Comparative Effectiveness of Biofeedback and Injectable Bulking Agents for Treatment of Fecal Incontinence: Design and Methods

Abstract

Fecal incontinence (FI), the involuntary passage of stool, is common and can markedly impair the quality of life. Among patients who fail initial options (pads or protective devices, bowel modifying agents, and pelvic floor exercises), the options are pelvic floor biofeedback (BIO), perianal injection with bulking agents (INJ), and sacral nerve electrical stimulation (SNS), which have not been subjected to head-to-head comparisons. This study will compare the safety and efficacy of BIO and INJ for managing FI. The impact of these approaches on quality-of-life and psychological distress, cost effectiveness, and predictors of response to therapy will also be evaluated. Six centers in the United States will enroll approximately 285 patients with moderate to severe FI. Patients who have 4 or more FI episodes over 2 weeks proceed to a 4-week trial of enhanced medical management (EMM) (ie, education, bowel management, and pelvic floor exercises). Thereafter, 194 non-responders as defined by a less than 75% reduction in the frequency of FI will be randomized to BIO or INJ. Three months later, the efficacy, safety, and cost of therapy will be assessed; non-responders will be invited to choose to add the other treatment or SNS for the remainder of the study. Early EMM responders will be re-evaluated 3 months later and non-responders randomized to BIO or INJ. Standardized, and where appropriate validated approaches will be used for study procedures, which will be performed by trained personnel. Prospectively collected data on care costs and resource utilization will be used for cost effectiveness analyses.

INTRODUCTION

Overall, approximately 10% of non-institutionalized adults have fecal incontinence (FI), which is the involuntary passage of solid or liquid stool or staining of under clothes¹. FI has a major impact on quality of life^{2, 3} being associated with symptoms of anxiety and depression and with avoidance of leaving home or inviting friends to visit⁴. FI increases the likelihood of admission to a nursing home⁵, and the prevalence in nursing homes is 48%⁶.

Options for managing FI include pads or protective devices, antidiarrheal or laxative drugs, pelvic floor biofeedback (BIO), injections with inert bulking agents (INJ), sacral nerve electrical stimulation (SNS), and surgical reconstruction⁷. Initial approaches (i.e., education, drugs to prevent diarrhea or constipation, and pelvic floor exercises to improve sphincter function) reduce the severity of FI by 41%; 21% of patients report “adequate relief”¹⁰. The next steps (i.e., BIO⁸, INJ of the bulking agent dextranomer⁹, and SNS¹⁰) are each more effective than control treatments in randomized clinical trials (RCTs). These three approaches differ substantially in their proposed pathophysiological basis, mode of administration, acceptance by patients and providers, safety, cost, and availability. Insurance coverage of the three treatments also vary considerably.

Direct comparisons between BIO, INJ, and SNS are unavailable, with one exception¹¹, there are no head-to-head comparisons between them. Inconsistencies between existing studies in design, inclusion criteria, and primary endpoints have hampered comparisons between treatment types and even between studies within the same treatment modality. Consequently, meta-analyses and systematic reviews of these options for treating FI are inconclusive^{7, 12–16}. A NIDDK Workshop concluded that the highest priority for research should be “trials comparing the effectiveness, safety, and cost of current therapies”⁷.

The primary aim of this randomized controlled study is to compare the effectiveness, safety, and cost of BIO and INJ for treatment at 3 months (primary endpoint) and at 6, 12, and 24 months (secondary endpoints). The secondary aims are to (i) assess the impact of the treatments on frequency and severity of FI, quality-of-life, and psychological distress; (ii) compare the cost effectiveness of the treatments; (iii) identify patient characteristics associated with responsiveness to each intervention; and (iv) examine the mechanistic basis for these treatments.

METHODS

Overall study design

Building on the core recommendations of the NIDDK Workshop for design of an ideal comparative effectiveness RCT⁷, this is an unmasked, randomized trial comparing the effectiveness, safety, and cost of two treatments in 194 patients with moderate to severe FI. Because no single medical center has the number of patients necessary to independently conduct this study, this is a multicenter study at 6 sites: Augusta University, GA, Colon and Rectal Surgery Associates, Minneapolis, MN, Mayo Clinic, Rochester MN, University of North Carolina, Chapel Hill, NC, University of Michigan, Ann Arbor, MI, and University of Alabama Medical Center, Birmingham, AL. RTI International, Research Triangle Park, NC serves as the data coordinating center for this trial. Participants who satisfy eligibility criteria complete a 2-week baseline bowel diary (Figures 1 and 2).

Figure 1. Study Timeline.

The study begins with a baseline period, which is followed by the EMM and randomization phases. Patients who have 4 or more FI episodes during a 2-week baseline period are eligible to proceed to the EMM phase, which lasts 4 weeks. Thereafter, responders continue with medical management while non-responders are randomized to one of 2 treatments. Non-responders to EMM after 3 additional months will also be randomized to one of the 2 treatment options. At 3 months after initiation of the randomized treatment, patients who have not responded have the option of choosing an alternative therapy. Study follow up lasts 2 years.

