Non-drug therapies for the management of chronic constipation in adults: the CapaCiTY research programme including three RCTs

Article history

The research reported in this issue of the journal was funded by PGfAR as project number RP-PG-0612-20001. The contractual start date was in July 2014. The final report began editorial review in October 2020 and was accepted for publication in May 2021. As the funder, the PGfAR programme agreed the research questions and study designs in advance with the investigators. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The PGfAR editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Copyright statement

Copyright © 2021 Knowles et al. This work was produced by Knowles et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This is an Open Access publication distributed under the terms of the Creative Commons Attribution CC BY 4.0 licence, which permits unrestricted use, distribution, reproduction and adaption in any medium and for any purpose provided that it is properly attributed. See: https://creativecommons.org/licenses/by/4.0/. For attribution the title, original author(s), the publication source – NIHR Journals Library, and the DOI of the publication must be cited.

2021 Knowles et al.

Background

The research programme

An evidence-based pathway for the management of CC in adults is currently lacking, although National Institute for Health and Care Excellence (NICE) guidance exists for the management of CC in children16–18 and for allied conditions (e.g. faecal incontinence) in adults. This arguably leads to variations in practice, particularly in specialist services. With a number of new drugs gaining NHS approval19–22 and new technologies at a horizon-scanning stage,23–26 it is timely that the currently limited evidence base is developed for resource-constrained NHS providers to have confidence that new and sometimes expensive investigations and therapies are appropriate and cost-effective. A cost-conscious pathway of care may help reduce health-care expenditures by appropriately sequencing the care provided while targeting more expensive therapies at those most likely to benefit from them. Such data could inform the development and commissioning of integrated care pathways. 27

The overall rationale of the Chronic Constipation Treatment Pathway (CapaCiTY) research programme, therefore, was to develop an evidence base for CC management through a series of three randomised controlled trials (RCTs) that answered some of the important questions for sequenced patient care (Figure 1). For each, the focus was on generating real-life evidence based on valid clinical outcome measures, patient acceptability and cost.

FIGURE 1.

Overview of the CapaCiTY research programme. Green arrows indicate studied pathways in CapaCiTY trials 1–3 (numbered circles). a, Alarm features excluded and secondary causes treated appropriately; b, in IBS-C, consider antispasmodics or neuromodulators in case constipation improves but abdominal pain persists and is dominant symptom; c, examples of overt prolapse include anterior (stage 3 cystocele), middle (stage 3 rectocele, uterovaginal) and posterior compartments (grade IV/V intussusception); d, if not performed previously; e, common adjuncts include sacrocolpopexy, hysterectomy, transvaginal tape and cystocele repair. FFD, functional defaecation disorder; IBS-C, constipation-predominant irritable bowel syndrome; MDT, multidisciplinary team; PEG, polyethylene glycol[0]; STARR, stapled rectal resection.

The specific objectives were as follows:

• work programme (WP)1

  • to develop a common methodological framework for subsequent studies

  • to recruit a UK cohort of adults with CC based on strict eligibility criteria and detail the baseline phenotype to identify disease risk factors, symptom burden, QoL and psychological morbidity.

• WP2 (CapaCiTY trial 1)

  • to determine whether or not standardised specialist-led habit training plus pelvic floor retraining using computer-assisted direct visual biofeedback (HTBF) is more clinically effective than standardised specialist-led habit training alone (HT)

  • to determine whether or not outcomes of such specialist-led interventions are improved by stratification to HTBF or HT based on prior knowledge of anorectal and colonic pathophysiology using standardised radiophysiological investigations (INVEST).

• WP3 (CapaCiTY trial 2)

  • to compare the impact of transanal irrigation (TAI) initiated with a low-volume and a high-volume system on patient disease-specific QoL after 3 months of treatment.

• WP4 (CapaCiTY trial 3)

  • to systematically review the evidence for all common surgical procedures used for adults with CC

  • to determine the clinical efficacy of laparoscopic ventral mesh rectopexy (lapVMR) compared with controls at short-term follow-up (24 weeks).

• WP5

  • to synthesise clinical outcome, patient acceptability and cost data from CapaCiTY trials and to develop an NHS pathway for the management of CC in adults based on data synthesis.

Common methodological framework

To permit the synthesis of all data at the end of the programme, we defined eligibility criteria, outcome measures and analytic methods that were common across all three trials. In addition, at the outset it was envisaged that some participants could move sequentially through more than one trial if a prior treatment had failed (although in practice this happened very infrequently because of delays in recruitment to all studies).

Patient recruitment

Specific background and rationale

In most UK practices, patients with CC are first referred to specialist nurses for a variety of nurse-led behavioural interventions to improve defaecatory function. A range of cohort studies,85 RCTs,86–91 reviews,92 guidelines93 and meta-analyses94 attest to the general success of this approach. However, opinion varies greatly concerning the complexity of intervention required and UK survey evidence (performed as part of this programme) indicates that there is remarkable variability of practice. 95

A simple form of behavioural therapy is habit training. This involves optimising dietary patterns to maximise gastrocolic response and the morning clustering of high-amplitude propagated colonic contractions that propel contents towards the rectum for subsequent evacuation. Dietary advice to optimise intake of liquid and fibre is given, as well as advice about frequency and length of toilet visits and posture. Patients are also instructed on basic gut anatomy and function, and gain an appreciation of how psychological and social stresses may influence gut functioning. Simple pelvic floor and balloon expulsion exercises are often included.

More complex forms of therapy include instrument-based biofeedback learning techniques. 85–91 Favoured in the USA and by about half of UK centres, these provide direct visual computer-based biofeedback of pelvic floor activity. Although small RCTs suggest an additive value of biofeedback over habit training alone in the management of selected patient subgroups of CC (e.g. dyssynergic defaecation),87,96–98 there has been no multicentre or adequately powered RCT in unselected patients, despite the uncertainty having significant resource implications. Aside from the issue that there is now considerable disagreement regarding the very diagnosis of dyssynergic defaecation (including from our own study using HRAM),99 most publications advocating biofeedback have come from specialist centres with considerable ‘investment’ in these techniques, with reports much less favourable when biofeedback is the ‘de-vested’ comparator. 100,101 The controversy following a Cochrane review that drew attention to the limitations of the evidence base102 attests to the polarity of opinion on this subject and the need for a more definitive (i.e. a larger, higher-quality) RCT.

This underpinned our first trial-specific research question:

• In unselected (i.e. no-INVEST group) adults with CC, is HTBF more clinically effective than HT as measured by PAC-QoL score at 6 months’ follow-up?

A further unanswered question regards the utility of behavioural therapies in different subgroups of patients with CC. Well-regarded international consensus (e.g. the Rome VI Criteria)103 supports a view that biofeedback significantly benefits only a subgroup of patients with a form of functional defaecation disorder termed ‘dyssynergic defaecation’ (see Figure 1). This poses the question whether or not further tests are required at an early stage (i.e. before behavioural therapy to select patients for biofeedback). There are significant differences of expert opinion on this subject: some advocate early complex and expensive investigations to guide treatment in most patients,8 whereas others advocate undertaking such tests only in resistant cases or in those progressing to surgery. 104 The advantage of guiding treatment105,106 is balanced against the invasive nature of some tests, radiation exposure, embarrassment and cost (≈ £600–1200 NHS tariff); most also necessitate an escalation of care (i.e. from secondary to tertiary centre). The need to resolve this question has been consistently highlighted38,107 but, to our knowledge, prior to the programme no RCT had stratified treatment selection on this basis.

This underpinned our second trial-specific research question:

• Is the impact of such specialist-led interventions improved by stratification to HTBF or HT based on prior knowledge of anorectal and colonic pathophysiology using INVEST as measured by PAC-QoL score at 6 months follow-up?

Both research questions were embedded in a single experimental design with three parallel trial groups and required a total sample size of 394 patients. For a full description of this trial, including interventions, trial-specific design procedures and all results and analyses, see Appendix 1. Only the main results and conclusions are summarised in this report.

Results

Recruitment started on 26 March 2015 (first intervention 21 May 2015) and ended 30 June 2018. A total of 182 participants (of the target 394 participants) were randomised out of 502 screened (36.3%) from 10 sites. Two sites opened but failed to recruit; the remainder randomised between 7 and 71 participants. A total of 68 participants were randomised to HT, 68 participants were randomised to HTBF and 46 participants underwent INVEST-guided therapy (HT or HTBF based on results) [the Consolidated Standards of Reporting Trials (CONSORT) flow diagram is shown in Figure 2]. A total of 178 participants provided PAC-QoL data at one or more time points (Figure 3). The primary outcome (PAC-QoL score) was available at both baseline and 6 months for only 103 participants. There was no evidence of an additive effect of HTBF over and above HT (Table 2).

FIGURE 2.

The CapaCiTY trial 1 CONSORT flow diagram.

FIGURE 3.

Overall PAC-QoL scores. (a) All participants (n = 178 at baseline); and (b) participants with no missing data to 12 months (n = 54). Both figure parts show reductions in score (i.e. improvement) over time.

A range of secondary outcomes covering symptoms and QoL improved in both the HT group and the HTBF group (e.g. mean PAC-SYM score reduced from 2.2 points at baseline to 1.5 points at 6 months and weekly laxative use reduced fourfold). Interventions led to small reductions in depression but no significant differences between the intervention types. Overall, about 65% of participants were globally satisfied or very satisfied with both interventions and this was reflected by participant experience reported at interview (i.e. similar proportions of participants liked, and a minority disliked, both interventions for a number of reasons).

Similar results were obtained for INVEST versus no-INVEST, with no evidence of a difference in primary outcome (see Table 2): participants provided reasons for liking INVEST (e.g. greater knowledge of their condition and knowing that it was not ‘all in their mind’) and disliking the invasiveness and embarrassment of the tests.

Given similar changes in EQ-5D-5L scores for all interventions, cost-effectiveness analyses favoured the simpler strategies (i.e. HT and no INVEST) as the dominant strategies. In both instances, cost increases were significant (HTBF vs. HT: £239, 95% CI £133 to £354; INVEST vs. no INVEST: £543, 95% CI £403 to £685) and QoL reduced (HTBF: –0.010 QALYs, 95% CI –0.053 to 0.03 QALYs; INVEST: –0.047 QALYs, 95% CI –0.093 to –0.001 QALYs). The probability that HT is cost-effective was a p-value of 0.83 at a WTP threshold of £30,000 per QALY.

Conclusions

Taking together the results from clinical effectiveness, cost-effectiveness and patient experience data, the following conclusions may be drawn while accepting the major caveat of under-recruitment:

• Included adults with CC had high levels of symptom burden and long durations of symptoms that had been refractory to previous treatments and could therefore be considered ‘hard to treat’. These symptoms were associated with a substantive effect on QoL and psychological well-being. Patient experience reflected the misery of the condition and fear that treatments would be ineffective.

• Our analysis of clinical effectiveness showed that all interventions trialled (i.e. HT, HTBF with INVEST and HTBF with no INVEST) reduced symptom burden and improved disease-specific QoL. The observed magnitude of these PAC-QoL score changes (≈ 0.8 points) can be considered to be clinically meaningful and represented a greater reduction than the minimum clinically important difference sought between groups by design (mean change 0.4 points). The findings from the primary outcome were coherent with a panel of secondary outcomes, and, overall, such improvements are unlikely to have occurred spontaneously in a condition that is generally considered chronic and stable.

• Confidence intervals rule out clinically important differences between HT and HTBF for the primary outcome and all main secondary outcomes. The same was true of INVEST and no INVEST.

