Teduglutide for treating short bowel syndrome

Short bowel syndrome is a chronic condition with limited treatment options

3.1 Short bowel syndrome (SBS) is a chronic and potentially life-threatening condition characterised by reduced absorption of nutrients, water and electrolytes. SBS is commonly caused by surgery that has been needed to remove abnormal bowel. In adults, this surgery may be needed for a range of conditions, including mesenteric ischaemia, Crohn's disease and radiation enteritis. In children, it is often because of necrotising enterocolitis in premature babies, or other conditions such as volvulus or gastroschisis. Some children can be born with a short bowel. SBS can lead to intestinal failure. This is when the length of intestine remaining means the intestinal functions drops below the necessary level for absorption of nutrients, water and electrolytes. Intestinal failure is classified as type 3 when it is chronic and people need to have parenteral support over months or years. Current treatment for SBS includes parenteral support, in which nutrients and fluids are given intravenously for an average of 10 to 14 hours a day for between 2 and 7 days a week. Most people have parenteral support at home using a permanent intravenous tube. While parenteral support is life-saving, it is very time-consuming, highly complex and its complications can be life-threatening. These include blood infections, blood clots, and kidney and liver failure. Clinical and patient experts both noted that there are currently supply issues with parenteral support, and people with SBS are having to adapt their care to manage this. They also highlighted that administering parenteral support is complicated and challenging for people with SBS and their carers. They emphasised a need for new treatments that offered more normality for those affected by SBS. The committee concluded that SBS is a chronic condition with limited treatment options that can cause further adverse complications.

People would welcome new treatment options for short bowel syndrome that reduce the number of days of parenteral support

3.2 SBS is not only burdensome for the person with the condition, but for their family too. This burden is often linked to the need to use parenteral support. Patient experts confirmed that parenteral support limits normality of all aspects of life, including family, work and social life. Most people with SBS have parenteral support at home, and either administer this themselves or with a carer's assistance. Patient experts highlighted that this places a huge burden on people with SBS and their carers, because the responsibility of doing complicated procedures at home can be overwhelming. They feel that the impact on carers is often overlooked, and that many families have not had a night off from giving parenteral support in many years. Carers also often feel guilt when people with SBS experience complications such as blood infections, because they feel that these would be avoided if they had done their caring duties correctly. Clinical experts highlighted that there is potential for teduglutide to allow some people with SBS to wean off parenteral support completely. This would greatly increase the quality of life of both people with SBS and their carers. They also noted that some people with SBS may be from areas with low socioeconomic status, where parenteral support is particularly difficult to administer. Therefore reducing the need for it would be very helpful. Both clinical and patient experts noted that there is an unmet need to reduce the time spent on parenteral support, to increase quality of life for people with SBS and their carers. The experts emphasised that this was a continuous burden, with little or no respite. They advised that any reduction in the number of nights of treatment would give people respite, which would give them and their family and carers some time to do other activities without the burden of parenteral support. The committee agreed that the impact of parenteral support on people with SBS and their carers was high. It concluded that people would welcome new treatment options for SBS that reduce the number of days of parenteral support.

Teduglutide is likely to be used once people become stable on parenteral support

3.3 Teduglutide's marketing authorisation is for the treatment of SBS in people 1 year and above who are stable after a period of intestinal adaptation. The company's positioning of teduglutide was aligned with the marketing authorisation. This was highlighted as a decision point in the pathway after the adaption phase. The adaption phase is estimated to be up to 2 years in adults but can be longer in children. Clinical experts noted that they would want to use teduglutide for any child who needs routine parenteral support, and they would not want to wait more than 2 years to start treatment. They also highlighted that starting teduglutide might reduce the need for intestinal transplant because transplants would only be indicated for people in whom parenteral support is not working. The committee concluded that teduglutide was likely to be used for people with SBS once their parenteral support needs had been established and stabilised on a regular schedule.

