3 Evidence

The diagnostics advisory committee considered evidence on SeHCAT (tauroselcholic [75 selenium] acid) for investigating bile acid diarrhoea from several sources. Full details of all the evidence are in the project documents on the NICE website.

3.1 The clinical and cost effectiveness of SeHCAT for investigating bile acid diarrhoea was assessed in people with:

  • chronic diarrhoea with an unknown cause, suspected or diagnosed diarrhoea-predominant irritable bowel syndrome (IBS‑D) or functional diarrhoea (primary bile acid diarrhoea)

  • Crohn's disease without ileal resection who have chronic diarrhoea (secondary bile acid diarrhoea).

Clinical effectiveness

3.2 A systematic review of clinical-effectiveness evidence by the external assessment group (EAG) identified 24 observational studies relevant to this assessment. No randomised controlled trials were identified. Of the 24 studies, 21 described outcomes only for some of the people who had a positive SeHCAT test result. The remaining 3 studies assessed how well the SeHCAT test predicts response to bile acid treatment (predictive accuracy). Heterogeneity between the studies was high and the quality of the studies was considered low.

Predictive accuracy of SeHCAT in suspected or diagnosed IBS-D or functional diarrhoea

3.3 The 3 small studies evaluating the predictive accuracy of SeHCAT were included in the previous assessment. They assessed the relationship between the SeHCAT test result and response to colestyramine treatment. Table 1 summarises the predictive accuracy estimates for the different SeHCAT thresholds that the studies reported. Because of the small number of studies, and differences in study characteristics and test thresholds, a pooled estimate of predictive accuracy was not calculated.

Table 1 Accuracy of SeHCAT for predicting response to bile acid sequestrants in suspected or diagnosed IBS-D or functional diarrhoea

Study

Study size

Threshold

Sensitivity

95% confidence interval (CI)

Specificity

95% CI

Merrick et al. (1985)

43

<8%

0.667

0.223 to 0.957

0.971

0.847 to 0.999

Merrick et al. (1985)

43

≤15%

1.000

0.541 to 1.000

0.912

0.763 to 0.981

Sciaretta et al. (1986)

13

<5%

0.857

0.421 to 0.996

1.000

0.541 to 1.000

Sciaretta et al. (1987)

46

<8%

0.950

0.751 to 0.999

0.962

0.804 to 0.999

Response to bile acid sequestrants after a positive SeHCAT test in suspected or diagnosed IBS-D or functional diarrhoea

3.4 In total, 8 studies evaluated the probability of response to bile acid sequestrants after a positive SeHCAT test at a 15% threshold. Only 2 of these were new studies found through the searches in this assessment. The median response rate in these 8 studies was 68% (range 38% to 86%). Between 70% and 100% of people had bile acid sequestrant treatment after a positive SeHCAT test. Because of the substantial differences between studies, meta-analysis of the response rate was considered inappropriate.

Effects of treatment on bowel symptoms in suspected or diagnosed IBS-D or functional diarrhoea

3.5 In addition to reporting the probability of response to treatment after a positive SeHCAT test, 3 of the studies described the effects of bile acid sequestrants on bowel symptoms. In these studies, colestyramine was described as improving stool consistency, reducing daily bowel movements and stool frequency, and removing the urgency of needing the toilet.

Tolerability of bile acid sequestrants in suspected or diagnosed IBS-D or functional diarrhoea

3.6 There were 8 studies reporting the proportion of people who found bile acid sequestrants difficult to tolerate or stopped their treatment for unclear reasons. Rates of intolerance and discontinuation were generally high (median 15%, range 4% to 27%). There was not enough information to determine whether these rates varied between the different types of bile acid sequestrants.

Health-related quality of life of people with suspected or diagnosed IBS-D or functional diarrhoea

3.7 There were 2 studies reporting changes in health-related quality of life in people who had bile acid sequestrants after a positive SeHCAT test result. One study evaluated quality of life using the SF‑36 questionnaire after 8 weeks of cholestyramine. There were improvements in the general pain domain in people with mild bile acid diarrhoea (defined as a positive SeHCAT test result at a threshold between 11% and 15%, p<0.05). There were also improvements across other domains (including emotional problems, energy or fatigue, emotional wellbeing, social functioning, general health and health change) in people with more severe bile acid diarrhoea (threshold 5% or less, p<0.05). Another study reported improvements in activity levels subscore (p=0.00998) using the EQ‑5D questionnaire in people who had colestyramine or colesevelam. This study did not report either the threshold used to define a positive SeHCAT test result or the duration of follow up.

