Discussion

Key Findings and Strength of Evidence

We found low-quality evidence to inform clinical decisionmaking for nonsurgical treatments for FI in adults in the United States. The evidence situation is worse for virtually all surgical treatments compared with nonsurgical therapies. The evidence of effectiveness is insufficient for all surgical treatments. More invasive surgical procedures are often associated with considerable complications. We were unable to conduct a meta-analysis because few studies examined the same treatment-outcome combination within similar timeframes and outcome measures varied widely. Table 18 summarizes the major findings of this review; supporting details of the strength of evidence assessments are provided in Appendix F, Table F10; risk of bias ratings for individual studies that informed the strength of evidence assessments are in Appendix F, Tables F11 and F12, respectively.

Table 18

Strength of evidence summary for nonsurgical treatments for fecal incontinence.

Low-strength evidence suggests that dietary fiber supplementation with psyllium decreases FI episode frequency by 2.5 occurrences per week after 1 month of use; that clonidine has no effect at 1 month; and that PFMT-BF with electrostimulation is no more effective than PFMT-BF on FI severity and changes in the FIQL instrument scores over 2 to 3 months.

Low-strength evidence at 6 months post-treatment suggests that dextranomer anal tissue-bulking injections are more effective than sham injections on FIQL, the number of FI-free days, and on the percent of patients with FI episode reduction of 50 percent or more from pre-injection levels, but no more effective than PFMT-BF with or without electrostimulation on FI severity and quality of life, and no more effective than sham injection on FI severity (CCFIS) or in reducing the number of FI episodes from baseline. The only anal sphincter tissue bulking agent examined in a randomized trial beyond 6 months was Durasphere® (off-label), which showed improvements in FI severity up to 6 months. However, gains with Durasphere® diminished slightly between 6 months and 1 year post-injections in two RCTs.^77,78

Although PFMT has been successful in addressing urinary incontinence,¹⁵⁴ the included PFMT literature focused mainly on refinements in treatment delivery to improve or prolong purported benefits of PFMT for FI rather than on establishing its benefits. Various iterations of PFMT produce similar improvements that appear to meet MID (Appendix E) when those measures were used (CCFIS,¹⁵⁵ FISI,¹⁵⁶ Vaizey,^155,157 and FIQL subscales¹⁵⁵). We found insufficient evidence that PFMT-BF offers any advantage over standard care (such as dietary fiber supplementation, stool-modifying drugs, and education) for FI. Assessing PFMT-BF training for FI was made difficult by the lack of standard protocols; no included studies used the same treatment protocol for timing, intensity, type, and duration of exercise. Some articles provided no information on exercise repetitions and duration, despite including intricate details regarding biofeedback sensors, probe placement, and patient positioning.

The evidence for FI treatment benefits was insufficient for all other nonsurgical and surgical interventions. Thus this literature provides little guidance for primary care providers and patients in their selection and sequencing of treatments for FI. Limitations in study conduct were common and generally avoidable. In particular, study reporting did not match the longstanding reporting recommendations of CONSORT.^158-160

FI treatment generally follows a longitudinal sequence, which complicates efforts to compare surgical and nonsurgical interventions. Patients earlier in their FI course typically receive nonsurgical treatment, and those who do not respond to nonsurgical treatments may then be offered surgery. Additionally, nonsurgical treatment is often used as an adjunct to surgery, whereby patients continue dietary modification, stool-modifying drugs, and sometimes PFMT after surgery.

Understanding the effectiveness of the range of FI treatments requires carefully defining the nature of the patients at risk in terms of underlying problems, clinical characteristics, and prior treatment history. Many included studies failed to provide this information. Nonsurgical studies focused on short-term or intermediate-term outcomes in the management of FI, leaving many unanswered questions about the durability and feasibility of interventions over time.

Aside from adults in nursing homes and those with spinal cord injuries, we were unable to report subgroup-specific outcomes due to the heterogeneity of FI etiologies in enrolled adults in the studies that reported etiology. The majority of enrolled adults were females and their FI etiologies were most often mixed or not reported.

