaneous electrodes (tNM). Although their use is common, there is a lack of evidence about their clinical efficacy, and the physiological targets they affect are not fully understood, often resulting in an empirical choice of treatment.6,7 Pelvic floor exercises or K, although a commonly recommended treatment for FI patients, have demonstrated their utility as a coadjutant treatment only, and not as a stand-alone one.8 Regarding BF therapy, this is considered a first-line treatment.9 However, studies which have tried to assess its efficacy do not show consistent results,10 and at least 1 study showed no advantage of BF over K.11 Regarding ES, significant differences vs controls have not been found, but a recent study12 claimed that a combination of BF + ES was more effective than monotherapy; overall however, there is not enough evidence to judge the efficacy of ES or to select the right patients for this treatment.13 Regarding peripheral NM, 2 systematic reviews concluded that NM can be an effective therapy for FI,14 but a recent multi-center study showed discouraging results.15 Overall there is a high degree of variability in study design and results, and furthermore, there is considerable heterogeneity in the translation of these results into clinical practice.
Additionally, the effect of these treatments on anorectal physiology is even less clearly understood. K and BF focus on the mechanical action of the pelvic floor muscles, but a classic study16 found no changes in mean resting pressure (MRP) or squeeze pressure (SP) in patients treated with BF; according to others, BF can indeed improve MRP,17 SP,18-20 as well as endurance of SP (ESP), recto-anal coordination, and rectal perception.18 With regard to ES, it has been suggested that it has an effect not only on slow and fast-twitch fibers of the muscle, but also on the very same neural pathways.13,21 Regarding NM, the action mechanism is little known, but it has been strongly suggested that it affects both motor and sensorial functions.22 Overall, the efficacy of the treatments is probably highly dependent on the pathophysiology underlying FI, hence the discordant results, so it is key on one hand to understand the pathophysiology, and on the other, to understand the action mechanisms of the treatments.
The main aim is to compare the clinical efficacy of 4 rehabilitation techniques (K, BF, ES, and tNM) in reducing the severity of FI symptoms in community-dwelling women and to explore their influence on anorectal physiology at a biomechanical and, secondarily at a neurophysiological level. We hypothesize that BF, ES, and tNM are superior treatments for FI compared to K, and that they affect specific aspects of impaired anorectal physiology.
Patients attending the gastrointestinal physiology unit from February 2013 to March 2017 with a history of more than 6 months of FI symptoms were consecutively screened. Patients with mild FI (Cleveland < 4), under 18 years of age, and those unable to follow the treatment properly were excluded. The study was conducted according to the principles laid down in the Declaration of Helsinki, and obtained Ethics Committee approval (code: 67/12). Written informed consent was obtained from the participants.
This is a randomized controlled trial (RCT) with a parallel design (TrialsGov: NCT03394794) with 4 treatment arms: K (base treatment, control group), BF + K, ES + K, and tNM + K, with a calculated sample of 45 patients in each group. Sample size was calculated to assess changes in severity (primary endpoint) of an active treatment compared to control (K) with Granmo software version 7.12 (IMIM, Barcelona, Spain) assuming an alpha risk of 0.05 and a beta risk of 0.20, a SD of 3.2 (for Cleveland score) and a minimum expected difference of 2. A simple randomization method was performed using QuickCalcs (GraphPad Software, San Diego, USA). The study had 2 stages: an initial 3-month stage with the patients receiving the assigned treatment to assess its efficacy (effect on clinical severity and QoL) and its effect on physiology, and a second 3-month stage, continuing only with K as a maintenance treatment to evaluate the persistence of the effects of the treatment (clinical severity and QoL) (Fig. 1).
Primary endpoint was the change before and after treatments in the severity score (Cleveland score); secondary outcomes were the effects on the anorectal physiology and neurophysiology.
Demographics and complete medical histories, comorbidities, risk factors, and medications were collected on Visit 1.
FI clinical severity was assessed using the Cleveland Incontinence Score,23 fecal consistency with the Bristol Stool Chart,24 and severity of urinary incontinence (UI) with the International Consultation on Incontinence Questionnaire (ICIQ) score.25 FI-related QoL and general health-related QoL were assessed using Fecal Incontinence Quality of Life (FIQL) score2 and EQ-5D,26 respectively.