Figure 2. Bowel Diary.

The bowel diary, which is a spiral notebook, records accidents on the left side and BMs and treatments on the right side. The notebook pages are printed front and back, and opposing pages are for the same day. The inside cover of the notebook has the Bristol Stool Form scale; the facing page provides instructions to patients for completing the diary.

Those who have 4 or more episodes of FI during this 2-week period enter the enhanced medical management (EMM) phase for 4 weeks. The EMM phase is designed to identify patients who respond to conservative measures. Throughout the study, a response is defined as 75% reduction in the frequency of FI compared to baseline¹⁷. Non-responders to EMM will be randomized to BIO or INJ. The randomization is performed by a computer program at the central Data Coordination Center and shared with the site only after participants are ready to be randomized. The primary assessment of efficacy, safety, and cost occurs 3 months later. Non-responders are invited to choose the other treatment or SNS as an adjunctive approach for the remainder of the study.

Patients who respond to EMM are also re-evaluated 3 months later. At this stage, non-responders will be randomized to BIO or INJ. Responders remain on EMM for the remainder of the study. Enrollment began in February 2019 (clinicaltrial.gov # NCT03811821).

Study Population

Ambulatory men and women aged 18 or older who have FI for 6 months or longer, two or more episodes of solid or liquid FI including soiling of under clothes per week, and fulfill criteria for treatment with BIO or INJ are eligible to participate in this study. Patients will be recruited from clinical practices and by advertisement, from the community. Patients who have another major disease that is responsible for FI (e.g., dementia, inflammatory bowel disease) and those in whom BIO or INJ is contraindicated or not feasible are excluded from enrollment (Table 1).

Table 1.

Exclusion criteria

Dementia (ie, score less than 6 on Six Item Screener)18 or psychosis

Obstetrical injury including third and fourth degree tears in the anal sphincter within 6 months.

Pregnant or planning pregnancy in next 2 years.

Internal anal sphincter separation >180 degrees on ultrasound imaging or MRI.

Spinal cord injury, spina bifida, or congenital malformation of anorectum.

Rectal prolapse or grade III/IV hemorrhoids.

History of previous anorectal surgery, such as stapled transanal rectal resection (STARR). Stapled hemorrhoidectomy is not an exclusion if performed more than 12 months previously. The FENIX procedure, artificial anal sphincter or transposed gracilis; surgical hemorrhoidectomy (other than stapled), and sphincteroplasty are permitted if performed more than 6 months previously and the patient meets inclusion criteria.

Established diagnosis of inflammatory bowel disease or current intestinal stoma.

History of pelvic radiation within previous 12 months or presence of active radiation proctitis.

Patients who have overflow diarrhea with rectal impaction with stool or an abnormal balloon expulsion test plus predominant symptoms of constipation.

Anatomic limitations to placement of dextranomer injections.

Presence or history of any medical disorder likely to require follow-up with MRI of the body (not head or neck), diathermy, microwave, or RF energy therapy.

Presence of existing implant in the anorectal region.

Allergy to hyaluronic acid based products.

Active anorectal conditions in the last 6 months including abscess, fissures, sepsis, significant bleeding, proctitis, colovaginal and rectovaginal fistulas, anorectal tumors, or other infections.

If the patient’s physician believes it is unsafe for the patient to temporarily stop anticoagulants for any test procedures and treatments associated with the study.

Patients who have 4 or more days with 4 or more bowel movements classed as a 6 or 7 on the Bristol Stool Scale per day during Baseline.

Patients with Parkinson’s disease, multiple sclerosis, severe diabetic neuropathy documented by EMG, and neurodegenerative disorder.

Immunotherapy or chemotherapy in the last 12 months.

Significant anal pain in the last 6 months.

If the participant is unwilling to stop using over-the-counter medications, herbal supplements, or prescribed medications for the purpose of modifying stool consistency, that are not included in the approved medications list (loperamide, laxatives, fiber supplements, and Questran are approved medications), for the duration of the research study.

Study Phases and Treatments

Baseline

At baseline, patients maintain a daily symptom diary for two weeks to (a) document that they have ≥ 4 episodes of FI (i.e., are eligible to participate in the study) and (b) provide a reference value for assessing treatment response at the end of enhanced medical treatment and at 3, 6, 12, and 24 months follow-up. For randomized participants, follow up is measured from the initiation date of the randomized treatment.

