Advertisement

Efficacy of Supervised Pelvic Floor Muscle Training and Biofeedback vs Attention-Control Treatment in Adults With Fecal Incontinence

Open AccessPublished:December 20, 2018DOI:https://doi.org/10.1016/j.cgh.2018.12.015

      Background & Aims

      Pelvic floor muscle training (PFMT) in combination with conservative treatment is recommended as first-line treatment for patients with fecal incontinence, although its efficacy is unclear. We investigated whether supervised PFMT in combination with conservative treatment is superior to attention-control massage treatment and conservative treatment in adults with fecal incontinence.

      Methods

      We performed a randomized, controlled, superiority trial of patients with fecal incontinence at a tertiary care center at a public hospital in Denmark. Ninety-eight adults with fecal incontinence were randomly assigned to groups that received supervised PFMT and biofeedback plus conservative treatment or attention-control treatment plus conservative treatment. The primary outcome was rating of symptom changes, after 16 weeks, based on scores from the Patient Global Impression of Improvement scale. Secondary outcomes were changes in the Vaizey incontinence score (Vaizey Score), Fecal Incontinence Severity Index, and Fecal Incontinence Quality of Life Scale.

      Results

      In the intention-to-treat analysis, participants in the PFMT group were significantly more likely to report improvement in incontinence symptoms based on Patient Global Impression of Improvement scale scores (unadjusted odds ratio, 5.16; 95% CI, 2.18–12.19; P = .0002). The PFMT group had a larger reduction in the mean Vaizey Score (reduction, -1.83 points; 95% CI, –3.57 to –0.08; P = .04). There were no significant differences in condition-specific quality of life. In the per-protocol analyses, the superiority of PFMT was increased. No adverse events were reported.

      Conclusions

      This randomized controlled trial of adults with fecal incontinence provides support for a superior effect of supervised PFMT in combination with conservative treatment compared with attention-control massage treatment and conservative treatment. We found that participants who received supervised PFMT had 5-fold higher odds of reporting improvements in fecal incontinence symptoms and had a larger mean reduction of incontinence severity based on the Vaizey Score compared with attention control massage treatment. Clinicaltrials.gov no: NCT01705535.

      Keywords

      Abbreviations used in this paper:

      EMG (electromyography), FIQL (Fecal Incontinence Quality of Life Scale), FISI (Fecal Incontinence Severity Index), MCID (minimum clinically important difference), PFMT (pelvic floor muscle training), PGI-I (Patient Global Impression of Improvement Scale), Vaizey Score (Vaizey Incontinence Score)

       Background

      Pelvic floor muscle training in combination with conservative treatment is recommended as first-line treatment for fecal incontinence. We performed a randomized controlled trial to determine the effects of supervised pelvic floor muscle training.

       Findings

      Participants receiving supervised pelvic floor muscle training had a 5-fold higher odds of reporting improvements of fecal incontinence and had a larger mean reduction in incontinence severity than an attention-control group.

       Implication for patient care

      Pelvic floor muscle training should be offered, in combination with conservative treatment, as first-line treatment for adults with fecal incontinence.
      Approximately 8% to 9% of the adult population is suffering from fecal incontinence.
      • Ng K.-S.
      • Sivakumaran Y.
      • Nassar N.
      • et al.
      Fecal incontinence: community prevalence and associated factors--a systematic review.
      The condition is surrounded by taboo
      • Brown H.W.
      • Wexner S.D.
      • Segall M.M.
      • et al.
      Quality of life impact in women with accidental bowel leakage.
      • Bharucha A.E.
      • Zinsmeister A.R.
      • Locke G.R.
      • et al.
      Prevalence and burden of fecal incontinence: a population-based study in women.
      ; it can have a devastating impact on quality of life
      • Brown H.W.
      • Wexner S.D.
      • Segall M.M.
      • et al.
      Quality of life impact in women with accidental bowel leakage.
      • Bharucha A.E.
      • Zinsmeister A.R.
      • Locke G.R.
      • et al.
      Prevalence and burden of fecal incontinence: a population-based study in women.
      • Olsson F.
      • Berterö C.
      Living with faecal incontinence: trying to control the daily life that is out of control.
      and lead to major limitations in daily life.
      • Bharucha A.E.
      • Zinsmeister A.R.
      • Locke G.R.
      • et al.
      Prevalence and burden of fecal incontinence: a population-based study in women.
      • Olsson F.
      • Berterö C.
      Living with faecal incontinence: trying to control the daily life that is out of control.
      Fecal incontinence often has multifactorial causes, and the recommended first-line treatment consists of a multimodal approach including information, fiber supplements, antidiarrheal medication, laxatives, pelvic floor muscle training (PFMT), and biofeedback training.
      • Norton C.
      Biofeedback and/or sphincter exercises for the treatment of faecal incontinence in adults.
      • Whitehead W.E.
      • Rao S.S.C.
      • Lowry A.
      • et al.
      Treatment of fecal incontinence: State of the Science Summary for the National Institute of Diabetes and Digestive and Kidney Diseases Workshop.
      PFMT is defined as systematic training of the pelvic floor muscles and the external anal sphincter with the aim of increasing muscle strength, endurance and/or coordination.
      • Norton C.
      Biofeedback and/or sphincter exercises for the treatment of faecal incontinence in adults.
      • Bo K.
      • Frawley H.C.
      • Haylen B.T.
      • et al.
      An International Urogynecological Association (IUGA)/International Continence Society (ICS) joint report on the terminology for the conservative and nonpharmacological management of female pelvic floor dysfunction.
      In most trials, PFMT is supplemented with biofeedback training, in which the patient receives feedback on a voluntary pelvic floor muscle contraction and/or rectal filling.
      • Norton C.
      Biofeedback and/or sphincter exercises for the treatment of faecal incontinence in adults.
      • Whitehead W.E.
      • Rao S.S.C.
      • Lowry A.
      • et al.
      Treatment of fecal incontinence: State of the Science Summary for the National Institute of Diabetes and Digestive and Kidney Diseases Workshop.
      • Bo K.
      • Frawley H.C.
      • Haylen B.T.
      • et al.
      An International Urogynecological Association (IUGA)/International Continence Society (ICS) joint report on the terminology for the conservative and nonpharmacological management of female pelvic floor dysfunction.
      Evidence from randomized controlled trials regarding the effect of PFMT for fecal incontinence is lacking.
      • Norton C.
      Biofeedback and/or sphincter exercises for the treatment of faecal incontinence in adults.
      A recent Cochrane systematic review
      • Norton C.
      Biofeedback and/or sphincter exercises for the treatment of faecal incontinence in adults.
      included 20 randomized controlled trials to investigate the efficacy of PFMT and/or biofeedback training for the treatment of fecal incontinence in adults. Most of the interventions were complex and included multiple active elements. With the exception of 2 trials, which evaluated vs surgery or evaluated PFMT and biofeedback training after surgery vs surgery alone, the included trials compared different training modalities and thus lacked a nontraining comparator.
      • Norton C.
      Biofeedback and/or sphincter exercises for the treatment of faecal incontinence in adults.
      Two recent trials published after the Cochrane review
      • Sjödahl J.
      • Walter S.A.
      • Johansson E.
      • et al.
      Combination therapy with biofeedback, loperamide, and stool-bulking agents is effective for the treatment of fecal incontinence in women - a randomized controlled trial.
      • Damon H.
      • Siproudhis L.
      • Faucheron J.-L.
      • et al.
      Perineal retraining improves conservative treatment for faecal incontinence: a multicentre randomized study.
      compared the effect of PFMT and biofeedback training with a group not receiving PFMT. However, neither of these trials controlled for the nonspecific trial effect associated with the attention given by the health care professional delivering the training interventions.
      • Sjödahl J.
      • Walter S.A.
      • Johansson E.
      • et al.
      Combination therapy with biofeedback, loperamide, and stool-bulking agents is effective for the treatment of fecal incontinence in women - a randomized controlled trial.
      • Damon H.
      • Siproudhis L.
      • Faucheron J.-L.
      • et al.
      Perineal retraining improves conservative treatment for faecal incontinence: a multicentre randomized study.
      This nonspecific trial effect is known to exist and may result in better outcomes for trial participants.
      • Braunholtz D.A.
      • Edwards S.J.
      • Lilford R.J.
      Are randomized clinical trials good for us (in the short term)? Evidence for a “trial effect.”.
      To evaluate the effect of PFMT, there is a need for a trial that uses a comparator to control for this nonspecific trial effect associated with the attention given by the health care professional. The aim of the current trial was to investigate if supervised PFMT in combination with conservative treatment (mainly information) is superior to attention-control massage treatment and conservative treatment in adults with fecal incontinence. This possible superiority effect was based on the primary outcome of changes in fecal incontinence symptoms after 16 weeks of treatment using the Patient Global Impression of Improvement Scale (PGI-I).