A Hitachi ultrasound (endoanal ultrasonography [EUS]) with an endoanal 10-MHz transducer (Hitachi Medical Corporation, Tokyo, Japan) was used to assess internal/external sphincters and puborectalis anatomic integrity.5
Anorectal function was studied using conventional anorectal manometry to measure MRP, SP, and ESP. To assess ESP, 4 common pressure profiles were identified when patients held SP for 10 seconds.5 Cough reflex, sensory thresholds, and rectoanal inhibitory reflex (RAIR) were also assessed. Data was acquired using a Polygraph ID (Medtronic, Minneapolis, MN, USA). All these procedures are fully described previously,5 and are those recommended by the Spanish Motility Group.
In a randomized subsample of 10 patients from each therapeutic group, the terminal efferent pathway was also studied with pudendal nerve terminal motor latency (PNTML) (Neuromodule, MMS, Enschede, the Netherlands) on both sides.5 Metrics recorded and analyzed were latency of the compound muscle action potential (CMAP), amplitude, duration, and area under the curve of the CMAP.27
In a second randomized subsample of 10 patients from each therapeutic group, sensory evoked potential (SEP) to anal (ASEP) and rectal (RSEP) electrical stimulation were studied. Anal and rectal mucosa were stimulated using a customized probe. Electroencephalographic signals were analyzed and SEP obtained in Cz (at the vertex scalp) for being the cortical site of maximal representation after anorectal stimulation. Latencies of the peaks p1, n1, p2, and n2 were analyzed, as well as amplitudes of p1-n1, n1-p2, and p2-n2.5
Group 1. K (control): Patients were given oral and written instructions on how to perform K at home. They had to exercise for 10 minutes 3 times a day for a 3-month period. The exercises included maximal fast and sustained squeeze exercises (Supplementary Fig. 1).
Group 2. BF + K: In addition to K, patients received six 45-minute BF sessions administered by a specialist nurse (Supplementary Fig. 2). BF training was focused on the strengthening of the EAS muscle28 and the coordination of EAS contraction with rectal distention. Sensory training was not performed. Patients laid down looking at a monitor that mirrored the tracings of a manometric BF unit. The type of exercises was the same as K.
Group 3. ES + K: In addition to K, patients were instructed on the home use of an electric stimulation unit (Elpha 3000 Conti; Danmeter A/S, Odense, Denmark) with a “Periform+” endovaginal probe (Neen Healthcare, Dereham, UK). The stimulator was to be used for 30 minutes a day, 5 days a week, set at a frequency of 35 Hz, pulse-width of 300 microseconds with cycles of 0.5-second ramp-up, 5 seconds on, 0.5-second ramp-down, and 5 seconds off. Patients were told to increase intensity until reaching their tolerance threshold.29
Group 4. tNM + K: In addition to K, patients were instructed on the home use of a neuromodulation unit (Elpha 3000 Conti; Danmeter A/S) provided with self-adhesive transcutaneous electrodes (one above tibial malleolus and 2 cm from the medial line, and a second one below the malleolus), and used for 30 minutes a day, 5 days a week; at 20 Hz with a pulse width of 200 microseconds. Stimulation was increased until a motor response of the first toe was observed, this being the therapeutic intensity for the patient.30
Continuous variables were expressed as means (SD) and compared with
Patients were divided into responders and non-responders to the treatments, the first being those where severity changed from Cleveland ≥ 9 to < 9 post treatment (those with Cleveland < 9 prior to treatment were not taken into account). The reason for this cut-off point for the assessment of the response to treatment derives from previous studies that demonstrated that in patients with Cleveland < 9 the impact on QoL was very low, compared with patients with Cleveland ≥ 9.2,32 To facilitate comparison with results from other studies, we also analyzed the results by considering as responders those patients with a reduction of ≥ 50% in the clinical severity of FI (Table).
One hundred and eighty patients were included in the study and randomly assigned to one of the treatment groups (Fig. 2). The mean age was 61.09 ± 12.17 years. Complettion of the study was 36 patients in the K, 36 in the BF + K, 39 in the ES + K, and 40 in the tNM + K group. Drop out was due to discomfort, an inability to self-administer the treatments, or the fact that patients simply did not return despite being sent reminders.
The full clinical characteristics and pathophysiology of FI in these patients have been previously published (Supplementary Tables 1 and 2).5 Most patients had a history of obstetric risk factors, and pathophysiology was mainly associated with mechanical sphincter dysfunctions, most showing urge FI (83.1%; from these, 44.7% also with passive FI), mainly due to EAS weakness because of muscular sphincteric damage and, to a lesser degree, to sphincter denervation. Passive FI alone was present in 17%. On the sensorial side, more than a quarter had rectal sensitivity impairments (mainly hyposensitivity), and moreover, impaired conduction through afferent anorectal pathways was also very prevalent. Clinical severity was moderate to severe (Cleveland 11.55 ± 3.80).
