SeHCAT (tauroselcholic [75selenium] acid) for the investigation of bile acid diarrhoea in adults: a systematic review and cost-effectiveness analysis

Diagnostic assessment

The initial investigation of patients with chronic diarrhoea should involve history-taking, an assessment of clinical symptoms and signs to exclude cancer, as indicated in NICE guideline 12, Suspected Cancer: Recognition and Referral. 5 The initial clinical assessment should also include blood and stool tests to exclude anaemia, coeliac disease, infection and inflammation, as recommended in the British Society of Gastroenterology (BSG) clinical guidelines. 3 The BSG guidelines position SeHCAT testing as part of secondary clinical assessment, following initial assessment/investigations to exclude coeliac disease (coeliac serology and upper gastrointestinal endoscopy and biopsy in people with suspected coeliac disease), common infections (stool examination for Clostridium difficile, ova, cysts and parasites) and colorectal cancer (colonoscopy in people with altered bowel habit and rectal bleeding, and faecal immunochemical testing to guide priority investigations in people with lower gastrointestinal symptoms and no rectal bleeding). 3

The BSG guidelines list BAD among the ‘common disorders’ to be investigated as part of secondary clinical assessment and state that a positive diagnosis of BAD should be made either using SeHCAT testing or by measuring the serum bile acid precursor 7-alpha-hydroxy-4-cholesten-3-one, depending on local availability. 3 The BSG guidelines also state that ‘there is insufficient evidence to recommend use of an empirical trial of treatment for bile acid diarrhoea rather than making a positive diagnosis’. 3 Referral to secondary care is required for investigation and diagnosis of BAD.

NICE clinical guideline 61, Irritable Bowel Syndrome in Adults: Diagnosis and Management,6 recommends considering a diagnosis of irritable bowel syndrome (IBS) for patients with abdominal pain or discomfort that is either relieved by defecation or associated with altered bowel frequency or stool form when the initial investigations are normal and at least two of the following symptoms are present: altered stool passage (straining, urgency, incomplete evacuation); abdominal bloating (more common in women than men), distension, tension or hardness; symptoms worsened by eating; and passage of mucus. The guideline also states that further tests such as colonoscopy or imaging are not necessary to confirm an IBS diagnosis. 6 Investigation of BAD may be useful among patients previously diagnosed with IBS-D; however, NICE clinical guideline 61 does not currently include any recommendations on the investigation of BAD. 6

Investigation of BAD may also be considered when diarrhoea persists, regardless of conventional treatment, in those conditions with which it may appear as a secondary condition. When chronic diarrhoea appears after ileal resection (removal of the terminal part of the small bowel to treat Crohn’s disease), BAD is so common (> 95% of cases)7 that a diagnostic test before treatment may not be considered necessary.

The use of SeHCAT in current clinical practice appears to vary, with some studies indicating that imaging tests and invasive investigations such as colonoscopy are often performed before SeHCAT. 4,8,9 Multiple interactions with different clinicians over many years often take place before BAD is investigated. 10

The manufacturer advises that SeHCAT testing is currently available at 85 hospitals across 74 out of 225 NHS acute trusts in England (data from August 2020). 11 According to the 2018/19 NHS National Cost Collection data,12 the trusts in which SeHCAT testing is available perform about 10,000 SeHCAT tests per year. The number of tests performed across trusts varies widely, ranging from < 50 tests per year to > 500 tests per year.

Management/treatment

The symptoms of BAD are most often controlled with BAS medication. BASs bind to bile acids in the small bowel and prevent them from irritating the colon; they may also slow transit time. The treatment may be long term.

There are currently three BASs available for use in the UK NHS: colestyramine, colestipol and colesevelam. Colestyramine and colestipol come in powder or granule form and colesevelam comes in tablet form. Use of both colestipol and colesevelam for BAD is currently off-label (NICE13). BASs can be difficult to tolerate: constipation and flatulence are commonly reported adverse events, some people find the taste and texture of colestyramine and colestipol very unpleasant and some patients have reported weight gain or weight loss. Increases in dose, addition of antidiarrhoea medication or changes in diet may also be needed to achieve adequate symptom control. Long-term use of colestyramine has been associated with reduced vitamin and folate levels. 14 However, 1–2 years of colestipol use has been reported to have no effect on vitamin A or folic acid levels, and only a small effect on vitamin D levels. 14 Colesevelam was not associated with significant reductions in the absorption of vitamins A, D, E or K in studies of up to 1 year. 14 Guidelines made no recommendation about routine monitoring of fat-soluble vitamins during long-term BAS therapy, while noting that approved product labels recommend supplementation of vitamins A, D and K only if deficiency occurs. 14

In current practice, in some UK secondary care settings (reported by specialist committee members), BAS treatment of BAD is started without a diagnostic test being performed (trial of treatment). The estimated time taken to ascertain the effectiveness of trial of treatment was between 4 and 12 weeks (clinical opinion of specialist committee members).

Search strategy

Search strategies used in the original report were updated in line with the NICE final scope. 11 Search strategies were based on target condition and intervention, as recommended in the CRD guidance for undertaking reviews in health care and the Cochrane Handbook for Systematic Reviews of Diagnostic Test Accuracy. 15,17,19,20

Searches were undertaken to identify studies of SeHCAT in the diagnosis of BAD. The search strategies combined relevant search terms comprising indexed keywords [e.g. medical subject headings (MeSH) and Emtree] and free-text terms; strategies were developed specifically for each database and the keywords adapted according to the configuration of each database. Only studies conducted with humans were sought. Searches were not limited by language or publication status (unpublished or published). The original 2011 strategies were adapted to incorporate changes to the preferred terminology and search methods for each resource. Owing to the time elapsed, some resources were no longer available, but additional resources were searched to maintain completeness.

Searches for studies on economic evaluations, costs and health-related quality of life (HRQoL) were also conducted (see Chapter 4, Identifying and reviewing published cost-effectiveness studies, for further details). To ensure that no relevant studies were missed, the results of the clinical effectiveness searches were also screened for records relevant to the cost-effectiveness evaluation, and all cost-effectiveness results, including guideline searches, were screened for studies relevant to the clinical effectiveness section.

The following databases were searched for relevant studies from inception:

• MEDLINE (Ovid) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, from 1946 to 30 November 2020

• EMBASE (Ovid), from 1974 to 25 November 2020

• Cochrane Database of Systematic Reviews (CDSR) (Wiley), up to November 2020, issue 11

• Cochrane Central Register of Controlled Trials (CENTRAL) (Wiley), up to November 2020, issue 11

• Database of Abstracts of Reviews of Effects (DARE) (CRD), up to March 2015

• Health Technology Assessment (HTA) database (CRD), up to March 2018

• Science Citation Index (SCI) (Web of Science), up to 27 November 2020

• KSR Evidence (https://ksrevidence.com/), up to 1 December 2020

• Latin American and Caribbean Health Sciences Literature (LILACS) (https://lilacs.bvsalud.org/en/), up to 27 November 2020

• National Institute for Health and Care Research (NIHR) HTA programme, up to 26 November 2020

• PROSPERO (international prospective register of systematic reviews) (www.crd.york.ac.uk/prospero/), up to 26 November 2020.

Completed and ongoing trials were identified by searches of the following resources:

• National Institutes of Health Clinical Trials database (www.clinicaltrials.gov/), up to 26 November 2020

• World Health Organization (WHO) International Clinical Trials Registry Platform (www.who.int/ictrp/en/), up to 2 December 2020

• EU Clinical Trials Register (www.clinicaltrialsregister.eu/ctr-search/), up to 2 December 2020.

Conference abstracts and proceedings were identified in a three-stage approach, conducted as follows:

• The main Ovid EMBASE search strategy was employed to include conference abstracts and proceedings.

• A second set of tailored searches was conducted on:

  • Northern Light Life Sciences Conference Abstracts (Ovid), from 2010 to December 2020 week 46

  • Conference Proceedings Citation Index – Science (CPCI-S) (Web of Science), from 1990 to 30 November 2020

• In addition, the 2020 United European Gastroenterology Week proceedings (not currently covered by EMBASE, Northern Light or CPCI-S) were searched manually.

Inclusion and exclusion criteria

Quality assessment

The methodological quality of included diagnostic accuracy studies was assessed using the quality assessment of diagnostic accuracy studies-2 (QUADAS-2) tool. 23 The methodological quality of observational studies, which reported treatment outcome only for those participants with a positive SeHCAT result, was assessed using a topic-specific adaptation of the quality assessment checklist by Wedlake et al. ,24 as used in our previous DAR,18 conducted to support the development of DG7;2 the use of this tool was carried forward to the current assessment to provide consistency. The results of the quality assessment are used for descriptive purposes to provide an evaluation of the overall quality of the included studies and to inform recommendations for the design of future studies. Quality assessment was undertaken by one reviewer and checked by a second reviewer (MW and ER); any disagreements were resolved by consensus or discussion with a third reviewer (NA).

The results of the quality assessments are summarised and presented in tables (see Study quality) and, for QUADAS-2 assessments, are presented in full, by study, in Appendix 3.

Methods of analysis/synthesis

Meta-analysis was considered inappropriate, owing to the small number of test accuracy studies, with varying diagnostic thresholds, and between-study heterogeneity with respect to population (prior investigations), treatment regimen, definition of response, follow-up period and SeHCAT administration; therefore, we employed a narrative synthesis. The clinical effectiveness results section of this report is structured by clinical application (diagnosis of primary BAD and diagnosis of secondary BAD in people with Crohn’s disease who have not undergone ileal resection). A detailed commentary on the major methodological problems or biases that affected the studies is also provided, together with a description of how this may have affected the individual study results.

For predictive accuracy studies (studies that reported sufficient data to support the calculation of the sensitivity and specificity of SeHCAT to predict response to treatment with BASs), the absolute numbers of true positive, false negative, false positive and true negative test results of SeHCAT, compared with the reference standard of treatment response, as well as sensitivity and specificity values, with 95% confidence intervals (CIs), are presented in Table 4. The results of individual studies were plotted in the receiver operating characteristic plane, with the diagnostic threshold used for the SeHCAT test indicated (see Figure 2).

The results of studies that reported treatment outcome only for those participants with a positive SeHCAT result (i.e. sufficient data to calculate PPV only) are presented in Table 5.

Overview of included studies

Based on the updated searches and inclusion screening described previously, and information taken from the assessment report conducted for DG7,18 a total of 24 studies,7,25–35,37–48 reported in 25 publications,7,25–48 were included in this review; the results section of this report cites studies using the primary publication only.

Fifteen of the included studies were published, in full, in peer-reviewed journals;7,30,34,35,37–45,47,48 eight were published as conference abstracts only;25–29,31–33 and one was an unpublished dissertation. 46 It should be noted that all eight studies that were published as conference abstracts only were new studies, identified during this assessment, that is the majority of the new evidence identified (eight out of nine studies) was not published, in full, in peer-reviewed journals.