      Methods

       Trial Design

      This was a randomized, controlled, superiority trial comparing supervised PFMT with attention-control massage treatment. Both treatments were given in combination with conservative treatment. The trial is reported in accordance with Consolidated Standards of Reporting Trials.
      • Moher D.
      • Hopewell S.
      • Schulz K.F.
      • et al.
      CONSORT 2010 explanation and elaboration: updated guidelines for reporting parallel group randomised trials.
      • Boutron I.
      • Altman D.G.
      • Moher D.
      • et al.
      CONSORT Statement for Randomized Trials of Nonpharmacologic Treatments: A 2017 Update and a CONSORT Extension for Nonpharmacologic Trial Abstracts.
      Reporting details can be found in the Supplementary Materials and Methods section, if indicated. The trial was preregistered with ClinicalTrials.gov (NCT01705535), and a detailed trial protocol, including the statistical analysis plan, was published before the data were analyzed and the trial was unblinded.
      • Ussing A.
      • Dahn I.
      • Due U.
      • et al.
      Supervised pelvic floor muscle training versus attention-control massage treatment in patients with faecal incontinence: statistical analysis plan for a randomised controlled trial.

       Participants and Setting

      The trial was conducted at Copenhagen University Hospital (Hvidovre, Denmark). From October 15, 2012, to December 15, 2015, consecutive patients referred for treatment of fecal incontinence at the Department of Surgical and Medical Gastroenterology were screened for eligibility.
      Inclusion criteria were fecal incontinence for at least 6 months and a minimum age of 18 years. Exclusion criteria were as follows: pregnancy, chronic diarrhea, severe neurologic disorders (multiple sclerosis, Parkinson’s disease, spinal cord injury, major stroke, or neuromuscular junction diseases), rectal prolapse, previous cancer surgery or radiotherapy in the lower abdomen, inadequate Danish, cognitive impairment, and having received more than 2 sessions of individually supervised PFMT in the previous 12 months. The trial was approved by the Regional Health Research Ethics Committee (ID: H-2-2012-067) and the Danish Data Protection Agency (ID: HVH-2012-031). Written informed consent was obtained from all participants.

       Randomization and Blinding

      Participants were allocated to 2 parallel groups (supervised PFMT or attention-control massage treatment) by simple randomization using a 1:1 ratio. Randomization and allocation were managed by staff not otherwise involved in the trial. A computer-generated sequence of random group assignments was produced and managed using sealed, opaque, numbered envelopes. The primary investigator (A.U.), who was blinded to the allocation sequence, enrolled the participants. It was not possible to blind the participants or the physiotherapists delivering the PFMT. Instead, we tried to blind the participants to the trial hypothesis.
      • Hróbjartsson A.
      • Boutron I.
      Blinding in randomized clinical trials: imposed impartiality.
      This was attempted by presenting the 2 treatments as equal both in the verbal and written information given at enrolment and at follow-up evaluation. All health care providers and outcome assessors were instructed not to reveal the trial hypothesis. The nurses delivering the conservative treatment and all outcome assessors were blinded.

       Interventions

       Conservative treatment program

      Both trial groups received the same conservative treatment program consisting of standard information about optimizing bowel emptying and the use of fiber supplements and if appropriate, information about antidiarrheal medication (loperamide) and/or laxatives with local effect in the rectum (glycerol). More details are provided in the Supplementary Materials and Methods section.

       Supervised pelvic floor muscle training and biofeedback

      The PFMT intervention has been described previously in detail.
      • Ussing A.
      • Dahn I.
      • Due U.
      • et al.
      Supervised pelvic floor muscle training versus attention-control massage treatment in patients with faecal incontinence: statistical analysis plan for a randomised controlled trial.
      The protocol for the PFMT intervention, including the complete training materials, is freely available for download (https://doi.org/10.1016/j.conctc.2017.07.006). In brief, the participants in the PFMT group received 6 individual treatments distributed over 16 weeks. Each treatment lasted 45 minutes and consisted of individually supervised PFMT. The participants were taught correct pelvic floor contractions verbally and by digital vaginal and rectal examination. At every session, they received an examination of their pelvic floor muscle function by digital vaginal and rectal examination and by intra-anal electromyography (EMG) biofeedback. They were instructed to perform an individually tailored home training program daily. The program progressed individually and consisted of 3 sets of 10 pelvic floor muscle contractions sustained for up to 10 seconds and 2 sets of 3 contractions sustained for up to 30 seconds. A training diary was used to quantify training adherence. More details are provided in the Supplementary Materials and Methods section.

       Attention-control massage treatment

      To match the attention associated with the supervised PFMT, the attention-control group received 6 individual treatments performed over 16 weeks. Each treatment lasted 30 minutes and consisted of massage of the neck and back. The participants were given no instructions in PFMT.

       Outcome Measures

      The primary outcome was improvement according to the PGI-I after the 16 weeks of treatment.
      • Yalcin I.
      • Bump R.C.
      Validation of two global impression questionnaires for incontinence.
      By using the PGI-I, the participants rated changes in incontinence symptoms after treatment using a 7-point scale (very much better, much better, a little better, unchanged, a little worse, much worse, or very much worse).
      • Yalcin I.
      • Bump R.C.
      Validation of two global impression questionnaires for incontinence.
      Supporting secondary outcomes were the Vaizey Incontinence Score (Vaizey Score),
      • Vaizey C.J.
      • Carapeti E.
      • Cahill J.A.
      • et al.
      Prospective comparison of faecal incontinence grading systems.
      the Fecal Incontinence Severity Index (FISI),
      • Rockwood T.H.
      • Church J.M.
      • Fleshman J.W.
      • et al.
      Patient and surgeon ranking of the severity of symptoms associated with fecal incontinence.
      and the Fecal Incontinence Quality of Life Scale (FIQL).
      • Rockwood T.H.
      • Church J.M.
      • Fleshman J.W.
      • et al.
      Fecal incontinence quality of life scale.
      All questionnaires were self-administered. Other secondary outcomes were anorectal manometry (resting pressure and maximum squeeze increment pressure), rectal capacity measurements (sensory threshold, urge sensation, and maximum tolerable volume), and fecal incontinence episodes recorded via 14-day bowel diary. For more details about the anorectal manometry and rectal capacity measurements and for the covariates collected at baseline, see the Supplementary Materials and Methods section. Outcome measures were assessed at baseline before randomization and within 14 days after completion of the intervention (primary end point). Data collection for the 36-month follow-up evaluation still is ongoing and results from this long-term follow-up evaluation will be reported separately. Deviations from the trial protocol and registration, together with explanations, have been described elsewhere.
      • Ussing A.
      • Dahn I.
      • Due U.
      • et al.
      Supervised pelvic floor muscle training versus attention-control massage treatment in patients with faecal incontinence: statistical analysis plan for a randomised controlled trial.