• Standardised specialist-led habit training alone and no INVEST are strongly supported by cost-effectiveness analyses. Despite under-recruitment (182 out of a planned 394 participants), participant-level data provided the most robust evidence to date on the first step of care for patients referred to hospital for CC. Neither a complex specialist-led intervention (e.g. pelvic floor retraining using biofeedback) nor stratification to complex or standardised therapy based on prior knowledge of anorectal and colonic pathophysiology (INVEST) were more cost-effective than HT. Analysis suggests that HT is the dominant strategy (i.e. lower cost and greater QoL) at a WTP threshold of £30,000 per QALY.

• All procedures were safe and well tolerated by patients.

• Interviews suggested that patient experience was mixed. Some regretted not being allocated to INVEST or HTBF (because they believed that they would have less knowledge of their condition and less likelihood of treatment success), whereas others felt that the tests and biofeedback were embarrassing and intrusive. Most participants reported a positive interaction with staff and at least some symptom benefit. Most would recommend trying the intervention they received to other people with CC.

• Staff were mostly supportive, but some found that adhering to the agreed intervention protocol constrained their clinical flexibility and they would have preferred to individualise the intervention. The biofeedback element added time to consultations, or limited what they could cover in HT.

• Because these patients had significant psychological comorbidity, and there is evidence that psychological treatments such as cognitive–behavioural therapy have significant and sustained benefits on symptoms, mood and QoL in people with irritable bowel syndrome (many of whom also experience severe constipation), future work should focus on incorporating psychological methods alongside HT. 108,109

• Taken together, the cost-effectiveness data (in the absence of differences in clinical effectiveness, patient experience and safety) promote the adoption of the simpler pathway (i.e. HT without INVEST). A revised prototype pathway is provided on this basis in the final conclusions of the synopsis (see Figure 8).

Specific background and rationale

Transanal irrigation, for which there are a variety of commercially available devices, has been rapidly disseminated internationally since about 2000, first in CC patients with neurological injury112,113 and subsequently in other groups CC groups. 114,115 Despite a lack of published data other than from small selected case series, TAI is now available on drug tariff (at a cost of ≈ £2500 per patient per annum) and is generally considered to be the next step for patients failing other nurse-led interventions. TAI has permeated the UK market without robust efficacy data and with ongoing concerns regarding longevity of treatment and complications. 112,116 Retrospective clinical audit data and review116,117 by the applicants suggest a continued response rate after 1 year of ≈ 50%, with patients thus avoiding or delaying surgical intervention, but an accurate assessment of response rate and acceptability of this intervention required confirmation in a trial. In addition, two alternative systems for delivery of TAI exist: low-volume systems delivering ≈ 70 ml per TAI, and high-volume systems delivering up to 2 l of TAI (although typically only 0.5–1.5 l is required per TAI). The theoretical benefit of higher-volume TAI is greater efficacy – simply put, more washout. However, the low-volume system is cheaper, costing ≈£750 per annum, based on alternate-day use, compared with a cost of £1400–1900 per annum for high-volume TAI, and it may also be less traumatic and more acceptable to patients.

This underpinned our first specific research question:

• In patients with CC who have failed conservative treatment (HT or HTBF), what is the impact on disease-specific QoL of TAI initiated with a low-volume and a high-volume system measured by PAC-QoL score at 3 months’ follow-up?

Transanal irrigation is an invasive therapy that requires, every day or every couple of days, insertion of the device into the anus followed by a period spent filling the rectum with fluid and then evacuating it. It is reasonable to suppose that patients would discontinue therapy that they felt was ineffective or unacceptable, or switch between low-volume and high-volume systems (permissible in the trial design).

This underpinned our second trial-specific research question:

• What is the survival rate of therapy of TAI initiated with a low-volume and a high-volume system and do patients prefer one system to another?

Both research questions were embedded in a single, pragmatic, two-group parallel design (Figure 4) and required a total sample size of 300 patients (150 per group). Patients used one system only (plus defined ‘rescue therapies’) for a minimum of 3 months. After this time point they could switch to the other system if their initial therapy was ineffective/unsatisfactory. This allowed identification of response rates to each system in the short term (3 months) and, thereafter, a comparison between treatment strategies (TAI initiated with low-volume therapy or high-volume therapy) rather than a pure comparison of the two techniques. This was a patient-centred study design aiming to limit the time that patients spent using ineffective therapy without being allowed to try an alternative. Reasons for switching were captured qualitatively.

FIGURE 4.

The CapaCiTY trial 2 CONSORT flow diagram.

For a full description of this trial, including interventions, trial-specific design procedures and all results and analyses, see Appendix 1. Only the main results and conclusions are summarised in this report.

Results

Survival rate of transanal irrigation therapy

In the absence of a sham control (impossible to devise), and with the inability to compare with standard therapy, it was reasonable to use treatment continuation as a marker of efficacy. The treatment is burdensome, and it can be argued that patients who are not receiving benefit would not continue with it. There was no encouragement from research or clinical staff for patients to continue to use ineffective therapy. The 1-year survival rate of the treatment of 76% (Figure 5) implied significant continued effect. A comparison of survival rate plots of low-volume and high-volume TAI that allows for crossover has not been presented at the time of publication. This analysis is planned but was outside the SAP. 118

FIGURE 5.

The CapaCiTY trial 2 survival rate of treatment as a surrogate of ongoing benefit: Kaplan–Meier plot for time to cessation of treatment.

Conclusions

Taking together the results from clinical, cost-effectiveness and patient experience data, the following conclusions may be drawn while accepting the major caveat of under-recruitment:

• The population studied represented a typical hospital-derived cohort of patients with severe CC that had not responded to conservative therapies.

• We did not carry out statistical significance tests owing to under-recruitment. Despite this, there is preliminary evidence that high-volume TAI may be more effective than low-volume TAI. Although effect sizes were small for both primary and secondary outcome measures (especially global improvement scores), dominant crossover from low-volume to high-volume TAI and health economic analysis point in the same direction of favouring high-volume TAI.

• Survival rate data were suggestive of a persisting benefit in the majority of patients, with three-quarters of patients still using TAI at 12 months. Overall, there were fewer patients on low-volume TAI at 12 months than on high-volume TAI. A survival rate analysis allowing for crossover is yet to be undertaken.

• Some AEs were reported, but, of six SAEs, none was related or life threatening. The most common AEs were rectal bleeding and anal pain (48 events in 16 patients).

• Despite under-recruitment, patient-level data from the trial still provided the most robust evidence currently available to inform the pathway of care. Initiating high-volume TAI did not increase overall cost (despite slightly higher basic device pricing) and resulted in higher patient QoL, suggesting a dominant interventional strategy (cost-effective at a WTP threshold of £30,000 per QALY; p = 0.99).

• Although cost-effectiveness base-case findings were robust to the complete-case and sensitivity analyses conducted, the findings should be treated with caution. Small absolute numbers of patients and high levels of missing data as follow-up progressed placed great reliance on the imputation process and output. As with CapaCiTY trial 1, although substantial efforts were made to deliver a robust imputation process, it not possible to prove that a MAR process has been attained. Actual level of TAI use (the predominant cost) was not directly recorded and was instead approximated from available variables.

• A final consideration in support of high-volume TAI concerns convergence. Health-care costs are similar comparing the two groups, suggesting that differences in health-care costs beyond 12 months are small. However, QoL diverges between groups over 12 months, suggesting that QALY gains may continue to accrue for some time beyond 12 months. Although speculative, any modelling of future cost and benefits would further increase the cost-effectiveness of high-volume TAI.

• Transanal irrigation often caused initial anxiety, was not always easy to learn and was time-consuming. Ongoing staff support was much appreciated, especially as results were not always immediate. Some found it of no benefit, but many persevered and found that, although not necessarily a pleasant experience, it had an impact (sometimes large) on their symptoms and QoL. Staff were possibly more enthusiastic than patients initially, and this enthusiasm was picked up by patients in some cases and helped them to persevere.

Reflections on work programme 3

Poor recruitment (65/300; 22%) was a disappointment and reduced the inferential quality of the study. As well as the usual difficulties as discussed for CapaCiTY trial 1, we failed to recognise an important constituent: that the patients offered TAI had (1) a severe burden of illness and (2) already tried all the usual conservative measures (which had failed). They were resorting to TAI often in a state of desperation. The difficulty, then, was that recruitment to the trial resulted in a delay in treatment while baseline assessments and investigations were undertaken. At the same time, patients were aware that they could access the treatment immediately if they remained outside the trial. With hindsight, it should have been clear that this would lead to difficulties. To overcome this would have required a waiting list for non-trial treatment or a pragmatic approach in which investigations and baseline diaries were dispensed with (as they are dispensed with in emergency medicine trials). Further issues, such as the failure to obtain funding for the therapy in several UK regions, proved beyond our control despite extensive discussion with relevant CRNs and at a national level. Ironically, we received letters stating that funding would not be provided because the therapy lacked evidence from a randomised trial.

Despite gross under-recruitment, we believe that the trial provided some meaningful evidence that high-volume TAI may be more effective than low-volume TAI for the majority of patients. The CI for the effect size appeared to rule out a clinically important difference of 0.4 in favour of low-volume TAI and not in favour of high-volume TAI. This finding promotes a care pathway in which high-volume TAI might be initiated as a default for patients when there is not a compelling reason or preference for low-volume TAI. This is a useful conclusion, but it lacks identification of a more definite means of stratifying patients to one treatment or the other based on baseline phenotype. Ideally, we would have liked to study the efficacy of TAI per se, but there was no way of designing a sham control. A waiting list comparison was considered but rejected because waiting times were not long enough to provide meaningful data. A cohort design was also proposed, but the grant review panel encouraged a RCT. The appeal of considering low-volume and high-volume TAI was the ability to relate responders in the different treatment cohorts to symptoms (urge vs. no urge) and selected physiology results (notably, presence or absence of blunted rectal sensation). We could potentially have learned something about the physiology and mechanism of action of these treatments by doing this. However, with very low recruitment numbers, these additional analyses (covariates of response to each or both therapies) could not be undertaken.

It was interesting to note from the staff interviews that specialist nurses providing the treatment already had strong views that high-volume TAI was a better treatment. It is possible that some bias might have been affected by these opinions.

Research recommendations

It does not seem necessary to recommend repeating the trial on the basis of inadequate recruitment because any future trial will face many of the same challenges, even with the changes proposed here. Furthermore, a more basic question still remains: what is the value of any type of anal irrigation? There have been a significant number of retrospective observational studies and occasional prospective ones. The difficulty would be in establishing a control, as already stated here. It may be possible to develop a prospective cohort study of patients with CC that, while being monitored prospectively, might adopt TAI as an alternative to usual care. A trials within cohorts (TwiCs) design could be considered if patients would consider secondary randomisation against continuing usual care.

The qualitative evidence has been very illuminating and further analysis of this will be undertaken to understand drivers for continuing and discontinuing therapy (to guide patient selection and encourage compliance). As new equipment is developed (including devices activated by mobile telephones) there may be a desire to assess equivalence and safety. In addition, there is a need to understand the merits of contraindications such as mild colitis and pregnancy.

Specific background and rationale

When non-surgical therapies fail, a decision must be made about whether or not to offer surgical interventions. Decision-making is greatly influenced by local expertise, commissioning and personal enthusiasm for particular interventions25,119,120 balanced against poor results in some patients. 25 Currently, there is large and difficult-to-justify variation in surgical practice according to need and type of procedure. The need to reduce variations in practice, based on available evidence, is a perpetual theme of national specialty group discussions, with various initiatives proposed. Multidisciplinary team (MDT) processes have been established in the UK with the aim of reducing potential for inadequately informed and potentially harmful interventions in poor surgical candidates. 25 However, when these MDT processes occur, decision-making is not helped by a near void of high-quality evidence.