No RCTs, CCTs or observational comparative studies were identified that met the inclusion criteria for this review (see Inclusion and exclusion criteria). Similarly, no observational studies were identified that reported the results of multivariable regression modelling with response to treatment with BAS as the dependent variable and index test (SeHCAT) result (continuous or categorical) as one of the independent variables. Finally, no new predictive accuracy studies (studies that reported sufficient data to support the calculation of the sensitivity and specificity of SeHCAT to predict response to treatment with BAS) were identified. All of the nine new studies included in this review25–33 were of the lowest level of evidence eligible for inclusion; these are observational studies that report some outcome data for patients treated with BASs, whereby only those patients with a positive SeHCAT test were offered treatment with BAS.

All 24 included studies provided some data about population 1: adults presenting with chronic diarrhoea with unknown cause, or FD, or suspected or diagnosed IBS-D (i.e. people with suspected primary BAD). Three of these studies,40,43,44 all of which were previously included in the assessment report conducted for DG7,18 provided limited predictive accuracy data for this population. The remaining 21 studies reported only information about the outcome of treatment with BASs for some or all of those participants who had a positive SeHCAT test result. 7,25–35,37–39,41–43,46–48

One study7 also provided data on population 2: adults presenting with chronic diarrhoea and a diagnosis of Crohn’s disease who have not undergone ileal resection (i.e. people with suspected secondary BAD). This study reported only information about the outcome of treatment with BASs for people with Crohn’s disease who had a positive SeHCAT test result, and was previously included in the assessment report conducted for DG7. 18 No new studies meeting the inclusion criteria for population 2 were identified for this assessment report.

All 21 studies for which information on geographic location was reported were conducted in Europe; 10 were conducted in the UK,7,27–30,40,45,46,48 five in Italy,25,33,39,43,44 three in Spain,37,38,41 two in Denmark34,47 and one each in Sweden42 and France. 35

Only three of the included studies provided any information about funding, and only one UK study40 reported receipt of any industry funding [SeHCAT test supplies were provided by Amersham International Ltd (Amersham, UK), which is now part of GE Healthcare]; details of all reported funding sources are provided in Table 1.

Further details of the characteristics of study participants and the technical details of the conduct of the index test (SeHCAT) and reference standard (BAS treatment regimen) are provided in Appendix 2.

Study quality

The three studies40,43,44 that provided predictive accuracy data (information on the ability of the SeHCAT test to predict response to treatment with BAS), all of which were previously included in the assessment report conducted for DG7,18 were assessed using the QUADAS-2 tool.

The included predictive accuracy studies were all published > 30 years ago and were generally poorly reported; all three studies were rated as having an ‘unclear’ risk of bias with respect to patient selection and reference standard (no study provided details of whether or not the assessment of response to treatment was conducted blind to the results of SeHCAT testing), and two of the three studies43,44 were also rated as having an ‘unclear’ risk of bias with respect to flow and timing because the duration of follow-up over which response to treatment was assessed was not reported. Merrick et al. 40 was rated as having a ‘high’ risk of bias for the ‘flow and timing’ domain of the QUADAS-2 tool because only patients with positive or equivocal SeHCAT test results received the reference standard (treatment with BAS); patients with a negative SeHCAT test result were managed with unspecified ‘simple conservative treatment’. Sciarretta et al. 43 was rated as having a ‘high’ risk of bias for the ‘index test’ domain of the QUADAS-2 tool because the threshold used to define a positive SeHCAT test result was not prespecified.

All three studies had at least one item of ‘high’ concern regarding applicability to this assessment. In some instances, the applicability issues identified are a consequence of the age of the studies. All three studies were rated as having ‘high’ or ‘unclear’ concerns regarding the applicability of the study population to that specified in the inclusion criteria for this review; all three studies included some participants with prior cholecystectomy and no study reported previous investigations equivalent to those specified in current BSG guidelines for the investigation of chronic diarrhoea. 3 All three studies were also rated as having ‘high’ concerns regarding the applicability of the index test; the age of the studies meant that no study used the current version of the SeHCAT test, manufactured by GE Healthcare, specified in the inclusion criteria for this assessment. Merrick et al. 40 was also rated as having ‘high’ concerns regarding the applicability of the reference standard, because the management of patients with a negative SeHCAT test was not considered likely to provide a reliable indication of whether or not these patients would have responded to treatment with BAS.

The results of the QUADAS-2 assessment are summarised in Table 2 and the full assessments are provided in Appendix 3.

The methodological quality of studies that reported treatment outcome only for those participants with a positive SeHCAT result (i.e. sufficient data to calculate PPV only) was assessed using a topic-specific adaptation of the quality assessment checklist by Wedlake et al. ,24 as used in our previous DAR. 18 The results of this assessment are summarised in Table 3. These studies represent the lowest level of evidence specified in the inclusion criteria for this assessment (see Inclusion and exclusion criteria) and were generally of poor methodological quality. No study in this group provided full outcome data for patients with a negative SeHCAT test result. Ten7,27,29,30,33,34,45–48 of the 21 studies7,25–35,37–39,41,42,45–48 of this type used a retrospective study design. Eleven studies provided no clear definition of chronic diarrhoea. 7,25,27–29,31–34,41,46 Ten studies did not provide sufficient information about the SeHCAT test used to allow the testing procedure to be reproduced. 25–29,31–33,45,47 Eight studies did not clearly describe how the decision to treat patients with BASs was made. 27,29,33–35,42,46,48 Nine studies provided no or an incomplete description of the BAS treatment provided to patients with a positive SeHCAT test result. 27–29,33–35,41,46,48 Finally, six studies did not report an objective measure of response to treatment. 25,27,29,30,33,46

Performance of the SeHCAT test for predicting response to treatment with bile acid sequestrant among patients with diarrhoea-predominant irritable bowel syndrome or functional diarrhoea

All 24 included studies provided some data about population 1: adults presenting with chronic diarrhoea with unknown cause, suspected or diagnosed IBS-D, or FD (i.e. people with suspected primary BAD). 7,25–35,37–48

Three of these studies,40,43,44 all of which were previously included in the assessment report conducted for DG7,18 provided limited data on the accuracy of the SeHCAT test for predicting response to treatment with BAS in this population. The results of these studies are summarised in Table 4. All three studies assessed the relationship between the SeHCAT test result and response to treatment with colestyramine.

Merrick et al. 40 reported sufficient data to allow the calculation of the performance of SeHCAT, for predicting treatment response, at two 7-day retention thresholds (< 8% and ≤ 15%). The estimated sensitivity of SeHCAT in predicting a positive response to treatment with colestyramine was 66.7% (95% CI 22.3% to 95.7%) using the < 8% threshold, and 100% (95% CI 54.1% to 100%) using the ≤ 15% threshold. The specificity estimates were 97.1% (95% CI 84.7% to 99.9%) and 91.2% (95% CI 76.3% to 98.1%) using the < 8% and ≤ 15% thresholds, respectively. 40 These results would appear to indicate that the use of the SeHCAT test with a threshold for 7-day retention of ≤ 15% (commonly used in UK clinical practice) could identify patients with IBS-D who may benefit from treatment with BASs. However, it should be noted that, although all 31 patients with a negative SeHCAT test result were classified as true negatives, this assessment was based on long-term follow-up: ‘None of the 31 patients with irritable bowel disease who retained more than 15% at seven days showed any evidence of small bowel disease, and none appeared during a follow up of at least 12, and in some up to 24 months. Simple conservative treatment resolved or eased most symptoms.’40 None of these 31 patients received treatment with colestyramine; therefore, it remains uncertain whether or not any of these patients could have benefited from treatment with BAS. One patient with a SeHCAT test result of < 8% and two with an equivocal result (8–15%) did not receive treatment with colestyramine; these patients were excluded from the analysis. 40 The remaining nine patients were treated with colestyramine; five of these had a SeHCAT test result of < 8%, one of whom did not respond to treatment, and four had an equivocal result (8–15%), two of whom responded to colestyramine and two of whom did not. 40

Sciarretta et al. 43 reported sufficient data to allow the calculation of the performance of SeHCAT, for predicting treatment response, at a threshold reported to be equivalent to a 7-day retention threshold of < 5%. The estimated sensitivity of SeHCAT in predicting a positive response to treatment with colestyramine was 85.7% (95% CI 42.1% to 99.6%) and the specificity was 100% (95% CI 54.1% to 100%). However, only 13 patients were included in this analysis. A subsequent study by Sciarretta et al. 44 estimated the sensitivity of SeHCAT in predicting a positive response to colestyramine as 95.0% (95% CI 75.1% to 99.9%) and the specificity as 96.2% (95% CI 80.4% to 99.9%), using a 7-day retention threshold of < 8% to define a positive SeHCAT test. It should be noted that there may have been overlap between the populations included in these two studies.

Figure 2 illustrates the variation in sensitivity and specificity with SeHCAT threshold, as reported in these three studies. 40,43,44

FIGURE 2.

Accuracy of the SeHCAT test to predict a response to treatment with colestyramine at different thresholds among patients with IBS-D. The centre dots represent the point estimates for sensitivity and specificity of SeHCAT in predicting response to treatment in the three studies at different cut-off values (5%, 8% and 15%). The vertical and horizontal lines represent the 95% CIs for sensitivity and specificity, respectively.

The between-study heterogeneity in these three studies was considerable. The principal diagnosis, method of SeHCAT administration, BAS treatment dose, definition of response to treatment and follow-up period were different between studies. Appendix 2 provides full details of study inclusion and exclusion criteria, participant characteristics, SeHCAT test methods, BAS treatment and definition of treatment response.

The remaining 21 studies7,25–35,37–39,41–43,46–48 reported information about the outcome of treatment with BAS for some or all of those participants who had a positive SeHCAT result only (i.e. sufficient information to estimate PPV), or other descriptive results.

As was the case for the predictive accuracy studies described previously, between-study heterogeneity for these studies was considerable. The principal diagnosis, threshold used to define a positive SeHCAT test, BAS treatment regimen, definition of response to treatment and follow-up period varied between studies. Appendix 2 provides full details of study inclusion and exclusion criteria, participant characteristics, SeHCAT test methods, BAS treatment and definition of treatment response, for the studies that reported this information.

Performance of the SeHCAT test for predicting response to treatment with bile acid sequestrant among patients with Crohn’s disease, who have not undergone ileal resection

One study7 (results are summarised in Table 6) provided data on population 2: adults presenting with chronic diarrhoea and a diagnosis of Crohn’s disease, who have not undergone ileal resection (i.e. people with suspected secondary BAD). This study reported only information about the outcome of treatment with BAS for people who had a positive SeHCAT result, and was included in our previous assessment report, conducted for DG7. 18 No new studies meeting the inclusion criteria for population 2 were identified for this assessment report. The single study that reported information about response to treatment with BAS among patients with Crohn’s disease provided only very limited information about response rates among patients with a positive SeHCAT test result (7-day retention value of < 10%) who were treated with colestyramine or colestipol. 7 Fewer than half (9/24) of the patients with a positive SeHCAT test result received treatment with BAS; the criteria used to decide whether or not to offer BAS were not reported. Most [8/9 (89%)] of the patients treated with BAS responded positively;7 however, the numbers treated with each BAS (colestyramine or colestipol) were not reported.