Enhanced medical management (EMM)

Patients will be educated about the basic physiological mechanisms of defecation. Dietary modifications and medications to normalize stool consistency will be tailored to the underlying bowel disturbance as evaluated by the 2-week baseline bowel diary. Options include dietary fiber supplementation, antidiarrheal medications (e.g., loperamide, cholestyramine, or colesevelam), and laxatives. When loperamide is recommended, it will be used regularly and/or on an as-needed basis (i.e., before meals and/or social outings). Pelvic floor exercises will be taught verbally and in writing⁸. This EMM protocol, which is detailed in a pamphlet that will be provided to patients, ensures that these measures are provided in a systematic manner. Consistent with current practice, only patients who do not respond to medical management are randomized to BIO or INJ, thereby increasing the likelihood of identifying differential effects of BIO and INJ. The predictors and the durability of response to EMM will be evaluated.

Biofeedback training

The BIO protocol, equipment, and manual are modified from that used to provide BIO in a multi-site FI study conducted by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) Pelvic Floor Disorders Network (PFDN)¹⁹,²⁰. BIO training and anorectal manometry will be conducted with the Medspira Anorectal Manometry system and MCompass software (Medspira Inc, Minneapolis, MN), which incorporates on-screen prompts to assist the therapist. Dr. Whitehead will individually train, then evaluate and certify all therapists with a live patient surrogate. Each BIO session will be audiotaped. Dr. Whitehead or another experienced therapist will assess adherence to protocol in 20% of the sessions performed by each therapist.

All patients will receive 5–6 sessions over 6 weeks. The BIO therapy is individualized to the specific pathophysiological deficit(s) in each patient including strength training for sphincter weakness, sensory discrimination training to improve the ability to detect weak rectal distension, and urge resistance training to improve the patient’s ability to retain stool in the rectum long enough to reach the toilet.

During strength training, patients are provided with visual feedback of rectal and anal pressure. The goal is to increase anal pressure to at least 100 mmHg while keeping rectal pressure below 20 mmHg. The therapist determines the normal squeeze pressure at the beginning of the session and then provides a series of trials with different target pressures. These target pressures begin at the patient’s initial level and are changed as the patient improves their ability to squeeze. The patient should be able to respond correctly 50–75% of the time. While contracting the anal sphincter, patients are reminded to relax their abdominal wall muscles because contraction of these muscles increases rectal pressure and likely predisposes to FI. The next phase is to focus on increasing the duration of voluntary anal contraction to 10 seconds. The squeeze duration is the duration for which the squeeze response is sustained above 50% of the maximum squeeze pressure at the beginning of the trial.

Sensory training is tailored to the sensory deficit; no visual feedback is provided. Therapy begins with a 50 ml distension which most patients can recognize. If the patient perceives the distention and contracts their sphincter appropriately, the next trial will be performed with a lower distending volume. By contrast, if 50ml is not perceived, the next distention is 10 ml larger. At each step, the adjustments are 10ml or less. The lowest volume that the patient appropriately recognizes is the sensory threshold, and attempts will be made to lower this threshold up to 20 ml. or less At each trial, patients receive feedback on their ability to recognize weaker distentions and to squeeze the anal sphincter in response.

Urge resistance training is based on the principle of systematic desensitization of a feared situation. It will also be performed without visual feedback. At first, the rectal balloon volume that evokes intolerable urgency is determined. Thereafter, to decrease the aversiveness of the sensation, the rectal balloon will be deflated, typically by 30–50 ml, then gradually reinflated, while patients are instructed to breathe deeply to relax. Over a series of 3–4 trials, the therapist cues patients to the target, which is the ability to tolerate rectal distention of 150 ml or greater. Typically, the tolerance for rectal distention begins to improve in the first session.

The patient’s symptoms also guide the type of training. For example, sensory discrimination training is more appropriate for passive FI while urge resistance training is more appropriate for urge FI. Strength training is provided during every BIO training session, but the therapist may place greater emphasis on maximum squeeze, squeeze duration, or avoidance of abdominal wall contractions based on the initial anorectal manometry.

Injection of dextranomer

This protocol is a modified version of the protocol used by Graf to ensure consistency across sites and safety of patients⁹. Patients are pretreated with oral antibiotics to minimize injection site infections. One ml of NASHA dextranomer is injected through an anoscope into each quadrant of the submucosa (4 ml total) approximately 5 mm above the dentate line. After each INJ, the needle is left in place for up to 10 seconds to avoid backward leakage of contrast through the injection channel. Following the INJ patients are counselled to use stool softeners (docusate sodium) until the first bowel movement and to use paracetamol as needed for INJ site discomfort. Six weeks after initial INJ, patients return for evaluation, and if the patient has achieved less than 75% reduction in FI, a second INJ is given.

Combination therapy

The primary assessment of efficacy is at 3 months after patients initiate randomized treatment (i.e., BIO or INJ). Non-responders to the randomly selected treatment will be invited to choose the other treatment or SNS as an adjunctive treatment. Participants who add a second treatment will continue to be monitored for the study duration. The study design for these patients emulates the clinical situation in which patients who have an unsatisfactory response to a treatment are offered a new treatment or an ancillary treatment. This design is prompted by a pilot feasibility survey of 187 FI patients at 3 clinical sites in which 57% of patients indicated that the ability to receive the treatment of their choice if the assigned treatment was not effective was important to their willingness to be randomized (unpublished data). Response to combinations of treatments will be evaluated in secondary analysis.

