5 Outcomes

The Diagnostics Advisory Committee (appendix A) considered evidence from a number of sources (appendix B), but primarily used the assessment performed by the External Assessment Group.

How outcomes were assessed

5.1 A de novo model was constructed by the External Assessment Group to assess the clinical and cost effectiveness of SeHCAT in people presenting with symptoms suggesting IBS-D and people with Crohn's disease without ileal resection. A systematic literature review was performed to obtain parameters for the model. For parameters for which no evidence was identified, a survey was carried out to elicit expert opinion.

Modelling approach

5.2 The model consisted of 2 parts: a decision model reflecting the diagnostic and initial treatment phase, and a Markov model to estimate long-term costs and effects. The model was used for both the IBS-D and Crohn's populations but with different parameter values.

5.3 In the population with IBS-D, the decision-tree component of the model used response to treatment, dependent on test outcome, to compare 2 strategies that did not use SeHCAT with 3 strategies that did use SeHCAT. The 3 strategies using SeHCAT had percentage reabsorption at cut-off points of 5%, 10% and 15%, with a normal (negative) test result being higher than the cut-off points. The 2 strategies that excluded the use of SeHCAT were 'no-SeHCAT' (meaning everybody was treated for IBS-D) and 'trial of treatment' with bile acid sequestrants.

5.4 In the population with Crohn's disease, the outline of the decision-tree component of the model was essentially the same as that for the IBS-D model with 2 exceptions. First, if SeHCAT was not used, if the test was negative, or if there was no response to 'trial of treatment', people were treated for Crohn's disease (that is, with therapy to control diarrhoea not caused by bile acid malabsorption). Second, the truncated model (for the population with Crohn's disease) did not include the SeHCAT strategy with a cut-off at 5% because no data on the probability of a positive test result were available for this cut-off point.

5.5 The Markov component of the model allowed the costs and effects of the various strategies to be assessed throughout the person's lifespan. People whose condition responded to treatment entered the Markov model in the 'no diarrhoea' state and people whose condition did not respond entered in the 'diarrhoea' state. Because the model is a lifetime horizon, the third state was death.

5.6 People could move between the 'no diarrhoea' state and the 'diarrhoea' state from 1 cycle to another, stay the same or die. The cycle length used was 6 months. The general Markov component of the model was given parameters according to treatment, which resulted in multiple models based on the same structure. This meant that, in the population with IBS-D, people who received bile acid sequestrants entered the bile acid sequestrant Markov model, and those who received IBS-D treatment entered the IBS-D Markov model. Likewise, in the population with Crohn's disease, people who received bile acid sequestrants entered the bile acid sequestrant Markov model, and people who received treatment for Crohn's disease with chronic diarrhoea entered the Crohn's Markov model.

5.7 Each Markov model included 6 transition probabilities. These were from 'diarrhoea' to 'no diarrhoea' and from 'no diarrhoea' to 'diarrhoea', from 'diarrhoea' to 'death' and from 'no diarrhoea' to 'death', and to 'stay in diarrhoea' or to 'stay in no diarrhoea'.

Systematic review for effect parameter values

5.8 Searches or systematic reviews were carried out to obtain evidence to estimate the parameter values for inclusion in the model. Twenty-four publications of 21 studies met the inclusion criteria and were included in the review. Of these, 20 studies provided data either on the accuracy of SeHCAT in predicting treatment response (in which treatment response was the reference standard), or on treatment effect in groups with positive or negative SeHCAT tests. Out of the 20 SeHCAT studies, 19 studies included people with chronic diarrhoea of unknown cause and 2 studies included people with Crohn's disease and chronic diarrhoea. The remaining study was a randomised placebo-controlled trial of the bile acid sequestrant colesevelam for IBS-D.

5.9 Three studies were identified that contained usable data on the accuracy of SeHCAT in assessing response to treatment in people with chronic diarrhoea. Merrick et al. (1985) estimated the sensitivity of SeHCAT in predicting a positive response to treatment to be 0.67 (95% confidence interval [CI] 0.22 to 0.96) and the specificity to be 0.97 (95% CI 0.85 to 0.10) using a cut-off of 8% for the test. Using a cut-off of 15%, the sensitivity was 1.00 (95% CI 0.54 to 1.00) and the specificity 0.91 (95% CI 0.76 to 0.98). Sciaretta et al. (1986) estimated the sensitivity of SeHCAT in predicting a positive response to be 0.86 (95% CI 0.42 to 0.10) and the specificity to be 1.00 (95% CI 0.541 to 1.000) using a cut-off of 5% for the test. However, only 13 people were included in this analysis. Sciaretta et al. (1987) estimated the sensitivity of SeHCAT in predicting a positive response to be 0.95 (95% CI 0.75 to 0.10) and the specificity to be 0.96 (95% CI 0.80 to 0.10) using a cut-off of 8% for the test.