At the time of starting the CapaCiTY research programme, lapVMR was considered a very effective procedure for the management of selected patients with CC. This procedure was first described for the treatment of external rectal prolapse in 2000121–123 and then for patients with internal prolapse and/or rectocele presenting with CC. 124–131 Its popularity, based on its lesser invasiveness and perceived small detrimental effect on bowel function,132 soared 15-fold in the UK between 2001 and 2012133 but rapidly declined during the CapaCiTY research programme (2016 to present) when concerns regarding the use of pelvic mesh surfaced in the media83 and the courts. 134

The evidence for efficacy of lapVMR was limited to observational data, the majority of which derived from single-centre case series. It is clear that complications can be limited by good technique and perhaps choice of mesh, but complications will not be eradicated. Thus, it can be argued that the future of lapVMR depends not on the very small observed differences in long-term mesh complications (e.g. in 1.0–2.0% of patients) but on a fundamental evaluation of whether or not the procedure is actually clinically beneficial (i.e. whether or not these complication rates would be deemed acceptable, provided patients are consented to the risk, if the patient benefit was sufficiently large).

This underpinned the specific research question:

• In selected adults with CC who have failed conservative treatment (i.e. HT or HTBF with or without TAI) and who have high-grade internal rectal prolapse, what is the clinical efficacy of lapVMR compared with controls at short-term (i.e. 24-week) follow-up.

In addition, this WP included the delivery of a series of high-quality systematic reviews for all main surgical procedures employed to treat CC to define benefits and harms and provide prototype graded practice recommendations (including for lapVMR).

We used a stepped-wedge randomised trial design to permit observer-masked data comparisons between patients awaiting intervention and patients who have undergone surgery. Contrary to most stepped-wedge trials, individual patients (as opposed to clusters) were randomised. In brief, eligible participants were randomised to three groups with different delays before surgery (Figure 6). In all groups there was a period of 4 weeks post eligibility assessment to arrange the logistics of surgery [time (weeks) since randomisation (T) –4 (T–4) to T0] and ensure that patients have returned to their normal life routine after various assessments. lapVMR was performed in group 1 at T0, in group 2 at T12 and in group 3 at T24. Unavoidably, participants were aware when surgery was undertaken and thus met one of the assumptions of a stepped-wedge design (i.e. no effect of treatment expected until surgery has been performed). Efficacy outcome data were collected at equally stepped time points (T0, T12, T24, T36 and T48). PAC-QoL and PAC-SYM total scores were analysed using a mixed linear regression model, adjusting for a random effect of participant and a fixed effect of time since randomisation (see Figure 6). The effects of the intervention at 12, 24, 36, 48, 60 and 72 weeks post surgery were modelled as fixed effects. For example, the intervention effect at 12 weeks post surgery was included in the regression model using a dummy variable set to 1 in those cells of Figure 6 where surgery occurred 12 weeks previously [i.e. at T12 in group 1 (when surgery was at T0), T24 in group 2 (when surgery was at T12) and T36 in group 3 (when surgery was at T24). This was the expected difference in outcome between a patient who had surgery 12 weeks previously and a patient who had not had surgery. This was similar for intervention effects at T24, T36, T48, T60 and T72 weeks post surgery. The analysis included a Kenward–Roger correction to correct for the small sample size, and was performed in Stata using the ‘mixed’ command with the options ‘reml dfmethod(kroger)’.

FIGURE 6.

The CapaCiTY trial 3 schematic stepped-wedge design.

The choice of design (in effect a modification of a standard, parallel-group, waiting-list control design) had several advantages. First, a stepped-wedge design is more efficient and thus improves recruitment feasibility (the major hurdle of nearly all surgical trials). Despite the multicentre approach of this study, the problems of recruitment were manifest. Simulation demonstrated that a parallel-group design required a much larger sample size than that proposed for the current study at the same power. Second, the trial design meant that there was only a one in three chance (rather than a one in two chance as in a parallel-group design) of waiting almost 6 months for surgery, which was more acceptable to patients (according to a 100-patient survey during the programme development phase). Using this design required a sample of 114 patients with 95% power (purposely chosen to reflect the magnitude and risk of intervention) at a 5% significance level.

For a full description of this trial, including interventions, trial-specific design procedures and all results and analyses, see Appendix 1. Only the main results and conclusions are summarised in this report.

Results

Stepped-wedge randomised trial

The first recruitment was on 1 March 2016, with the first intervention on 15 June 2016. Recruitment ended on 31 January 2019. A total of 28 patients (target 114 patients) were randomised out of 81 screened (34.5%) from nine sites. Two sites opened but failed to recruit and one site failed to randomise; the remainder (n = 6) randomised 1–11 patients each. Nine patients were randomised to immediate surgery, 10 were randomised to a 12 weeks’ waiting time and nine were randomised to a 24 weeks’ waiting time. The CONSORT flow diagram is shown in Figure 7. One patient was randomised but did not undergo surgery. However, this patient remained in the study on ITT principles. Two patients dropped out of the study before the primary end point and a further five failed to complete the primary outcome (PAC-QoL score at 24 weeks), which was therefore undertaken in 19 patients.

FIGURE 7.

The CapaCiTY trial 3 CONSORT flow diagram. a, One patient did not undergo surgery; this patient continued to participate and was included in analysis on ITT principles.

Laparoscopic ventral mesh rectopexy resulted in substantial improvement in symptoms, with the mean PAC-QoL score reducing from 2.63 points at baseline to 1.26 points at 24 weeks, and a similar reduction in PAC-SYM score from 2.24 points at baseline to 1.19 points at 24 weeks (Table 4). Secondary outcomes also indicated improvement over time, with PAC-QoL and PAC–SYM scores showing maintained reductions compared with baseline up to the 72-week time point (accepting potential attrition bias). Improvements in total score reflected improved scores across all domains of the instruments. Positive directional effects were observed for nearly all other secondary outcomes, with some quite substantial improvements in measures [e.g. > 25% scalar improvements in psychological measures (e.g. PHQ-9, GAD-7), St Mark’s Incontinence Score and EQ-VAS]. Global patient satisfaction was 2.7 points at 24 weeks (nearest to ‘very satisfied’), although this dropped to 2.2 points (‘moderately satisfied’) at 48 weeks. This result was mirrored in the global patient improvement score (EQ-VAS score 0–100 points between ‘no effect’ and ‘complete cure’), which was 72.2 points at 24 weeks and 56.5 points at 48 weeks.

TABLE 4 - Total PAC-QoL and PAC-SYM scores at baseline and follow-up points post surgery, with 95% CI and p-value for change from baseline to each follow-up point

All estimates are adjusted for time.

Time point	Number completing outcome evaluations	Observed mean score (points)	Estimated change from baseline score (points)	95% CI for change in score (points)	p-value for change in score (points)
PAC-QoL
Baseline	26	2.63	–	–	–
12 weeks	23	1.35	–1.04	–1.54 to –0.55	0.0001
24 weeks	19	1.26	–1.09	–1.76 to –0.41	0.0019
36 weeks	19	1.47	–0.98	–1.87 to –0.10	0.0296
48 weeks	17	1.43	–1.07	–2.16 to 0.02	0.0552
60 weeks	9	1.22	–1.26	–2.56 to 0.05	0.0587
72 weeks	5	1.11	–1.38	–2.94 to 0.19	0.0840
PAC-SYM
Baseline	26	2.24	–	–	–
12 weeks	23	1.15	–0.97	–1.41 to –0.53	0.0000
24 weeks	18	1.19	–0.92	–1.52 to –0.32	0.0029
36 weeks	19	1.25	–1.03	–1.80 to –0.26	0.0094
48 weeks	17	1.36	–0.97	–1.92 to –0.02	0.0444
60 weeks	9	1.19	–1.16	–2.28 to –0.03	0.0448
72 weeks	5	0.82	–1.51	–2.87 to –0.16	0.0289

Comparing pre-surgery and post-surgery periods, intervention led to a significant increase in cost (£5012, 95% CI £4446 to £5322), similar to the cost of surgery, and a modest and imprecise increase in QoL (0.043 QALYs, 95% CI –0.005 to 0.093 QALYs). The ICER was £115,512 per QALY. At the NICE threshold of £30,000 per QALY, the probability that surgery is cost-effective within 48 weeks of surgical intervention is 0%. The EVPI (per subject) reflects the opportunity loss of ongoing uncertainty and suggests no requirement for further research.

Laparoscopic ventral mesh rectopexy was shown to be a safe procedure. There were 30 AEs reported by 16 patients, 20 of which were considered to have possible causality related to surgery and none of which had any long-term sequelae. There were five SAEs, of which four were deemed to be related to surgery. Three of these were for postoperative pain, which was entirely to be expected in a proportion of patients; one was for a chest infection and none resulted in long-term patient harm.

Patient experience of lapVMR was, on the whole, positive. Some patients did not find surgery to be the miracle cure that they were looking for and some reported negative experiences in the perioperative period, but for the most part these were not related to the operation itself. Some patients also found benefit from the dietary and behavioural changes that they initiated as a result of advice that they were given as part of the perioperative care package.

The media scrutiny of the use of mesh undoubtedly affected both patient and surgeon perception and willingness to take part in the study. Some centres paused or abandoned lapVMR totally in the light of the mesh scandal, and there was a perception that for others the heightened scrutiny of practice in the protocol also negatively affected recruitment.

Conclusions

Taking together the results from clinical effectiveness, cost-effectiveness and patient experience data, the following conclusions may be drawn while accepting the major caveat of under-recruitment:

• Included patients had a high symptom burden and long duration of symptoms that had been refractory to previous treatments, including a minimum of bowel habit training by a specialist practitioner. Patients had been thoroughly investigated and, therefore, could be considered both ‘hard to treat’ and ‘carefully selected’ for surgical intervention.

• Our analysis of clinical effectiveness showed a reduced symptom burden and improved disease-specific QoL. The magnitude of the effect of surgery (estimated reduction of 1.09 points in PAC-QoL at 24 weeks) was greater than the minimum clinically important difference sought by design (mean change 1.0 points) and this change was statistically significant. In addition, significant and clinically important improvements in PAC-SYM score and bowel frequency provided further evidence of the benefit of surgery.

• The findings of the primary outcome showed a continued improvement for the duration of the study period (estimated 1.38-point reduction in PAC-QoL at 72 weeks), and this finding was supported by a panel of secondary outcome measures, accepting inferential limitations posed by potential attrition bias.

• Serious under-recruitment limited the scope to conduct economic analysis. To explore the potential cost-effectiveness of lapVMR some assumptions were necessary: (1) the effect of surgery was not affected by allocation (and thus delay to surgery); (2) growth curve analysis is a reliable method to characterise pre-surgery and post-surgery costs and QoL; (3) pre-surgery non-surgery costs and EQ-5D-5L scores are stable, making it possible to construct a single pre-surgery follow-up covering the 48 weeks before surgery; and (4) benefits following surgery are limited to the first 48 weeks following surgery. The weakest of these assumptions might be limiting benefit from surgery to 48 weeks; any future QALY gain and reduced burden from decreased use of the health-care system for symptoms of CC would make surgery more cost-effective. Potential longer-term complications of mesh insertion and removal were not considered, but if emergent would reduce cost-effectiveness.