Adverse effects from nonsurgical interventions are uncommon and tend to be minor. In contrast, AEs from surgical interventions are common and often substantial. For some procedures, complications may occur years after the surgery. The severity of complications increases with invasiveness of the treatment. Most of the surgical adverse effects were identified from case series. However, we felt confident using case series for surgical complications because these problems were extremely unlikely to arise among controls who did not receive surgery. Complications from ACE, sphincter repair, and sphincter replacement were most severe. SNS complications were less severe, but all of these treatments may require further surgery. Removal of SNS was required in up to one in four recipients. The highest complications of any surgical procedure for FI were reported for sphincter replacement (ABS). The ABS required surgical removal (explant) in 20-81percent of patients; infections were common and some patients ultimately required permanent colostomy. Significant complications are important to consider when providers are counseling patients with severe FI.

Findings in Relationship to What Is Already Known

We examined the comparative effectiveness of treatments for FI across the range of treatments available to adults in the United States. In contrast, prior systematic reviews typically examined evidence within single modes of FI treatment, such as such as surgery or drugs.^{6-9,13,14,16,17,161-163} Similar to our findings, single-treatment-mode systematic reviews found weak evidence for most FI treatments, and similar literature limitations (small number of studies, small patient samples, and substantial methodological limitations), leaving little definitive evidence to support specific treatments for FI. This review adds unique comparative information to assist providers and patients in clinical decisions among several treatment options.

We found FI treatment guidelines from two professional societies: the American College of Gastroenterology (ACG),¹⁶⁴ and a recent guideline available from the American Society of Colon and Rectal Surgeons (ASCRS).¹⁶⁵

Appendix F, Table F13 provides a table that contrasts the recommendations of these guideline groups and the findings in this review. Injectable tissue bulking injections received weak support by ASCRS and ACG, which is consistent with the findings of this review. No other professional society recommendations could be supported by the results of this review. Both societies supported combined nonsurgical treatments (diet, antidiarrheal drugs, education). Colorectal surgeons more strongly favored surgical approaches than did the gastroenterologists; both groups supported SNS, which had insufficient evidence in this review. Many treatments examined in this review were not mentioned in either guideline (dietary fiber [alone], other drugs, PFMT versus other comparators, PFMT-BF with electrostimulation, electrostimulation without PFMT, rectal irrigation [alone], and interventions for older adults in nursing homes).

Applicability and Limitations of the Evidence Base

Several important characteristics limit the generalizability and applicability of the studies reviewed. Overall, the evidence base would benefit from better compliance with CONSORT¹⁵⁸ and greater efforts to avoid compromising study integrity by analyzing only completers or those with perfect compliance, or by aggregating data from those whose condition deteriorated with those who remained stable.

The large number of outcome measures in RCTs alone impeded comparability across studies and the ability to conduct meta-analysis. The field would benefit from using a more consistent set of outcome measures to facilitate comparability. In cases where a new assessment tool is used, simultaneously including a validated, commonly used measure would facilitate interpretation. The wide heterogeneity in current measures leaves the field with many unique, often underpowered studies for a particular intervention and/or subgroup, which provides insufficient evidence to inform clinical decisions.

Common outcome measures need standardized labels across all disciplines that treat adults with FI. Measures that underwent several iterations, including changes in content and scale, were variably identified and often mislabeled, even in recent literature. For example, the Vaizey FI score (0-24²⁶) was sometimes labeled as “St. Mark's” (0-13²⁸), yet baseline or outcomes values, or the reference (when cited) for the measure, made it obvious that the Vaizey score had been used.^75,81-83 The Cleveland Clinic Fecal Incontinence Score (CCFIS)²⁴ was also variably labeled as CCFIS, Wexner or Jorge/Wexner in the articles we reviewed.

More uniformity in both how FI episodes are defined and graded for severity would improve comparability across studies. Definitions of FI episodes were particularly difficult to compare across studies (soiling versus solid stool versus solid plus liquid stool versus liquid only). FI severity was defined in numerous ways (episode frequency, CCFIS or other scale at screening, etc.), and was often used as a sample selection criteria in clinical studies. Mild to severe grading is problematic because FI severity grading is not standardized. Moreover, clinicians and patients sometimes disagree on FI severity ratings for a given patient situation.²⁵ Definitions of urgency also varied in the few studies that measured it. Input from adults with FI may suggest ways to identify and quantify aspects of FI that can capture what matters most to patients in outcome measures. Issues with urgency may be just as problematic to patients as actual FI episodes, since the uncertainty and fear of accidental bowel leakage surrounding urgency require, at minimum, the same prompt behavior: finding a toilet.