Cleveland score showed a significant reduction in clinical severity of 31.68% in the K group, and more than 40.00% in the BF + K (41.70%), ES + K (46.20%), and tNM + K (47.50%) groups, although there was no statistical difference between them. Data is shown in Table. The FIQL score showed that QoL improved significantly in all 4 dimensions (life style, coping, embarrassment, and depression) for patients from all treatment groups. There were no statistical differences between the groups. Regarding general health-related QoL, EQ5D showed no differences in any of the score categories after any particular treatment. Transforming qualitative data from the questionnaire into a summary index (0-1), the health status of patients in the BF + K and ES + K groups improved significantly (Table).
According to our definition of responders as those achieving a Cleveland score < 9, the response rate was 38.50%, 46.70%, 65.60%, and 56.30% for, K, BF + K, ES + K, and NM + K, respectively (difference between K and ES is significant,
In the 3-month follow-up period after the treatments in which all patients continued to perform Kegel exercises only, the observed positive effect of treatments on clinical severity was sustained in all cases; BF + K patients further improved their symptoms. Regarding QoL, post-treatment effects were also steadily maintained during the entire follow-up phase (Fig. 3).
There was a statistically significant decrease in the ICIQ score at 3 months post-treatment in patients with UI treated with K, ES + K and tNM + K (Supplementary Table 5). These results were maintained at the 3-month follow-up stage.
Fecal consistency was evaluated before, during, and after the treatment period. We observed that the percentage of patients in the tNM + K group with very loose stools (Bristol 6 and 7) diminished after the 3-month active treatment, that is, fecal consistency changed to Bristol ≤ 5 (32.7% pre-treatment and 7.5% post-treatment;
The rehabilitation treatments assessed in this study for FI in community-dwelling women (K, BF + K, ES + K, and tNM + K) showed strong and similar therapeutic efficacy in terms of clinical severity and QoL. However, they affect specific and distinct pathophysiological mechanisms. Previous studies have tried to assess the therapeutic efficacy of individual treatments compared to control groups,8,10,11,13,15,19,20,33 with varying results, but to our knowledge no study has compared the efficacy of these 4 therapies and their effects on so many different aspects of anorectal physiology.
All the treatments for FI strongly improved clinical severity, especially in the ES + K and tNM + K groups compared to control group (K), achieving a decrease of almost 50% in the Cleveland score. This data is similar to that obtained by Healy et al34 using ES, and Boyle et al35 using NM (percutaneous); there are few studies of tNM, and one30 showed almost no improvement. Regarding BF, the improvement in our patients was more pronounced than in most other studies.8,11,36 QoL improved for all patients regardless of the treatment applied; FI-related QoL is associated not only with the severity of FI, but also with previous mood disorders, age and gender,3 which implies that in the management of FI consideration of the patient’s individual characteristics must be included. The 3-month follow-up post-treatment using only K in all groups shows that both the improvement in clinical severity and the recovery of QoL achieved during each treatment were maintained. So, although K proved to be less effective during the treatment period, it would appear reasonable to recommend it as a maintenance therapy during follow up.
In patients treated with K and ES + K, MRP increased minimally, probably insufficient to counterbalance FI symptoms associated with EAI impairment. The small increments in MRP observed may be attributable to an improvement in EAS tonic contraction and their modest contribution to the anal canal resting pressure. In contrast, the effect of BF, ES, and tNM on the voluntary squeeze (striated muscle) was of much greater magnitude allowing patients to delay defecation for longer, and minimize urge-FI symptoms. K exercises only did not cause the expected changes in EAS contractility, probably because patients may not have been fully aware of which muscle group they were exercising, hence the importance of monitoring with BF. The ineffective ESP profiles prior to treatment changed to a much better sustained pressure in a significant percentage of patients treated with BF and with ES, suggesting that both treatments are able to increase slow-twitch fiber contractility.
The improvements in SP observed in patients treated using tNM cannot be attributed to a direct effect on sphincters, but to an effect outside the anorectal region37; patients treated with tNM showed a tendency to shorter PNTML, suggesting that a higher proportion of muscle fibers in the EAS is activated after a potential re-innervation. On the other hand, through the stimulation of afferent pathways,38 tNM may be able to induce changes that occur in CNS and, in turn, improve motor control of the pelvic muscles. To sum up, BF and ES would be useful techniques to help patients with EAS insufficiency and intact efferent pathways, while tNM seems a reasonable treatment for those with pudendal motor nerve impairment, although more research needs to be carried out.