       Sample Size

      Based on previous findings in the literature, we hypothesized that 30% in the attention-control group
      • Heymen S.
      • Scarlett Y.
      • Jones K.
      • et al.
      Randomized controlled trial shows biofeedback to be superior to pelvic floor exercises for fecal incontinence.
      • Norton C.
      • Chelvanayagam S.
      • Wilson-Barnett J.
      • et al.
      Randomized controlled trial of biofeedback for fecal incontinence.
      and 60% in the PFMT group
      • Norton C.
      Biofeedback and/or sphincter exercises for the treatment of faecal incontinence in adults.
      • Solomon M.J.
      • Pager C.K.
      • Rex J.
      • et al.
      Randomized, controlled trial of biofeedback with anal manometry, transanal ultrasound, or pelvic floor retraining with digital guidance alone in the treatment of mild to moderate fecal incontinence.
      would achieve improvement of their fecal incontinence symptoms. We consider this difference to represent the minimum clinically important difference (MCID). Improvement was defined as a score of very much better, much better, or a little better at the PGI-I (primary outcome). Based on a binomial distribution, we needed a sample size of 84 to show this difference in treatment effects between groups using a statistical power of 80% and a significance level of 5% (2-tailed). To account for a drop-out rate of up to 16%, we included 102 participants.

       Statistical Analyses

      The statistical analysis plan has been described previously in full.
      • Ussing A.
      • Dahn I.
      • Due U.
      • et al.
      Supervised pelvic floor muscle training versus attention-control massage treatment in patients with faecal incontinence: statistical analysis plan for a randomised controlled trial.
      The primary outcome, the PGI-I ratings after 16 weeks of treatment, were analyzed by logistic regression using the intention-to-treat principle. For the primary analysis, the PGI-I outcome was dichotomized into symptom improvement (3 upper values on the PGI-I) or unchanged or worsening symptoms (4 lower values at PGI-I). Missing data for the PGI-I were imputed with a score of unchanged in both groups.
      As a sensitivity test, we conducted a multiple logistic regression of the primary outcome. The model was adjusted for use of fiber supplements, use of antidiarrheal medication, stool consistency, incontinence type, presence of urinary incontinence, and confirmed sphincter injuries as described in the statistical analysis plan. In the review process the sex and age variables were added to the adjusted analyses. The covariates were added one at a time and subsequently all at once. As a goodness-of-fit, the same variables were examined as modifiers one by one, by adding them in interaction terms with the group variable to the fully adjusted model.
      Finally, per-protocol logistic regression analyses were conducted following the principles outlined earlier. These analyses included participants who had attended at least 4 of 6 PFMT consultations, and had documented home training on at least 70% of the possible training days.
      The FIQL, FISI, Vaizey Score, anorectal manometry data, rectal capacity measurements, and number of fecal incontinence episodes were analyzed using general linear regression. The primary analyses were unadjusted. As sensitivity analyses, we conducted adjusted analyses adding all the covariates at once. The analyses were conducted using the per-protocol principle as described earlier. For the FIQL, FISI, and Vaizey Score, the analyses also were conducted using the intention-to-treat principle. Missing values for total scores were imputed by multiple imputation using the monotone method, except for the bowel diary, for which we used the Markov Chain Monte Carlo method. A total of 10 imputed data sets were used. The imputation models included baseline values for the missing end point value and group. All general linear regression models were controlled for goodness-of-fit by visual evaluation of residual plots to check the assumptions of linearity, variance homogeneity, and normal distribution of the residuals.
      All analyses were conducted in SAS version 9.4 (SAS institute Inc, Cary, NC). All estimates are reported along with 95% CIs and actual P values (2-sided). The level of significance was set at .05. All authors had access to the trial data and reviewed and approved the final manuscript.

      Results

       Participants, Baseline Characteristics, and Attrition

      Between October 15, 2012, and December 15, 2015, we screened 458 patients for eligibility. In total, 98 participants were randomized, 49 to supervised PFMT and 49 to attention-control massage treatment. The overall patient flow including attrition can be seen in Supplementary Figure 1. The baseline characteristics are shown in Table 1. The Participants in the PFMT group were slightly older than the participants in the attention-control group, with median ages of 65 and 58 years, respectively. There were 6 men in the attention-control group and 3 in the PFMT group.
      Table 1Baseline Characteristics
      VariablePFMT (n = 49)Attention-control (n = 49)
      Demographics
       Age, y, median (range)65.13 (34.30–77.29)58.43 (26.80–89.23)
       Women46 (93.9%)43 (87.8%)
       Ethnicity, North Europe46 (93.9%)47 (95.9%)
       Body mass index, kg/m2, median (range)24.5 (17.3–36.1)23.2 (18.3–47.6)
      Incontinence history
       Duration of fecal incontinence, y, median (range)3.25 (0.67–50)4.0 (0.75–24)
      Stool consistency
       Loose/mushy19 (38.8%)14 (28.6%)
       Soft, but formed11 (22.5%)13 (26.5%)
       Hard, firm3 (6.1%)1 (2.0%)
       Varying16 (32.7%)21 (42.9%)
      Incontinence type
       Passive10 (20.8%)16 (33.3%)
       Urgency18 (37.5%)14 (29.2%)
       Mixed, both20 (41.7%)18 (37.5%)
      Use of medication/fiber
       Fiber supplements24 (49%)22 (44.9%)
       Antidiarrheal medication4 (8.2%)5 (10.2%)
       Laxatives5 (10.2%)7 (14.3%)
      Medical history
       Previous urogynecological surgery30 (61.2%)21 (43.8%)
       Previous anal surgery16 (33.3%)21 (43.8%)
       Presence of urinary incontinence27 (55.1%)24 (49%)
       Diabetes mellitus2 (4.1%)4 (8.2%)
       Inflammatory bowel diseases2 (4.1%)4 (8.2%)
       Other diseases21 (43.8%)27 (55.1%)
      Obstetric history
       Number of women who delivered44 (95.7%)36 (83.7%)
       Parity among women, median (range)2 (0–4)2 (0–4)
      Self-reported questionnaires
       Vaizey Score, median (range)14 (6–22)14 (2–20)
       FISI, median (range)27 (8–57)28 (8–50)
       FIQL Lifestyle, median (range)3.40 (1.20–4.00)3.20 (1.46–4.00)
       FIQL Coping/behavior, median (range)2.00 (1.11–4.00)2.00 (1.00–4.00)
       FIQL Depression/self-perception, median (range)3.19 (1.43–3.81)2.91 (1.48–3.81)
       FIQL Embarrassment, median (range)2.33 (1.00–3.67)2.33 (1.00–3.67)
      14-day bowel dairy
       Number of fecal incontinence episodes, median (range)2.0 (0–23)1.5 (0–32)
      Anorectal manometry
       Resting pressure, mm Hg, median (range)67.6 (38.6–124.0)72.2 (34.4–163.5)
       Maximum squeeze increment pressure, mm Hg, median (range)109.6 (49.2–216.3)124.4 (55.6–256.30)
      Rectal capacity measurements
       First sensation, mL, median (range)30 (10–60)20 (10–90)
       Urge sensation, mL, median (range)60 (20–180)50 (20–170)
       Maximum tolerable volume, mL, median (range)110 (40–350)100 (30–400)
      Endoanal ultrasound, assessed at 16 weeks
       Anal sphincter injuries19 (38.8%)14 (30.4%)
       Defect <120°8 (42.1%)2 (14.3%)
       Defect ≥120°11 (57.9%)12 (85.7%)
      NOTE. Values are numbers (percentages) unless stated otherwise.
      FIQL, Fecal Incontinence Quality of Life Scale (range, 1–4); FISI, Fecal Incontinence Severity Index (range, 0–61); PFMT, pelvic floor muscle training; Vaizey Score, Vaizey Incontinence Score (range, 0–24).