Appendix 2 provides all reported details of the inclusion and exclusion criteria, participant characteristics, SeHCAT test methods, BAS treatment and definition of treatment response, for this study. 7

Pooled estimates of treatment response rates for inclusion in cost-effectiveness modelling

Meta-analysis of test accuracy estimates (i.e. sensitivity and specificity) was considered inappropriate in this assessment, owing to the small number of test accuracy studies, with varying diagnostic thresholds, and between-study heterogeneity with respect to population (prior investigations), treatment regimen, definition of response, follow-up period and SeHCAT administration. However, to provide input parameters for cost-effectiveness modelling, some pooled estimates were calculated using the inverse-variance method on the logit scale, for the probability of testing positive at the 15% threshold (Table 7) and the probability of achieving a positive response to treatment, given a positive test at the 15% threshold (Table 8). The random-effects analysis was chosen because of the high heterogeneity, qualitatively assessed, in accordance with the Cochrane Handbook, section 10.10.4.1. 20

Model structure

Conceptual model description

The structure of the health economic model is in line with that developed for the previous assessment of SeHCAT. 18 Thus, the model consists of two parts:

1. a short-term decision-analytic model reflecting the diagnostic pathway and initial response to treatment (assumed to be the first 6 months)

2. a long-term (Markov) model that estimates the lifetime costs and effects for patients receiving subsequent treatment.

An outline of the short-term model structure for the population of adults with suspected primary BAD (population 1) is presented in Figure 3. The main difference with respect to the model developed for the previous assessment of SeHCAT18 is the potential inclusion of the colonoscopy investigation in the model, based on discussions during the scoping phase suggesting that SeHCAT could be used to avoid unnecessary colonoscopies. Thus, our base-case scenario for population 1 places colonoscopy after SeHCAT testing, in accordance with most clearly expressed clinical expert opinion and BSG guidelines whereby colonoscopy is required for investigation of cancer and not for ruling out IBD. As a secondary scenario for population 1, we assumed that no colonoscopy would occur after SeHCAT testing, as this would have already occurred in the clinical pathway. Note that, in practice, colonoscopy can be excluded from the model by setting this probability equal to zero (i.e. at the colonoscopy branch all patients will follow the ‘no colonoscopy’ path and, subsequently, will be treated as IBS-D patients).

FIGURE 3.

Decision-analytic model, population 1. TOT, trial of treatment; Tx, treatment.

In the SeHCAT strategy, patients may have a positive or a negative test result. If the test is positive (i.e. the percentage of whole-body retention of bile acids is below a certain cut-off point), patients are treated with BASs and they may or may not respond to that treatment. Patients with a positive SeHCAT result and an initial response to BASs are at risk of treatment discontinuation because of BAS intolerance. In this case, patients do not go through further testing, because, given the positive SeHCAT result, it is assumed that these patients will be treated as having BAD. If the result of the SeHCAT test is negative, a proportion of patients are investigated for IBD with a colonoscopy. If, after the colonoscopy, patients are diagnosed as having IBD, then they are treated accordingly. Otherwise, patients are treated as having IBS-D. Patients testing SeHCAT negative and not undergoing colonoscopy are diagnosed as having IBS-D and are treated accordingly. All end points of the SeHCAT-negative branch are thus determined depending on whether or not patients respond to IBS-D or IBD treatment. The no-SeHCAT strategy assumes that all patients follow the same paths as for the SeHCAT-negative test. Thus, patients may be investigated for IBD with a colonoscopy, may be treated for IBD or IBS-D and may or may not respond to treatment. Finally, in the trial-of-treatment strategy, all patients receive BASs at the beginning. If patients do not respond to BASs, they follow the same paths as for the SeHCAT-negative and the no-SeHCAT strategies. Patients with an initial response to BASs are also at risk of treatment discontinuation, as in the SeHCAT-positive branch of the model. Treatment discontinuation may vary between patients with a positive SeHCAT result and those not tested.

The short-term model for population 2 (adults with chronic diarrhoea and a diagnosis of Crohn’s disease, who have not undergone ileal resection) is shown in Figure 4. The main difference with respect to the short-term model in Figure 4 is that Crohn’s disease patients are not expected to undergo colonoscopy, because it is assumed that these patients would already have had colonoscopy to diagnose their Crohn’s disease. Therefore, all end points of the decision-analytic model are determined depending on whether or not patients respond to treatment (BAS or diarrhoea treatments for Crohn’s disease patients). This is the same structure as assumed in the previous assessment of SeHCAT. 18

FIGURE 4.

Decision-analytic model, population 2. TOT, trial of treatment; Tx, treatment; w/o, without.

To assess the long-term costs and effects of the various strategies across both populations, patients are assumed to enter a simple three-state Markov model, as shown in Figure 5.

FIGURE 5.

Markov model, populations 1 and 2.

Patients who had a treatment response in the short-term model start in the ‘no-diarrhoea’ health state and patients who did not respond to treatment in the short-term model start in the ‘diarrhoea’ health state. Because the model has a lifetime time horizon, the third state included is ‘death’. In the previous assessment of SeHCAT, no link with increased mortality was found. 56 Therefore, because there is no new evidence to suggest that this has changed, only background mortality was considered in the economic model. Transitions between the ‘diarrhoea’ and ‘no-diarrhoea’ health states were informed by clinical expert opinion, as clinical data regarding the long-term effectiveness of BASs and IBD and IBS-D treatments were not identified. The cycle length is 6 months, as in the previous assessment of SeHCAT. 18 In consultation with clinical experts, it was agreed that a period of 6 months, in general, would be sufficient to capture initial response to treatment (i.e. in the decision-analytic model). Six months was also deemed as an appropriate cycle length for the Markov model as this would represent a plausible time to reassess response to treatment in the long term. Long-term adverse events, such as constipation, and treatment discontinuation were not included in the Markov model owing to lack of data. The Markov model is then parameterised according to treatment.

Model parameters and implementation

This section describes the parameters used in the decision-analytic and the Markov models and how their values were estimated. When possible, input for the models was based on our systematic review (described in Chapter 3), other published literature and UK databases. When such evidence was not available, expert opinion was used. We sent out a questionnaire to the specialist committee members of this assessment and their answers were used to inform the input parameters for which data were lacking. In the absence of better evidence, it was assumed that the experts’ responses would adequately reflect the uncertainty in the elicited evidence, and the case mix observed by each expert would reflect the relevant patient population in UK clinical practice. The evidence elicitation process did not follow any specific guidance or methodology. It was considered that this approach would better reflect the current uncertainty than, for example, seeking consensus answers. The same approach was followed in the previous assessment of SeHCAT. 18 However, given the limited number of answers obtained, translating the experts’ responses into model parameters was often challenging. How inconsistent evidence was dealt with is described throughout the report, mostly as footnotes in tables, but also in the main text. When experts were unable to provide estimates, modelling assumptions were made. The model was implemented in R (the most recent version used was 4.1.0) using RStudio (the most recent version used was 2021.09.1 Build 372) (The R Foundation for Statistical Computing, Vienna, Austria). At the time of writing this report, the model is not publicly available, but it can be requested through NICE.

Summary input parameters

The input parameters used in the health economic models for populations 1 and 2 are summarised in Tables 35 and 36, respectively.

Overview of main model assumptions

Secondary analysis

The base-case scenario was built under the assumption that colonoscopy was not part of the clinical pathway before patients enter the model. This analysis is, however, likely to deviate from current guidelines. As explained previously, this was nevertheless chosen as the base-case analysis following scoping discussions. Given its importance, a scenario in which colonoscopy is not included in the cost-effectiveness model (e.g. assumed to occur before SeHCAT testing or trial of treatment with BAS) was presented separately.

Scenario analyses

A series of scenario analyses were conducted to explore the most important areas of uncertainty in the model described above. Summaries of the scenario analyses conducted are given in the following sections.

Results of the base-case analysis, population 1

The results of the base-case analysis for population 1 are shown in Table 43. It can be seen that the BAS trial-of-treatment strategy was dominated by the SeHCAT 15% strategy (the latter provided more QALYs at lower costs). Therefore, the relevant comparison for the ICER calculation was SeHCAT 15% versus no SeHCAT, for which the ICER was £9688, which is below the commonly used threshold of £20,000 per QALY gained. The SeHCAT 15% strategy was estimated to provide 0.23 additional QALYs at an incremental cost of £2236, compared with the no-SeHCAT strategy. The base-case analysis also revealed that, in the short term, the SeHCAT 15% strategy is the one with more colonoscopies avoided per patient (65%) and with the highest response rate for any type of medication (68%), but that these come at the highest initial costs (£786), because of the costs of the SeHCAT test. The cost per colonoscopy avoided was the lowest for the SeHCAT 15% strategy (£786 ÷ 0.65 = £1209) and the cost per response was the lowest for the BAS trial-of-treatment strategy (£507 ÷ 0.65 = £780). Life-years were 19.96 for all three strategies. Because no difference in mortality across strategies was assumed in the model, the same life-years were expected to be estimated for the three strategies.

Results of the secondary analysis, population 1

As explained previously, the secondary analysis was based on the assumption that patients had undergone colonoscopy before entering the model. Thus, in practice, this scenario was run by removing all colonoscopy branches from Figure 3, that is by setting the probability of colonoscopy equal to 0.

The results of the secondary analysis for population 1 are summarised in Table 44. The SeHCAT 15% strategy was estimated to provide the highest QALYs at the highest costs, but, unlike the base-case analysis, the BAS trial-of-treatment strategy was no longer dominated by the SeHCAT 15% strategy. We observed that the ICER of BAS trial of treatment versus no SeHCAT and the ICER of SeHCAT 15% versus BAS trial of treatment are close to (but <) the £20,000 per QALY threshold. The secondary analysis also showed that, in the short term, the SeHCAT 15% strategy was the one with the highest response rate for any type of medication (67%), but that this comes at the highest initial costs (£553), because of the costs of the SeHCAT test. The cost per response was the highest for the SeHCAT 15% strategy (£553 ÷ 0.67 = £825). For the other two strategies, this cost was nearly the same: £134 for no SeHCAT and £135 for BAS trial of treatment. Life-years (not shown in Table 44) were also 19.96 for all strategies, as in the base-case analysis, as no changes in mortality were assumed.

Results of the probabilistic sensitivity analyses, population 1

The base-case PSA cost-effectiveness results can be seen in Table 45. These aligned well with the deterministic results; the BAS trial-of-treatment strategy was dominated by the SeHCAT 15% strategy and the ICER of SeHCAT 15% versus no SeHCAT was £9661.