• The base case suggests that lapVMR is not cost-effective (p = 0.00 at a WTP threshold of £30,000 per QALY). However, because this finding comes from an observational analysis of a stepped-wedge design, and for the reasons described, it remains vulnerable to bias that might increase or decrease cost-effectiveness.

• Although some adverse effects were reported, lapVMR was safe and well tolerated overall.

• Some patients described how their bowel function was better immediately after surgery and had continued success, and these patients noted a better QoL both mentally and physically; others found the benefit to be short-lived or did not feel any change. Pain was a significant issue for some patients. The mesh controversy dominated staff experience.

Reflections on work programme 4

The study was severely hampered by under-recruitment. Difficulties in attracting centres to recruit pre-dated the zenith of the mesh scandal and reflected wide variation in practice across the UK in terms of both patient selection and lapVMR operative technique. The mesh scandal undoubtedly compounded the recruitment problems faced by the study and yet paradoxically emphasised the need for such a study to take place. It was, for instance, very clear that the attention we paid to strict inclusion and exclusion criteria [actually based on guidance from the Pelvic Floor Society (London, UK)134] led to increased scrutiny in many centres where such surgery was being undertaken without rigorous application of these criteria. This regrettably (for the trial but not the patients concerned) led to a rapid revision (manifest as a drop-off) in the number of patients available at several centres for recruitment. Unfortunately, with the evolution of the media storm against placement of pelvic mesh and worldwide class actions against surgeons and manufacturers of mesh, the issue of patient selection has now made its way to the courts. Therefore, although it might be argued that the study design was overzealous in its approach to inclusion and exclusion, the decision of our expert panel to rigidly maintain the selection and procedural standards laid out in the protocol was undoubtedly the right one.

With the media storm blowing up and the Cumberlege report142 in preparation, there was never a time when the results of this trial were more needed, and we were very disappointed by our failure to keep the trial going. This was not for want of trying. With the strong support of the Association of Coloproctology of Great Britain and Ireland, the Pelvic Floor Society wrote on our behalf to NHS England seeking its support for a mandate to surgeons in England and Wales to continue surgery only in the context of research indemnity. This request, if granted, would have helped push recruitment and prevented some hospitals from abandoning the operation, thus denying patients access to lapVMR and the scientific community an answer to the main trial questions.

However, despite these setbacks, the main aim of the trial, namely to determine the effect size of surgery for the first time in a high-quality experimental design, and thus improve on the level IV evidence provided by 18 case series (as outlined in our systematic review),140 was addressed, albeit at a lower than desirable level of statistical power. Being to our knowledge the only high-quality evidence in the field, our results show substantial symptomatic benefit (more than we sought by design) to a cohort of highly selected patients from lapVMR performed to a standardised technique.

Cost-effectiveness analysis was disappointing and, as with all expensive (surgical) interventions, we did not show a cost benefit in the short term (to 48 weeks). Enthusiasts will say that the benefit of mesh placement is economically apparent only in the long term. It is also difficult to demonstrate cost-effectiveness in our multiply comorbid patients whose overall sense of well-being (as measured by the EQ-5D-5L) is only partially affected by improvement in bowel-associated symptoms.

Research recommendations

The current COVID-19 pandemic has created a natural hiatus in the provision of nearly all non-urgent benign surgery. This has coincided with the publication of the Cumberlege report142 and together these provide a ‘pause for thought’ on the future use of lapVMR in patients with CC. Despite the underpowering of our study, it is unlikely that such a trial will be repeated (at least in the UK). Rather, it is likely that enthusiasts will cite the effect size in our study as mirroring that in observational trials, balancing this against the very real risk of harm (notably the small proportion of patients with long-term mesh complications). The reduction in anxiety surrounding the use of mesh as time passes and the production of updated consensus guidance on patient selection and operative technique may make further study in this area feasible.

Reflections on the whole research programme and overall conclusions

How will the international community receive such a pathway? We feel that the answer to this in the UK is with general positivity. Over the 6 years that the CapaCiTY research programme has run, from initial meetings at national society conferences to the present, we have witnessed hugely positive engagement from experts in the field through to more general audiences. Our ability to bring together the specialist community, both specialists whose centres recruited and specialists whose centres were unable, has proved a major triumph of the programme. Although this is not measurable in adequately powered clinical-effectiveness data, it can be inferred by how many units have standardised their approach to diagnostics and therapy by education through the programme. Furthermore, conclusions from our systematic reviews have already entered major textbooks and have received national and international platforms for presentation. Dissemination activity is ongoing, and there is no doubt that we will be invited for national and international presentation of the conclusions of the three CapaCiTY trials. These ‘softer’ end points will be the main enduring legacy of the CapaCiTY research programme and the lives of people living with CC will be the better for it.

Lessons learned for future research

The considerable challenges to recruitment for all three trials have been noted. Table 5 outlines some insights about how these problems might be mitigated in the future.

Patient and public involvement

Patient representation in the CapaCiTY research programme supported all research activities. The focus of the CapaCiTY research programme from a PPI perspective was to convene a Constipation Research Advisory Group (CRAG) comprising eight patients and two carers from London, UK, and Durham, UK. This group had geographical diversity (covering both north and south England) and a disease-appropriate demographic (eight female and two male participants). PPI was managed by two co-applicants and the CRAG had an active ‘contributory’ rather than ‘representative’ role. The overarching functions of the CRAG included:

• managing the research (e.g. steering/advisory group)

• developing participant information resources

• advising on protocol revisions (in particular on recruitment and participant burden)

• disseminating of research findings.

Although CRAG members had patient knowledge of living with severe CC, they did not have the same knowledge of clinical trial and medical research terminology as the research team. Because CRAG members were invited (in rotation) to sit on the Programme Steering Committee (PSC), this put them at a disadvantage. Therefore, training was provided to CRAG members at the start of the project in 2015 that included taking part in an activity in identifying barriers to recruitment in clinical trials and potential solutions. This enabled the CRAG members to review all patient-facing documents including the patient information sheet, patient diaries and patient journals at the beginning of the programme for all three trials. It also enabled them to feel more comfortable when they attended a PSC meeting.

The CRAG was actively involved and effective in providing lay advice and guidance in areas of trial feasibility, design, management, marketing, analysis, recruitment and reporting. Furthermore, CRAG members reviewed all patient-facing material during the substantial amendment in 2016. The CRAG was also consulted about reducing the length of follow-up from 24 to 12 months.

The CRAG reported and made recommendations to the PSC. It was agreed that normally CRAG meetings would be held every 6 months prior to the PSC (ideally 1 month before). The CRAG would feed recommendations back to the PSC. Therefore, the CRAG had a significant role in strategic decision-making. For example, in 2017 the CRAG was provided with a detailed presentation as per the PSC’s recommendation. The PSC asked that the CRAG answer whether or not the CRAG deemed CapaCiTY trial 2 futile owing to under-recruitment. Although it was noted that the inclination of the PSC was to continue, they were keen to consult with the CRAG. The group discussed at length the key areas under consideration, which included the following: was it safe for the patients participating in the trial? Was it cost-effective to continue? Would the research questions be answered?

Though the group accepted that, ideally, more patients should be recruited to improve the accuracy and validity of the data results, it was generally agreed that 100 patients was an acceptable milestone.

The CRAG also discussed financial matters. However, the main focus of the in-depth discussion was whether or not the research questions would be answered. Owing to the smaller sample size it was recognised that the primary research question could be compromised. However, the CRAG recognised the importance of being able to answer many other questions in relation to high-volume and low-volume TAI treatments. The CRAG discussed the value and importance of the secondary data that had been captured to date and data that would continue to be collected if the trial was to continue. It was evident that from a patient perspective these data were of significance. It was noted that the CRAG wished to avoid any negative messaging and in particular did not wish to put off future patients from entering a trial because of the possibility of cancellation.

This consultative approach between the PSC and CRAG, seeking to direct and support the three trials, ensured that both researchers and patients had a mutual understanding of what was required. This ensured that the findings of the three trials would be of benefit to participants in those clinical trials. In addition, three members of the CRAG held a teleconference with Christine Norton and Shiva Taheri to contribute to the interpretation of the interview data and to agree and support the main messages for dissemination.

Overall, patient representation made a major contribution to design, conduct and recruitment of the CapaCiTY research programme. The CRAG was involved in all major decisions that were made during the programme, which meant that patient benefit was always put first in everything we did. The CRAG will continue to support the CapaCiTY research programme in dissemination activities, including at the Bowel Research UK annual Big Bowel Event.

Programme Steering Committee

Ian McCurrach and Louisa Smalley (PPI representatives) and Daniel Altman and Melanie Smuk (programme statisticians).

Data Monitoring Committee

David Jayne (chairperson), Rupert Pearse (independent clinician) and Neil Corrigan (independent statistician).

Contributions of authors

Charles H Knowles (https://orcid.org/0000-0001-9854-6754) (Professor of Surgery) was chief investigator of the CapaCiTY programme

Lesley Booth (https://orcid.org/0000-0001-5812-4549) (Patient and Bowel Research UK Charity Research Director) was lead for PPI.

Steve R Brown (https://orcid.org/0000-0002-0980-2793) (Professor of Colorectal Surgery) was co-principal investigator for CapaCiTY trial 3.

Samantha Cross (https://orcid.org/0000-0002-0501-0289) (Lecturer in Medical Statistics) conducted statistical analyses and summarised results in tables for all three CapaCiTY trials.

Sandra Eldridge (https://orcid.org/0000-0001-5638-2317) (Professor of Statistics) provided senior oversight of trial design and all statistical analyses.

Christopher Emmett (https://orcid.org/0000-0002-6022-8475) (Clinical Research Fellow, Gastroenterology) assisted with recruitment and provided clinical support for CapaCiTY trial 2.

Ugo Grossi (https://orcid.org/0000-0001-5372-2873) (Clinical Research Fellow, Surgery) assisted with recruitment and provided clinical support for CapaCiTY trials 1 and 3.

Mary Jordan (https://orcid.org/0000-0002-0497-8634) (Senior Research Fellow, Health Economics) conducted the analysis of economic models.

Jon Lacy-Colson (https://orcid.org/0000-0002-5151-8358) (Consultant Colorectal Surgeon) was co-principal investigator for CapaCiTY trial 3.

James Mason (https://orcid.org/0000-0001-9210-4082) (Professor of Health Economics) provided senior oversight of health economic analyses and modelling.

John McLaughlin (https://orcid.org/0000-0001-6158-5135) (Professor of Gastroenterology) served as chairperson of the PSC and contributed to the study write up.

Rona Moss-Morris (https://orcid.org/0000-0002-2927-3446) (Professor of Psychology) conducted analyses and write-up of all psychological outcomes.

Christine Norton (https://orcid.org/0000-0003-2259-0948) (Professor of Nursing) was co-principal investigator for CapaCiTY trial 1 and oversaw all qualitative studies.

S Mark Scott (https://orcid.org/0000-0002-7997-1533) (Senior Health Scientist) was co-principal investigator for CapaCiTY trial 1 and oversaw analysis of all INVEST interventions.

Natasha Stevens (https://orcid.org/0000-0002-3910-7863) (Senior Trials Manager) provided programme management for all three trials (during years 1 and 2) and assisted development of the CapaCiTY research programme.

Shiva Taheri (https://orcid.org/0000-0003-0879-0601) (Trials Manager) provided programme management for all three trials (during years 3–5).

Yan Yiannakou (https://orcid.org/0000-0001-8437-2925) (Professor of Gastroenterology) was principal investigator for CapaCiTY trial 2.