       Outcomes

       Primary outcome

      In the PFMT group, 35 participants (74.5%) reported improvement of their incontinence symptoms vs 16 participants (35.5%) in the attention-control group. The exact distribution of the participant PGI-I ratings is shown in Table 2.
      Table 2Number and Percentages of the PGI-I Ratings in the 2 Groups
      PGI-I ratingsPFMT (n = 47), frequency (%)Attention-control (n = 45), frequency (%)
      Very much better5 (10.6)2 (4.4)
      Much better13 (27.7)6 (13.3)
      A little better17 (36.2)8 (17.8)
      Improvement
      Boldface shows PGI-I scores were dichotomized as follows: improvement: a score of very much better, much better, or a little better; unchanged/deterioration: a score of unchanged, a little worse, much worse, or very much worse. Across all 6 cut-off points on the PGI-I scale, the PFMT group had significantly higher odds for reporting their symptom change in a higher/better category than the attention-control group (P = .0003).
      35 (74.5)16 (35.6)
      Unchanged12 (25.5)26 (57.8)
      A little worse0 (0)1 (2.2)
      Much worse0 (0)2 (4.4)
      Very much worse0 (0)0 (0)
      Unchanged or deterioration
      Boldface shows PGI-I scores were dichotomized as follows: improvement: a score of very much better, much better, or a little better; unchanged/deterioration: a score of unchanged, a little worse, much worse, or very much worse. Across all 6 cut-off points on the PGI-I scale, the PFMT group had significantly higher odds for reporting their symptom change in a higher/better category than the attention-control group (P = .0003).
      12 (25.5)29 (64.4)
      Totaln = 47 (100)n = 45 (100)
      PFMT, pelvic floor muscle training; PGI-I, Patient Global Impression of Improvement Scale.
      a Boldface shows PGI-I scores were dichotomized as follows: improvement: a score of very much better, much better, or a little better; unchanged/deterioration: a score of unchanged, a little worse, much worse, or very much worse. Across all 6 cut-off points on the PGI-I scale, the PFMT group had significantly higher odds for reporting their symptom change in a higher/better category than the attention-control group (P = .0003).
      The unadjusted and adjusted results for the primary outcome and the per-protocol analyses are shown in Table 3. The intention-to-treat analyses showed an unadjusted odds ratio of 5.16 (95% CI, 2.18–12.19) for reporting improvement in the PFMT group compared with the attention-control group (P = .0002). When only a score of very much better or much better was defined as improvement the result still was significant, with an unadjusted odds ratio of 2.98 (95% CI, 1.15–7.73; P = .025).
      Table 3Primary Outcome: Odds Ratios for Reporting Improvement on the PGI-I in the PFMT Group After 16 Weeks of Treatment
      ITT analysis, unadjusted (n = 98)
      Imputed 6 values for missing PGI-I values.
      ITT analysis, adjusted
      Adjusted for age, sex, antidiarrheal medication at baseline, fiber supplements at baseline, incontinence type, presence of urinary incontinence, stool consistency, and anal sphincter injuries (at 16-week follow-up evaluation).
      (n = 92)
      No imputations were made because of simultaneous missing values for both outcome and covariates.
      Per-protocol analysis, unadjusted (n = 82)
      Imputed 4 values for missing PGI-I values.
      Per-protocol analysis, adjusted
      Adjusted for age, sex, antidiarrheal medication at baseline, fiber supplements at baseline, incontinence type, presence of urinary incontinence, stool consistency, and anal sphincter injuries (at 16-week follow-up evaluation).
      (n = 78)
      No imputations were made because of simultaneous missing values for both outcome and covariates.
      OR (95% CI)5.16 (2.18–12.19)7.26 (2.55–20.69)7.66 (2.75–21.38)7.97 (2.39–26.55)
      P value.0002.0002.0001.0007
      NOTE. PGI-I scores were dichotomized as participants reporting either improvement of their incontinence symptoms (a score of very much better, much better, or a little better) or unchanged/deterioration (a score of unchanged, a little worse, much worse, or very much worse).
      ITT, intention-to-treat; OR, odds ratio; PFMT, pelvic floor muscle training; PGI-I, Patient Global Impression of Improvement Scale.
      a Imputed 6 values for missing PGI-I values.
      b Adjusted for age, sex, antidiarrheal medication at baseline, fiber supplements at baseline, incontinence type, presence of urinary incontinence, stool consistency, and anal sphincter injuries (at 16-week follow-up evaluation).
      c No imputations were made because of simultaneous missing values for both outcome and covariates.
      d Imputed 4 values for missing PGI-I values.
      There were no signs of interaction between the groups and any of the examined covariates.
      Secondary and explorative analyses of the primary outcome including a subgroup analysis according to incontinence severity at baseline are shown in Supplementary Table 1.

       Secondary outcomes

      The intention-to-treat analysis of the Vaizey Score showed that the PFMT group had a statistically significantly larger mean reduction in incontinence severity than the attention-control group, the unadjusted mean difference in change being -1.83 points (95% CI, -3.57 to -0.08; P = .04) (Table 4). There were no statistically significant differences between groups in condition-specific quality of life or in incontinence severity measured with FISI (Table 4). The per-protocol analysis showed statistically significant differences in favor of PFMT both for the Vaizey Score and the 2 FIQL subscales of Lifestyle and Coping/behavior (Table 4).
      Table 4Supportive Outcomes
      Change from baseline to follow-up evaluation, mean change (SD)16-week follow-up evaluation

      Mean between-group difference in change from baseline to follow-up evaluation, mean difference (95% CI), P value
      ITT analysesPer-protocol analyses
      OutcomesPFMT (n = 47)
      No imputations.
      Attention-control (n = 45)
      No imputations.
      Unadjusted (n = 98)
      Imputed 6 values for missing end point values for each of the secondary outcomes.
      Adjusted
      Adjusted for age, sex, antidiarrheal medication at baseline, fiber supplements at baseline, incontinence type, presence of urinary incontinence, stool consistency, and anal sphincter injuries (at 16-week follow-up evaluation).
      (n = 92)
      No imputations were made because of simultaneous missing values for both outcome and covariates.
      Unadjusted (n = 82)
      Imputed 4 values for missing end point values for each of the secondary outcomes.
      Adjusted
      Adjusted for age, sex, antidiarrheal medication at baseline, fiber supplements at baseline, incontinence type, presence of urinary incontinence, stool consistency, and anal sphincter injuries (at 16-week follow-up evaluation).
      (n = 78)
      No imputations were made because of simultaneous missing values for both outcome and covariates.
      Vaizey Score range, 0–24−2.57 (4.51)−0.67 (4.05)−1.83 (−3.57 to −0.082),

      .040
      −1.87 (−3.88 to 0.13),

      .067
      −2.08 (−4.09 to −0.07),

      .042
      −2.24 (−4.57 to 0.097),

      .060
      FISI range, 0–61−4.23 (10.86)−0.40 (13.05)−3.80 (−8.65 to 1.061),

      .13
      −4.64 (−10.029 to 0.76),

      .092
      −5.12 (−10.61 to 0.37),

      .067
      −5.88 (−12.16 to 0.40),

      .067
      FIQL
      Lifestyle0.16 (0.48)−0.031 (0.50)0.19 (−0.0068 to 0.38),

      .059
      0.18 (−0.033 to 0.40),

      .010
      0.22 (0.011–0.44),

      .040
      0.24 (−0.0017 to 0.49),

      .052
      Coping/behavior0.20 (0.59)0.073 (0.43)0.12 (−0.084 to 0.33),

      .24
      0.08 (−0.15 to 0.31),

      .48
      0.23 (0.01–0.45),

      .037
      0.18 (−0.059 to 0.43),

      .14
      Depression/self-perception−0.017 (0.46)0.014 (0.48)−0.031 (−0.22 to 0.16),