The cost-effectiveness plane and cost-effectiveness acceptability curves (CEACs) resulting from the PSA are shown in Figures 6 and 7, respectively. Note that the cost-effectiveness plane shows the results of pairwise comparisons versus the no-SeHCAT strategy only. It can be seen that the vast majority of the simulations are in the eastern quadrants, in which both BAS trial of treatment and SeHCAT 15% are more effective than no SeHCAT. Most of the simulations are in the north-eastern quadrant of the cost-effectiveness plane, where BAS trial of treatment and SeHCAT 15% are also more costly than no SeHCAT. The CEACs show that, at lower values of the threshold ICER, no SeHCAT is the strategy with the largest probability of being cost-effective, given that its costs are the lowest. However, this probability rapidly decreases as the threshold ICER increases, and SeHCAT 15% becomes the strategy most likely to be deemed as cost-effective, with a 67% probability of being cost-effective at a threshold ICER of £20,000 per QALY gained, and a 73% probability at a threshold ICER of £30,000 per QALY gained.

FIGURE 6.

The cost-effectiveness plane from the PSA base-case results, population 1.

FIGURE 7.

The CEACs from the PSA base-case results, population 1.

A PSA was also run under the assumptions of the secondary analysis, whereby no colonoscopy was included in the model. The PSA results can be seen in Table 46. In this case, they are also in line with the results of the deterministic analysis. The SeHCAT 15% strategy was estimated to provide the most QALYs at the highest costs, but no strategy was dominated. The ICER of BAS trial of treatment versus no SeHCAT and the ICER of SeHCAT 15% versus BAS trial of treatment were close to the £20,000 per QALY threshold, but, unlike the deterministic analysis, the ICER of SeHCAT 15% versus BAS trial of treatment was now above this threshold.

The cost-effectiveness plane and CEACs resulting from the PSA in the secondary analysis are shown in Figures 8 and 9, respectively. The cost-effectiveness plane shows, again, the results of pairwise comparisons versus no SeHCAT only. It can be seen that the vast majority of the simulations are in the north-eastern quadrant of the cost-effectiveness plane, in which both BAS trial of treatment and SeHCAT 15% are more effective and more costly than no SeHCAT. The difference in uncertainty between the two comparisons is remarkable, especially on the costs side. However, this can be explained by the distribution of costs, which is notably wider in the SeHCAT strategy. The CEACs show that, at lower values of the threshold ICER, no SeHCAT is the strategy with the largest probability of being cost-effective. This probability decreases as the threshold ICER increases; at a threshold ICER of approximately £20,000 per QALY gained, SeHCAT 15% and no SeHCAT have nearly the same probability of being cost-effective. At larger values of the threshold ICER, SeHCAT 15% is the strategy most likely to be considered cost-effective. In particular, at a threshold ICER of £30,000 per QALY gained, this probability is approximately 50%.

FIGURE 8.

The cost-effectiveness plane from PSA secondary analysis results (no colonoscopy), population 1.

FIGURE 9.

The CEACs from PSA secondary analysis results (no colonoscopy), population 1.

Results of the scenario analyses, population 1

Results of the base-case analysis, population 2

The results of the base-case analysis for population 2 are shown in Table 50, where it can be seen that the no-SeHCAT strategy was dominated by the BAS trial-of-treatment strategy. Therefore, the relevant comparison for the ICER calculation is SeHCAT 15% versus BAS trial of treatment, for which the ICER was £1127. The SeHCAT 15% strategy was estimated to provide 0.1071 additional QALYs at an incremental cost of £185, compared with BAS trial of treatment. The base-case analysis also indicated that, in the short term, the SeHCAT 15% strategy is the one with the highest response rate for any type of medication (71%), but that this comes at the highest initial costs (£1061), owing to the inclusion of the SeHCAT test. The cost per response is the lowest for the BAS trial-of-treatment strategy (£756 ÷ 0.6 = £1260). Life-years (not shown in Table 50) were 18.696 for all strategies. As no difference in mortality across strategies was assumed in the model, the same life-years were expected to be estimated for the three strategies.

Results of the probabilistic sensitivity analyses, population 2

The base-case PSA results for population 2 can be seen in Table 51. These aligned well with the deterministic results, but now the SeHCAT 15% strategy is dominant because the lowest total costs are estimated for this strategy. In general, PSA costs are higher than the deterministic ones. This can be explained by the skewness of the triangular distributions chosen to parameterise the cost inputs of the model.

The cost-effectiveness plane and CEACs resulting from the PSA are shown in Figures 10 and 11, respectively. Note the cost-effectiveness plane shows the results of pairwise comparisons versus no SeHCAT. It can be seen that all simulations (except one) are in the eastern quadrants, in which both BAS trial of treatment and SeHCAT 15% are more cost-effective than no SeHCAT. Approximately half of the simulations are in the south-eastern quadrant of the cost-effectiveness plane, where BAS trial of treatment and SeHCAT 15% are dominant, compared with no SeHCAT. The CEACs show that, for any positive value of the threshold ICER, SeHCAT 15% is the strategy with the largest probability of being cost-effective. In particular, at a threshold ICER of £20,000 per QALY gained, the estimated probability of being cost-effective is 89%; at a threshold ICER of £30,000 per QALY gained, it is 92%.

FIGURE 10.

Cost-effectiveness plane from base-case PSA results, population 2.

FIGURE 11.

The CEACs from the base-case PSA results, population 2.

Results of the scenario analyses, population 2

Clinical effectiveness

The evidence base relating to the use of SeHCAT testing among adults presenting with chronic diarrhoea with unknown cause (FD), or suspected or diagnosed IBS-D (i.e. people with suspected primary BAD), and among adults presenting with chronic diarrhoea and a diagnosis of Crohn’s disease, who have not undergone ileal resection (i.e. people with suspected secondary BAD) has not changed substantively since our previous systematic review,18 which was conducted to support the development of DG7. 2 The search yield increased considerably in this update assessment; searches of bibliographic databases (from inception to November 2020) identified a total of 5518 unique references, after deduplication against the EndNote library from our previous systematic review,18 compared with the total of 4240 unique references identified for the period from inception to April 2012 covered by the searches conducted for our previous systematic review. 18 However, despite the large number of records retrieved, only nine new studies were identified25–33 that met the inclusion criteria for this assessment. Most (eight out of nine) of these studies were published as conference abstracts only. 25–29,31–33 In addition, six publications,50–54,71 which were included in our previous systematic review,18 did not meet the inclusion criteria for this assessment. This current assessment includes a total of 25 publications relating to 24 studies, compared with the 24 publications relating to 21 studies included in our previous systematic review. 18

No RCTs, CCTs or observational comparative studies that met the inclusion criteria for this assessment (see Chapter 3, Inclusion and exclusion criteria) were identified. Similarly, no multivariable regression models were identified with response to treatment with BAS as the dependent variable and index test (SeHCAT) result (continuous or categorical) considered as one of the independent variables. Finally, no new predictive accuracy studies (studies that reported sufficient data to support the calculation of the sensitivity and specificity of the SeHCAT test to predict response to treatment with BAS) were identified. All of the nine new studies included in this review25–33 were of the lowest level of evidence eligible for inclusion; these were observational studies that reported some outcome data for patients treated with BAS, whereby only those patients with a positive SeHCAT test were offered treatment with BAS.

All 247,25–35,37–48 of the studies included in this assessment provided some data for population 1 and one study7 also provided data on population 2.

Three studies,40,43,44 all of which were included in our previous assessment report,18 conducted for DG7,7 provided limited data on the accuracy of the SeHCAT test for predicting response to treatment with BAS for population 1. Merrick et al. 40 reported sufficient data to allow the calculation of the performance of the SeHCAT test for predicting treatment response at two 7-day retention thresholds (< 8% and ≤ 15%). The estimated sensitivity of the SeHCAT test in predicting a positive response to treatment with colestyramine, using the < 15% threshold (commonly used in UK clinical practice),4,72 was 100% (95% CI 54.1% to 100%) and the corresponding specificity estimate was 91.2% (95% CI 76.3% to 98.1%). 40 These results would appear to indicate that the use of the SeHCAT test, with a 15% threshold, could identify patients with IBS-D who may benefit from treatment with BAS. However, it should be noted that the CIs around the sensitivity estimate were wide and, although all 31 patients with a negative SeHCAT test result were classified as true negatives, this assessment was based on long-term follow-up; none of the patients with a negative SeHCAT test result, at the > 15% retention threshold, exhibited evidence of small bowel disease and none of these patients developed evidence of small bowel disease during follow-up. Simple conservative treatment improved or resolved most symptoms. 40 None of these 31 patients received treatment with colestyramine; therefore, it remains uncertain whether or not any of these patients could have benefited from treatment with BAS. The remaining two studies43,44 provided data for 7-day SeHCAT retention thresholds of 5% and 8%.

Eight studies reported information about the rate of positive response to treatment with BAS, using a threshold of < 15% or ≤ 15%, for population 1. 27,29,34,42,45–48 The median proportion of patients with a positive SeHCAT test result who received treatment with BAS was 86% (range 70–100%) and the median response rate was 68% (range 38–86%). The equivalent data from the predictive accuracy study by Merrick et al. 40 indicated a treatment response rate of 67% among patients with 7-day SeHCAT retention values of ≤ 15%; in this study, 9 out of 12 (75%) patients with SeHCAT retention values of ≤ 15% received treatment with colestyramine. Eleven studies reported information about the rate of positive response to treatment with BAS, using a threshold of < 10% or ≤ 10%. 7,26,31,32,34,35,37,38,41,42,48 The median proportion of patients with a positive SeHCAT test result who received treatment with BAS was 100% (range 52–100%) and the median response rate was 85% (range 67–100%).

The single study that reported information about response to treatment with BAS for population 2 provided only limited information about response rates among patients with a positive SeHCAT test result (7-day retention value of < 10%) who were treated with colestyramine or colestipol. 7 Only 9 out of 24 patients with a positive SeHCAT test result received treatment with BAS and the numbers receiving each drug were not reported; eight out of nine (89%) patients treated with BAS responded positively. 7

Cost-effectiveness

We have assessed the cost-effectiveness of SeHCAT testing in the two populations described in the previous section. For both populations, the cost-effectiveness of SeHCAT compared with no SeHCAT and compared with trial of treatment with BAS was assessed. For the SeHCAT option, only the strategy based on the 15% cut-off point was included in the cost-effectiveness analyses for both populations. The main reason for this was that, in the clinical expert elicitation exercise to inform parameters for which data are lacking (the majority of parameters included in the model), all clinical experts consulted provided estimates for the 15% cut-off value only.

For each population, the following two models were combined:

1. A short-term decision-analytic model reflecting the diagnostic pathway and initial response to treatment (assumed to be the first 6 months).

2. A long-term (Markov) model that estimates the lifetime costs and effects for patients receiving subsequent treatment. The Markov model is parameterised according to treatment; thus, in practice, there is one Markov model for each type of medication included in the analyses [i.e. BAS (colestyramine and colesevelam), and IBS-D, IBD and diarrhoea medication for Crohn’s disease patients].