Publications

Data-sharing statement

All data requests should be submitted to the corresponding author for consideration. Access to anonymised data may be granted following review.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, CCF, NETSCC, PGfAR or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the PGfAR programme or the Department of Health and Social Care.

CapaCiTY trial 1: habit training compared with habit training with direct visual biofeedback in adults with chronic constipation – randomised controlled trial

Recruitment

Recruitment started on 26 March 2015 (first intervention 21 May 2015) and ended 30 June 2018. A total of 182 patients (target 394 patients) were randomised out of 502 patients screened (36.3%) from 10 sites. Reasons for screen failure are shown in Figure 9. Two sites opened but failed to recruit; the remainder randomised between 7 and 71 patients. A total of 182 patients were randomised: 68 to HT, 68 to HTBF and 46 to INVEST-guided therapy (HT or HTBF based on results). A total of 79 participants dropped out of the study before the primary end point, leaving 103 participants at 6 months. The CONSORT flow diagram is shown in Figure 2.

FIGURE 9.

The CapaCiTY trial 1 screen failures. SNS, sacral nerve stimulation.

Results: clinical effectiveness

Primary outcome

Table 8 shows the results for the primary outcome.

TABLE 8 - Mean PAC-QoL score by randomised group at 6 months and mean differences between HT and HTBF and between no-INVEST and INVEST groups for those included in the final analysis model

Adjusted for sex, baseline PAC-QoL score and breakthrough medication (i.e. use of oral and/or rectal laxatives).

Intervention	Mean score (points)		Treatment differencea (95% CI)	p-value
	Baseline (SD)	6 months (SD)
HT vs. HTBF
HT (n = 38)	2.26 (0.69)	1.49 (0.85)	Reference
HTBF (n = 30)	2.41 (0.81)	1.65 (1.03)	–0.03 (–0.33 to 0.27)	0.8445
No INVEST vs. INVEST
No INVEST (n = 68)	2.33 (0.74)	1.56 (0.93)	Reference
INVEST (n = 22)	2.36 (0.78)	1.81 (1.03)	0.22 (–0.11 to 0.55)	0.1871

Secondary outcomes

Table 9 shows changes in PAC-QoL score analysed as binary variables (adjusted analyses) and Table 10 shows data for individual PAC-QoL domains. All other secondary outcomes are shown in Tables 11 and 12. Figure 10 provides a summary of the similar decrease in PAC-QoL score in all intervention groups.

TABLE 9 - Binary PAC-QoL score at 3, 6 and 12 months by randomised group and ORs for HT and HTBF and no-INVEST and INVEST groups for those included in the final analysis model

OR, odds ratio.

Odds ratio comparing HTBF to HT or INVEST to No-INVEST, adjusted for sex, baseline PAC-QoL score and breakthrough medication (i.e. use of oral and/or rectal laxatives).

Intervention	PAC-QoL score (points), mean (SD)		Patients with ≥ 0.4-point reduction, n (%)	OR for ≥ 0.4-point reductiona (95% CI)	Patients with ≥ 1.0-point reduction, n (%)	OR for ≥ 1.0-point reductiona (95% CI)
	Baseline	Follow-up
3 months
HT (N = 43)	2.3 (0.7)	1.7 (0.8)	26 (60.5)	Reference	15 (34.9)	Reference
HTBF (N = 34)	2.3 (0.9)	1.7 (1.0)	18 (52.9)	0.74 (0.3 to 1.9)	11 (32.4)	0.94 (0.3 to 2.7)
No INVEST (N = 77)	2.3 (0.8)	1.7 (0.9)	44 (57.1)	Reference	15 (34.9)	Reference
INVEST (N = 26)	2.5 (0.8)	1.9 (1.0)	15 (57.7)	0.95 (0.4 to 2.4)	11 (32.4)	0.34 (0.1 to 1.1)
6 months
HT (N = 38)	2.3 (0.7)	1.5 (0.8)	26 (68.4)	Reference	14 (36.8)	Reference
HTBF (N = 30)	2.4 (0.8)	1.7 (1.0)	21 (70.0)	1.09 (0.4 to 3.2)	10 (33.3)	0.91 (0.3 to 2.8)
No INVEST (N = 68)	2.3 (0.7)	1.6 (0.9)	47 (69.1)	Reference	14 (36.8)	Reference
INVEST (N = 22)	2.4 (0.8)	1.8 (1.0)	12 (54.5)	0.48 (0.2 to 1.4)	10 (33.3)	0.66 (0.2 to 1.9)
12 months
HT (N = 29)	2.2 (0.7)	1.5 (0.8)	16 (55.2)	Reference	9 (31.0)	Reference
HTBF (N = 23)	2.2 (0.9)	1.4 (1.2)	16 (69.6)	1.68 (0.5 to 5.6)	7 (30.4)	0.84 (0.2 to 3.0)
No INVEST (N = 52)	2.2 (0.8)	1.4 (1.0)	32 (61.5)	Reference	9 (31.0)	Reference
INVEST (N = 13)	2.4 (0.8)	1.7 (0.9)	9 (69.2)	1.60 (0.4 to 6.3)	7 (30.4)	1.37 (0.3 to 6.5)

TABLE 10 - Overall PAC-QoL score for individual domains at 3, 6 and 12 months by randomised group and mean differences between HT and HTBF and between no-INVEST and INVEST groups for those included in each analysis model at the relevant time point

Observed baseline PAC-QoL score (mean and SD) for those with data at both baseline and follow-up time point. (Statistics only for those included in model.)

Adjusted for sex, baseline PAC-QoL score and breakthrough medication (i.e. use of oral and/or rectal laxatives).

Note

The results under each time point summarise only the data from participants with recorded outcomes at both baseline and that follow-up time point.

Intervention	3 months				6 months				12 months
	n	Baseline, mean (SD)a	Follow-up, mean (SD)	Treatment differenceb (95% CI)	n	Baseline, mean (SD)a	Follow-up, mean (SD)	Treatment differenceb (95% CI)	n	Baseline, mean (SD)a	Follow-up, mean (SD)	Treatment differenceb (95% CI)
Overall score (points)
HT	43	2.3 (0.1)	1.7 (0.1)	Reference	38	2.3 (0.1)	1.5 (0.1)	Reference	29	2.2 (0.1)	1.5 (0.1)	Reference
HTBF	34	2.3 (0.2)	1.7 (0.1)	–0.02 (–0.3 to 0.3)	30	2.4 (0.1)	1.7 (0.1)	–0.03 (–0.3 to 0.3)	23	2.2 (0.2)	1.4 (0.2)	–0.01 (–0.4 to 0.4)
No INVEST	77	2.3 (0.1)	1.7 (0.1)	Reference	68	2.3 (0.1)	1.6 (0.1)	Reference	52	2.2 (0.1)	1.4 (0.1)	Reference
INVEST	26	2.5 (0.2)	1.9 (0.1)	0.09 (–0.2 to 0.4)	22	2.4 (0.2)	1.8 (0.1)	0.22 (–0.1 to 0.5)	13	2.4 (0.2)	1.7 (0.2)	–0.02 (–0.4 to 0.4)
Dissatisfaction score (points)
HT	43	3.1 (0.1)	2.1 (0.1)	Reference	38	3.1 (0.1)	2.2 (0.0)	Reference	29	3.1 (0.1)	2.2 (0.0)	Reference
HTBF	34	3.1 (0.1)	2.2 (0.1)	0.17 (–0.3 to 0.6)	30	3.3 (0.1)	2.3 (0.1)	–0.04 (–0.5 to 0.4)	22	3.2 (0.1)	2.0 (0.1)	–0.17 (–0.7 to 0.3)
No INVEST	77	3.1 (0.1)	2.2 (0.0)	Reference	68	3.2 (0.1)	2.2 (0.1)	Reference	51	3.2 (0.1)	2.1 (0.1)	Reference
INVEST	26	3.4 (0.1)	2.4 (0.1)	0.09 (–0.3 to 0.5)	22	3.3 (0.1)	2.3 (0.1)	0.04 (–0.4 to 0.5)	13	3.2 (0.2)	2.0 (0.2)	–0.31 (–0.9 to 0.2)
Physical discomfort score (points)
HT	43	2.5 (0.1)	1.8 (0.1)	Reference	38	2.5 (0.1)	1.7 (0.1)	Reference	29	2.4 (0.1)	1.6 (0.1)	Reference
HTBF	34	2.3 (0.2)	1.8 (0.2)	0.10 (–0.3 to 0.5)	30	2.4 (0.2)	1.7 (0.2)	0.06 (–0.3 to 0.5)	23	2.2 (0.2)	1.5 (0.2)	0.14 (–0.4 to 0.7)
No INVEST	77	2.4 (0.1)	1.8 (0.1)	Reference	68	2.5 (0.1)	1.7 (0.1)	Reference	52	2.3 (0.1)	1.6 (0.1)	Reference
INVEST	26	2.5 (0.2)	2.1 (0.1)	0.14 (–0.2 to 0.5)	22	2.4 (0.2)	1.9 (0.2)	0.27 (–0.1 to 0.7)	13	2.3 (0.3)	1.8 (0.2)	0.10 (–0.4 to 0.6)
Psychosocial discomfort score (points)
HT	43	1.8 (0.1)	1.3 (0.1)	Reference	38	1.7 (0.2)	1.1 (0.1)	Reference	29	1.5 (0.2)	1.1 (0.1)	Reference
HTBF	34	1.8 (0.2)	1.3 (0.2)	–0.05 (–0.4 to 0.3)	30	1.9 (0.2)	1.4 (0.1)	0.07 (–0.3 to 0.4)	23	1.8 (0.2)	1.1 (0.2)	–0.17 (–0.6 to 0.2)
No INVEST	77	1.8 (0.1)	1.3 (0.1)	Reference	68	1.8 (0.1)	1.2 (0.1)	Reference	52	1.6 (0.1)	1.1 (0.1)	Reference
INVEST	26	1.9 (0.2)	1.6 (0.2)	0.21 (–0.1 to 0.6)	22	1.7 (0.2)	1.4 (0.2)	0.24 (–0.1 to 0.6)	13	1.9 (0.3)	1.3 (0.3)	0.02 (–0.4 to 0.5)
Worries and concerns score (points)
HT	43	2.3 (0.1)	1.7 (0.1)	Reference	38	2.2 (0.2)	1.4 (0.1)	Reference	29	2.3 (0.2)	1.3 (0.1)	Reference
HTBF	34	2.3 (0.2)	1.6 (0.1)	–0.08 (–0.4 to 0.3)	30	2.4 (0.2)	1.6 (0.2)	–0.01 (–0.4 to 0.4)	23	2.2 (0.2)	1.4 (0.2)	0.15 (–0.3 to 0.6)
No INVEST	77	2.3 (0.1)	1.7 (0.1)	Reference	68	2.3 (0.1)	1.5 (0.1)	Reference	52	2.2 (0.1)	1.4 (0.1)	Reference
INVEST	26	2.5 (0.2)	2.0 (0.2)	0.08 (–0.3 to 0.4)	22	2.4 (0.2)	1.9 (0.2)	0.33 (–0.1 to 0.7)	13	2.4 (0.3)	1.7 (0.2)	0.12 (–0.4 to 0.6)

TABLE 11 - Other continuous secondary outcomes at 3, 6 and 12 months by randomised group and mean differences between HT and HTBF and between no-INVEST and INVEST groups for those included in each of the final analysis models at the relevant time point

Note

The results under each time point summarise only the data from participants with recorded outcomes at both baseline and that follow-up time point.