Inconsistencies in the labeling of PFMT were particularly confusing. Clinical studies and one guideline¹⁶⁵ labeled this entire group of treatments as biofeedback, which is a vehicle by which PFMT is enhanced, not the treatment itself. Given that biofeedback is used to enhance many types of treatments, efforts to standardize labels used for the various iterations of PFMT in the literature (PFMT, PFMT-BF, PFMT-BF with electrostimulation) would be helpful for readers.

The value of intermediate physiologic measures is unclear given the lack of a well- established link between physiologic measures and patient-centered outcomes. Manometric and other physiologic measures are overabundant, but far more information is needed about typical patient demographics, clinical features, and status at baseline. The latter data would better contextualize study results and help to inform which treatments work best in which patients.

Although FI is a chronic problem, most evidence is only short or intermediate term; longer term information on both benefits and adverse effects would better inform clinical decisions for chronic FI management.

Although we had hoped to use etiology as a basis for assessing FI treatments, we could not because the material on etiology was often unclear, incomplete, or absent. Often no dominant etiology was described. Multiple etiologies may contribute to FI, and etiologies were variably reported or implied in the literature. One-third of RCTs provided no etiologic information, while other authors provided great detail of nonmutually exclusive contributing factors. No study provided information about the frequency of multiple FI etiologies per enrolled adult in baseline patient information tables, such as summary counts per patient or common etiologic combinations. Baseline testing was commonly conducted to ascertain the presence and degree of anal sphincter tearing, but further etiologic identification was less commonly reported. In addition to FI severity at baseline, etiologic multiplicity information could advance understanding of which etiologic factors respond best to given treatments or treatment combinations. Additionally, the term neurogenic FI would benefit from standardization. Aside from its use in the presence of significant nervous system pathology, neurogenic FI appears to be a catch-all term for any FI etiologies in the absence of identified structural pathology. Nonetheless, such distinctions were unclear. Careful descriptions of patients in clinical studies, including baseline characteristics, comorbid conditions (including urinary incontinence) and FI etiologies, would improve understanding of the applicability of results from individual studies and facilitate future literature syntheses.

Well-designed and conducted prospective cohort studies are underused in FI and may better identify baseline patient, FI severity, and etiologic factors more than highly selected RCT samples and also help to determine how such factors affect outcomes from various approaches over time. Most of the observational studies with comparators that we reviewed had extensive study limitations that rendered invalid any treatment conclusions about differences between groups. Common limitations within individual studies were noncomparable intervention and control groups that differed on important prognostic factors at baseline (such as prior surgery or age), and inconsistent timing of outcomes assessments (ranged from months to years and often varied by study group), with no or inadequate efforts to adjust for these differences.

We did not find RCT or OBS evidence for all available FI treatments. The studies included in this review may not reflect the frequency of which specific treatments are used in clinical practice. For example, the easiest treatments to study (drugs) are not necessarily those that are used most often. According to our TEP, topical medications, narcotics, and one or two surgical procedures are no longer commonly used but are still FDA-approved for use in the United States.

Finally, a segment of the FI literature we reviewed lacked baseline patient information that described enrolled adults in person-centered terms. This was especially true for (but not limited to) most SNS studies. Aside from limited treatment metrics of interest to investigators, baseline information surrounding patients and their FI experience (etiology, duration, and severity) was missing; enrolled adults were identified largely by their physiologic (sphincter) metrics. The lack of baseline patient information in a segment of the FI literature was unexpected, given the longstanding recommendations of CONSORT.^158-160

Limitations of the Review Process

Meta-analysis was not possible because numerous outcomes were used.

We were unable to report potential differences in treatment effectiveness within FI etiologic subgroups because FI is often multifactorial. Most studies included adults with mixed FI etiologies. In many instances, little information on etiology was provided at baseline.

Outcomes assessments were often timed at unusual intervals, necessitating our aggregation of evidence into short-term (less than 3 months), intermediate-term (3 to 6 months), or long-term (more than 6 months) effects.

While this review was limited to English-language publications, the possibility of missing clinical trials for FDA-approved treatments in the United States is remote.^166,167

We did not examine the FDA Adverse Event Reporting System for drug harms.

We did not contact authors for missing data or clarification of ambiguous or indeterminable table and text information.

Research Gaps

The overall strength of evidence for treatments for FI in adults was low or insufficient, suggesting that future studies with higher quality could change the conclusions of this review. Many research gaps are identified above in Applicability and Limitations of the Evidence Base. We first provide overarching comments that could advance the field of FI research given the information we noted during this review, followed by specific research gaps that we identified.