Sacral nerve stimulation

The SNS protocol is based on minor modifications to a multisite study protocol²¹. A barbed quadripolar temporary stimulation electrode is inserted into the sacral nerve plexus at S2, S3, or S4 with local anesthesia to the back. The electrode position is selected by adjusting the electrode position until electrical stimulation from an external stimulator produces a distinct contraction of the external anal sphincter and is felt by the patient. The temporary stimulation electrodes are then attached to an external stimulator for 10–14 days. If patient diaries show a reduction of at least 50% in the number of FI episodes per week and/or number of incontinence days per week, the stimulation leads are attached to a permanent stimulator implanted in the patient’s buttocks. The patient is scheduled for return clinic visits at one and 3 months to adjust stimulation parameters if necessary and to screen for adverse events. Adherence of surgeons to the procedure manual for SNS is monitored by review of surgical notes.

Assessments

Table 2 provides a timeline of the patient-reported outcomes (PROs) and anorectal assessments. Except for the bowel diaries, which are adapted from previous studies, and the Out_of_Pocket Questionnaires, these assessments have been extensively validated previously.

Table 2.

Timeline of Study Assessments

Schedule of assessments	Baseline	Randomization	3 months 1	6 months 1	12 months 1	18 months TC	24 months 1
Socio-demographics and examination	X
Bowel diary	X	X	X	X	X		X
Bristol stool form scale	X	X	X	X	X		X
Endoanal ultrasound (3 mo for dextranomer only)	X		X 2
Anorectal manometry (ARM)		X	X 2
Magnetic Evoked Potential (MEP)		X	X 2
Fecal Incontinence Severity Scale (FISS)	X	X	X	X	X		X
Fecal Incontinence Quality of Life (FIQOL)22	X	X	X	X	X		X
Credibility/ Expectancy Questionnaire23	X	X
PHQ-12 Somatization scale 24, 25		X	X	X	X		X
PROMIS Anxiety-7 26		X	X	X	X		X
PROMIS Depression-8 26		X	X	X	X		X
PROMIS Self-Efficacy Symptom Management 49		X	X	X	X		X
Assessment of AEs, SAEs	X	X	X	X	X	X	X
Cost (treatments, # of visits, out of pocket)	X	X	X	X	X	X	X
Work Productivity and Impairment (WPAI)	X	X	X	X	X		X
EuroQol 5D (EQ-5D)	X	X	X	X	X		X

Abbreviation;TC = telephone call

¹This visit may also be conducted by phone for patients who are unwilling to travel due to COVID-19.

²Optional assessment

The Out-of-Pocket Cost Questionnaire

This is a new, unvalidated questionnaire that is designed to collect information on the amounts patients pay for non-prescription medications and absorbent pads purchased for the self-management of FI (Table 3). The rationale for the survey is derived from an anonymous internet survey in which people with FI indicated that they had coped with FI by wearing pads, taking antidiarrheal medicines, and limiting food intake when they had to leave the house, in that order²⁷; half of these respondents had not discussed this problem with any physician. This suggests that purchase of pads and medications for diarrhea or constipation may be a major expense for patients.

Table 3.

Out of Pocket Costs for Treating Accidental Bowel Leakage

The questions below ask about expenses you had to pay during the last month out of your own pocket for treating your accidental bowel leakage. These are expenses that most insurance policies do not cover.
No.	Question
1	Do you use protective pads to prevent soiling your clothes because of accidental bowel leakage?
	□ Yes □ No
2	If yes, Which of the pictures* below are most like the pads you use? Select all that apply.
3	Did you wear these pads for protection from urine leakage as well as bowel leakage?
	□ Yes, I wore pads to protect me from both urine and bowel leakage
	□ No, pads were worn only for bowel leakage
4	How many pads did you use on a typical day? __ __ __
5	How much money did you spend in a typical week for pads? __ __ __ dollars
6	Did your insurance reimburse you for some of these expenses for pads?
	□ Yes. How much did you receive from insurance? __ __ __ dollars
	□ No, all these expenses were paid by me personally
7	Do you take medicines or fiber supplements to control diarrhea because diarrhea makes your accidental bowel leakage worse?
	□ Yes □ No
8	How much do you spend in a typical week for fiber or medicines to control diarrhea? __ __ __ dollars
9	Do you take medicines or fiber supplements to control constipation because constipation makes your accidental bowel leakage worse?
	□ Yes □ No
10	How much do you spend in a typical week for laxatives or fiber supplements to control constipation? __ __ __ dollars
11	Did your insurance company reimburse you for some of the money you spent for medications or fiber supplements to treat diarrhea or constipation?
	□ Yes. How much did you receive from insurance? __ __ __ dollars
	□ No, all these expenses were paid by me personally
12	Did you have any other expenses caused by having accidental bowel leakage?
	□ Yes □ No
13	Please tell us below what you paid for and approximately how much you spent per week.
	________________________________________________________________________
	________________________________________________________________________

^*Portions of the graphics included in this document were developed as part of a study funded by the National Institute of Nursing Research, NIH, R03806M38261.