The main difference with respect to the model developed for the previous assessment of SeHCAT18 is that, for population 1, our model places colonoscopy after the SeHCAT test, in accordance with most clearly expressed clinical expert opinion and BSG guidelines in which colonoscopy is required for investigation of cancer and not for ruling out IBD. In practice, colonoscopy can be excluded from the model by setting this probability equal to zero. In the decision-analytic model, the number of responders, the expected costs and the number of colonoscopies avoided (when applicable) were calculated for each comparator. In the Markov models, lifetime expected (quality-adjusted) life-years and expected costs per patient were calculated for each comparator.

When possible, input parameters for the model were estimated based on our systematic review, other published literature and UK databases. When such evidence was not available, expert opinion was sought. The impact of parameter uncertainty was explored through PSAs and scenario analyses. ICERs were estimated as additional cost per additional QALY. Other outcomes included in the analyses were short-term costs, response to treatment and, for population 1, the percentage of colonoscopies avoided. These three outcomes were calculated in the decision-analytic model (thus, assumed to be in the first 6 months of the simulation).

For both populations, the SeHCAT 15% strategy dominated the other two strategies or resulted in ICERs below the common thresholds of £20,000 and £30,000 per QALY gained. Dominance or cost-effectiveness was led, in general, by treatment response as the SeHCAT 15% was the strategy with the highest response rate in the majority of the scenarios explored, including the base-case scenario for both populations. In scenarios in which the other two strategies were estimated to provide higher response rates than SeHCAT, the scenarios are likely to be based on unrealistic assumptions regarding response to no SeHCAT or BAS trial of treatment. Even in those scenarios in which overall response in the BAS trial-of-treatment strategy was higher than in the SeHCAT 15% strategy, the ICERs for the comparison of BAS trial of treatment versus SeHCAT 15% were well above the thresholds of £20,000 or £30,000 per QALY gained. SeHCAT 15% was also the strategy in which more colonoscopies were avoided.

Clinical effectiveness

Extensive literature searches were conducted in an attempt to maximise retrieval of relevant studies. These included electronic searches of a variety of bibliographic databases, as well as screening of clinical trials registers and conference abstracts to identify unpublished studies. Because of the known difficulties in identifying test accuracy studies using study design-related search terms,73 no study design filters were used, to maximise sensitivity at the expense of reduced specificity. Thus, large numbers of citations were identified and screened; however, the search yield (proportion of studies identified that met the inclusion criteria for this assessment) was very low.

The possibility of publication bias remains a potential problem for all systematic reviews. Considerations may differ for systematic reviews of test accuracy studies. It is relatively simple to define a positive result for studies of treatment, for example a significant difference between the treatment and control groups that favours treatment. This is not the case for test accuracy studies, which measure agreement between an index test and a reference standard. It would seem likely that studies finding greater agreement (i.e. high estimates of sensitivity and specificity) will be published more often. In addition, test accuracy data are often collected as part of routine clinical practice, or by retrospective review of records; test accuracy studies are not subject to the formal registration procedures applied to RCTs and are therefore more easily discarded when results appear unfavourable. It would seem likely that similar considerations would apply to the type of observational study [studies in which only participants with a positive index test (SeHCAT) result receive the reference standard (treatment with BAS)] that comprises most of the evidence in this assessment. The extent to which publication bias occurs in studies of test accuracy remains unclear; however, simulation studies have indicated that the effect of publication bias on meta-analytic estimates of test accuracy is minimal. 74 Formal assessment of publication bias in systematic reviews of test accuracy studies remains problematic and reliability is limited. 74 We did not undertake a statistical assessment of publication bias in this review. However, our search strategy included a variety of routes to identify unpublished studies and resulted in the inclusion of a number of conference abstracts.

Clear inclusion criteria were specified in the protocol for this review; the review has been registered on PROSPERO (CRD42020223877) and the protocol is available online. 1 The eligibility of studies for inclusion is therefore transparent. In addition, we have provided specific reasons for exclusion for all of the studies that were considered potentially relevant at initial citation screening and were subsequently excluded on assessment of the full publication (see Appendix 4). The review process followed recommended methods to minimise the potential for error and/or bias;15 studies were independently screened for inclusion by two reviewers, and data extraction and quality assessment were undertaken by one reviewer and checked by a second (MW and ER). Any disagreements were resolved by consensus.

There is no accepted reference standard method for diagnosing BAD, although the SeHCAT test is widely used in UK clinical practice. This being the case, the only options for evaluating the clinical effectiveness of the SeHCAT test involve methods of evaluating the impact of the test on clinical outcomes (e.g. RCTs, CCTs or observational comparative studies in which outcomes among patients who received SeHCAT testing are compared with outcomes among those who did not), or study designs (e.g. multivariable prediction modelling or predictive accuracy studies) that assess the ability of the SeHCAT test to predict patients’ response to standard treatments (BAS) for BAD. It can be argued that RCTs, CCTs and multivariable models (in which treatment response is the dependent variable) represent a higher level of evidence than conventional test accuracy studies, in that they provide a direct link between the test and the clinical outcome of interest and can incorporate controls for factors (other than the test) that may affect clinical outcome. A predictive accuracy study, in which response to treatment or a longer-term outcome is treated as the reference standard, also makes a direct link between the test and clinical outcome, but has the weakness that it does not account for factors other than the test that may affect outcome.

Three studies40,43,44 included in the review were classified as predictive accuracy studies (studies that provided data on the sensitivity and specificity of the SeHCAT test for predicting response to treatment with BAS). The methodological quality of these studies was assessed using a modification of the QUADAS-2 tool,23 which is recommended by the Cochrane Collaboration. 17 The QUADAS-2 tool is structured into four key domains covering participant selection, index test, reference standard and the flow of participants through the study (including timing of tests). Each domain is rated for risk of bias (low, high or unclear); the participant selection, index test and reference standard domains are also separately rated for concerns regarding the applicability (low, high or unclear) of the study to the review question. For continuity, the methodological quality of studies that reported treatment outcome only for those participants with a positive SeHCAT result (i.e. sufficient data to calculate PPV only) was assessed using a topic-specific adaptation of the quality assessment checklist by Wedlake et al. ,24 as used in our previous DAR. 18

As was the case for our previous systematic review on this topic,18 the main limitations for this assessment are the paucity of data (only nine new studies were identified for this update assessment), the level of evidence (21/24 of the included studies were of the lowest level of evidence specified in the inclusion criteria) and the generally poor quality of the included studies (see Chapter 3, Study quality). Studies that reported information about the rate of response to treatment with BAS among participants who had a positive SeHCAT test appeared not to be using the SeHCAT test result alone to determine treatment decisions, as not all participants with a positive SeHCAT test received BAS; other reasons for deciding whether or not to offer BAS were not reported. There were substantial differences between studies included in the review (studies were generally poorly reported and there was variation in the SeHCAT test methods and thresholds, BAS treatment regimens and definition of response to treatment). The applicability of the included studies to the review question was unclear; previous investigations were generally poorly reported and not equivalent to those specified in current BSG guidelines for the investigation of chronic diarrhoea. 3

Cost-effectiveness

The main objective of this assessment was to update the previous assessment of SeHCAT conducted in 2012/2013. 18 As mentioned in the previous section, the evidence base relevant for this assessment has not changed substantively since the previous one. Therefore, current strengths and limitations are similar to those discussed in the previous report.

This report presents a full economic evaluation study in two populations of interest: (1) adult patients presenting with chronic diarrhoea with unknown cause (FD), or suspected or diagnosed IBS-D (i.e. people with suspected primary BAD); and (2) adults with chronic diarrhoea and a diagnosis of Crohn’s disease, who have not undergone ileal resection (i.e. people with suspected secondary BAD). Short- and long-term consequences were assessed both in costs and effects of using the SeHCAT test at a 15% cut-off threshold, compared with no SeHCAT and trial of treatment with a BAS (colestyramine or colesevelam). For both patient populations, a linked evidence approach was used to model cost and consequences, to combine outcomes of the SeHCAT test and the related changes in treatment decisions and final health outcomes. Our model and analyses distinguish between the initial diagnostic phase (treatment responder vs. non-responder, and colonoscopies avoided, at 6 months) and the long-term projection of treatment response into final health outcomes (lifetime costs and consequences, the latter expressed in QALYs).

There are several differences with respect to the analyses in the previous assessment of SeHCAT. 18 As mentioned in the previous section, for the SeHCAT option, only the strategy based on the 15% cut-off point was included in the cost-effectiveness analyses because of the lack of data to inform other relevant SeHCAT strategies identified in the literature (e.g. 5% or 10%). Although including just one SeHCAT strategy could be seen as a limitation, it could be argued that it is an advantage because including other SeHCAT strategies for which data are also lacking would only increase the uncertainty around the plausibility of the cost-effectiveness results.

Another important difference in this assessment is that the health economic model extends the model previously developed by placing colonoscopy after the SeHCAT test for population 1. This was assumed in accordance with most clearly expressed clinical expert opinion and BSG guidelines in which colonoscopy is required for the investigation of cancer and not for ruling out IBD. In practice, colonoscopy can be excluded from the model by setting this probability equal to zero; in that case, the updated model can be seen as equivalent to the model in the previous assessment of SeHCAT. 18

The impact of using SeHCAT was included in the analyses in terms of BAS treatment response as reported in peer-reviewed papers. For this purpose, we selected only those papers that fulfilled our quality criteria as presented in Chapter 3. In all models developed, we have used the best available evidence to inform input parameters that were relevant for the UK. When evidence was not available through published studies or databases, we used the most likely and plausible values as reported by clinical experts. For this purpose, we sent out an updated questionnaire in which new questions were used to target the evidence gaps from the previous models. The lack of evidence was handled by performing PSAs and a wide range of scenario analyses. Unlike in the previous assessment of SeHCAT,18 this time it was preferred to have a base-case scenario for each population that was defined based on the assumptions that were deemed more plausible by the modelling team based on the available evidence and clinical experts’ feedback.

In the updated assessment we were able to incorporate patients switching from treatment with colestyramine to colesevelam using clinical experts’ inputs. Unfortunately, it was not possible to translate it into changes in response rates, but at least changes in costs and HRQoL due to treatment-switching were included in the new analyses.

A strength of the HRQoL evidence is that EQ-5D utility values were identified for patients with IBS with and without diarrhoea. However, assumptions had to be made to estimate these utilities among patients with Crohn’s disease and to estimate the utility of patients who respond to treatment with BAS in both populations, all of which represent important limitations in the HRQoL evidence.