The only appropriate comparator for teduglutide is parenteral support combined with best supportive care

3.4 For people with SBS, the company submission compared teduglutide with established clinical management including parenteral support with best supportive care (antimotility and antisecretory agents, fluid restriction and dietary optimisation). Further surgical procedures can be done after parenteral support stabilisation to attempt to increase the length of remaining bowel that is in direct continuity. However, because these are rarely done in practice, they were not included as a relevant comparator. Intestinal transplant can also be done if all other treatments fail. This is done in some children, but not routinely. Surgery was not considered a comparator in this appraisal. But avoiding the need for a transplant, which has lifelong consequences, could be a benefit of teduglutide. The committee concluded that parenteral support with best supportive care was the only appropriate comparator for teduglutide.

The clinical trial evidence used in the company model is appropriate, but the patient support programme data is less certain

3.5 The key clinical evidence for teduglutide came from 1 clinical trial and its long-term open-label extension study, and 1 non-interventional real-world study in adults with SBS:

The company submitted clinical evidence for adults from an extension to STEPS‑2 and 1 further clinical trial and its extension that were not included in the economic model:

The company decided not to include STEPS‑3 data in its model because the relevant cohort of patients only contained 5 people. 004 and 005 were not included because the results have weak external validity due to the restrictive parenteral support weaning algorithm used as part of the study protocol. The company also submitted evidence from 8 non-interventional real-world studies in adults that was not included in the model.

The company submitted clinical evidence in children from 2 open-label studies and their extensions, but did not include this evidence in the economic model:

No clinical data from studies of children was used in the modelling. The company's justification for this was that the trials of children had small sample sizes and non-continuous treatment across follow-on studies. The ERG was satisfied with the company's choice of clinical evidence used in the model. It agreed that STEPS was of good methodological quality. It highlighted the small sample size of STEPS‑3 and accepted that the weaning algorithm used in 004 and 005 was not closely matched to clinical practice. It also recognised the small sample numbers in C13, C14 and their extensions. The committee concluded that the clinical trial evidence chosen by the company for use in the model was appropriate.

Results from trials of children show that teduglutide reduces parenteral support needs, but they are not generalisable to the NHS

3.6 The company provided efficacy data for children from 2 clinical trials and their long-term extensions. This data was provided to show that the efficacy of teduglutide in children is similar to, and may even exceed, its efficacy in adults. In C14, the primary outcome was the percentage of people who had a 20% to 100% reduction in parenteral volume from baseline at week 24. This was defined as a 'clinical response'. Reduction in days of parenteral support per week was measured as a secondary outcome. C14 found that more children with SBS had a clinical response on teduglutide over the standard care group (69% compared with 11%). Children having teduglutide also had a reduction in days of parenteral support per week from baseline, while those on standard care did not. (‑1.3 days compared with 0 days). The primary outcome for C13 was the reduction in days of parenteral support per week. The results of this are confidential and cannot be disclosed here. Reduction in parenteral support volume was measured as a secondary outcome. Results from C13 also showed reductions in parenteral volume from baseline at 12 weeks. Results from the extension studies are confidential and cannot be disclosed here. C13, C14 and their extensions were not included in the model because they had small sample sizes and non-continuous treatment with teduglutide (see section 3.5). Instead, the company used adult data to inform both the adult and child base cases. The committee concluded that the results for children did indicate that teduglutide has clinical benefits for children. But these were not generalisable to the NHS because of their limitations in study design.

Clinical evidence from adult trials shows that teduglutide reduces parenteral support needs

3.7 In STEPS and STEPS‑2, the primary outcome was the percentage of people with SBS who had a 20% to 100% reduction of parenteral volume from baseline at week 20 maintained to week 24. This was defined by the company as a 'clinical response'. A key secondary outcome was the change in days per week of parenteral support from baseline. Both STEPS and STEPS‑2 assessed the safety of teduglutide. Outcomes considered from PSP were a 20% to 100% reduction in parenteral support volume from baseline ('clinical response') and change in days per week of parenteral support from baseline. The ERG noted the definition of 'clinical response' differed slightly between STEPS and PSP but did not feel that it would affect trial results. Both clinical and patient experts agreed that reduction in days on parenteral support per week is more valuable to patients than reducing parenteral volume. People in the STEPS trial had an average number of days on parenteral support of 5.6 and 5.9 days (teduglutide and placebo arms respectively) before treatment. After treatment, significantly more people with SBS had a 1‑day or more reduction in weekly parenteral support on teduglutide over placebo (53.8% compared with 23.1%). The results from PSP are confidential and cannot be disclosed here. The committee concluded that the clinical evidence indicates that teduglutide reduces parenteral support needs in adults with SBS.