Evidence in Crohn's disease

3.8 No evidence was found for the predictive accuracy of SeHCAT or for patient-reported outcomes in people with Crohn's disease without ileal resection who have chronic diarrhoea.

3.9 Only 1 small study (Smith et al. 2000) evaluated the probability of response to bile acid sequestrants (colestyramine or colestipol) after a positive SeHCAT test in 44 people with Crohn's disease. This study was included in the previous assessment. The threshold used to define a positive SeHCAT test was 10%. In this study, 24 (55%) people had a positive SeHCAT test result at a 10% threshold. But only 9 of these 24 (38%) people had bile acid sequestrants. This treatment was considered to work for 8 of these 9 people (89%).

Cost effectiveness

Economic model

3.10 The EAG developed 2 de novo economic models to assess SeHCAT's cost effectiveness for investigating and diagnosing bile acid diarrhoea in the populations in the scope of this assessment. The models used a lifetime (50 years) time horizon to estimate outcomes in terms of quality-adjusted life years (QALYs) and costs from an NHS perspective.

3.11 The models included:

  • a short-term decision analytic model that captured the diagnostic pathway and initial response to treatment (first 6 months)

  • a long-term Markov model that estimated the lifetime costs and effects for people having treatment.

3.12 SeHCAT testing, using a 15% threshold value for a positive test, was compared with these strategies:

  • no SeHCAT testing and no bile acid sequestrants

  • no SeHCAT testing and a trial of bile acid sequestrants.

Model inputs for suspected or diagnosed IBS-D or functional diarrhoea

3.13 When possible, model inputs were based on the clinical effectiveness systematic review or other published literature. When such evidence was not available, expert opinion was used. This was obtained from a questionnaire the EAG sent to the clinical expert specialist committee members for this assessment.

3.14 The probability of a positive SeHCAT test result in the base case was a pooled estimate of 45.4%, calculated from the 8 studies in the systematic review that used SeHCAT at a 15% threshold.

3.15 People who were offered bile acid sequestrants had either colestyramine or colesevelam. Based on the responses to the EAG's questionnaire, it was assumed that 50% of people in the SeHCAT strategy and 85% of people in the trial of treatment strategy started colestyramine.

3.16 The probability of response to bile acid sequestrants after a positive SeHCAT test result was a pooled estimate of 63.8%, from the 8 studies using SeHCAT at a 15% threshold. In the trial of treatment strategy, based on expert opinion, the treatment response was estimated as 30%. Based on the available evidence, it was not possible to distinguish between the response to colestyramine and colesevelam.

3.17 Expert opinion suggested that colestyramine may be difficult to tolerate so people may be offered colesevelam instead. In both the SeHCAT strategy and the trial of treatment strategy, the probability of switching from colestyramine to colesevelam was estimated to be 50% based on expert opinion.

3.18 The probability of having a colonoscopy in the model was estimated based on expert opinion:

  • For people who had a negative SeHCAT test result or when bile acid sequestrant treatment after a positive SeHCAT test did not work, the probability was 49%.

  • For people who had not had a SeHCAT test or a trial of bile acid sequestrants, the probability was 74%.

  • For people who had a trial of bile acid sequestrants that had not worked, the probability was 90%.

3.19 The probability of being diagnosed with inflammatory bowel disease (IBD) after colonoscopy was likely to be very low. This was supported by expert opinion and a study (Patel et al. 2015) that reported the number of people with IBS‑D‑like symptoms who were eventually diagnosed with IBD. Based on this, the proportion of people who would be diagnosed with IBD after colonoscopy was estimated to be 5.3%. Based on expert opinion, the probability of response to IBD treatment was estimated to be 72%.

3.20 The estimated probabilities of having a colonoscopy and a diagnosis of IBD (5.3% of the people having a colonoscopy) meant that most people, about 96%, would be offered IBS‑D treatment. The probability of IBS‑D treatment response, after a colonoscopy that ruled out IBD, was estimated based on expert opinion:

  • For people who had a negative SeHCAT test result, the probability was 56%.

  • For people who had not had a SeHCAT test or a trial of bile acid sequestrants, the probability was 46%.

  • For people whose bile acid sequestrant treatment had not worked, the probability was 50%.

    Based on these estimates, the EAG calculated that the probability of IBS‑D treatment response would be slightly lower for people who had not had a colonoscopy to rule out IBD:

  • For people who had a negative SeHCAT test result, the probability was 53%.