Unit costs were retrieved from appropriate sources and were based on the 2020 costing year. Most of the information needed to calculate costs (e.g. medication use, dosage, proportion of patients requiring each type of medication) was based on experts’ opinion. The costs estimated might be considered uncertain, as questions were filled in by a maximum of four experts, and some questions were answered by just one or two. For example, if, for one medication, different dosages were given by different experts, the average of the experts’ answers was used. It is uncertain whether or not these averages would approach the true value, given the small sample size. When the full information of a certain medication was not available from the questionnaire, this medication was excluded from the model, for example vedolizumab for IBD patients. As these costs are expected to be high, the current estimated IBD costs might be underestimated. Similar issues were encountered for infliximab and for psychological treatment in the IBD population. Most notably, when clinical experts were asked about treatment of diarrhoea among Crohn’s disease patients, their answers suggested that this was similar to treatment of BAD with BAS, as colestyramine and colesevelam were mentioned. However, assuming these as treatment of diarrhoea among Crohn’s disease patients would result in no distinction between the no-SeHCAT and the BAS trial-of-treatment strategies. Therefore, diarrhoea treatment for patients with Crohn’s disease was assumed to be the same as in the previous SeHCAT report. 18

One of the main limitations of this study is still that the studies used to estimate the probability of a positive SeHCAT test result and the probability of BAS response, after a positive SeHCAT test, were based on populations other than the ones defined in this evaluation. Most IBS-D studies included patients for whom various tests had been performed and no organic cause of the diarrhoea was found. This is in contrast with the population defined in this assessment, which is patients with symptoms suggestive of functional disease for whom only basic blood tests have been performed. It is therefore likely that, in our population, the prevalence of BAD is lower than the prevalence observed in the published studies.

Another limitation that was already present in the previous assessment of SeHCAT concerns the modelling of non-responders. It is assumed in the model that non-responders would only use loperamide for some symptomatic relief. It might be likely that, for example in the IBS-D population (i.e. patients for whom no diagnostic testing other than initial blood work has been performed), some non-responders will be referred for diagnostic testing to check for organic causes of the chronic diarrhoea.

The most important limitation is still the lack of data on various important parameters of the model. This is most notably the case for patients after testing negative for SeHCAT and for the BAS trial-of-treatment strategy. The necessity to rely on expert opinion was still high because the majority of parameters included in the model were informed by the answers provided to our questionnaire.

Clinical effectiveness

Two systematic reviews, published since the publication of NICE DG7,2 have provided estimates of the prevalence of BAM (as defined by a 7-day SeHCAT retention value of < 10%) among adults with IBS-D (defined by the Rome I, II or III criteria)75 or adults with IBS-D or FD with no organic explanation. 76 The pooled prevalence estimates from these two systematic reviews were 28.1% (95% CI 22.6% to 34.0%), based on data from six studies (n = 908),75 and 30.8% (95% CI 24.7% to 37.7%), based on data from 24 studies (total number of participants unclear). 76 These data support the idea that BAM may be a significant underlying pathology in a substantial proportion of patients with IBS-D or FD and, by extension, that ‘underdiagnosis’ of BAM in this population could result in patients not receiving potentially beneficial treatment with BAS, or experiencing delays in treatment. A web-based survey of 227 UK nuclear medicine departments, published in 2013 shortly after NICE DG7,2 reported that 73 out of 129 (57%) responding centres were using SeHCAT, of which 51 out of 73 (70%) reported an increase in workload over the preceding 3 years. 72 Although this study is approximately 8 years old, and hence cannot be taken as a reliable representation of current service provision, it may be worth noting that responding centres reported a very wide range of annual patient workloads, median 30 studies per year (range 1–300), indicating substantial geographic variation in service provision. 72 A subsequent prospective survey of 38 UK centres providing SeHCAT testing, published in 2016, reported that the total number of SeHCAT tests conducted by participating centres over a 6-month period was 1070;4 this study did not provide a breakdown of test numbers by centre.

Despite the apparent significance of BAM in the adult IBS-D/FD population, and the expansion of provision of SeHCAT testing in the UK, there remains a surprising lack of evidence linking the use of SeHCAT testing to patient-perceived outcomes. As described in Chapter 3, Performance of the SeHCAT test for predicting response to treatment with bile acid sequestrant among patients with diarrhoea-predominant irritable bowel syndrome or functional diarrhoea, and Performance of the SeHCAT test for predicting response to treatment with bile acid sequestrant among patients with Crohn’s disease, who have not undergone ileal resection, the available evidence is largely limited to studies that describe the proportion of patients with a positive SeHCAT test result who respond positively to treatment with BAS. The thresholds used by these studies to define a positive SeHCAT test varied and, although some studies did evaluate multiple thresholds, data were sparse and the optimal SeHCAT decision threshold, to define presence of BAM and select patients for treatment with BAS, remains unclear. For example, two studies reported information about treatment response rates for three 7-day SeHCAT retention thresholds (5%, 10% and 15%). 34,48 The results of both studies indicated that, if a 5% or 10% threshold were applied, some patients with a negative SeHCAT result that might be considered to be ‘borderline’ or ‘equivocal’ (i.e. 7-day retention values of between 5% and 15% or between 10% and 15%), who may benefit from treatment with BAS, would be missed (see Chapter 3, Performance of the SeHCAT test for predicting response to treatment with bile acid sequestrant among patients with diarrhoea-predominant irritable bowel syndrome or functional diarrhoea). Furthermore, there is apparent variation in UK practice with respect to the threshold used to define a positive SeHCAT test result; the 2013 survey of UK practice found that 42 out of 72 (58%) centres providing SeHCAT tests reported using a 7-day retention value of > 15% to define an ‘unequivocally normal’ test result, with 19% using a lower threshold and 22% using a higher threshold. 72 The 2016 survey of UK centres found that the majority [22/32 (69%)] of reporting centres used a ‘normal’ threshold of ≥ 15%. 4 However, variation in practice remained, with ‘normal’ threshold values ranging from ≥ 10% to ≥ 20%; the key findings of this study included the statement that ‘there was a high level of heterogeneity in practice, with no standardised protocol, and no consistently defined diagnostic threshold values of SeHCAT retention. 4 In summary, UK practice varies with respect to the threshold used to define a ‘normal’ SeHCAT test result, and the extent to which patients with ‘borderline’ or ‘equivocal’ 7-day SeHCAT retention values (e.g. between 10% and 15%) could benefit from treatment with BAS remains unclear, and the extent to which patients with 7-day retention values of > 15% may benefit from treatment with BAS is unknown.

Given the uncertainty regarding the optimal SeHCAT decision threshold to define presence of BAM and select patients for treatment with BAS, the potential for intra-individual variation in SeHCAT retention values (e.g. arising from variation in dietary intake before the test or concomitant medication use) may be an important consideration for the implementation of SeHCAT testing in clinical practice. The 2013 survey of UK practice found that 45 out of 72 (62%) responding centres providing SeHCAT tests reported issuing no specific instructions to patients regarding pre-test fasting and 31 out of 72 (42%) gave no specific instructions regarding medication. 72 This information was not reported in the 2016 survey. 4

‘Trial of treatment’ with BAS without testing is sometimes advocated as an alternative approach to investigating BAM as a potential undiagnosed cause of symptoms among patients with IBS-D,77,78 and ‘trial of treatment’ is a comparator for the cost-effectiveness modelling included in this assessment. However, it should be noted that a positive response to treatment with BAS cannot be considered to be 100% specific for a diagnosis of BAM, as these drugs can slow gut transit irrespective of any effect on bile acid metabolism. We identified a German-language study that reported the author’s experience (1991–2017) of using a ‘trial of treatment’ with colestyramine among patients with chronic diarrhoea for whom organic causes had been excluded. 78 This study did not meet the inclusion criteria for our systematic review, as only patients with a positive response to colestyramine were offered SeHCAT testing, and it did not provide data to inform cost-effectiveness modelling, as the total number of patients who received a ‘trial of treatment’ (and hence the proportion who responded) was not reported. 78 However, this study did report the proportion of people [8/60 (13%)] who responded positively to colestyramine and received a SeHCAT test, for whom that test was negative for BAM (7-day retention value of ≥ 20%); this may be considered indicative of the proportion of patients with unexplained chronic diarrhoea who respond positively to BAS, for whom there is no evidence of BAM. 78 In support of SeHCAT testing, it has been suggested (scoping discussions for this assessment) that SeHCAT testing and the assignment of a diagnosis of BAM may improve adherence to treatment with BASs, which are generally considered to be poorly tolerated; when reported, rates of intolerance/discontinuation in the studies included in this assessment were generally high (median 15%, range 4–27%), and the 2016 survey of SeHCAT provision and practice in the UK found that 20 out of 101 (20%) patients who were prescribed BAS reported side effects, including bloating, diarrhoea, constipation, nausea/vomiting, urticarial rash, pain and intolerance to tablets. 4 When information about the BAS treatment was provided, most (16/20) studies included in this assessment reported the use of colestyramine alone. 25,31,32,34,35,37–45,47,48 Three studies reported more than one option for treatment with BAS: colestyramine or colesevelam,28 colestyramine or colestipol,7 and colestyramine or colesevelam or colestipol;30 none of these studies reported either the numbers of patients treated with each drug or the criteria used to select treatment. Similarly, the 2016 survey of UK practice did not provide a breakdown of side effects/intolerance by type of BAS received. 4 There was insufficient information to determine whether or not levels of intolerance varied between colestyramine, colestipol and colesevelam, and no study reported information about patient preferences. We did not identify any studies that reported information about patient preferences with respect to SeHCAT testing versus ‘trial of treatment’ without testing.

Finally, there is a lack of evidence about the efficacy and safety of BAS for the treatment of patients diagnosed with BAM; the clinical effectiveness searches conducted for this assessment identified only three treatment RCTs,79–81 of which only one used a positive SeHCAT test (7-day retention value of < 10%) to define BAM and select patients for inclusion. 81 This was a very small (n = 19) placebo-controlled RCT,81 evaluating two doses (250 mg and 1 g twice daily) of a colonic release formulation of colestyramine, which found no significant effect on the primary outcome (mean daily bowel movement at week 2 of treatment), but reported reductions in instances of diarrhoea and improvements in stool consistency in the treated groups. Although outside the scope of this assessment, it should be noted that our searches identified an ongoing systematic review on the effectiveness of non-pharmacological therapies in the management of BAD among adults. 82

Cost-effectiveness

The main uncertainties in the cost-effectiveness analyses are still caused by a lack of essential data. The majority of the input parameters of the model were informed by clinical experts. In particular, evidence is especially limited as to what occurs after a negative SeHCAT result, for the BAS trial-of-treatment strategy and for the Crohn’s disease population in general. Therefore, a substantial number of assumptions had to be made to make it possible to perform the cost-effectiveness analyses.

As in the previous assessment of SeHCAT,18 the lack of evidence on the accuracy of the SeHCAT test based on a reference test implied that, in the diagnostic decision-analytic models, the most common way of modelling test accuracy, using sensitivity and specificity of the test, was not feasible. Thus, it was not possible to indicate false-positive and false-negative probabilities of testing. The accuracy of SeHCAT testing was thus based on the test result in combination with response to BAS treatment. It might occur that patients responding to BAS are true positive (patients with a true response), but they may also be false-positive patients with a placebo response.