The weaning algorithm in STEPS may have affected generalisability of the results in both arms

3.8 People with SBS taking teduglutide can potentially reduce parenteral support and gradually move onto an oral diet through a process known as weaning. The STEPS trial used a weaning algorithm to decide if people with SBS should have their parenteral support reduced. A feature of the STEPS results is the apparent efficacy of placebo, with 23.1% of the placebo arm reducing their days of parenteral support by 1 day or more. The company commented that this is unrealistic, and that reductions in days of parenteral support per week in the placebo arm were unlikely to be because of improved intestinal function. It advised that people entering STEPS had parenteral support optimisation and stabilisation before starting teduglutide or placebo. So, it stated that any differences in parenteral support over time in the placebo arm were not attributable to further optimisation of care. It suggested that the observed placebo effect is instead a result of the strict weaning algorithm used in the trial that was solely based on urine output. The ERG noted that the company's argument was plausible, and that changes in parenteral support do not rely on urine output alone in clinical practice. The company continued by stating that people in the placebo arm lost weight over the course of the trial and this indicates that they were not having adequate parenteral support. The committee queried whether this placebo effect reflected clinical practice. The clinical experts explained that in clinical practice clinicians do not know urine output on a day-to-day basis, and so weaning is less precise than in STEPS. They added that the weaning algorithm relying entirely on urine output was unrealistic and that maintaining weight should also be part of the weaning criteria. They clarified that while it may be possible for adults with SBS to reduce their parenteral support while having current standard care, this would only be for a small period and would not be sustainable if medical advice was followed correctly. They added that it is highly unlikely that an adult with SBS would reduce their parenteral support needs without further surgery. This means that the placebo effect seen in STEPS did not reflect clinical practice. Conversely, the company also commented that the weaning algorithm used in STEPS in the teduglutide arm underestimated the extent to which parenteral support could be reduced, thereby underestimating the effect of teduglutide. The company's view was that the strict nature of this algorithm was therefore likely to underestimate the relative treatment effect of teduglutide compared with placebo. The committee discussed if it would be likely for teduglutide to allow some people with SBS to stop needing parenteral support. Clinical experts commented that they would expect some people to come off parenteral support entirely, and that they have had experiences in practice when that has been the case. They stated that some children may be able to wean off parenteral support naturally as their bowel matures and they grow. But teduglutide allows them to improve faster and avoid both clinical and social challenges associated with parenteral support. The committee considered whether the absolute or relative benefits of teduglutide were important to consider when assessing the clinical effectiveness of teduglutide. The ERG did scenario analyses to explore this but agree with the company's arguments relating to the weaning algorithm. The ERG also received clinical feedback that people on standard care were unlikely to reduce their parenteral support. Also, the weaning algorithm constrained how quickly people were able to wean off parenteral support if they improved on teduglutide. The ERG highlighted that while the weaning algorithm is restrictive and may not reflect clinical practice, it was applied to both arms of the STEPS trial. Therefore, the internal validity of the results could be considered robust, but the absolute effects of teduglutide and placebo may not be valid. The committee was unable to conclude whether the company's and ERG's interpretation of the trial (that the results of the teduglutide arm were underestimated while the results of the placebo arm were overestimated) was valid. This is because it is aware that placebo-controlled trials are usually appropriately designed to determine the true treatment effect of a new drug. It concluded that the weaning algorithm may have affected generalisability of the results from both arms of STEPS, which made the true relative treatment effect of teduglutide compared with placebo uncertain.