Effectiveness

The bowel habits and frequency of FI are recorded in a daily paper diary. For each accidental bowel leakage, patients record the amount of leakage, the rectal sensation prior to the event (i.e., no awareness, normal warning, or strong urge), and the consistency of the leaked stool. Likewise, patients record the consistency (i.e., Bristol Stool Form Scale) for every bowel movement. The diary also records the medications used to prevent bowel leakage and the number of pads (perineal protective devices) used.

Safety

Using a combination of checklists (for expected AEs) and open -ended questions, adverse events will be collected by questionnaires during visits, review of medical records, and by interviewing participants. Events will be classified as serious adverse events (SAE) if they meet conventional criteria. The severity of AEs will be graded by common terminology criteria for adverse events (CTCAE) criteria.

Cost of care

During the study, biofeedback therapy and dextranomer are provided at no cost to patients or payors (insurance). SNS is covered by insurance and is offered to participants who fail the treatment to which they were randomized. The study has a provision to defray the patient’s expense (i.e., deductible) for SNS and for medications.

The cost for each treatment will be evaluated from the perspectives of patients, payers, and society. Patient costs will be obtained from the Out-of-Pocket Cost Questionnaire that gathers information on the expense of non-prescription medications and absorbent pads (Table 3). The expense incurred by patients is derived from the number and type of treatment visits multiplied by Medicare authorized reimbursement rates. The societal cost is derived from the sum of the payer costs, patient costs, and patients’ productivity losses associated with FI as evaluated by the Work Productivity and Impairment questionnaire (WPAI)²⁸. The WPAI is a widely used measure of work absenteeism, impairment while at work, and interference with daily activities^{28, 29}. While it has not been used to measure the impact of FI on work productivity, it has been used in studies of overactive bladder, urinary incontinence³⁰, and irritable bowel syndrome³¹.

The metric used for comparing cost effectiveness is the quality-adjusted life years, a measure that accounts for both duration and quality of life. This calculation requires a health-related quality of life index, which is a generic health related quality of life scale that is used to compare the morbidity-related impact of different disease states. This study will use the EQ-5D, which is correlated with the severity of FI. This questionnaire has been used to discriminate between SNS and another surgical treatment for FI³².

Translumbosacral anorectal Magnetic stimulation test

The lumbar and sacral motor evoked potentials (MEP) will be tested at the randomization visit, and optionally also at the 3 month follow-up visit, at four treatment sites (Mayo Clinic; Augusta University; Colon and Rectal Surgery Associates in Minneapolis, MN; and University of North Carolina) to assess the presence of underlying neuropathy and its influence on or changes after therapy^{33, 34}. This test uses a specially designed rectal probe with 2 pairs of bipolar steel ring electrodes, each 2 cm apart, mounted on a catheter (Gaeltec Ltd, Isle of Skye). The proximal pair are located 8–10 cm from the anus and the distal pair at 1–2 cm from the anus. The electrodes are used to record motor evoked potentials after magnetic stimulation. Magnetic stimulation of the lumbar (TLMS) and sacral (TSMS) nerve roots will be performed using commercially available 90 mm coil magnetic stimulator (Magstim 200²; MAGSTIM, Whitland, UK). With the subject in the prone position, TLMS and TSMS stimulation will be performed to the left and right of the spine respectively at L₂-L₃ and S₂-S₃ vertebrae. As a control, the EMG responses from the anterior tibialis muscle of one leg will be recorded. The MEP responses will be recorded into a 4-channel amplifier with a built in a-d interface (Nihon Kohden, Japan) with filter settings of 5–2000 Hz and at a sampling rate of 4–8 kHz, fed into a computer for display and analysis. An optimal response will be defined as an anal and rectal MEP of >10μV and tibialis response of 100μV on at least 3 of 6 consecutive trials (Figure 3). Five MEP responses will be obtained on each side and the average of the best three responses will be used to measure MEP response. The latencies of wave forms (Figure 3) on both the TLMS and TSMS-induced MEP responses will be analyzed on the right and left sides. The first and most prominent negative or positive deflection will be designated as the MEP response. For each of 8 (i.e., left and right lumbo- and sacro-, rectal and anal) responses, the latency will be averaged across 8 measurements. Patients will be classified as having a neuropathy if ≥1 of 8 average latencies are abnormal^{33, 34}.

Figure 3. Motor evoked potentials in response to magnetic stimulation of the lumbar spine.