Continuous outcome	3 months				6 months				12 months
	n	Baseline, mean (SD)	Follow-up, mean (SD)	Treatment difference (95% CI)	n	Baseline, mean (SD)	Follow-up, mean (SD)	Treatment difference (95% CI)	n	Baseline, mean (SD)	Follow-up, mean (SD)	Treatment difference (95% CI)
PAC-SYM score (points)
Overall
HT	43	2.2 (0.8)	1.6 (0.8)	Reference	38	2.2 (0.8)	1.5 (0.8)	Reference	29	2.0 (0.8)	1.3 (0.7)	Reference
HTBF	34	2.1 (0.9)	1.7 (1.1)	–0.1 (–0.5 to 0.3)	30	2.1 (0.8)	1.5 (1.1)	–0.1 (–0.5 to 0.3)	23	2.1 (0.9)	1.3 (1.1)	0.0 (–0.5 to 0.5)
No INVEST	77	2.2 (0.8)	1.6 (0.9)	Reference	68	2.1 (0.8)	1.5 (0.9)	Reference	52	2.1 (0.8)	1.3 (0.9)	Reference
INVEST	26	2.3 (0.9)	1.7 (0.9)	–0.1 (–0.5 to 0.3)	22	2.2 (0.9)	1.7 (0.8)	–0.2 (–0.6 to 0.2)	13	2.0 (0.9)	1.4 (0.8)	–0.1 (–0.6 to 0.4)
Stool symptoms
HT	43	2.6 (0.9)	1.8 (0.9)	Reference	38	2.5 (0.9)	1.8 (1.0)	Reference	29	2.5 (0.7)	1.5 (0.7)	Reference
HTBF	34	2.5 (0.8)	1.8 (1.2)	–0.0 (–0.5 to 0.5)	30	2.5 (0.9)	1.7 (1.3)	0.1 (–0.4 to 0.6)	23	2.5 (0.9)	1.4 (1.2)	0.1 (–0.4 to 0.6)
No INVEST	77	2.6 (0.9)	1.8 (1.0)	Reference	68	2.5 (0.9)	1.8 (1.1)	Reference	52	2.5 (0.8)	1.5 (0.9)	Reference
INVEST	26	2.7 (1.0)	2.0 (1.1)	–0.2 (–0.7 to 0.3)	22	2.6 (1.0)	2.0 (1.0)	–0.3 (–0.8 to 0.2)	13	2.4 (1.2)	1.6 (0.8)	–0.1 (–0.7 to 0.5)
Abdominal symptoms
HT	43	2.2 (1.0)	1.6 (1.1)	Reference	38	2.2 (0.9)	1.3 (0.9)	Reference	29	2.0 (1.0)	1.4 (1.0)	Reference
HTBF	34	2.0 (1.1)	1.7 (1.1)	–0.1 (–0.6 to 0.4)	30	2.0 (1.0)	1.5 (1.2)	–0.2 (–0.7 to 0.3)	23	1.9 (1.1)	1.3 (1.3)	0.1 (–0.6 to 0.8)
No INVEST	77	2.1 (1.0)	1.6 (1.1)	Reference	68	2.1 (1.0)	1.4 (1.0)	Reference	52	2.0 (1.0)	1.4 (1.1)	Reference
INVEST	26	2.2 (1.0)	1.9 (1.0)	–0.2 (–0.7 to 0.3)	22	2.0 (1.1)	1.7 (1.0)	–0.3 (–0.8 to 0.2)	13	1.7 (0.9)	1.3 (1.0)	0.0 (–0.7 to 0.7)
Rectal symptoms
HT	42	1.5 (1.1)	1.2 (0.9)	Reference	37	1.5 (1.1)	1.0 (0.8)	Reference	29	1.3 (1.0)	0.7 (0.7)	Reference
HTBF	34	1.6 (1.2)	1.3 (1.3)	–0.1 (–0.6 to 0.4)	30	1.6 (1.1)	1.2 (1.2)	–0.2 (–0.7 to 0.3)	23	1.5 (1.1)	0.9 (1.0)	–0.1 (–0.6 to 0.4)
No INVEST	76	1.5 (1.2)	1.2 (1.1)	Reference	67	1.5 (1.1)	1.1 (1.0)	Reference	52	1.4 (1.0)	0.8 (0.8)	Reference
INVEST	26	1.8 (1.2)	1.2 (1.0)	0.0 (–0.5 to 0.5)	22	1.7 (1.3)	1.1 (0.9)	0.0 (–0.5 to 0.5)	13	1.7 (1.1)	1.1 (1.0)	–0.3 (–0.8 to 0.2)
Diary data
Bowel frequency: mean number of attempts to empty bowels over 2 weeks
HT	42	27.9 (22.4)	22.4 (11.9)	Reference	34	29.0 (24.1)	21.8 (11.0)	Reference	25	31.6 (26.1)	24.2 (12.7)	Reference
HTBF	33	23.4 (12.6)	21.7 (12.9)	0.7 (–5.0 to 6.4)	29	26.6 (18.1)	22.6 (13.8)	–0.7 (–7.0 to 5.6)	22	27.4 (19.7)	24.5 (15.0)	–0.3 (–8.4 to 7.8)
No INVEST	75	25.9 (18.7)	22.1 (12.3)	Reference	63	27.9 (21.4)	22.2 (12.3)	Reference	47	29.6 (23.2)	24.4 (13.7)	Reference
INVEST	26	29.9 (15.0)	26.3 (13.6)	–4.2 (–9.9 to 1.5)	22	28.4 (15.6)	23.0 (11.5)	–0.8 (–6.8 to 5.2)	12	24.4 (12.4)	23.0 (11.2)	1.4 (–7.2 to 10.0)
Bowel frequency: mean number of times stool was passed over 2 weeks
HT	42	15.1 (10.9)	15.9 (10.8)	Reference	34	15.8 (11.7)	14.2 (8.3)	Reference	25	16.2 (11.9)	17.6 (11.1)	Reference
HTBF	33	14.8 (7.6)	16.5 (10.8)	–0.6 (–5.6 to 4.4)	28	14.8 (8.9)	15.4 (11.3)	–1.2 (–6.2 to 3.8)	22	16.2 (9.1)	18.6 (13.0)	–1.0 (–8.1 to 6.1)
No INVEST	75	14.9 (9.5)	16.1 (10.7)	Reference	62	15.3 (10.4)	14.8 (9.7)	Reference	47	16.2 (10.6)	18.1 (11.9)	Reference
INVEST	26	15.0 (9.7)	16.7 (10.9)	–0.5 (–5.4 to 4.4)	22	13.8 (7.7)	14.3 (8.2)	0.5 (–4.1 to 5.1)	12	15.2 (11.4)	17.0 (10.7)	1.1 (–6.4 to 8.6)
Nature of bowel movement: mean number of days (out of 14) laxatives used
HT	42	4.3 (5.2)	1.0 (3.0)	Reference	34	3.6 (4.5)	0.5 (1.6)	Reference	25	3.8 (4.9)	1.3 (2.5)	Reference
HTBF	33	3.6 (5.0)	1.0 (3.1)	–0.0 (–1.4 to 1.4)	29	3.5 (4.8)	1.0 (2.7)	–0.4 (–1.5 to 0.7)	22	3.6 (5.0)	0.9 (2.2)	0.4 (–1.0 to 1.8)
No INVEST	75	4.0 (5.1)	1.0 (3.0)	Reference	63	3.5 (4.6)	0.7 (2.2)	Reference	47	3.7 (4.9)	1.1 (2.3)	Reference
INVEST	26	6.2 (6.0)	1.7 (4.1)	–0.6 (–2.1 to 0.9)	22	6.2 (6.1)	0.6 (3.0)	0.1 (–1.1 to 1.3)	12	5.1 (5.9)	2.6 (5.4)	–1.5 (–3.5 to 0.5)
Nature of bowel movement: mean number of days (out of 14) glycerine suppositories used
HT	42	0.8 (2.5)	1.1 (2.3)	Reference	34	0.2 (0.7)	0.8 (2.0)	Reference	25	0.1 (0.3)	0.5 (1.1)	Reference
HTBF	32	0.5 (1.5)	0.9 (2.3)	0.2 (–0.9 to 1.3)	29	0.9 (2.1)	1.2 (2.8)	–0.4 (–1.6 to 0.8)	22	0.6 (1.9)	0.1 (0.3)	0.4 (–0.1 to 0.9)
No INVEST	74	0.7 (2.1)	1.0 (2.3)	Reference	63	0.5 (1.5)	1.0 (2.4)	Reference	47	0.4 (1.3)	0.3 (0.9)	Reference
INVEST	26	1.2 (3.3)	0.8 (2.0)	0.2 (–0.8 to 1.2)	22	1.3 (3.6)	1.2 (3.1)	–0.3 (–1.6 to 1.0)	12	1.3 (4.2)	2.2 (4.0)	–1.8 (–3.0,–0.6)
EQ-VAS score (points)
HT	43	68.4 (21.0)	71.0 (21.6)	Reference	39	69.9 (19.7)	69.0 (23.8)	Reference	29	69.7 (18.1)	73.8 (20.2)	Reference
HTBF	34	66.7 (20.1)	71.2 (21.1)	–0.2 (–10.0 to 9.6)	29	66.1 (20.1)	71.0 (20.1)	–2.0 (–12.9 to 8.9)	23	68.8 (17.4)	74.7 (18.8)	–0.9 (–11.9 to 10.1)
No INVEST	77	67.6 (20.5)	71.1 (21.2)	Reference	68	68.3 (19.8)	69.8 (22.1)	Reference	52	69.3 (17.6)	74.2 (19.4)	Reference
INVEST	26	65.3 (17.4)	66.8 (17.6)	4.3 (–4.9 to 13.5)	22	68.1 (16.6)	71.1 (17.3)	–1.3 (–11.6 to 9.0)	13	71.5 (17.4)	71.2 (19.9)	3.0 (–9.1 to 15.1)
PHQ-9 score (points)
HT	43	8.4 (7.0)	7.8 (7.1)	Reference	38	7.5 (6.5)	7.7 (7.4)	Reference	29	6.7 (6.8)	7.2 (6.5)	Reference
HTBF	34	6.8 (6.4)	6.4 (6.2)	1.4 (–1.7 to 4.5)	30	6.8 (6.4)	7.2 (6.9)	0.5 (–3.0 to 4.0)	23	6.4 (6.8)	5.9 (7.6)	1.3 (–2.6 to 5.2)
No INVEST	77	7.7 (6.8)	7.2 (6.7)	Reference	68	7.2 (6.4)	7.5 (7.1)	Reference	52	6.6 (6.8)	6.6 (6.9)	Reference
INVEST	26	10.0 (6.0)	9.5 (7.4)	–2.3 (–5.4 to 0.8)	22	8.8 (4.8)	8.8 (5.6)	–1.4 (–4.7 to 1.9)	13	8.2 (7.0)	8.2 (7.1)	–1.6 (–5.9 to 2.7)
GAD-7 score (points)
HT	43	7.4 (6.8)	7.1 (6.9)	Reference	38	6.9 (6.4)	5.8 (6.3)	Reference	29	6.4 (6.4)	6.5 (6.5)	Reference
HTBF	34	7.6 (6.5)	6.6 (6.1)	0.6 (–2.4 to 3.6)	30	7.0 (6.4)	6.4 (6.0)	–0.6 (–3.6 to 2.4)	23	6.2 (6.0)	5.1 (6.3)	1.4 (–2.2 to 5.0)
No INVEST	77	7.5 (6.6)	6.9 (6.5)	Reference	68	7.0 (6.4)	6.1 (6.1)	Reference	52	6.3 (6.2)	5.9 (6.4)	Reference
INVEST	26	8.8 (6.5)	9.2 (6.5)	–2.3 (–5.2 to 0.6)	22	8.4 (6.0)	8.7 (6.4)	–2.6 (–5.6 to 0.4)	13	6.8 (6.1)	7.4 (7.1)	–1.5 (–5.6 to 2.6)
Global patient satisfaction score (points)
HT	42	–	2.5 (1.0)	Reference	37	–	2.8 (0.9)	Reference	29	–	2.7 (0.8)	Reference
HTBF	32	–	2.4 (1.1)	0.1 (–0.4 to 0.6)	29	–	2.3 (1.3)	0.4 (–0.1 to 0.9)	23	–	2.5 (1.5)	0.2 (–0.5 to 0.9)
No INVEST	74	–	2.5 (1.0)	Reference	66	–	2.6 (1.1)	Reference	52	–	2.6 (1.2)	Reference
INVEST	25	–	2.3 (1.4)	0.2 (–0.3 to 0.7)	21	–	2.5 (1.1)	0.1 (–0.4 to 0.6)	13	–	2.6 (1.1)	–0.0 (–0.7 to 0.7)
Global patient improvement score (points)
HT	43	–	61.5 (27.8)	Reference	38	–	65.8 (23.1)	Reference	29	–	69.8 (23.2)	Reference
HTBF	34	–	44.9 (34.3)	16.7 (2.6 to 30.8)	28	–	52.7 (35.2)	13.0 (–1.4 to 27.4)	23	–	59.8 (38.0)	10.0 (–7.2 to 27.2)
No INVEST	77	–	54.2 (31.7)	Reference	66	–	60.2 (29.4)	Reference	52	–	65.4 (30.7)	Reference
INVEST	26	–	45.0 (30.8)	9.2 (–5.0 to 23.4)	22	–	45.7 (30.2)	14.5 (0.0 to 29.0)	13	–	62.2 (32.3)	3.2 (–16.0 to 22.4)