The frequency of adverse events is broadly similar between teduglutide and placebo for adults

3.9 The company provided safety evidence for adults, pooled from STEPS, STEPS‑2, 004 and 005. The ERG found that the pooling of safety data from these trials was appropriate. Adults on teduglutide most commonly reported abdominal pain (38.5% compared with 27.1%), gastrointestinal stoma complications (37.8% compared with 13.6%), upper respiratory tract infections (27.5% compared with 13.6%) and abdominal distension (16.5% compared with 1.7%). The frequency and severity of these adverse events were broadly similar between the people having teduglutide and placebo, except for abdominal distension. The company commented that the observed adverse events were likely to be because of pro-absorptive and intestinotrophic effects of teduglutide, insufficient parenteral support weaning or the underlying nature of SBS. The ERG accepted this reasoning following advice from clinical experts. The committee concluded that the overall frequency and severity of adverse events resemble those of the placebo group.

The safety profile of teduglutide in children is similar to adults

3.10 The company also provided safety data for children, pooled from C13, SHP633‑303, C14 and SHP633‑304. Children most commonly reported vomiting (51.7%), pyrexia (43.8%), upper respiratory tract infection (41.6%), cough (33.7%) and device-related (central venous catheter) infection (29.2%). The ERG commented that the safety profile for teduglutide in children is like that seen in adults. Upper respiratory events, pyrexia, vomiting and catheter complications (including infections) were more common in children compared with adults. The company stated that these adverse events would be expected to be more frequent in children. The committee commented that children are often admitted to hospital with catheter complications, because frequent diarrhoea can mean it is difficult for carers to always keep the catheter completely clean. The committee concluded that the safety profile of teduglutide is similar for both adults and children.

The estimation of health-state transition probabilities within the model is a source of uncertainty

3.11 The economic model was developed using a Markov structure, comprising 9 health states defined by the days of parenteral support per week (from 7 days to parental support independence or to death). The company included a treatment stopping rule so that modelled teduglutide use would reflect its use in clinical practice as closely as possible. The summary of product characteristics recommends that treatment should be stopped if there is no overall improvement in the condition. It recommends that adults should have an evaluation after 6 months, with treatment continuation being reconsidered if there is no treatment benefit by 12 months. The model reflected this by assuming that those who had not had a reduction of at least 1 day of parenteral support per week at 12 months, compared with baseline, stop teduglutide. Once treatment is stopped, they immediately reverted to their baseline parenteral support state before teduglutide. Teduglutide is modelled to affect both cost and quality-adjusted life years (QALYs):

Costs:

QALYs:

To calculate transition probabilities for teduglutide, the company pooled clinical data from the teduglutide arms of STEPS and STEPS‑2 and data from the PSP when estimating the reductions in parenteral support for the teduglutide group. It explained that it took this approach rather than using the relative treatment effect from the trial because the weaning algorithm in STEPS and STEPS‑2 underestimates parenteral support reductions for teduglutide (see section 3.8). The company supported this claim by doing an analysis comparing the percentage of people stopping parenteral support entirely while taking teduglutide between STEPS, PSP, and a combination of other real-world studies. The company also assumed that there is no change in parenteral support in the standard care arm and applied the STEPS baseline parenteral support requirement over the time horizon in the standard care arm of the model. The reasoning for this was that people need to have a stable parenteral support requirement before teduglutide, and reductions in parenteral support would not be expected in clinical practice without teduglutide (see section 3.3). The ERG confirmed that the model structure is appropriate. It advised that the company's explanation for underestimation of teduglutide effectiveness in the STEPS and STEPS‑2 trials was plausible. But any comparison of effects between observational studies and randomised controlled trials should be interpreted with caution. The committee expressed some concern around the company's methodology for estimating transition probabilities. This was specifically around breaking randomisation when pooling the real-world and teduglutide arm trial data while disregarding the placebo effect seen in STEPS. The ERG stated that it had done scenario analysis exploring the relative treatment effect of teduglutide from the STEPS data alone. This had a substantial upwards impact on the incremental cost-effectiveness ratio (ICER). But, because it received clinical expert feedback that people having standard care would not be expected to reduce their parenteral support needs, the ERG considered this scenario to be conservative and did not incorporate it into its base case. The committee concluded that the company's use of STEPS, STEPS‑2 and PSP data to model health state transitions is uncertain and may bias results in favour of teduglutide.