      .75
      −0.086 (−0.30 to 0.13),

      .43
      0.022 (−0.18 to 0.22),

      .83
      −0.016 (−0.25 to 0.22),

      .90
      Embarrassment0.19 (0.80)0.19 (0.66)−0.0056 (−0.30 to 0.29),

      .97
      0.042 (−0.29 to 0.37),

      .80
      0.14 (−0.17 to 0.45),

      .38
      0.095 (−0.27 to 0.46),

      .61
      FIQL, Fecal Incontinence Quality of Life Scale (a condition-specific quality-of-life scale consisting of 4 subscales: Lifestyle, Coping/behavior, Depression/self-perception, and Embarrassment; subscales range from 1–4, with 1 = worst quality of life, 4 = best quality of life); FISI, Fecal Incontinence Severity Index (with 0 = complete continence, 61 = complete incontinence); ITT, intention-to-treat; PFMT, pelvic floor muscle training; Vaizey Score, Vaizey Incontinence Score (with 0 = complete continence, 24 = complete incontinence).
      a No imputations.
      b Imputed 6 values for missing end point values for each of the secondary outcomes.
      c Adjusted for age, sex, antidiarrheal medication at baseline, fiber supplements at baseline, incontinence type, presence of urinary incontinence, stool consistency, and anal sphincter injuries (at 16-week follow-up evaluation).
      d No imputations were made because of simultaneous missing values for both outcome and covariates.
      e Imputed 4 values for missing end point values for each of the secondary outcomes.
      There were no significant differences in change between groups in the anorectal manometry measurements, the rectal capacity measurements, or in the number of fecal incontinence episodes as recorded in the 14-day bowel diary (Supplementary Table 2).
      The median number of consultations was 5 (range, 0–6) in both groups. The PFMT group documented a median of 95 home training days (range, 0–112 d) in the training diaries, corresponding to 84.8% of the possible training days. In the PFMT group, 43 (87.8%) attended at least 4 of the 6 consultations, and 33 (67.4%) documented home training on at least 70% of the possible training days. These 33 participants were included in the per-protocol analyses. No adverse events were reported.

      Discussion

      After 16 weeks of treatment, the participants in the PFMT group had 5.16 times higher odds of reporting improvement of incontinence symptoms compared with the attention-control group.
      Two other trials have reported results comparable with our results.
      • Damon H.
      • Siproudhis L.
      • Faucheron J.-L.
      • et al.
      Perineal retraining improves conservative treatment for faecal incontinence: a multicentre randomized study.
      • Heymen S.
      • Scarlett Y.
      • Jones K.
      • et al.
      Randomized controlled trial shows biofeedback to be superior to pelvic floor exercises for fecal incontinence.
      Damon et al
      • Damon H.
      • Siproudhis L.
      • Faucheron J.-L.
      • et al.
      Perineal retraining improves conservative treatment for faecal incontinence: a multicentre randomized study.
      reported an odds ratio of 2.34 for successful improvement after PFMT and biofeedback in combination with conservative treatment. However, in contrast to our findings, this was not reflected in reduced severity. This may be explained by different severity measures. Heymen et al
      • Heymen S.
      • Scarlett Y.
      • Jones K.
      • et al.
      Randomized controlled trial shows biofeedback to be superior to pelvic floor exercises for fecal incontinence.
      showed symptom relief for 76% after supervised PFMT and sensitivity biofeedback vs 41% in a nonsupervised PFMT group. This was accompanied by a significant reduction in the FISI. In contrast, we were not able to show significant changes in severity on the FISI, but all mean between-group differences in the FISI score, however, were in favor of PFMT and greater than the MCID of -3.56.
      • Jelovsek J.E.
      • Chen Z.
      • Markland A.D.
      • et al.
      Minimum important differences for scales assessing symptom severity and quality of life in patients with fecal incontinence.
      Contrary to our findings, Norton et al
      • Norton C.
      • Chelvanayagam S.
      • Wilson-Barnett J.
      • et al.
      Randomized controlled trial of biofeedback for fecal incontinence.
      did not show additional benefit from offering PFMT and/or biofeedback to conservative treatment. This may be explained by the fact that the conservative treatment contained an urge-resistance program with instructions of resistance to fecal urgency by holding sustained sphincter contractions for progressively longer time durations (up to 10 minutes). Whether these contractions for urge-resistance differ from voluntary pelvic floor muscle contractions and whether the control group reflected a true control group could be questioned. The conclusions of Norton et al
      • Norton C.
      • Chelvanayagam S.
      • Wilson-Barnett J.
      • et al.
      Randomized controlled trial of biofeedback for fecal incontinence.
      of no additional benefit from PFMT could be subject for discussion.
      Sjödahl et al
      • Sjödahl J.
      • Walter S.A.
      • Johansson E.
      • et al.
      Combination therapy with biofeedback, loperamide, and stool-bulking agents is effective for the treatment of fecal incontinence in women - a randomized controlled trial.
      did not show an effect of supervised PFMT in combination with strength biofeedback compared with conservative treatment. This result could be owing to small sample size, minimal supervision, and/or the chosen outcome measures (bowel diary and anorectal manometry). When PFMT and conservative treatment were given in combination, the trial showed a significant reduction in incontinence episodes from baseline.
      In the current trial, the mean between-group differences in changes on the Vaizey Score were between -1.83 and -2.24 points, this is comparable with results reported by other investigators
      • Norton C.
      • Chelvanayagam S.
      • Wilson-Barnett J.
      • et al.
      Randomized controlled trial of biofeedback for fecal incontinence.
      ,
      • Bols E.M.J.
      • Hendriks H.J.M.
      • Berghmans L.C.M.
      • et al.
      Responsiveness and interpretability of incontinence severity scores and FIQL in patients with fecal incontinence: a secondary analysis from a randomized controlled trial.
      • Johannessen H.H.
      • Wibe A.
      • Stordahl A.
      • et al.
      Do pelvic floor muscle exercises reduce postpartum anal incontinence? A randomised controlled trial.
      • Young C.J.
      • Zahid A.
      • Koh C.E.
      • et al.
      A randomised control trial of four different regimes of biofeedback program in the treatment of faecal incontinence.
      and may represent the MCID.
      • Bols E.M.J.
      • Hendriks H.J.M.
      • Berghmans L.C.M.
      • et al.
      Responsiveness and interpretability of incontinence severity scores and FIQL in patients with fecal incontinence: a secondary analysis from a randomized controlled trial.
      • Johannessen H.H.
      • Wibe A.
      • Stordahl A.
      • et al.
      Do pelvic floor muscle exercises reduce postpartum anal incontinence? A randomised controlled trial.
      • Bols E.M.J.
      • Hendriks E.J.M.
      • Deutekom M.
      • et al.
      Inconclusive psychometric properties of the Vaizey score in fecally incontinent patients: a prospective cohort study.
      Similar to other investigators,
      • Damon H.
      • Siproudhis L.
      • Faucheron J.-L.
      • et al.
      Perineal retraining improves conservative treatment for faecal incontinence: a multicentre randomized study.
      • Heymen S.
      • Scarlett Y.
      • Jones K.
      • et al.
      Randomized controlled trial shows biofeedback to be superior to pelvic floor exercises for fecal incontinence.
      we were not able to show the effect of PFMT on the FIQL in the intention-to-treat analysis. We found a significant effect on the Lifestyle and Coping/behavior subscales only in the per-protocol analyses. Our sample size was calculated in relation to the primary outcome and it is possible that we did not have enough power to show significant differences on the FIQL.
      In the current trial, more than one third of patients had sphincter injuries confirmed at endoanal ultrasound, this reflects the tertiary setting of our trial. However, our results may be highly relevant in a primary setting because there is an unmet need for treatment of fecal incontinence in primary health care,
      • Whitehead W.E.
      • Palsson O.S.
      • Simren M.
      Treating fecal incontinence: an unmet need in primary care medicine.
      and the interventions do not necessarily need to be conducted at specialized centers.
      • Whitehead W.E.
      • Palsson O.S.
      • Simren M.
      Treating fecal incontinence: an unmet need in primary care medicine.
      The number of sphincter injuries may have contributed to the fact that we were not able to show any significant differences at the anorectal manometry. Another reason could be that the median baseline resting and maximum squeeze pressure in the current trial were comparable with the median for healthy controls measured with comparable equipment.
      • Gundling F.
      • Seidl H.
      • Scalercio N.
      • et al.
      Influence of gender and age on anorectal function: normal values from anorectal manometry in a large Caucasian population.
      Similar to the current trial, 2 other trials found no significant increase in squeeze pressure after PFMT and conservative treatment,
      • Sjödahl J.
      • Walter S.A.
      • Johansson E.
      • et al.
      Combination therapy with biofeedback, loperamide, and stool-bulking agents is effective for the treatment of fecal incontinence in women - a randomized controlled trial.
      • Johannessen H.H.
      • Wibe A.
      • Stordahl A.
      • et al.
      Do pelvic floor muscle exercises reduce postpartum anal incontinence? A randomised controlled trial.
      whereas most other trials have found that PFMT improved maximum squeeze pressure.
      • Heymen S.
      • Scarlett Y.
      • Jones K.
      • et al.
      Randomized controlled trial shows biofeedback to be superior to pelvic floor exercises for fecal incontinence.
      • Norton C.
      • Chelvanayagam S.
      • Wilson-Barnett J.
      • et al.
      Randomized controlled trial of biofeedback for fecal incontinence.
      • Solomon M.J.
      • Pager C.K.
      • Rex J.
      • et al.
      Randomized, controlled trial of biofeedback with anal manometry, transanal ultrasound, or pelvic floor retraining with digital guidance alone in the treatment of mild to moderate fecal incontinence.
      • Ilnyckyj A.
      • Fachnie E.
      • Tougas G.
      A randomized-controlled trial comparing an educational intervention alone vs education and biofeedback in the management of faecal incontinence in women.
      • Dobben A.C.
      • Terra M.P.
      • Berghmans B.
      • et al.
      Functional changes after physiotherapy in fecal incontinence.
      However, improvement in squeeze pressure is not always correlated with improvement in clinical outcomes.
      • Solomon M.J.
      • Pager C.K.
      • Rex J.
      • et al.
      Randomized, controlled trial of biofeedback with anal manometry, transanal ultrasound, or pelvic floor retraining with digital guidance alone in the treatment of mild to moderate fecal incontinence.
      • Dobben A.C.
      • Terra M.P.
      • Berghmans B.
      • et al.
      Functional changes after physiotherapy in fecal incontinence.
      It is not surprising that we did not find any changes in the rectal capacity measurement because EMG biofeedback is not expected to change sensory thresholds, as one would expect from, for example, sensitivity biofeedback.
      We were not able to show any differences between groups in the number of fecal incontinence episodes. We had much missing data in the bowel diaries and we can only guess what the result would have been if the data had been more complete. Electronic assessment of incontinence episodes could be a way to reduce the amount of missing data in future trials.
      In our trial all outcome assessors and researchers were blinded, and we did try to blind the participants to the trial hypothesis. It was not possible to blind the participants or the physiotherapists to the treatment allocation. The lack of blinding may have resulted in a higher placebo response in the PFMT group
      • Hróbjartsson A.
      • Boutron I.
      Blinding in randomized clinical trials: imposed impartiality.
      owing to higher expectations of treatment effect. We did not measure to what extent the participants believed that they were receiving effective treatment, so we cannot tell if differences in expectations between groups influenced the subjective outcomes. However, all primary and secondary results were in favor of supervised PFMT, which makes it less possible that our findings should be owing solely to higher expectations in the PFMT group.
      The attention-control massage treatment was chosen as a physiologically inactive control treatment to control for the nonspecific trial effect
      • Braunholtz D.A.
      • Edwards S.J.
      • Lilford R.J.
      Are randomized clinical trials good for us (in the short term)? Evidence for a “trial effect.”.
      related to the attention given by the health care professional. We could have chosen a sham training intervention instead, but given the fact that many patients are unable to conduct a correct pelvic floor muscle contraction, we found it unethical to teach the participants sham contractions with the risk of rehearsing a wrong contraction that may be difficult to correct later.