5.10 Two studies were identified that contained data on the accuracy of SeHCAT for assessing response to treatment in people with Crohn's disease. Neither study presented data for people with a negative SeHCAT test. Moreover, neither study explained why certain people received treatment and others did not.

5.11 For the effectiveness of bile acid sequestrants in treating chronic diarrhoea, 1 controlled clinical trial compared colesevelam with placebo for people with IBS-D. No controlled trials that assessed the effectiveness of bile acid sequestrants in terms of bowel function in people with chronic diarrhoea of unknown cause were identified. In addition to the randomised controlled trial, data on the effectiveness of bile acid sequestrants for treating bile acid malabsorption were also obtained from the 19 studies described above. All 19 studies provided data on the effectiveness of bile acid sequestrants given after a positive SeHCAT test. Only 3 of the studies provided data on the effectiveness of bile acid sequestrants given after a negative SeHCAT test.

5.12 Two studies looked at the effectiveness of bile acid sequestrants in treating diarrhoea in people with Crohn's disease. Neither study presented data for people with a negative SeHCAT test. In addition, it was not clear why certain people were treated and others were not. From both studies, those with a positive SeHCAT test had a response rate of 95% at a cut-off of 5% and 87% at a cut-off of 10%.

Additional model parameters

5.13 Most of the model parameters used represent the probability of events occurring in the care pathway. The model was populated with estimates of these probabilities, which were obtained from the systematic review when data were available, supplemented with expert opinion when objective data could not be found. Based on the data retrieved, a random effects meta-analysis was performed to find a pooled estimate for each of the 3 cut-off values (5%, 10% and 15%).

5.14 The probabilities of a positive SeHCAT test in the population with IBS-D for cut-off points of 5%, 10% and 15% were 0.22, 0.36 and 0.38 respectively.

5.15 For the population with Crohn's disease, there were no studies identified with a SeHCAT 5% cut-off point. The External Assessment Group only estimated the probability of a positive SeHCAT test in this population for cut-off points of 10% and 15%, which were 0.54 and 0.63 respectively.

5.16 In the population with IBS-D, the probability of a positive response to bile acid sequestrants was 0.88 for SeHCAT 5%, 0.76 for SeHCAT 10% and 0.73 for SeHCAT 15%.

5.17 For the population with Crohn's disease, no data were available on response to bile acid sequestrants in people with a positive SeHCAT test. The External Assessment Group therefore assumed that the response rate for this population would be the same as that in people with IBS-D (0.76 for SeHCAT 10% and 0.73 for SeHCAT 15%).

5.18 Rates of response to treatment in people with IBS-D who had not had a SeHCAT test were not found in the literature. A questionnaire was sent to clinical specialists asking what percentage of people would eventually be successfully treated with the usual IBS-D treatment options, and what would be a plausible range for this percentage. Based on the responses received, a pooled mean of a 52% response to IBS-D treatment was calculated.

5.19 In the strategies in which a SeHCAT test was performed, only those with a negative SeHCAT test received IBS-D treatment. This implies that most people who have bile acid malabsorption are no longer part of the group receiving IBS-D treatment. The External Assessment Group assumed that the response rate to IBS-D treatment in the population with a negative SeHCAT test may therefore be higher than in the population who did not have SeHCAT test. In the base-case analysis, an assumption was made that, at cut-off points of 15%, 10% and 5%, the response rates to IBS-D treatment in the SeHCAT-negative population are higher by 10%, 8% and 5% respectively. In the 'trial-of-treatment' strategy, it was assumed that the response rate to IBS-D treatment is the same as in the strategy with a SeHCAT cut-off of 15%.

5.20 For people with Crohn's disease without ileal resection, there were no data in the literature about how often the condition responds to treatment for diarrhoea. The External Assessment Group asked clinical specialists what percentage of people would eventually be treated. Based on responses received, a pooled mean of 62% response to treatment was calculated.