Two levels of research improvements would likely advance the field: 1) Clinical research needs to be properly conducted and accurately reported in accordance with CONSORT criteria. For example, it is essential to report data from all randomized adults to minimize attrition bias. Eliminating data from adults who did not respond favorably to treatment, were lost to followup, or had suboptimal treatment compliance is not acceptable. 2) Moving the field to a higher level of research quality may require establishing academic research/clinical centers that will allow for a more structured team approach to research question development, study design selection, enhanced patient input into outcome measure development, the assessment of simultaneous treatments, improved FI etiologic classification, better co-intervention tracking, and the minimization of losses to followup. Funding mechanisms such as the P01 or P50 program grants from the National Institutes of Health could support such clinical research activities. Such centers could be regional centers that do high volume work in FI or they could be research centers that coordinate multicenter studies, providing strong research designs and assuring fidelity to treatment.

Validated outcome measures that capture the FI impact features most meaningful to patients are critical, in addition to the standardized labeling of such measures across studies (see Applicability and Limitations of the Evidence Base); only some of the current outcome measures solicited patient input during instrument construction.

Some specific aspects of FI treatment deserve more attention, including the durability of treatment effects over time. Short-term, easy-to-use treatments, such as drugs and fiber supplements, may be important for planning around important social events, but it is unclear whether their beneficial effects are sustained longer term. Little information was available on rectal irrigation for adults with FI unrelated to spinal cord injury, yet rectal irrigation may prove to be a viable management tool, at least in the short term.

Few if any treatments can entirely cure FI; therefore, information on treatment combinations would benefit the evidence base. This is especially true since many interventions, once initiated, are continued long term. Dietary fiber, intermittent stool modifying drugs, and PFMT-BF may all be used pre- and post-surgery, but patients who would best benefit from combined therapies are not well identified.

Further research is needed to establish what elements of PFMT-BF work for FI, and for how long. Intervention specifics including the optimal type of exercise, duration, number of repetitions, frequency, and specific patients and FI etiologies for which PFMT-BF is effective or ineffective are lacking. Long-term exercise compliance with PFMT-BF for FI is unknown.

Since the benefit of surgical interventions, including sacral nerve stimulation, may diminish over time, more work is needed to determine which additional interventions should be undertaken and when they should be initiated to enhance or prolong the durability of surgical benefits.

We do not know whether the degree of external sphincter defect predicts the outcome of sacral nerve stimulation or nonsurgical treatment. Older studies excluded patients with extensive tears. However, lower-quality observational studies report that even patients with extensive tears improved with SNS up to 1 year.^168,169

Information is limited about the results of treatment options chosen after failed surgical treatments.¹⁰⁰

Better comparison of the benefit-to-harm ratio of FI treatments is needed, especially for invasive and surgical interventions. Substantial and life-altering adverse events occur post-surgery for FI, and these were under identified in RCTs alone.

The long-term effects of injected anal bulking agents are unclear, including their effects on adjacent normal tissues and the location of the injected substance itself.

More work is needed to identify ways to improve outcomes for adults with FI and spinal cord injuries and for older adults in nursing homes. Interventions for nursing home residents with FI focused on the prevention of fecal impaction, though staff-implemented interventions that gave greater attention to fluid, diet, and toileting measures, none of which improved FI outcomes.

Studies of FDA and non-FDA approved interventions in ClinicalTrials.gov that may eventually mitigate some of these research gaps include, but are not limited to, interventions for older adults (multicomponent [behavioral, education, medication] FI intervention delivered by home health nurses to frail elderly patients, effect of a nursing home staff education program on FI in residents, surgically-placed TOPAS mesh sling [pelvic floor] for women with FI, pelvic floor muscle training, botulinum toxin A injections on FI and urgency, long-term safety and efficacy of Solesta [dextranomer injection], percutaneous tibial nerve stimulation for FI, plus several studies of injections of biologics including stem cells). Also, a case series was recently published for a new nonsurgical vaginal bowel control device.¹²

Implications for Clinical and Policy Decisionmaking

The current FI literature base lacks high-quality research evidence to inform clinical practice or policy. Given the clinical complexity of many adults with FI, potential new centers that could generate better research evidence and manage patients in multidisciplinary settings may be the next best step to advance both research and patient care (see Research Gaps above). In the absence of such centers, many adults with FI coordinate their own care between multiple disciplines and multiple sites, making managing FI and FI care a full-time job, especially for more severely afflicted individuals.