  • For people who had not had a SeHCAT test or a trial of bile acid sequestrant, the probability was 44%.

  • For people whose bile acid sequestrant treatment had not worked, the probability was 47%.

3.21 The long-term Markov model included health states for 'diarrhoea', 'no diarrhoea' and 'death'. Assumptions about people moving between the 'diarrhoea' and 'no diarrhoea' health states were informed by expert opinion. This suggested that, in general, the response to bile acid sequestrants and to IBS‑D treatment is expected to last without relapses. So, no movement from the 'no diarrhoea' to the 'diarrhoea' health state in the long term should be expected. With IBD treatment, the experts noted that relapses are expected to occur after the initial response to treatment. Therefore, in the long-term Markov model it was assumed that people having IBD treatment would move between the 'diarrhoea' and 'no diarrhoea' health states. Based on expert opinion, the EAG assumed that people having IBD treatment have an average of 1 relapse every 5 years. So, the base-case probability of people having IBD treatment moving from the 'no diarrhoea' to 'diarrhoea' health state was estimated as 0.45%.

3.22 The base case assumed that no excess mortality was associated with bile acid diarrhoea.

Health-related quality of life in suspected or diagnosed IBS-D or functional diarrhoea

3.23 The utility values used were the same as in the previous assessment (see table 2). Because colestyramine can be difficult to tolerate, it was assumed that people for whom colestyramine worked had a slightly lower (75%) utility gain from their treatment. The model did not include utility loss for colonoscopy.

Table 2 Utility values used in the base case for people with suspected or diagnosed IBS-D or functional diarrhoea

Health state

Subpopulation

Utility value

Source

No diarrhoea

People for whom colesevelam, diarrhoea-predominant irritable bowel syndrome (IBS-D) or inflammatory bowel disease (IBD) treatment works (treatment response)

0.776

Pooled estimate from Mearin et al. (2004) and Spiegel et al. (2009)

No diarrhoea

People for whom colestyramine treatment works (treatment response)

0.760

Assumption

Diarrhoea

People for whom bile acid sequestrant, IBS-D or IBD treatment does not work (no treatment response)

0.712

Pooled estimate from Mearin et al. (2004) and Spiegel et al. (2009)

Costs for suspected or diagnosed IBS-D or functional diarrhoea

3.24 The company's cost for SeHCAT was £195 per capsule. The cost of administering it in the NHS was £282 per test, taken from the NHS national tariff for 2021 to 2022. Therefore, the total cost of a SeHCAT test in the base case was £477 per test.

3.25 Other costs considered in the model included the costs of bile acid sequestrants, IBS‑D treatment, IBD treatment and colonoscopy (see table 3).

Table 3 Costs of medications, additional treatments and colonoscopy in the model for people with suspected or diagnosed IBS-D or functional diarrhoea

Resource

Cost per person per day

Sources

Bile acid sequestrants: colestyramine

£0.35

BNF, expert opinion

Bile acid sequestrants: colesevelam

£2.56

BNF, expert opinion

IBS-D treatment: medication

£0.06

BNF, expert opinion

IBS-D treatment: diet therapy

£12.24

NHS national tariff, expert opinion

IBS-D treatment: psychological therapy

£35.74

NHS national tariff, expert opinion

IBD treatment: medication

£21.73

BNF, expert opinion

IBD treatment: diet therapy

£149.00

NHS national tariff, expert opinion

IBD treatment: psychological therapy

£289.33

NHS national tariff, expert opinion

Colonoscopy

£175.75

NHS national tariff, expert opinion (90% conventional colonoscopy and 10% CT colonoscopy)

Base-case assumptions in suspected or diagnosed IBS-D or functional diarrhoea

3.26 These key assumptions were applied in the base-case analysis:

  • People whose condition responds to bile acid sequestrant treatment have bile acid diarrhoea.

  • Treatment for bile acid diarrhoea includes only bile acid sequestrants, either colestyramine or colesevelam.

  • Some people will switch to colesevelam early in the treatment because colestyramine may be difficult to tolerate.

  • People for whom bile acid sequestrant or IBS‑D treatment works in the short term will continue and will benefit from it for the rest of their life.

  • People who take colesevelam will have better quality of life than people who take colestyramine.

  • Some people who have not had a SeHCAT test, or who have a negative SeHCAT test result, or for whom bile acid sequestrants have not worked in the short term, will have a colonoscopy to detect IBD.

  • Some people for whom IBD treatment works in the short term will have relapses throughout their life.