The assumptions and data sources are very similar between the models for adults and children

3.12 No clinical study data from studies of children was used in the modelling (see section 3.5). The company considered that children would gain more benefit from teduglutide and so using adult data would give a conservative cost-effectiveness estimate for children. Clinical experts confirmed that children have more possibility for intestinal development and their SBS may be more responsive to treatment. The committee noted that offering teduglutide to children would reduce the likelihood of repeated line infections, because it is more difficult to avoid contamination of the catheter in children. The company decided to model the 2 populations separately. When modelling for children, treatment is allowed to continue beyond the age of 18. After the age of 18, adult model assumptions are applied to this population. The ERG agreed that the 2 populations should be considered separately. The assumptions and data sources are very similar between the models for adults and children. The differences in model assumptions to reflect children included:

The cost-effectiveness results for children are much more favourable than for adults. The ERG clarified that this is because of the younger starting age and longer time horizon in the model for children. Teduglutide also reduces the costs associated with parenteral support (see section 3.11), and these 2 attributing factors mean that QALYs and cost benefits accrue for longer in the model. The committee concluded that the difference seen between the ICERs for adults and children are feasible.

The committee needs further analyses and justification for the choice of starting age and efficacy in the adult base case

3.13 The company modelled adults and children separately. To do this, they used different starting ages for both populations. The starting age for the adult base case was 50 years. The company's justification for using this age was that it was the average age of the STEPS trial population. The committee commented that it was unsure how much this age reflected the average age of adults with SBS in the real world. The committee also had ongoing concerns about not using any of the control arm data in the model. It appreciated that there are potential issues with the weaning algorithm in the trial, but better justification of the exclusion of the placebo data in the adult analysis was needed. As a result, the committee concluded that it wants to see further justification for the company's choice of starting age, a set of scenarios with different plausible starting ages, and an analysis that uses the placebo arm data from STEPS rather than assuming a steady state for those not on teduglutide.

The company's choice of the log-normal distribution to extrapolate overall survival in adults is acceptable

3.14 There were very few deaths during the STEPS trial, so the company explained it was not able to extrapolate overall survival for adults from this data. Instead, overall survival in the adult model is based on extrapolation of published Kaplan–Meier data for people with SBS on long-term parenteral support (Salazar et al. 2021). An alternative source of survival data is used for modelling overall survival in children (see section 3.12). The company chose to use the log-normal curve in its adult base case, based on both statistical fit and predicted hazard functions compared with Salazar et al. 2021. The survival probabilities were adjusted using life tables for England from the Office of National Statistics to ensure extrapolations did not cause the mortality rate to fall below that of the general population. The ERG questioned if it was plausible for a proportion of people with SBS on long-term parenteral support to have the same mortality as the general population. The ERG explored this using an exponential curve because this retains mortality at a higher level than general population for longer. However, it used the log-normal curve in its base case after accepting that it provides a better fit to the data than the exponential curve. Clinical experts commented that people with SBS have near normal life expectancy once weaned off parenteral support. The committee therefore considered the company's assumptions around life expectancy for people on parenteral support to be appropriate. The committee concluded that the log-normal distribution was acceptable for extrapolating overall survival.

The company's approach to modelling complications associated with parenteral support is acceptable for decision making

3.15 Parenteral support can cause complications including intestinal failure associated liver disease (IFALD) and chronic kidney disease (CKD). In the company model, the risk of developing IFALD and CKD increases with days of parenteral support per week. The company explained that teduglutide reduces the incidence of these complications by reducing the required number of days of parenteral support per week. Clinical experts agreed that teduglutide should reduce the risk of IFALD and CKD by reducing days of parenteral support and improving intestinal fluid absorption. The company did not model a mortality risk for IFALD and CKD. This was because clinical feedback stated that deaths because of these conditions in SBS are very rare, and the real-world data used to inform mortality already includes death from complications. The ERG highlighted that a lack of structural link in the model between the proportion of people with SBS living with complications and risk of death may lead to overestimation of IFALD and CKD over time. It stated that this could cause bias in both directions by overestimating costs and utility losses related to living with IFALD or CKD, and failing to capture the small, expected survival benefit for teduglutide. The committee concluded that the company's approach to modelling these complications was acceptable for decision making.