Another unresolved uncertainty regarding the trial-of-treatment strategy relates to the placebo response that may be expected in the true IBS-D patients receiving BAS. It is well known that patients with IBS-D are likely to show high placebo responses to treatment. 83 Clinical experts pointed out that long-term inappropriate treatment with BAS could have implications for absorption of other drugs and vitamins. These long-term undesired consequences were not included in the modelled trial-of-treatment strategy. Clinical experts also indicated that a response to BAS is not helpful diagnostically because BASs are constipating drugs in any event, as known from when they were used for lowering cholesterol levels in people with no bowel problems. Therefore, using BAS as a diagnostic would be no better than using loperamide as a diagnostic test for any form of diarrhoea. In addition, transitions between the ‘diarrhoea’ and ‘no-diarrhoea’ health states might not be the same for BAS patients having a positive SeHCAT result and for patients responding to a BAS trial of treatment, as patients without a positive diagnosis may be less inclined to accept the side effects of BAS (colestyramine).

The uncertainties in the Markov model are still also unresolved. The diarrhoea health state was valued by cost and utilities irrespective of the cause of the symptom. However, there is no evidence to confirm whether or not this is true. For the increase in utility when patients become responders, we made the same previous assumption that patients responding to BAS (colestyramine) would get only 75% of the utility benefit of becoming a responder. It is unknown to what extent this assumption of 75% is realistic. However, scenario analyses showed that the impact of this assumption on the model results is minor.

Transition probabilities in the Markov model remain uncertain as well. Patients can, in theory, transition between the ‘diarrhoea’ and ‘no-diarrhoea’ health states. Being in the latter health state can be understood as an improvement of the symptoms, but relapse (i.e. transition to the ‘diarrhoea’ health state) is also possible. In the base-case analysis, it is assumed that patients responding to BAS and IBS-D treatment remain in the ‘no-diarrhoea’ health state, which can be seen as a ‘cure’ assumption. Although this can be a simplistic assumption, it was based on the opinion of the clinical experts, who indicated that, in general, patients initially responding to treatment are expected to continue responding in the long term. Making more realistic assumptions would have required more data, which were not available. Scenario analyses were conducted to test the impact of transition assumptions on the results (see, e.g., scenario analyses 7 and 14). These analyses showed that the impact on the model results was minor. This time it was also not possible to include a health state of ‘constipation’ or other adverse events in the long-term Markov model, given the lack of data. Threshold or any other type of exploratory analyses on adverse events were not considered for pragmatic reasons. The rationale for this was the same as explained previously regarding the selection of only the SeHCAT 15% cut-off point for the cost-effectiveness analyses. Scenario analyses on adverse events would not be evidence based; therefore, the relevance of such scenarios could be questionable. Thus, it was preferred to focus on other (of the many) uncertainty areas around the cost-effectiveness analyses that were deemed more important at this stage than the modelling of adverse events.

Finally, it is uncertain how the cost-effectiveness results would change should other SeHCAT strategies be included in the analyses. The available clinical evidence regarding the cut-off values defining a positive SeHCAT test shows that the various cut-off values influence test-accuracy estimates expressed in BAS treatment response. The cost-effectiveness analyses included in this report have shown that response to treatment is a key driver of the cost-effectiveness results. The strategy with the highest response rate is likely to be the preferred one in terms of health benefits, but it remains uncertain whether or not this will be translated into cost-effectiveness.

EMBASE (Ovid)

Date range searched: 1974 to 25 November 2020.

Date searched: 26 November 2020.

(SeHCAT OR BAS) + BAD (No A)

1. tauroselcholic acid/ (233)

2. (tauroselcholic or selenohomocholyltaurine or 75018-71-2).ti,ab,ot,hw,rn,tn. (397)

3. (SeHCAT or Se-HCAT or 75SeHCAT or Se-75 or 75-SeHCAT or SE75).ti,ab,ot,hw,tn. (1596)

4. (23-seleno-25-homo-tauro-cholic acid or selenium homocholic acid taurine or 23-selena-25-homocholyltaurine or 23-selena-25-homotaurocholate or 23- selena-25-homotaurocholic-acid or selenium radioisotopes or tauroselenocholic acid or 75Se-homotaurocholate).ti,ab,ot,hw,tn. (52)

5. (selenium adj3 “75”).ti,ab,ot,hw,tn. (860)

6. or/1-5 (2179)

7. bile acid sequestrant/ (1459)

8. ((bile adj3 (acid or salt) adj3 sequest$) or BAS).ti,ab,ot,hw,rn. (19,061)

9. Colestipol/ or (Colestipol or cholestabyl or cholestipol or colestid or diethylenetriamine-epichlorohydrin-copolymer or diethylenetriamine-polymer-with-1-chloro-2,3-epoxypropane or epichlorohydrin-copolymer-with-diethylenetriamine or flavored-colestid or lestid or u-26,597a or u-26597-a or u-26597a or u-26,597a or 25085-17-0 or 37296-80-3 or 50925-79-6).ti,ab,ot,hw,rn,tn. (2940)

10. Colestyramine/ or (colestyramine or chol-less or choles or cholesthexal or cholestyramin or cholestyramine or cholybar or cholytar or colestepril or colestiramina or colestran or colestrol or colestyramin or cuemid or lipocol-merz or lismol or locholest or prevalite or quantalan or questran or resincoles-tiramina or resincolestiramina or vasosan-p-granulat or vasosan-s-granulat or 11041-12-6 or 58391-37-0).ti,ab,ot,hw,rn,tn. (11,381)

11. Colesevelam/ or (Colesevelam or cholestagel or gt-31-104 or gt-31-104hb or gt-31-104 or gt-31-104hb or gt31-104 or gt31-104hb or gt31-104 or gt31-104hb or welchol or lodalis or 182815-43-6 or 182815-44-7).ti,ab,ot,hw,rn,tn. (1406)

12. aluminum hydroxide/ or (aluminum hydroxide or Ageldrate or al u creme or alcid or aldrox or algeldraat or algeldrate or algelox or alhydrogel or alkagel or alocol or alokreem or alterna gel or alu cap or alu-cap or alu-tab or alucol or aludrox or alugelibys or alumigel or alumina gel or alumina trihydrate or aluminium hydroxide or aluminium hydroxyde or aluminoid or aluminox or aluminum hydrate or aluminum hydroxide gel or aluminum oxide trihydrate or aluminum trihydrate or alutab or amphogel or amphojel or amphotabs or antiphos or bayerite or chefarox or collumina or collumol or colodral or colugel or creamalin or cremorin or diplogel or f 1000 or f1000 or fluagel or gastracol or gastrosetarderm or gelumina or hoemigel or hycolal or hydracoll or hydrated alumina or hydrolum or hydronal or hydroxal or lactalumina or neutroxide or palliacol or pepsamar or ulcerin-p or vanogel or 21645-51-2 or 1330-44-5 or 80206-84-4 or brasivil or rocgel or alugel or hydrated alumina or basalgel or dialume or nephrox).ti,ab,ot,hw,rn,tn. (10,894)

13. or/7-12 (41,822)

14. 6 or 13 (43,803)

15. (BAM or I-BAM or IBAM or PBAM or BSM or BAD).ti,ab,ot,hw. (60,921)

16. bile acid diarrh?ea$.ti,ab,ot,hw. (227)

17. chronic diarrhea/ or bile acid/ or bile salt/ (36,292)

18. ((chronic or explosive or smelly or watery or recur$ or persist$ or protract$ or continual$ or continuous$ or sustain$ or constant$ or relentless$ or unrelent$ or functional or aggressive) adj2 (diarrh?e$ or diarrea$ or f?eces)).ti,ab,ot,hw. (16,726)

19. (malabsorb$ or mal-absorb$ or malabsorp$ or mal-absorp$).ti,ab,ot,hw. (24,005)

20. ((bile or biliary) adj3 (acid$ or salt$)).ti,ab,ot,hw. (50,197)

21. or/15-20 (148,024)

22. 14 and 21 (5860)

23. animal/ (1,492,379)

24. animal experiment/ (2,624,468)

25. (rat or rats or mouse or mice or murine or rodent or rodents or hamster or hamsters or pig or pigs or porcine or rabbit or rabbits or animal or animals or dogs or dog or cats or cow or bovine or sheep or ovine or monkey or monkeys).mp. (6,912,383)

26. or/23-25 (6,912,383)

27. exp human/ (21,744,415)

28. human experiment/ (528,150)

29. or/27-28 (21,746,205)

30. 26 not (26 and 29) (5,288,236)

31. 22 not 30 (4797)

MEDLINE(Ovid) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations and Daily

Date range searched: 1946 to 30 November 2020.

Date searched: 1 December 2020.

1. (tauroselcholic or selenohomocholyltaurine or 75018-71-2).ti,ab,ot,hw,rn. (10)

2. (SeHCAT or Se-HCAT or 75SeHCAT or Se-75 or 75-SeHCAT or SE75).ti,ab,ot,hw,rn. (375)

3. (23-seleno-25-homo-tauro-cholic acid or selenium homocholic acid taurine or 23-selena-25-homocholyltaurine or 23-selena-25-homotaurocholate or 23- selena-25-homotaurocholic-acid or selenium radioisotopes or tauroselenocholic acid or 75Se-homotaurocholate).ti,ab,ot,hw,rn. (373)

4. (selenium adj3 “75”).ti,ab,ot,hw. (185)

5. or/1-4 (788)

6. ((bile adj3 (acid or salt) adj3 sequest$) or BAS).ti,ab,ot,hw,rn. (5844)

7. Colestipol/ or (Colestipol or cholestabyl or cholestipol or colestid or diethylenetriamine-epichlorohydrin-copolymer or diethylenetriamine-polymer-with-1-chloro-2,3-epoxypropane or epichlorohydrin-copolymer-with-diethylenetriamine or flavored-colestid or lestid or u-26,597a or u-26597-a or u-26597a or u-26,597a or 25085-17-0 or 37296-80-3 or 50925-79-6).ti,ab,ot,hw,rn. (551)

8. Cholestyramine Resin/ or (colestyramine$ or chol-less or choles or cholesthexal or cholestyramin or cholestyramine$ or cholybar or cholytar or colestepril or colestiramina or colestran or colestrol or colestyramin or cuemid$ or lipocol-merz or lismol or locholest or prevalite or quantalan or questran$ or resincoles-tiramina or resincolestiramina or vasosan-p-granulat or vasosan-s-granulat or 11041-12-6 or 58391-37-0 or mk 135 or mk135).ti,ab,ot,hw,rn. (3644)

9. Colesevelam Hydrochloride/ or (Colesevelam or cholestagel or gt-31-104 or gt-31-104hb or gt-31-104 or gt-31-104hb or gt31-104 or gt31-104hb or gt31-104 or gt31-104hb or welchol or lodalis or 182815-43-6 or 182815-44-7).ti,ab,ot,hw,rn. (302)

10. Aluminum Hydroxide/ or (aluminum hydroxide or Ageldrate or al u creme or alcid or aldrox or algeldraat or algeldrate or algelox or alhydrogel or alkagel or alocol or alokreem or alterna gel or alu cap or alu-cap or alu-tab or alucol or aludrox or alugelibys or alumigel or alumina gel or alumina trihydrate or aluminium hydroxide or aluminium hydroxyde or aluminoid or aluminox or aluminum hydrate or aluminum hydroxide gel or aluminum oxide trihydrate or aluminum trihydrate or alutab or amphogel or amphojel or amphotabs or antiphos or bayerite or chefarox or collumina or collumol or colodral or colugel or creamalin or cremorin or diplogel or f 1000 or f1000 or fluagel or gastracol or gastrosetarderm or gelumina or hoemigel or hycolal or hydracoll or hydrated alumina or hydrolum or hydronal or hydroxal or lactalumina or neutroxide or palliacol or pepsamar or ulcerin-p or vanogel or 21645-51-2 or 1330-44-5 or 80206-84-4).ti,ab,ot,hw,rn. (6299)