  • People for whom none of the treatments offered have worked in the short term are assumed to take loperamide for the rest of their life.

  • All the resource-use estimates are based on expert opinion.

Model inputs for Crohn's disease

3.27 The clinical effectiveness systematic review found only 1 study that reported the probability of a positive SeHCAT test in people with Crohn's disease without ileal resection who have chronic diarrhoea. This probability of 55% was used in the base case.

3.28 People who were offered bile acid sequestrants started either colestyramine or colesevelam. Based on expert opinion, it was assumed that 63% of people in the SeHCAT strategy and 58% of people in the trial of treatment strategy started colestyramine.

3.29 The probability of response to treatment after a positive SeHCAT test result at a 15% threshold was estimated as 89% in the base case. This came from the same small study that provided data for the probability of a positive SeHCAT test result. It was higher than the maximum 70% probability of response estimated by the clinical experts. In the trial of treatment strategy, based on expert opinion, the treatment response was estimated as 33%. Based on the available evidence, it was not possible to distinguish between the response to colestyramine and colesevelam.

3.30 In both the SeHCAT and the trial of treatment strategies, the probability of switching from colestyramine to colesevelam because of poor tolerability was estimated to be 44%, based on expert opinion.

3.31 Treatment options for diarrhoea in Crohn's disease may vary depending on whether they treat diarrhoea because of relapse or prevent diarrhoea during remission. Because of this, it was not possible to find evidence on how well diarrhoea treatment in Crohn's disease might work. Therefore, the probability of Crohn's disease treatment response was estimated based on expert opinion:

  • For people who had a negative SeHCAT test result, the probability was 42%.

  • For people who had not had a SeHCAT test or a trial of bile acid sequestrants, the probability was 40%.

  • For people whose bile acid sequestrant treatment had not worked, the probability was 41%.

3.32 Assumptions about people moving between the 'diarrhoea' and 'no diarrhoea' health states were informed by expert opinion. This suggested that, in general, the response to bile acid sequestrants is expected to last. So, no relapses and no movement from the 'no diarrhoea' to the 'diarrhoea' health state in the long term should be expected. For diarrhoea treatment for Crohn's disease, the experts expected that relapses would occur after the initial response to treatment. Therefore, in the long-term Markov model it was assumed that people having this treatment would move between the 'diarrhoea' and 'no diarrhoea' health states. As with people having IBD treatment in the IBS‑D or functional diarrhoea model, it was assumed that people having diarrhoea treatment for Crohn's disease would have an average of 1 relapse every 5 years. So, the base-case probability of people on this treatment moving from the 'no diarrhoea' to 'diarrhoea' health state was estimated as 0.575%.

3.33 The base case assumed that no excess mortality was associated with bile acid diarrhoea. A pooled standardised mortality ratio estimate from a meta-analysis of mortality in Crohn's disease by Canavan et al. (2007) was applied to the overall UK mortality estimates.

Health-related quality of life in Crohn's disease

3.34 No studies on health-related quality of life in people with Crohn's disease and diarrhoea were found. The estimate from a study providing utilities for people with active Crohn's disease (Buxton et al. 2007) was assumed to also reflect quality of life in the diarrhoea health state. To estimate the utility gain for people for whom the treatment worked, it was assumed that the utility loss because of diarrhoea was the same as for people with IBS‑D or functional diarrhoea. As in the IBS‑D or functional diarrhoea model, it was assumed that the utility gain from colestyramine would be slightly lower than from the other treatments. Table 4 summarises the utility values used.

Table 4 Utility values used in the base case in people with Crohn's disease

Health state

Subpopulation

Utility value

Source

No diarrhoea

People for whom colesevelam or treatment of diarrhoea in Crohn's disease works (treatment response)

0.764

Assumption

No diarrhoea

People for whom colestyramine treatment works (treatment response)

0.748

Assumption

Diarrhoea

People for whom bile acid sequestrants or treatment of diarrhoea in Crohn's disease does not work (no treatment response)

0.700

Estimate from Buxton et al. (2007)

Costs for Crohn's disease

3.35 The total cost of SeHCAT in the base-case model was £477 per test, the same as in the IBS‑D or functional diarrhoea model.

3.36 The costs of treating bile acid diarrhoea with bile acid sequestrants were £0.35 per person per day for colestyramine and £2.56 per person per day for colesevelam. These were the same as in the IBS‑D or functional diarrhoea model.

3.37 The cost of the medication for treating diarrhoea in Crohn's disease was £5.76 per person per day. This was estimated using BNF prices, and the average dosages and proportion of people having different types of medication reported by the experts in the previous assessment.