Typical examples of a normal motor evoked potential (MEP) response with a normal latency time in a healthy control subject (A), and an abnormal MEP response in a patient with fecal incontinence showing prolonged latency and smaller amplitude response (B).

Anorectal ultrasound

This will be measured at baseline in all subjects enrolled in the study, and optionally at 3 months, in the dextranomer treated patients. Anorectal ultrasound probes and protocols differ at different institutions but every ultrasonographer is expected to provide a report that includes the presence and estimated size of defects in the IAS and EAS. All sites will use BK Probes for anal ultrasound.

Anal manometry and rectal sensation

This will be measured at baseline in all subjects, and optionally at 3 months. After up to 2 sodium phosphate enemas (Fleet saline enemas, C.B. Fleet Company, Inc), anorectal pressures will be evaluated in the left lateral position with manometry (Mcompass, Medspira Inc, Minneapolis, MN)^{35, 36}. Pressures will be measured at rest [20s], during squeeze [3 maneuvers 20s each], Valsalva [20s] and evacuation without and with rectal distention [50 ml]). Rectal sensory thresholds for first sensation, desire, urgency, and pain were recorded by progressively distending the rectal balloon in 20 ml increments from 0–180 ml.

Balloon expulsion test

The time required to expel a standardized rectal balloon filled with 50 mL of water will also be evaluated ³⁵.

Primary and Secondary Outcomes

There are three primary outcomes -- efficacy, safety, and cost -- at 3 months after initiation of treatment with BIO or INJ. The primary efficacy outcome will be the treatment response, which is defined by a 75% reduction in the average weekly frequency of FI episodes between the baseline (prior to the EMM treatment period) and the 3 month diaries. The 75% threshold is based on the observation that 50% of FI patients in a survey reported that a reduction of 70–80% rather than 50% was necessary for a treatment to be considered successful¹⁷. For safety, the primary outcome is the proportion of patients with AEs of pelvic pain of grade II or higher based on CTCAE criteria, treatment site infection, or SAEs requiring hospitalization within 3 months following the start of randomized treatment. The primary cost outcome is the cost of each randomized treatment through 3 months. Secondary outcomes include assessments at 6, 12, and 24 months for all these measures. Additional secondary outcomes include a validated FI severity scale^{2, 37}, a quality of life scale²², and psychological distress³⁸.

The severity of FI will be assessed with the Fecal Incontinence Severity Scale (FISS) which is derived from the Fecal Incontinence and Constipation Assessment questionnaire³⁹. This is a validated FI severity scale which incorporates the frequency of different types of stool loss (solid, liquid, and both), the circumstances surrounding FI (urgency, passive, combined, or neither), and volume of leakage⁴. The psychometric properties of this instrument have been validated in community-based studies, and it has been used in a therapeutic trial^{2, 4, 39, 40,41}. The severity of FI evaluated by this scale is strongly correlated with the impact of FI on quality of life².

To facilitate comparisons with previously published studies that have used a 50% responder definition^{9, 21}, we will also calculate the proportion of patients who have at least a 50% reduction in FI episodes. The proportion of patients in each treatment arm who are continent at each follow up visit will be reported as an additional secondary outcome defined as absence of FI episodes on the two-week bowel diary, self-report of no ABLs for the last month, and self-report that they “no longer have this problem.”

Potential moderator variables to be assessed include the PHQ12 (Somatization)^{24, 25,42} FI-specific quality of life (unpublished data), the credibility/expectancy questionnaire (CEQ)^23,43, sex ⁴⁴, subtype of FI (i.e., passive vs. urge-related FI)⁴⁵, Bristol Stool Form Scale (BSFS) rating of stool consistency⁴⁶, ⁴⁷ and anorectal pressures and rectal sensory thresholds measured with manometry. Anal resting and anal squeeze pressures, sphincter squeeze endurance, rectal sensory thresholds (bag volume and intra-bag pressure at the first report of sensation, desire to defecate, urgency to defecate, and maximum tolerable volume), and rectal compliance will be measured at 3 months follow up for EMM responders or at 3 months following randomized treatment^{35, 36}.

Sample Size Estimates

The study was designed to detect a difference of 20% between groups in the proportion of responders at 3 months because this was considered to be the Minimum Clinically Important Difference (MCID). Conservatively assuming an average response rate of 50% across the groups, 97 patients per group provides 80% power to test for differences of this magnitude between the 2 treatment groups at an alpha of 0.05. Since the planned analysis will use an intention to treat (ITT) approach in which all participants will have a defined efficacy outcome regardless of study completion, this sample size was not inflated to account for potential drop out.