TABLE 12 - Binary secondary outcomes (EQ-5D-5L) at 3, 6 and 12 months by randomised group and mean differences between HT and HTBF and no-INVEST and INVEST groups for those included in each of the final analysis models at the relevant time point

Note

The results under each time point summarise only the data from participants with recorded outcomes at both baseline and that follow-up time point.

Binary outcome	N	Indicating problems at baseline, n (%)	3 months				6 months				12 months
			Indicating problems at follow-up, n (%)	Difference in proportions (95% CI)	N	Indicating problems at baseline, n (%)	Indicating problems at follow-up, n (%)	Difference in proportions (95% CI)	N	Indicating problems at baseline, n (%)	Indicating problems at follow-up, n (%)	Difference in proportions (95% CI)
Mobility
HT	43	14 (32.6)	12 (27.9)	Reference	38	11 (28.9)	12 (31.6)	Reference	29	7 (24.1)	5 (17.2)	Reference
HTBF	34	9 (26.5)	12 (35.3)	–0.1 (–0.3 to 0.1)	30	9 (30.0)	12 (40.0)	–0.1 (–0.3 to 0.1)	23	6 (26.1)	8 (34.8)	–0.2 (–0.4 to 0.1)
No INVEST	77	23 (29.9)	24 (31.2)	Reference	68	20 (29.4)	24 (35.3)	Reference	52	13 (25.0)	13 (25.0)	Reference
INVEST	26	9 (34.6)	8 (30.8)	0.0 (–0.2 to 0.2)	22	8 (36.4)	7 (31.8)	0.0 (–0.2 to 0.3)	13	5 (38.5)	4 (30.8)	–0.1 (–0.3 to 0.2)
Self-care
HT	43	8 (18.6)	8 (18.6)	Reference	38	4 (10.5)	4 (10.5)	Reference	29	2 (6.9)	4 (13.8)	Reference
HTBF	34	4 (11.8)	7 (20.6)	–0.0 (–0.2 to 0.2)	29	4 (13.8)	5 (17.2)	–0.1 (–0.2 to 0.1)	23	4 (17.4)	3 (13.0)	0.0 (–0.2 to 0.2)
No INVEST	77	12 (15.6)	15 (19.5)	Reference	67	8 (11.9)	9 (13.4)	Reference	52	6 (11.5)	7 (13.5)	Reference
INVEST	26	5 (19.2)	4 (15.4)	0.0 (–0.1 to 0.2)	22	4 (18.2)	3 (13.6)	–0.0 (–0.2 to 0.2)	13	1 (7.7)	0 (0.0)	0.1 (0.0 to 0.2)
Usual activities
HT	43	14 (32.6)	20 (46.5)	Reference	38	10 (26.3)	15 (39.5)	Reference	29	6 (20.7)	8 (27.6)	Reference
HTBF	34	14 (41.2)	15 (44.1)	0.0 (–0.2 to 0.2)	30	13 (43.3)	11 (36.7)	0.0 (–0.2 to 0.3)	22	8 (36.4)	9 (40.9)	–0.1 (–0.4 to 0.1)
No INVEST	77	28 (36.4)	35 (45.5)	Reference	68	23 (33.8)	26 (38.2)	Reference	51	14 (27.5)	17 (33.3)	Reference
INVEST	26	11 (42.3)	14 (53.8)	–0.1 (–0.3 to 0.1)	22	8 (36.4)	8 (36.4)	0.0 (–0.2 to 0.3)	13	6 (46.2)	3 (23.1)	0.1 (–0.2 to 0.4)
Pain/discomfort
HT	43	37 (86.0)	32 (74.4)	Reference	38	33 (86.8)	27 (71.1)	Reference	29	25 (86.2)	20 (69.0)	Reference
HTBF	34	28 (82.4)	27 (79.4)	–0.0 (–0.2 to 0.1)	30	25 (83.3)	24 (80.0)	–0.1 (–0.3 to 0.1)	23	18 (78.3)	13 (56.5)	0.1 (–0.1 to 0.4)
No INVEST	77	65 (84.4)	59 (76.6)	Reference	68	58 (85.3)	51 (75.0)	Reference	52	43 (82.7)	33 (63.5)	Reference
INVEST	26	24 (92.3)	22 (84.6)	–0.1 (–0.2 to 0.1)	22	19 (86.4)	17 (77.3)	–0.0 (–0.2 to 0.2)	13	12 (92.3)	12 (92.3)	–0.3 (–0.5 to –0.1)
Anxiety/depression
HT	43	26 (60.5)	24 (55.8)	Reference	38	23 (60.5)	20 (52.6)	Reference	29	18 (62.1)	16 (55.2)	Reference
HTBF	33	15 (45.5)	14 (42.4)	0.1 (–0.1 to 0.4)	30	12 (40.0)	14 (46.7)	0.1 (–0.2 to 0.3)	23	8 (34.8)	11 (47.8)	0.1 (–0.2 to 0.3)
No INVEST	76	41 (53.9)	38 (50.0)	Reference	68	35 (51.5)	34 (50.0)	Reference	52	26 (50.0)	27 (51.9)	Reference
INVEST	26	17 (65.4)	17 (65.4)	–0.2 (–0.4 to 0.1)	22	15 (68.2)	16 (72.7)	–0.2 (–0.4 to –0.0)	13	8 (61.5)	8 (61.5)	–0.1 (–0.4 to 0.2)

FIGURE 10.

Average PAC-QoL score over time in months for (a) all participants (n = 178); and (b) those with no missing data (n = 54). Both show reductions in score (i.e. improvement) over time.

Supplementary analyses for secondary outcomes included:

1. Binary responses (i.e. responder vs. non-responder) to treatment defined as a ≥ 0.4-point reduction in PAC-QoL score (at 3, 6 and 12 months post end of treatment). A logistic regression model was fitted using the ‘xtmelogit’ (with logit link) command in Stata. Two stratification variables (sex and study site) are included in the model; sex was included as a fixed-effect covariate in the regression model and site as a random effect. Other covariates in the model were fixed effects for intervention, baseline PAC-QoL score and breakthrough medication (i.e. use of oral and/or rectal laxatives).

2. Binary responses to treatment defined as a ≥ 1.0-point reduction in PAC-QoL score (at 3, 6 and 12 months post end of treatment). A logistic regression model was fitted in Stata using the ‘xtmelogit’ (with logit link) command. Two stratification variables (sex and study site) were included in the model; sex was included as a fixed-effect covariate in the regression model and site as random effect. Other covariates included the model were fixed effects for intervention, baseline PAC-QoL score and breakthrough medication (i.e. use of oral and/or rectal laxatives).

Subgroup analyses

Planned subgroup analyses showed no difference in response (e.g. reduction in PAC-QoL score) between patients with or without depression (baseline PHQ-9 score ≥ 5 points), anxiety (baseline GAD-7 score ≥ 4.5 points) or joint hypermobility (baseline joint hypermobility syndrome score ≥ 2 points). However, there was a significant interaction between depression and the INVEST treatment and anxiety and the INVEST treatment.

Safety analyses

The study (not being of a medicinal product) did not record unrelated AEs. Serious adverse events (SAEs) and related AEs were small in number, with only two (both unrelated) SAEs reported (Table 13).

TABLE 13 - The CapaCiTY trial 1 AEs and SAEs

None of the SAEs was identified as related to a treatment; no unresolved SAEs were reported.

Variable	HT (N = 68)	HTBF (N = 68)	No INVEST (N = 136)	INVEST (N = 46)	Total (N = 182)
Related AEs
Number of patients reporting AE	4	4	8	5	13
Number of AEs reported by category
Abdominal pain	0	0	0	0	0
Anal/rectal pain or discomfort	0	2	2	1	3
Bloating	0	0	0	0	0
Constipation	0	0	0	0	0
Haemorrhoids	0	0	0	0	0
Loose motions	0	2	2	0	2
Rectal bleeding	3	2	5	5	10
Vaginal/perineal bulge	0	0	0	0	0
Miscellaneous	1	0	1	2	3
Severity
Mild	2	2	4	5	9
Moderate	2	4	6	3	9
Severe	0	0	0	0	0
Action
No action taken	1	3	4	5	9
Withdrawal	0	0	0	1	1
Concomitant medication	1	1	2	2	4
Non-drug therapy	1	2	3	0	3
SAEsa
Number of patients reporting SAE	0	0	0	2	2
Number of SAEs reported that require hospitalisation or prolongation of existing hospitalisation	0	0	0	2	2
Causality
Unlikely to be related	0	0	0	2	2
Related	0	0	0	0	0
Expectedness
Expected	0	0	0	0	0
Unexpected	0	0	0	2	2
Action taken
Withdrawal	0	0	0	1	1
None	0	0	0	1	1

Results: cost-effectiveness

Summary and completeness of data

Participants were allocated randomly to HT, HTBF or INVEST-guided treatment. The average duration of training and support visits for HTBF was greater than for HT, as expected, with the INVEST group falling between these two, reflecting the split allocation. Patients allocated to HTBF completed an average of 2.1 biofeedback training sessions; 43 (93%) patients allocated to INVEST completed the INVEST procedure. The cost of intervention procedures was the predominant determinant of overall cost (Table 14) and the only variable differing significantly between groups [one-way analysis of variance (ANOVA), p < 0.001]. Use of prescription drugs and community health-care was relatively low and similar between groups. Hence, cost differences were driven by the intervention costs. Societal costs reflected a minority of patients (≈ 20%) reporting substantial time away from work. Completeness of cost data diminished as follow-up progressed from 86% in the first period to 36% in the last period. Quality-of-life scores at each follow-up point and summary QALYs were similar between groups. Similarly, completeness of EQ-5D-5L data diminished as follow-up proceeded from 98% to 35%. An apparent trend of improving quality-of-life in all groups may be confounded by increasing missingness of data.