Base-case assumptions in Crohn's disease

3.38 Except for the assumption about colonoscopy, the key assumptions used in the base-case analysis for people with suspected or diagnosed IBS‑D or functional diarrhoea were also applied in the base-case analysis for people with Crohn's disease without ileal resection who have chronic diarrhoea.

3.39 These key assumptions were also applied:

  • Everyone has had a colonoscopy to diagnose Crohn's disease.

  • People who have not had a SeHCAT test, or who have a negative SeHCAT test result, or whose bile acid sequestrant treatment has not worked in the short term, will be offered treatment for diarrhoea in Crohn's disease.

  • Some people with Crohn's disease for whom the diarrhoea treatment works in the short term will have relapses throughout their life.

Base-case results in suspected or diagnosed IBS-D or functional diarrhoea

3.40 The SeHCAT strategy was more effective and less expensive (dominant) than the strategy of offering a trial of bile acid sequestrants. It was also more effective but more expensive than the strategy in which bile acid diarrhoea was not investigated or treated. The incremental cost-effectiveness ratio (ICER) for the SeHCAT strategy compared with this strategy was £9,661 per QALY gained (probabilistic base-case analysis).

3.41 In the short term, the SeHCAT strategy had the lowest rate of colonoscopies and the lowest cost per avoided colonoscopy. It also had the highest rate of treatment response (any type of treatment). The initial costs of the SeHCAT strategy were the highest because of the costs of the SeHCAT test. The results of the deterministic and probabilistic analyses were similar.

Secondary analysis in suspected or diagnosed IBS-D or functional diarrhoea

3.42 In this analysis, it was assumed that colonoscopy was not offered to people:

  • who had no SeHCAT test

  • who had a negative SeHCAT test result

  • whose bile acid sequestrant treatment did not work.

3.43 As with the base-case analysis, the SeHCAT strategy provided the highest QALYs. But in this analysis, it was more expensive than the strategy in which no testing and no bile acid sequestrant was offered and the strategy in which a trial of treatment was offered. The ICER for the SeHCAT strategy compared with the trial of treatment strategy was £21,036 per QALY gained (probabilistic base-case analysis).

3.44 In the short term, as in the base-case scenario, the SeHCAT strategy had the highest rate of treatment response. Initial costs of the SeHCAT strategy were again the highest because of the costs of the SeHCAT test. The results of the deterministic and probabilistic analyses were similar.

Analysis of alternative scenarios in suspected or diagnosed IBS-D or functional diarrhoea

3.45 Robustness of the cost-effectiveness results to alternative model assumptions was considered in several scenario analyses. In nearly all the scenarios, the cost-effectiveness results were similar to the base case, or the SeHCAT strategy produced ICERs around or below £20,000 per QALY gained. In the scenarios in which another strategy could be considered the most cost-effective option (when the probability of colonoscopy is set to 0, and the probability of response to IBS‑D treatment is lower in the SeHCAT arm), the model assumptions were likely to be unrealistic.

Base-case results in Crohn's disease

3.46 In both the deterministic and probabilistic base-case analyses, the SeHCAT strategy was the most cost effective. In the deterministic analysis, it was more expensive but also more effective than the strategy of offering a trial of a bile acid sequestrant. The ICER for the SeHCAT strategy compared with this strategy was £1,727 per QALY gained (deterministic base-case analysis). In the probabilistic analysis, it was both more effective and less expensive than the trial of treatment strategy. In both analyses, the strategy in which bile acid diarrhoea was not investigated or treated was more expensive and less effective than the other strategies. The total costs of all the strategies in the probabilistic analysis were higher than the total costs in the deterministic analysis. This was because it was assumed in the probabilistic analysis that, within the possible cost range, the costs would more often be higher than lower.

3.47 In the short term, the SeHCAT strategy had the highest treatment response rate to any type of medication. But the initial costs were higher than in the trial of treatment strategy because of the costs of the SeHCAT test. Cost per response was the lowest for the trial of treatment strategy.

Analysis of alternative scenarios in Crohn's disease

3.48 Robustness of the cost-effectiveness results to alternative model assumptions and parameters was considered in several scenario analyses. In nearly all the scenarios, the cost-effectiveness results were similar to the base case or SeHCAT-produced ICERs at below £9,500 per QALY gained. In the scenarios in which another strategy could be considered the most cost-effective option, the model assumptions were likely to be unrealistic.

  • National Institute for Health and Care Excellence (NICE)