      Conclusions

      This trial provides support for a superior effect of supervised PFMT in combination with conservative treatment compared with attention-control massage treatment and conservative treatment in adults with fecal incontinence. Participants in the PFMT group had a 5-fold higher odds of reporting improvements in fecal incontinence symptoms and had a larger mean reduction of incontinence severity at the Vaizey Score. Based on the results, PFMT in combination with conservative treatment should be offered as first-line treatment for adults with fecal incontinence.

      Acknowledgments

      The authors thank all of the participants who contributed to the trial. The authors thank the staff at the Department of Physiotherapy and Occupational Therapy and the Department of Surgical and Medical Gastroenterology. The authors especially thank the physiotherapists Cathrine Blegvad Stenz and Susanne Høgsberg Knudsen for conducting the supervised pelvic floor muscle training, and Sarah Le Basson and Louise Hein Sommer for providing the attention-control massage treatment. The authors thank the nurses involved in the screening of eligible participants and the conduct of the conservative treatment. Finally, the authors thank the staff at the Clinical Research Centre for their highly valued support and academic discussions and contributions.

      Supplementary Material and Methods

       Detailed Description of the Conservative Treatment Program

      Both trial groups received a conservative treatment program consisting of standard information about optimizing bowel emptying and the use of fiber supplements and, if appropriate, information about the use of antidiarrheal medication (loperamide) and/or laxatives with a local effect in the rectum (glycerol).
      Participants were informed about regulation of stool consistency, and following a healthy and varying the diet with regular meals and high-fiber content was recommended. In addition, they were recommended fiber supplements such as psyllium. Psyllium was recommended as a powder or capsules. Participants were recommended to have 2 teaspoons of powder (3 g) or 6 capsules (3 g), both to be consumed twice a day. As an alternative to or supplement to psyllium, participants were informed that they could have a breakfast cereal with high-fiber content. They were recommended to drink 2.5 L liquid every day. Participants were informed that if they were bothered by thin stool, they could use antidiarrheal medicine (loperamide). They were recommended to start with half a tablet (equaling 1 mg) and then increase the dose until their stool was formed and they avoided accidents, but did not become constipated. Participants were informed that they may achieve more complete bowel emptying by using a laxative with a local effect in the rectum such as glycerol. They were informed that suppositories can be used to achieve more complete emptying after a spontaneous bowel movement or to initiate a bowel movement.
      The information in the conservative treatment program was delivered by 1 of 3 nurses who was specialized in fecal incontinence and was initiated on the first visit to the outpatient clinic at the Department of Surgical and Medical Gastroenterology before inclusion in the trial. The information was followed up by a telephone call from the nurse after approximately 1 month, thus the conservative treatment in both groups was continued parallel to the allocated intervention after the randomization.
      In addition to the conservative treatment program, the participants in both groups received their allocated interventions within 14 days of their randomization. The interventions in both groups were delivered as individual face-to-face visits at the Department of Physiotherapy and Occupational Therapy and were distributed over 16 weeks, with treatment offered in treatment weeks 0, 2, 5, 8, 12, and 16 (±1 week).

       Detailed Description of the Supervised Pelvic Floor Muscle Training Intervention

      Participants in the PFMT group received 6 individual treatments of 45 minutes consisting of individually supervised PFMT. The treatments were given by 1 of 2 physiotherapists who was specialized in treatment of fecal incontinence. The participants were taught correct pelvic floor contractions verbally and by digital vaginal and rectal examination. The verbal instruction was “squeeze around the anus as if you were about to hold back air or stool passing from the bowel.” At every session, the participant received an examination of their pelvic floor muscle function by digital vaginal and rectal examination and by intra-anal EMG biofeedback. The EMG biofeedback was used to give the participants visual and auditory feedback on a voluntary contraction to enhance awareness, strength, and endurance of a voluntary contraction and to support the performance of correct pelvic floor muscle contractions in different body positions and during movements. The participants were instructed to perform an individually tailored daily home training program. The program consisted of 3 sets of 10 pelvic floor muscle contractions sustained for up to 10 seconds, and 2 sets of 3 contractions sustained for up to 30 seconds. The participants also were taught to contract the pelvic floor muscle in response to fecal urgency and in situations with increased abdominal pressure. A training diary was used as a motivational tool and to quantify training adherence. Progressive overload was achieved by increasing the duration of the contractions and by performing the exercises against gravity in different body positions and during movements. The PFMT intervention was standardized in a protocol, and protocol adherence was recorded by the physiotherapists filling in case report forms.