Sensorial perception in the rectum and anal canal is decisive in preserving continence. BF and ES improved the first rectal perception threshold, the most objective sensorial landmark measured with anorectal manometry.39 While BF promotes immediate sensorial learning generated by the motor task, direct activation of sensory axons by ES21,34 probably induces more complex central learning mechanisms leading to proprioception. Patients treated with tNM also showed improvement in the first perception (near to statistical significance), the results being in line with those obtained in previous studies with SNS,40 changes that must also be linked to neuroplasticity phenomena.
From the neurophysiological perspective, after treatment with ES, patients exhibited a shortening of the latency of p1 of the ASEP; this finding suggests an influence of ES on afferent pathways, probably resulting in better integration of sensory inputs and control of the pelvic floor muscles.
In the tNM group the percentage of patients with loose stools (Bristol > 5) diminished very significantly post-treatment, suggesting that tNM can modulate colonic activity41 and that the change in fecal consistency could explain in part the improvement in many patients with FI. Likewise, is also remarkable that patients after treatment with tNM required smaller volumes of distension to initiate RAIR suggesting an influence on the mechanisms that regulate rectal compliance.
As expected, patients also suffering from UI and treated with K, ES + K, and tNM + K experienced a significant improvement in the severity of the clinical symptoms of UI. It is noteworthy that patients treated solely with K improved UI symptoms but not FI, which reinforces the argument that some treated with this therapy for FI probably focus on the contractile activity on urogenital muscles other than the EAS and the puborectalis.
In order to identify which patients responded to a particular treatment, the rule of a decrease in symptoms of more than 50% is commonly used. This criterion, in our opinion, implies the acceptance of decreases in symptoms after a treatment that may be no more than marginal. Changes from grave clinical severity before treatment (Cleveland ≥ 9) to mild or moderate severity (Cleveland < 9) as a positive response seems to us more reasonable as the impact on QoL in one range or another is highly significant.2 In these terms, the percentage of responders was higher for ES + K (65.5%) and tNM + K (56.3%) than for the other 2 treatments.
Regardless of the treatment assigned, only a small percentage of patients with rectal hypersensitivity (20.0%), loose stools (26.7%), or passive FI (29.4%) responded to treatment. The latter was an independent factor associated with a negative response to the treatment. Rectal hypersensitivity and passive FI are challenging conditions that require further research, as no current treatment appears to really mitigate the symptoms. On the other hand, the influence of fecal consistency in response of the treatment is critical,4 being essential to treat diarrhea before attempting any rehabilitation.
Our study has some limitations. One is the absence of a sham control groups. Kegel was intended to act as control but it does not omit the therapeutic element of the research, and bias remain possible because of the unblinded design. Another is the limited sample of patients in each group, which has prevented the clarification of certain aspects, especially those regarding the study of neurophysiology. In RSEP recording, it was difficult to ensure intrarectal probe positioning—although contact with mucosa was confirmed—in order to ensure the repeatability of the procedures. In addition, we could not include the FI episodes as a primary outcome, because data was not strictly reliable.
To conclude, our research highlights the strong therapeutic effect of the therapies studied in those patients that complete the protocol treatment, and their distinct pathophysiological targets. This will help us to define a multimodal algorithm for treatment with distinct pathophysiological phenotypes of patients with FI that need a specific treatment and approach, and should begin with the treatment of diarrhea, if present. Future studies should assess the clinical therapeutic efficacy of multimodal treatment algorithms to further improve patients’ symptoms, and help them to regain QoL and return to their normal way of life.
All authors thank Dr Mateu Serra and Elisabet Palomera for their statistical support; Jane Perkins for assistance with English; Daniel Alvarez for his help with some of the images; and Jordi Roig from DPM for his support on Elpha stimulators.
Part of this research was funded through 2 PERIS grants from the Catalonian Health Department (SLT002/16/00214 and SLT008/18/00168). CIBERehd is funded by Instituto de Salud Carlos III, Barcelona, Spain.
Study conception and design: Lluís Mundet, Pere Clavé, and Laia Rofes; data collection and analysis: Lluís Mundet, Omar Ortega, Christopher Cabib and Laia Rofes; writing of the manuscript: Lluís Mundet, Christopher Cabib, and Pere Clavé; and final approval: Pere Clavé. All authors approved the final version of the article, including authorship list.