TABLE 14 - The CapaCiTY trial 1 economic analysis variables (£, 2018)

Notes

Subtotals may not sum to totals because of different numbers of participants contributing.

Intervention occurred in the 3 months preceding the trial baseline, when follow-up commenced.

Days from work included in societal cost but not NHS cost.

The QALY estimates reported cover 15 months and are undiscounted in this table.

Variable	HT (N = 68 patients)			HTBF (N = 68 patients)			INVEST (N = 46 patients)
	Mean	SD	n	Mean	SD	n	Mean	SD	n
Resource use
Intervention
Duration of visits (minutes)	140	80	57	152	102	61	143	85	39
Number of biofeedback sessions	–	–	–	2.09	1.43	67	0.93	1.35	44
Invest procedure completed	–	–	–	–	–	–	0.93	0.33	46
Number of prescriptions	2.00	1.98	48	1.43	1.97	46	1.53	1.62	40
0–3 months
Number of prescriptions	0.48	0.70	44	0.83	1.42	36	0.52	1.09	27
GP visits	0.02	0.15	44	0.14	0.35	36	0.11	0.32	27
District nurse visits	0.02	0.15	44	0.00	0.00	36	0.00	0.00	27
Pharmacist visits	0.05	0.21	44	0.06	0.23	36	0.15	0.46	27
A&E visits	0.00	0.00	44	0.03	0.17	36	0.00	0.00	27
Outpatient visits	0.20	0.51	44	0.08	0.28	36	0.15	0.46	27
Days from work	0.98	2.65	44	0.64	1.73	36	2.59	5.80	27
Inpatient visits	0.00	0.00	43	0.11	0.68	35	0.00	0.00	27
4–6 months
Number of prescriptions	0.54	1.00	39	0.37	0.61	30	0.32	0.78	22
GP visits	0.18	0.60	39	0.17	0.59	30	0.14	0.47	22
District nurse visits	0.00	0.00	39	0.00	0.00	30	0.00	0.00	22
Pharmacist visits	0.08	0.35	39	0.17	0.59	30	0.05	0.21	22
A&E visits	0.00	0.00	39	0.00	0.00	30	0.00	0.00	22
Outpatient visits	0.18	0.72	39	0.10	0.31	30	0.14	0.35	22
Days from work	3.51	12.72	39	0.86	2.26	29	1.95	5.72	22
Inpatient visits	0.03	0.16	39	0.00	0.00	30	0.00	0.00	21
7–12 months
Number of prescriptions	0.48	0.89	31	0.33	0.64	24	1.15	1.46	13
GP visits	0.27	0.64	30	0.29	0.91	24	0.46	1.66	13
District nurse visits	0.07	0.37	30	0.00	0.00	24	0.00	0.00	13
Pharmacist visits	0.03	0.18	30	0.00	0.00	24	0.54	1.66	13
A&E visits	0.10	0.55	30	0.00	0.00	24	0.08	0.28	13
Outpatient visits	0.10	0.55	30	0.42	1.32	24	0.38	1.39	13
Days from work	2.30	6.72	30	1.08	3.08	24	4.00	11.48	13
Inpatient visits	0.00	0.00	30	0.04	0.21	23	0.00	0.00	13
Health-care cost
–3 to 0 months (intervention)	326	186	57	493	322	61	804	351	39
0–3 months	42	87	43	55	137	35	35	75	27
4–6 months	59	152	39	26	51	30	25	51	21
7–12 months	57	142	30	132	320	23	147	283	13
Overall	506	205	19	815	411	17	988	269	9
Societal cost
Overall	1187	1866	19	1231	1016	17	1312	665	9
EQ-5D-5L
–3 months	0.664	0.308	68	0.693	0.247	66	0.679	0.223	45
3 months	0.652	0.329	43	0.708	0.261	33	0.665	0.211	26
6 months	0.733	0.251	38	0.681	0.272	29	0.642	0.256	22
12 months	0.737	0.262	29	0.734	0.289	22	0.713	0.136	13
QALYs
–3 to 12 months	0.874	0.357	23	0.842	0.345	19	0.897	0.171	11

Cost-effectiveness analyses

The base-case model is reported in Table 15. Compared with HT, both HTBF and INVEST-guided treatment were estimated to be associated with higher costs and lower QoL, making HT the dominant strategy. In both instances cost increases were significant (HTBF vs. HT: £239, 95% CI £133 to £354; INVEST vs. HT: £543, 95% CI £403 to £685), and QoL decrements were small for HTBF (–0.010 QALYs, 95% CI –0.053 to 0.03 QALYs) and more significant for INVEST (–0.047 QALYs, 95% CI –0.093 to –0.001 QALYs), compared with HT. These findings are presented in Figure 11. On the ICER plane, credible regions are plotted around median ICERs reflecting where 95% of bootstrapped estimates lie. Incremental differences comparing HTBF and INVEST are not presented because both are dominated by HT. NMB values were negative for HTBF and INVEST across the commonly used range of WTP thresholds, indicating lack of cost-effectiveness (see Table 15). These findings are presented as CEACs at varying WTP thresholds, showing the probability of each intervention being cost-effective. At the NICE threshold of £30,000 per QALY, the probability that HT is the most cost-effective alternative is 83%. The EVPI values the opportunity loss of ongoing uncertainty. At a WTP threshold of £30,000 per QALY, the EVPI per subject is £59. There are no reliable British figures for the incidence, prevalence or chronic idiopathic constipation requiring hospital management or a future timeline for relevant new research. Assuming that an approximate cost of a further definitive RCT is £1M, the research would need to benefit only ≈ 17,000 patients, which might indicate that further research is merited, particularly if the findings have international relevance.

TABLE 15 - The CapaCiTY trial 1 cost-effectiveness analysis (£, 2018)

Probability cost-effective at WTP threshold of £15,000 per QALY.

Probability cost-effective at WTP threshold of £30,000 per QALY.

Denotes a dominated strategy: increase in cost and decrease in QALYs.

Non-recoverable cost assumption.

Non-participants (zero intervention cost patients) excluded.

Notes

Bootstrapped 95% CIs; median ICERs.

Base-case and sensitivity analyses: imputed with 60 draws; assuming MAR.

Complete case: includes only participants with complete cost and QALY data; assuming missing completely at random.

Variable	ICER (95% CI)	IC (95% CI)	IQ (95% CI)	CEAC p-valuea	CEAC p-valueb	NMBa (95% CI)	NMBb (95% CI)	EVPIb
Base case
HTBF vs. HT	c (7506 to c)	239 (133 to 354)	–0.010 (–0.053 to 0.03)	0.098	0.168	–402 (–1058 to 219)	–565 (–1858 to 665)	59
INVEST vs. HT	c (c to c)	543 (403 to 685)	–0.047 (–0.093 to –0.001)	0.001	0.002	–1241 (–1962 to –539)	–1939 (–3338 to –564)	–
Complete case
HTBF vs. HT	c (3777 to c)	340 (135 to 578)	–0.016 (–0.121 to 0.076)	0.142	0.174	–620 (–2206 to 798)	–897 (–4026 to 1906)	609
INVEST vs. HT	c (3848 to c)	542 (351 to 764)	–0.005 (–0.156 to 0.124)	0.230	0.304	–670 (–2934 to 1357)	–793 (–5262 to 3259)	–
Sensitivity analysis 1d
HTBF vs. HT	c (73,105 to c)	615 (536 to 695)	–0.031 (–0.074 to 0.008)	0.001	0.005	–1092 (–1761 to –468)	–1570 (–2885 to –330)	1
INVEST vs. HT	c (c to c)	744 (624 to 875)	–0.069 (–0.113 to –0.021)	0.000	0.000	–1777 (–2454 to –1079)	–2809 (–4149 to –1394)	–
Sensitivity analysis 2e
HTBF vs. HT	c (11,439 to c)	246 (136 to 365)	–0.024 (–0.073 to 0.019)	0.041	0.081	–621 (–1382 to 64)	–996 (–2468 to 335)	25
INVEST vs. HT	c (c to c)	532 (407 to 652)	–0.053 (–0.102 to –0.002)	0.000	0.001	–1325 (–2077 to –566)	–2119 (–3609 to –617)	–

FIGURE 11.

The CapaCiTY trial 1 cost-effectiveness analysis: base case. (a) ICER plane [the average ICER for each intervention (compared with HT) is shown as a line from the origin and the credible regions (dashed circles) show where 95% of estimates lie]; (b) incremental NMB (the shaded regions are the corresponding 95% CIs); (c) CEACs and CEAF; and (d) EVPI.

In Figure 11a the ICER plane shows the bootstrapped distribution of estimated incremental costs and QALYs comparing HTBF or INVEST with HT. Bootstrapping is a method of resampling the trial data to mimic conducting many trials; the range of findings reflects the uncertainty in the trial. The average ICER for each intervention (compared with HT) is shown as a line from the origin. The credible regions (dashed circles) show where 95% of estimates lie. Compared with HT, average costs increase for both HTBF and INVEST and QALYs decrease, meaning that HT is the dominant treatment strategy. In Figure 11b the incremental NMB describes the resource gain (or loss) when investing in HTBF or INVEST. The shaded regions are the corresponding 95% CIs. Investing in HTBF or INVEST at any threshold WTP would impose a loss to the health system when compared with other alternative choices. In Figure 11c the presentation of CEACs re-presents the incremental NMB findings as the probability that each intervention is cost-effective at different WTP thresholds. The CEAF is a necessary further analysis when three or more interventions are compared. Although HT is the dominant strategy the finding is uncertain, being estimated to be correct with a probability of 83% at a WTP threshold of £30,000 per QALY. In Figure 11d the EVPI indicates the value to a health-care system of further research to eliminate uncertainty, as WTP varies.

The complete-case analysis is also presented in Table 15. Although the numbers contributing are small, the findings are similar to the base case, if less precise. The base-case analysis costed the intervention received by each participant in terms of the number of scheduled visits attended, time taken and equipment used. This assumes no resource losses due to non-attendance or cancellation. A first post hoc sensitivity analysis costed each intervention fully per protocol, assuming that these were non-recoverable costs. The consequence is that the incremental cost of HTBF and INVEST increases, increasing the probability that HT is cost-effective (p > 99% at a WTP threshold of £30,000 per QALY). A minority of patients (12%) had no record of engaging in the trial intervention. In a second post hoc sensitivity analysis, these subjects were removed to explore a participant analysis and reduce the imputation burden. Findings were similar to and thus supportive of the base-case findings.

CapaCiTY trial 2: pragmatic randomised trial of low-volume compared with high-volume initiated transanal irrigation therapy in adults with chronic constipation

Recruitment

First recruitment and intervention was on 11 November 2015; recruitment ended 30 June 2018. A total of 65 patients (target 300 patients) were randomised out of 150 screened (21.7%) from seven sites. Reasons for screen failure are shown in Figure 12. Three sites opened but failed to recruit; the remainder randomised 1–33 patients, with approximately half of the patients recruited from secondary care and half from tertiary care. A total of 65 patients were randomised. A total of 14 patients withdrew from the study before the primary end point (51 completed the 3-month follow-up). A total of 30 patients were randomised to low-volume TAI and 35 to high-volume TAI. The CONSORT flow diagram is shown in Figure 4.

FIGURE 12.

The CapaCiTY trial 2 screen failures. SNS, sacral nerve stimulation.