       Detailed Description of the Anorectal Manometry, Rectal Capacity Measurements, and Endoanal Ultrasonography

       Anorectal manometry

      The investigation was conducted with the water-perfused catheter technique. Resting pressure and maximum squeeze increment pressure were determined. Both measures were recorded in the high-pressure zone using the pull-through technique. Maximum squeeze increment pressure was measured as the average of 3 squeezes using the pull-through technique (1 mm/s).

       Rectal capacity measurements

      A catheter was inserted into the rectum and then slowly inflated with air. The sensory threshold was determined as the minimum volume of filling at which a rectal sensation was perceived. Urge sensation was determined as the volume associated with the initial urge to defecate, and the maximum tolerated volume was determined as the volume at which the participants felt a strong desire to defecate and felt pain or discomfort.
      Anorectal manometry and rectal capacity measurements were measured by a trained nurse with more than 10 years of experience in conducting these investigations.

       Descriptive Variables and Covariates Collected at Baseline

      In addition to the outcome measures, the baseline data consisted of demographic background information, a medical history with an emphasis on the history of fecal incontinence, and known risk factors for fecal incontinence, and for women an obstetric history as well. An endoanal ultrasonography also was conducted (at 16 weeks) to detect internal and/or external anal sphincter defects. Defects were defined as a gap in the muscle ring or a loss of muscle substance in a range of more than 60°. The extension of sphincter defect was measured in degrees.

      Results

       Secondary Analyses of the Primary Outcome

      An intention-to-treat proportional odds model was fitted with the PGI-I scores as outcome using all 7 categories.
      The model showed that the participants in the PFMT group had 4.45 times higher odds (95% CI, 2.02–9.78) for rating their symptom changes in a higher (better) PGI-I category across all 6 cut-off points of the PGI-I scale (P = .0002.)
      During the review process, it was suggested to conduct a subgroup analysis of the effect at the PGI-I according to incontinence severity. We did this by dividing the participants into 3 subgroups according to the Vaizey Score at baseline (Vaizey Score subgroups: 1–8, 9–16, and 17–24). The subgroup effect was examined in an intention-to-treat logistic regression model in which the severity variable was added as an interaction term with the group variable. The model was unadjusted because the trial was not powered to conduct an adjusted subgroup analyses. The analysis showed significant interaction between baseline severity and group (P = .0076). There was no significant effect for the group with the lowest severity at baseline (subgroup 1), whereas both subgroups 2 and 3 had a significant effect of supervised PFMT. The effect was most pronounced for the group with the highest severity (subgroup 3) (Supplementary Table 1).
      Supplementary Table 1Odds Ratios for Reporting Improvement on the PGI-I in the PFMT Group After 16 Weeks of Treatment by Severity Subgroups
      ITT analysis, unadjustedIncontinence severity at baseline measured with the Vaizey Score
      Subgroup 1, Vaizey Score between 1 and 8Subgroup 2, Vaizey Score between 9 and 16Subgroup 3, Vaizey Score between 17 and 24
      OR (95% CI)1.2 (0.12–11.87)6.9 (2.16–22.098)14.4 (1.38–150.81)
      P value n.88 n = 15
      Imputed 1 value for a missing PGI-I value.
      .0011 n = 60
      Imputed 4 values for missing PGI-I values.
      .026 n = 23
      Imputed 1 value for a missing PGI-I value.
      NOTE. PGI-I scores were dichotomized as participants reporting either improvement of their incontinence symptoms (a score of very much better, much better, or a little better) or unchanged/deterioration (a score of unchanged, a little worse, much worse, or very much worse).
      ITT, intention-to-treat; OR, odds ratio; PFMT, pelvic floor muscle training; PGI-I, Patient Global Impression of improvement; Vaizey Score, Vaizey Incontinence Score.
      a Imputed 1 value for a missing PGI-I value.
      b Imputed 4 values for missing PGI-I values.
      Supplementary Table 2Explorative Outcomes
      Change from baseline to follow-up evaluation16-week follow-up evaluation

      Mean between group difference in change from baseline to follow-up evaluation
      Per protocol
      No imputations.
      Per-protocol analyses, unadjusted (n = 82)
      Imputed 3 values.
      Per-protocol analyses, adjusted
      Adjusted for age, sex, antidiarrheal medication at baseline, fiber supplements at baseline, incontinence type, presence of urinary incontinence, stool consistency, and anal sphincter injuries (at 16-week follow-up evaluation).
      (n = 78)
      No imputations were made because of simultaneous missing values for both outcome and covariates.
      PFMT (n = 33)Attention-control (n = 46)
      Mean change (SD)Mean change (SD)Mean difference in change (95% CI)P valueMean difference in change (95% CI)P value
      Anorectal manometry
       Resting pressure, mm Hg−4.49 (14.38)3.21 (18.55)−7.07 (−14.75 to 0.60).07−7.06 (−15.69 to 1.58).11
       Maximum squeeze increment pressure, mm Hg−0.72 (23.00)1.90 (24.53)−1.77 (−12.55 to 9.004).753.16 (−8.07 to 14.39).58
      Rectal capacity measurements
       Sensory threshold, mL−1.82 (17.22)−1.96 (14.39)−0.10 (−7.10 to 6.89).981.14 (−6.75 to 9.025).77
       Urge sensation, mL−3.33 (22.59)−4.13 (25.70)0.32 (−10.55 to 11.19).95−0.67 (−12.97 to 11.63).91
       Maximum tolerated volume, mL−9.70 (46.53)−9.78 (37.86)0.19 (−18.19 to 18.56).98−5.77 (−26.13 to 14.59).57
      14-day bowel diary
       Fecal incontinence episodes−0.65 (3.87) n = 170.15 (4.91) n = 27
       Multiple imputation0.70 (−2.99 to 4.40)

      (n = 82)
      .711.12 (−2.97 to 5.21)

      (n = 78)
      .59
       Imputation with zero−0.15 (−2.53 to 2.22)

      (n = 79)
      Imputed 35 values, missing values were imputed with zero.
      .90−0.71 (−3.66 to 2.25) (n = 75)
      Imputed 32 values, missing values were imputed with zero. The results from the 14-day bowel diary should be interpreted with great caution because of the large amount of missing data.
      .63
       Complete case−0.80 (−3.63 to 2.044)

      (n = 44)
      .57−1.73 (−5.08 to 1.63) (n = 43).30
      NOTE. As described in the statistical analysis plan,
      • Ussing A.
      • Dahn I.
      • Due U.
      • et al.
      Supervised pelvic floor muscle training versus attention-control massage treatment in patients with faecal incontinence: statistical analysis plan for a randomised controlled trial.
      we conducted a sensitivity analysis for the bowel diary because the missing data exceeded 10% (only 44 of the 82 participants in the per-protocol analysis had complete outcomes for fecal incontinence episodes). In this analysis, missing values for incontinence episodes were imputed with a value of zero because missing values predominantly were judged not to be missing at random but seemed to be negative answers. As an extra analysis, we conducted a per-protocol analysis without imputations, only including the 44 participants with complete outcomes. This complete case analysis was not prespecified but included because of the large amount of missing data. For the same reason, we did not analyze the number of stools recorded in the bowel diary, as had been prespecified.
      a No imputations.
      b Imputed 3 values.
      c Adjusted for age, sex, antidiarrheal medication at baseline, fiber supplements at baseline, incontinence type, presence of urinary incontinence, stool consistency, and anal sphincter injuries (at 16-week follow-up evaluation).
      d No imputations were made because of simultaneous missing values for both outcome and covariates.
      e Imputed 35 values, missing values were imputed with zero.
      f Imputed 32 values, missing values were imputed with zero. The results from the 14-day bowel diary should be interpreted with great caution because of the large amount of missing data.
      Figure thumbnail fx1
      Supplementary Figure 1Flow diagram describing patient flow through each stage of the trial. We screened 458 patients for eligibility, of whom 45 did not meet the inclusion criteria. Among those who met the inclusion criteria, 145 were excluded because of the presence of exclusion criteria and 166 declined participation. The main reasons for declining were long distance to the hospital or unwillingness to attend the required 9 appointments. A total of 102 participants signed the informed consent form, but 4 dropped out before randomization. Thus, 98 participants were randomized, 49 to supervised PFMT and 49 to attention-control massage treatment. Of the 98 randomized participants, 93 were seen at the 16-week follow-up evaluation and the primary end point was assessed in 92 participants (93.9%), except for the anorectal manometry and rectal capacity measurements, which were measured in 93 participants. Thus, end point data for the PGI-I, the Vaizey Score, the FISI, and the FIQL were missing for 6 participants (6.1%). ITT, intention-to-treat; PFMT, pelvic floor muscle training; PGI-I, Patient Global Impression of Improvement Scale.