11. or/6-10 (15,925)

12. 5 or 11 (16,640)

13. (BAM or I-BAM or IBAM or PBAM or BSM or BAD).ti,ab,ot,hw. (40,139)

14. bile acid diarrh?ea$.ti,ab,ot,hw. (111)

15. diarrhea/ (48,230)

16. ((chronic or explosive or smelly or watery or recur$ or persist$ or protracted or continual$ or continuous$ or sustain$ or constant$ or relentless$ or unrelent$ or functional or aggressive) adj2 (diarrh?e$ or diarrea$)).ti,ab,ot,hw. (10,235)

17. "Bile Acids and Salts"/ (22,496)

18. (malabsorb$ or mal-absorb$ or malabsorp$ or mal-absorp$).ti,ab,ot,hw. (17,246)

19. ((bile or biliary) adj3 (acid$ or salt$)).ti,ab,ot,hw. (40,041)

20. or/13-19 (147,664)

21. 12 and 20 (2978)

22. animals/ not (animals/ and humans/) (4,727,656)

23. 21 not 22 (2282)

Cochrane Central Register of Controlled Trials (Wiley) and Cochrane Database of Systematic Reviews (Wiley)

Searched up to November 2020, issue 11.

Date searched: 26 November 2020.

#1 (tauroselcholic or selenohomocholyltaurine or “75018-71-2”) 5

#2 SeHCAT or “Se-HCAT” or 75SeHCAT or “Se-75” or “75-SeHCAT” or SE75 3000

#3 “ 23-seleno-25-homo-tauro-cholic acid” or selenium homocholic acid taurine or “23-selena-25-homocholyltaurine” or “23-selena-25-homotaurocholate” or “23- selena-25-homotaurocholic-acid” or selenium radioisotopes or tauroselenocholic acid or “75Se-homotaurocholate” 6

#4 selenium near “75” 33

#5 #1 OR #2 OR #3 OR #4 3033

#6 ((bile near (acid or salt) near sequest*) or BAS) 4488

#7 MeSH descriptor: [Colestipol] explode all trees 90

#8 Colestipol or cholestabyl or cholestipol or colestid or “diethylenetriamine-epichlorohydrin-copolymer” or “diethylenetriamine-polymer-with-1-chloro-2,3-epoxypropane” or “epichlorohydrin-copolymer-with-diethylenetriamine” or “flavored-colestid” or lestid or “u-26,597a” or “u-26597-a” or “u-26597a” or “u-26,597a” or “25085-17-0” or “37296-80-3” or “50925-79-6” 177

#9 MeSH descriptor: [Cholestyramine Resin] explode all trees 275

#10 (colestyramine or “chol-less” or choles or cholesthexal or cholestyramin or cholestyramine or cholybar or cholytar or colestepril or colestiramina or colestran or colestrol or colestyramin or cuemid or “lipocol-merz” or lismol or locholest or prevalite or quantalan or questran or “resincoles-tiramina” or resincolestiramina or “vasosan-p-granulat” or “vasosan-s-granulat” or “11041-12-6” or “58391-37-0”) 556

#11 MeSH descriptor: [Colesevelam Hydrochloride] explode all trees 107

#12 (Colesevelam or cholestagel or “gt-31-104” or “gt-31-104hb” or “gt-31-104” or “gt-31-104hb” or “gt31-104” or “gt31-104hb” or “gt31-104” or “gt31-104hb” or welchol or lodalis or “182815-43-6” or “182815-44-7”) 177

#13 MeSH descriptor: [Aluminum Hydroxide] explode all trees 579

#14 (aluminum hydroxide or Ageldrate or al u creme or alcid or aldrox or algeldraat or algeldrate or algelox or alhydrogel or alkagel or alocol or alokreem or alterna gel or alu cap or “alu-cap” or “alu-tab” or alucol or aludrox or alugelibys or alumigel or alumina gel or alumina trihydrate or aluminium hydroxide or aluminium hydroxide or aluminoid or aluminox or aluminum hydrate or aluminum hydroxide gel or aluminum oxide trihydrate or aluminum trihydrate or alutab or amphogel or amphojel or amphotabs or antiphos or bayerite or chefarox or collumina or collumol or colodral or colugel or creamalin or cremorin or diplogel or f 1000 or f1000 or fluagel or gastracol or gastrosetarderm or gelumina or hoemigel or hycolal or hydracoll or hydrated alumina or hydrolum or hydronal or hydroxal or lactalumina or neutroxide or palliacol or pepsamar or “ulcerin-p” or vanogel or “21645-51-2” or “1330-44-5” or “80206-84-4” or brasivil or rocgel or alugel or hydrated alumina or basalgel or dialume or nephrox) 7070

#15 #6 or #7 or #8 or #9 or #10 or #11 or #12 or #13 or #14 11,461

#16 #5 or #15 12,601

#17 (bile acid near (diarrhoe* or diarrhe* or diarrea*)):ti,ab,kw 39

#18 (chronic or explosive or smelly or watery or recur* or persist* or protract* or continual* or continuous* or sustain* or constant* or relentless* or unrelent* or functional or aggressive) near (diarrhoe* or diarrhe* or diarrea*):ti,ab,kw 1364

#19 (malabsorb* or “mal-absorb*” or malabsorp* or “mal-absorp*”):ti,ab,kw 1045

#20 (BAM or “I-BAM” or IBAM or PBAM or BSM or BAD):ti,ab,kw 2863

#21 ((bile or biliary) near (acid* or salt*)):ti,ab,kw 2196

#22 MeSH descriptor: [Bile Acids and Salts] explode all trees 1193

#23 MeSH descriptor: [Diarrhea] this term only 3119

#24 #17 or #18 or #19 or #20 or #21 or #22 or #23 10,503

#25 #16 and #24 539.

CDSR retrieved 131 records.

CDSR Protocols retrieved three records.

CENTRAL retrieved 404 records.

Database of Abstracts of Reviews of Effects (Centre for Reviews and Dissemination) and Health Technology Assessment database (Centre for Reviews and Dissemination) (www.crd.york.ac.uk/CRDWeb/)

Searched up to March 2015 and March 2018, respectively.

Date searched: 26 November 2020.

1. (tauroselcholic or selenohomocholyltaurine or 75018-71-2) 3

2. (SeHCAT or Se-HCAT or 75SeHCAT or Se-75 or 75-SeHCAT or SE75) 3

3. (23-seleno-25-homo-tauro-cholic acid or selenium homocholic acid taurine or 23-selena-25-homocholyltaurine or 23-selena-25-homotaurocholate or 23- selena-25-homotaurocholic-acid or selenium radioisotopes or tauroselenocholic acid opr 75Se-homotaurocholate) 0

4. (selenium near “75”) 5

5. #1 OR #2 OR #3 OR #4 5

6. (((bile near (acid or salt) near sequest*) or BAS)) 30

7. MeSH DESCRIPTOR Colestipol EXPLODE ALL TREES 3

8. ((Colestipol or cholestabyl or cholestipol or colestid or diethylenetriamine-epichlorohydrin-copolymer or diethylenetriamine-polymer-with-1-chloro-2,3-epoxypropane or epichlorohydrin-copolymer-with-diethylenetriamine or flavored-colestid or lestid or u-26,597a or u-26597-a or u-26597a or u-26,597a or 25085-17-0 or 37296-80-3 or 50925-79-6)) 21

9. MeSH DESCRIPTOR Cholestyramine Resin EXPLODE ALL TREES 6

10. ((colestyramine or chol-less or choles or cholesthexal or cholestyramin or cholestyramine or cholybar or cholytar or colestepril or colestiramina or colestran or colestrol or colestyramin or cuemid or lipocol-merz or lismol or locholest or prevalite or quantalan or questran or resincoles-tiramina or resincolestiramina or vasosan-p-granulat or vasosan-s-granulat or 11041-12-6 or 58391-37-0)) 37

11. MeSH DESCRIPTOR Colesevelam Hydrochloride EXPLODE ALL TREES 1

12. (Colesevelam or cholestagel or gt-31-104 or gt-31-104hb or gt-31-104 or gt-31-104hb or gt31-104 or gt31-104hb or gt31-104 or gt31-104hb or welchol or lodalis or 182815-43-6 or 182815-44-7) 4

13. MeSH DESCRIPTOR Aluminum Hydroxide EXPLODE ALL TREES 4

14. ((aluminum hydroxide or Ageldrate or al u creme or alcid or aldrox or algeldraat or algeldrate or algelox or alhydrogel or alkagel or alocol or alokreem or alterna gel or alu cap or alu-cap or alu-tab or alucol or aludrox or alugelibys or alumigel or alumina gel or alumina trihydrate or aluminium hydroxide or aluminium hydroxide or aluminoid or aluminox or aluminum hydrate or aluminum hydroxide gel or aluminum oxide trihydrate or aluminum trihydrate or alutab or amphogel or amphojel or amphotabs or antiphos or bayerite or chefarox or collumina or collumol or colodral or colugel or creamalin or cremorin or diplogel or f 1000 or f1000 or fluagel or gastracol or gastrosetarderm or gelumina or hoemigel or hycolal or hydracoll or hydrated alumina or hydrolum or hydronal or hydroxal or lactalumina or neutroxide or palliacol or pepsamar or ulcerin-p or vanogel or 21645-51-2 or 1330-44-5 or 80206-84-4 or brasivil or rocgel or alugel or hydrated alumina or basalgel or dialume or nephrox)) 10

15. #6 OR #7 OR #8 OR #9 OR #10 OR #11 OR #12 OR #13 OR #14 86

16. ((chronic or explosive or smelly or watery or recur* or persist* or protract* or continual* or continuous* or sustain* or constant* or relentless* or unrelent* or functional or aggressive) near (diarrhoe* or diarrhe* or diarrea*)) 50

17. ((malabsorb* or mal-absorb* or malabsorp* or mal-absorp*)) 44

18. ((bile acid near (diarrhoe* or diarrhe* or diarrea*))) 0

19. ((BAM or I-BAM or IBAM or PBAM or BSM or BAD)) 76

20. ((bile or biliary) near (acid* or salt*)) 38

21. MeSH DESCRIPTOR Diarrhea EXPLODE ALL TREES 228

22. MeSH DESCRIPTOR Bile Acids and Salts EXPLODE ALL TREES 49

23. #16 OR #17 OR #18 OR #19 OR #20 OR #21 OR #22 442

24. #1 OR #15 88

25. #23 AND #24 17

26. (#25) IN DARE 13

27. (#25) IN HTA 3