For the primary safety outcome, we will use the Wilson Score interval method⁴⁸ to estimate 95% confidence intervals for the percentage of participants in each randomized treatment group who experience an AE of pelvic pain of grade II or greater severity, treatment site infection, or an SAE requiring hospitalization between randomization and the 3 month visit. Based on previous studies, we assumed the percentages of participants with a qualifying AE or SAE to be approximately 9.8% (95% confidence interval (CI) 5.7 to 16.3) in the INJ group and 0% (95% CI 0 to 3.6) in the BIO group. These confidence intervals are narrow enough to reveal clinically meaningful differences in the safety profiles of the treatment groups. Reports of SAEs at later time points will be described but not tested between groups.

To perform power calculations for the difference in 3 month treatment costs, we used cost estimates for treatment provided in Bernstein⁴⁹ and assumed a standard deviation of 0.5 (UB-LB). Cost of BIO treatment was estimated at $265.45 ($212.36–318.55) per visit, and we conservatively assumed treatment for 5 visits. Physician and device costs (range) for each INJ treatment was estimated at $281.03 ($224.82–337.23) and $4,900 ($2,940–6,860). We assumed two injections per participant in the first 3 months. Societal costs were estimated by adding treatment costs and 3 month medical and productivity costs associated with FI taken from Xu⁵⁰. Given the large differences in costs between treatments, the power to find pairwise differences between groups with an alpha of 0.05 was estimated to be near 100% for the payer, patient, and societal perspectives.

Randomization and Masking

Randomization will occur in a ratio of 1:1 using randomly permuted blocks with sizes known only to the data coordinating center and will be stratified by clinical site and sex (male and female). Patients cannot be masked in a trial comparing behavioral, surgical, and medical therapy; thus, participants will be aware of the treatment to which they are assigned and the nature of alternative treatment to which they were not assigned. Most outcomes are patient-reported through bowel diaries and questionnaires, so outcome assessment is also unmasked. Should subjects have an a priori preference for one of the treatments, their expectation of benefit could be biased based on whether they were randomized to their preferred treatment. However, random assignment to the treatments should balance the number of people with a positive expectancy in each treatment group. Moreover, we will assess individual differences in expectation of benefit by administering the Credibility/Expectancy Questionnaire at baseline and immediately after randomization.

Data Analysis Plan

For the primary efficacy analysis, a generalized linear model will be constructed to predict treatment response at 3 months post-randomization. Treatment group and the stratification factors of clinical site and sex will be included in the model as categorical independent variables. An intention-to-treat approach will be used in which participants will be analyzed based on their assigned treatment group regardless of actual treatment received. Participants without a valid bowel diary at 3 months due to drop out or other reasons will be classified as non-responders. Using this model, we will test the null hypothesis that the proportion of responders is the same in the treatment groups at 3 months following randomization; the test will be considered significant at an alpha of <0.05.

For the primary safety outcome through 3 months, we will estimate 95% confidence intervals in each of the treatment groups using the Wilson Score interval method; we will compare the groups descriptively based on those confidence intervals. Participants who are missing safety data due to drop out or other reasons will be excluded from this analysis.

Mean and median costs per participant will be estimated for each of the treatment groups and from the perspectives of payers, patients, and society. We will analyze whether differences in costs through 3 month follow up between the intervention arms across cost perspectives are statistically significant using parametric t-tests or using non-parametric bootstrapping methods. Differences will be considered significant at an alpha of <0.05.

Efficacy through 24 months will be assessed using a longitudinal extension of the model described for the primary outcome. Because patients who are non-responders at 3 months may add another FI treatment, they will be retained as treatment failures in the long-term analysis of the comparative effectiveness of the initial randomized treatments. Safety and cost outcomes through 24 months will be evaluated using the same methods as the primary outcomes. Categorical secondary outcomes through 24 months will be analyzed using longitudinal generalized linear models adjusted for clinical site and sex, and changes from baseline in continuous outcomes will be estimated using analogous general linear models. Responder rates for participants treated with EMM, BIO, INJ, and combinations of treatments will be estimated at time points up to 24 months using Wilson Score confidence intervals.

Cost effectiveness will be evaluated based on the average cost per QALY gained for each intervention and the incremental cost per QALY gained.

DISCUSSION

Prompted by the recommendations of a NIDDK Workshop ⁷, this trial compares the effectiveness, safety, and cost of current treatments (i.e., BIO and INJ) in FI patients who do not respond to education, pelvic floor exercises, and medications to manage bowel disturbances. In non head-to-head comparisons, these approaches are approximately of comparable efficacy. While BIO is the least expensive and safest option, most studies of pelvic floor biofeedback therapy for FI were conducted at single centers and results vary considerably among studies. The outcomes of BIO are highly dependent on the skills and experience of the therapist. Hence, from a public health perspective, a rigorous comparison of these BIO and INJ options across multiple sites is critical.

There are several strengths to this study design. To reduce the contribution of non-specific benefits unrelated to the primary therapy, only patients who fail to respond to EMM will be randomized to BIO or INJ. Patients who do not respond to the initial randomized therapy option will be invited to choose the other treatment (BIO or INJ) or SNS, which provides exploratory data on combining two active treatments. Likewise, patients who initially respond to EMM (ie, at 4 weeks) but relapse on EMM 3 months later will be randomized to BIO or INJ. These patients will be pooled with the other patients randomly assigned to the same treatment for the primary analyses. Both features may encourage some patients to participate in the study.