      References

        • Ng K.-S.
        • Sivakumaran Y.
        • Nassar N.
        • et al.
        Fecal incontinence: community prevalence and associated factors--a systematic review.
        Dis Colon Rectum. 2015; 58: 1194-1209
        • Brown H.W.
        • Wexner S.D.
        • Segall M.M.
        • et al.
        Quality of life impact in women with accidental bowel leakage.
        Int J Clin Pract. 2012; 66: 1109-1116
        • Bharucha A.E.
        • Zinsmeister A.R.
        • Locke G.R.
        • et al.
        Prevalence and burden of fecal incontinence: a population-based study in women.
        Gastroenterology. 2005; 129: 42-49
        • Olsson F.
        • Berterö C.
        Living with faecal incontinence: trying to control the daily life that is out of control.
        J Clin Nurs. 2015; 24: 141-150
        • Norton C.
        Biofeedback and/or sphincter exercises for the treatment of faecal incontinence in adults.
        Cochrane Database Syst Rev. 2012; 7: CD002111
        • Whitehead W.E.
        • Rao S.S.C.
        • Lowry A.
        • et al.
        Treatment of fecal incontinence: State of the Science Summary for the National Institute of Diabetes and Digestive and Kidney Diseases Workshop.
        Am J Gastroenterol. 2015; 110: 138-146
        • Bo K.
        • Frawley H.C.
        • Haylen B.T.
        • et al.
        An International Urogynecological Association (IUGA)/International Continence Society (ICS) joint report on the terminology for the conservative and nonpharmacological management of female pelvic floor dysfunction.
        Int Urogynecology J. 2017; 28: 191-213
        • Sjödahl J.
        • Walter S.A.
        • Johansson E.
        • et al.
        Combination therapy with biofeedback, loperamide, and stool-bulking agents is effective for the treatment of fecal incontinence in women - a randomized controlled trial.
        Scand J Gastroenterol. 2015; 50: 965-974
        • Damon H.
        • Siproudhis L.
        • Faucheron J.-L.
        • et al.
        Perineal retraining improves conservative treatment for faecal incontinence: a multicentre randomized study.
        Dig Liver Dis. 2014; 46: 237-242
        • Braunholtz D.A.
        • Edwards S.J.
        • Lilford R.J.
        Are randomized clinical trials good for us (in the short term)? Evidence for a “trial effect.”.
        J Clin Epidemiol. 2001; 54: 217-224
        • Moher D.
        • Hopewell S.
        • Schulz K.F.
        • et al.
        CONSORT 2010 explanation and elaboration: updated guidelines for reporting parallel group randomised trials.
        BMJ. 2010; 340: c869
        • Boutron I.
        • Altman D.G.
        • Moher D.
        • et al.
        CONSORT Statement for Randomized Trials of Nonpharmacologic Treatments: A 2017 Update and a CONSORT Extension for Nonpharmacologic Trial Abstracts.
        Ann Intern Med. 2017; 167: 40
        • Ussing A.
        • Dahn I.
        • Due U.
        • et al.
        Supervised pelvic floor muscle training versus attention-control massage treatment in patients with faecal incontinence: statistical analysis plan for a randomised controlled trial.
        Contemp Clin Trials Commun. 2017; 8: 192-202
        • Hróbjartsson A.
        • Boutron I.
        Blinding in randomized clinical trials: imposed impartiality.
        Clin Pharmacol Ther. 2011; 90: 732-736
        • Yalcin I.
        • Bump R.C.
        Validation of two global impression questionnaires for incontinence.
        Am J Obstet Gynecol. 2003; 189: 98-101
        • Vaizey C.J.
        • Carapeti E.
        • Cahill J.A.
        • et al.
        Prospective comparison of faecal incontinence grading systems.
        Gut. 1999; 44: 77-80
        • Rockwood T.H.
        • Church J.M.
        • Fleshman J.W.
        • et al.
        Patient and surgeon ranking of the severity of symptoms associated with fecal incontinence.
        Dis Colon Rectum. 1999; 42: 1525-1531
        • Rockwood T.H.
        • Church J.M.
        • Fleshman J.W.
        • et al.
        Fecal incontinence quality of life scale.
        Dis Colon Rectum. 2000; 43: 9-16
        • Heymen S.
        • Scarlett Y.
        • Jones K.
        • et al.
        Randomized controlled trial shows biofeedback to be superior to pelvic floor exercises for fecal incontinence.
        Dis Colon Rectum. 2009; 52: 1730-1737
        • Norton C.
        • Chelvanayagam S.
        • Wilson-Barnett J.
        • et al.
        Randomized controlled trial of biofeedback for fecal incontinence.
        Gastroenterology. 2003; 125: 1320-1329
        • Solomon M.J.
        • Pager C.K.
        • Rex J.
        • et al.
        Randomized, controlled trial of biofeedback with anal manometry, transanal ultrasound, or pelvic floor retraining with digital guidance alone in the treatment of mild to moderate fecal incontinence.
        Dis Colon Rectum. 2003; 46: 703-710
        • Jelovsek J.E.
        • Chen Z.
        • Markland A.D.
        • et al.
        Minimum important differences for scales assessing symptom severity and quality of life in patients with fecal incontinence.
        Female Pelvic Med Reconstr Surg. 2014; 20: 342-348
        • Bols E.M.J.
        • Hendriks H.J.M.
        • Berghmans L.C.M.
        • et al.
        Responsiveness and interpretability of incontinence severity scores and FIQL in patients with fecal incontinence: a secondary analysis from a randomized controlled trial.
        Int Urogynecology J. 2013; 24: 469-478
        • Johannessen H.H.
        • Wibe A.
        • Stordahl A.
        • et al.
        Do pelvic floor muscle exercises reduce postpartum anal incontinence? A randomised controlled trial.
        BJOG. 2017; 124: 686-694
        • Young C.J.
        • Zahid A.
        • Koh C.E.
        • et al.
        A randomised control trial of four different regimes of biofeedback program in the treatment of faecal incontinence.
        Colorectal Dis. 2018; 20: 312-320
        • Bols E.M.J.
        • Hendriks E.J.M.
        • Deutekom M.
        • et al.
        Inconclusive psychometric properties of the Vaizey score in fecally incontinent patients: a prospective cohort study.
        Neurourol Urodyn. 2010; 29: 370-377
        • Whitehead W.E.
        • Palsson O.S.
        • Simren M.
        Treating fecal incontinence: an unmet need in primary care medicine.
        N C Med J. 2016; 77: 211-215
        • Gundling F.
        • Seidl H.
        • Scalercio N.
        • et al.
        Influence of gender and age on anorectal function: normal values from anorectal manometry in a large Caucasian population.
        Digestion. 2010; 81: 207-213
        • Ilnyckyj A.
        • Fachnie E.
        • Tougas G.
        A randomized-controlled trial comparing an educational intervention alone vs education and biofeedback in the management of faecal incontinence in women.
        Neurogastroenterol Motil. 2005; 17: 58-63
        • Dobben A.C.
        • Terra M.P.
        • Berghmans B.
        • et al.
        Functional changes after physiotherapy in fecal incontinence.
        Int J Colorectal Dis. 2006; 21: 515-521