5.21 The External Assessment Group assumed that response to treatment for diarrhoea may be higher in people with Crohn's disease who had a negative SeHCAT test than in the population who did not have a SeHCAT test. No data were available to test this assumption. The External Assessment Group made the same assumption as it did in the population with IBS-D, which is that the response rate is 10% higher than in the population who did not have a SeHCAT test for a cut-off of 15% and 8% higher for a cut-off of 10%.

5.22 The External Assessment Group assumed that there is no excess mortality associated with IBS-D and bile acid malabsorption, and concluded that the overall mortality in the UK population is relevant. Data on mortality were derived from England and Wales Interim Life Tables 1980–82 to 2008–10. In addition, the average age and gender distribution in the model cohort was 47 years and the ratio of men to women was 0.71.

5.23 For the population with Crohn's disease, no reports were found that suggested the chronic diarrhoea caused by the condition leads to excess mortality. However, people with Crohn's disease have a shorter life expectancy compared with the general population. This shorter life expectancy was included in the modelling and no additional reduction was assumed for those people who also have chronic diarrhoea.

5.24 No data were available about the transition probabilities from 'diarrhoea' to 'no diarrhoea' and vice versa. A range of plausible scenarios were used to show the impact of the assumptions on the model estimates of health and cost outcomes.

Quality of life

5.25 For the population with IBS-D, the following utilities were estimated by the External Assessment Group based on multiple studies:

  • for the health state 'diarrhoea', utility was 0.71

  • for the health state 'no diarrhoea' (IBS-D), utility was 0.78

  • for the health state 'no diarrhoea' (bile acid malabsorption), utility was 0.76.

5.26 For the population with Crohn's disease, the following utilities were similarly estimated:

  • for the health state 'diarrhoea', utility was 0.70

  • for the health state 'no diarrhoea' (IBS-D), utility was 0.76

  • for the health state 'no diarrhoea' (bile acid malabsorption), utility was 0.74.

Economic analysis

Costs

5.27 Table 1 shows the cost estimates that were used in the model.

Table 1: Cost estimates used in the model

Description

Cost (£)

Cost per day of IBS-Da medication

0.17

Diet costs per 6 months associated with IBS-D

46.05

Psychological costs per 6 months associated with IBS-D

129.81

Cost per day of BASb medication (IBS-D model)

0.63

Cost of SeHCAT capsule

195.00

Cost for administering SeHCAT test

186.00

Maintenance and service costs of SeHCAT test

0.00

Cost per day associated with health state 'diarrhoea'

0.06

IBS-D medication cost per day associated with health state 'no diarrhoea'

0.17

BAS cost per day associated with health state 'no diarrhoea' (IBS-D model)

0.63

Cost per day of Crohn's medication

1.78

Cost per day of BAS medication (Crohn's model)

0.63

Crohn's medication cost per day associated with health state 'no diarrhoea'

1.08

BAS cost per day associated with health state 'no diarrhoea' (Crohn's model)

0.63

a IBS-D, diarrhoea-predominant irritable bowel syndrome

b BAS, bile acid sequestrant

Cost effectiveness

Results of analysis for IBS-D

5.28 No information was available to estimate the transition probabilities in the Markov models other than for all-cause mortality. Therefore, a range of scenarios were used to show the impact of the assumptions on the model estimates of health and cost.

5.29 In some scenarios, the use of SeHCAT was cost effective. In others, it was not cost effective. The results of the economic evaluation showed that there is considerable uncertainty about the cost effectiveness of SeHCAT testing for people diagnosed with IBS-D.

5.30 An additional analysis was carried out to explore the impact of the cost of additional testing for people whose disease does not respond to IBS-D treatment in all of the scenarios. This additional testing resulted in no change in any of the scenarios in terms of whether SeHCAT testing was found to be cost effective.

Results of analysis for Crohn's disease

5.31 No information was available to estimate the transition probabilities in the Markov models other than for all-cause mortality. Therefore, a range of scenarios were used to show the impact of the assumptions on the model estimates of health and cost.

5.32 Again, in some scenarios, the use of SeHCAT was cost effective. In others, it was not cost effective. The results of the economic evaluation showed that there is considerable uncertainty about the cost effectiveness of SeHCAT testing for people diagnosed with Crohn's disease who have not had ileal resection.

  • National Institute for Health and Care Excellence (NICE)