The eligibility criteria are designed to recruit patients with moderately severe FI. However, patients with FI due to severe diarrhea or associated with selected diseases (e.g., selected neurological disorders, active anorectal conditions, or inflammatory bowel disease) which would obviate treatment with BIO or INJ are not eligible to participate. Treatments will be administered in accordance with standardized protocols and manuals, each of which was written by one of the principal investigators and finalized by mutual consensus of all the investigators and study team members. Also, all investigators performing therapeutic interventions will have undergone standardized training in the same. Recognizing that FI patients have a spectrum of bowel disturbances⁴, the medical management of FI will be tailored to underlying bowel disturbance, meticulously characterized by daily diaries, and with approved medications. Loperamide effectively reduces diarrhea but can lead to constipation⁵¹. Constipation will be managed per standard guidelines, beginning with fiber supplementation, and adding osmotic and stimulant laxatives as necessary⁵².

Several measures have been taken to standardize biofeedback therapy. The software used to provide biofeedback therapy was especially developed and used in a recent study conducted by the Pelvic Floor Disorders Network²⁰. This program has in-built protocols that guide users while they provide biofeedback therapy. The biofeedback therapy will be individualized to the specific anorectal deficit. One of the PIs serves as the lead therapist and has trained all therapists at each site. The therapist’s skills are monitored.

While clinicians may have individual preferences, there is limited comparative data to inform decisions regarding the management of FI in patients who have failed initial treatment. The approved therapies vary considerably in their availability, cost including insurance coverage, and adverse effects. Hence, the common practice is to start with the least invasive option (BIO) and progress to other options if BIO is unsuccessful. Finally, from a public health perspective efficacy must be balanced against financial cost and adverse effects. The results of this study may optimize the management of FI and answer these public health questions. In addition, exploratory analyses will assess whether there are subsets of patients that are more likely to respond to certain treatments or treatment combinations.

Stewarded by the National Institutes of Health (NIH), this clinical trial embraces multifaceted approaches to facilitate quality and efficiency⁵³. All investigators have completed Good Clinical Practices (GCP) training. Avoiding duplicative reviews, oversight is provided by a single Institutional Review Board (IRB) at the University of North Carolina, Chapel Hill. The study protocol and progress are updated on the clinicaltrials.gov website (NCT03811821). Progress is tracked by a Data Safety and Monitoring Board that is coordinated by the NIH. The study investigators have monthly meetings with program officials from NIH.

However, this trial faces challenges that are shared with other clinical trials, especially randomized-controlled trials that compare behavioral with medical or surgical therapy^54–56. Arguably, suboptimal recruitment is the major risk; between 22% to 70% of clinical trials do not meet accrual goals⁵⁷. We anticipate that the use of treatments that are of proven efficacy, and also approved by the Food and Drug Administration for managing FI, will increase the willingness of participants to enroll in this study. While participants receive a stipend for study visits, adapting to the restrictions on travel during the COVID-19 pandemic⁵⁸, the protocol was modified to allow study visits to be conducted remotely when feasible. Potential study participants are identified by multipronged approaches, primarily from the clinical practice and by public advertisement. While anorectal ultrasound will be performed by experienced clinicians with a similar instrument at all sites, the manner in which the procedure is performed and the interpretation of images may vary among sites. However, to ensure more uniformity, all data will be collected on a standardized form and the studies will be performed by experienced operators. The SNS procedure and dextranomer injections will also be performed by experienced operators using standardized protocols. Biofeedback therapy will be provided with a user-friendly program by therapists who had variable experience with providing biofeedback therapy before the study and were trained by an expert, albeit with one patient. Because of the nature of interventions, the participants cannot be masked, but we will assess their level of preference for some treatments with the Credibility/Expectancy Questionnaire, which will be administered at baseline and after the first treatment session.²³ Conceivably, participants may prefer one or more treatments. The statistical analysis will adjust ascertain if there is a residual preference for one treatment over another after adjusting for a priori preferences. The investigators will not be masked to which treatment each participant receives and this may introduce bias in the assessment of success for one treatment over another. However, the primary and key secondary outcome measures are based on diaries and questionnaires completed by participants that are designed to overcome potential investigator bias. By incorporating patient feedback, we strived to design a user-friendly bowel diary, which was validated in a cohort of 40 healthy people and 90 patients with FI (data not shown). However, we recognize that reporting the detailed information in bowel diaries may be challenging for some patients.

In conclusion, the design, interventions, and outcome measures of the FIT trial are designed to provide valid and reliable estimates of the efficacy, safety and cost-effectiveness of medical management for FI and of BIO and INJ in patients who fail medical therapy for FI.