The company's approach to modelling adverse events is appropriate for decision making

3.16 The company considered 2 time periods when modelling adverse events, based on STEPS (first 6 months) and STEPS‑2 (after 6 months). When modelling adverse events in the teduglutide arm, observed adverse events in the teduglutide arm of STEPS were used to estimate a rate per person for the length of STEPS (6 months). This was then divided by 6 to get a per-cycle rate of individual adverse events. A similar method was used when modelling adverse events in the teduglutide arm from 6 months onwards, except the rate per person for the length of STEPS‑2 (24 months) was divided by 24 to obtain a per-cycle rate of individual adverse events. When modelling adverse events in the standard care arm, observed adverse events in the placebo arm of STEPS were used. These analyses showed that adverse event rates decreased substantially from 6 months onwards in people taking teduglutide (0.98 adverse events per cycle per person to 0.43 adverse events per cycle per person). The company stated that this was because people became more tolerant to teduglutide as treatment progressed. People on teduglutide also needed less parenteral support so any adverse events relating to this would be reduced in the teduglutide arm compared with the standard care arm. The ERG highlighted that there was no standard care safety data available beyond 6 months to validate these results. However, it was satisfied with the company's explanations of reduced adverse events in the teduglutide arm. Clinical experts agreed that they would expect the rate of adverse events to decrease over time with teduglutide, but only in people who had reductions in days of parenteral support per week. The committee concluded that the company's approach to modelling adverse events was appropriate for decision making.

The company's approach to modelling the incidence and costs of line sepsis is appropriate for decision making

3.17 Another complication of parenteral support is line sepsis. The company model assumes that health-state costs related to line sepsis increase with the number of days of parenteral support per week. The company explained that time spent on a catheter is recognised as being linked to sepsis incidence. Days of parenteral support per week is equivalent to days when the catheter will need to be manipulated to administer the treatment, and so it is appropriate to vary rates of line sepsis according to this in the model. The ERG commented that its understanding of catheter days is the number of days a catheter is inserted for access, not the number of days it is used for delivering parenteral support. However, it supports the plausibility of a relationship between the number of days of parenteral support per week and risk of line sepsis. Clinical experts agreed that they would expect the risk of line sepsis to be greater when parenteral support is administered more frequently. Also, they would expect some people on teduglutide to not need a central venous catheter at all, meaning related complications would be reduced. The committee concluded that the company's approach to modelling the incidence and costs of line sepsis is appropriate for decision making.

The health-state utilities from STEPS do not reflect the quality of life of people with SBS

3.18 When considering impact on quality of life, patients and clinical experts both agree that reduction in days of parenteral support per week is the most relevant outcome of teduglutide treatment. STEPS used the SBS-quality-of-life scale, a disease-specific tool, to measure quality of life in adults with SBS. People with SBS were asked to rate the influence of SBS on 17 items, including general wellbeing, leisure activities, work life and social life. The company stated that the health-related quality-of-life data collected in STEPS did not show statistically significant quality-of-life differences between teduglutide and standard care after 24 weeks of treatment. The company also argued that the STEPS health-related quality-of-life data showed an inconsistent relationship between days of parenteral support and health-state utilities. When the quality-of-life data was stratified and mapped to utility values, the highest utility values were seen for 4 days of parenteral support per week. The company explained that this lacked face validity because there was no gain in quality of life for fewer days per week of parenteral support. As a result, the company used values used in health-state vignettes (Ballinger et al. 2018) instead of the quality-of-life data from STEPS in its base case. It also assumed that carer utilities are related to days of parenteral support per week. The ERG accepted the company's use of the vignette utilities, but explored uncertainty through various scenario analyses. It stated that the company's approach may exaggerate the quality-of-life benefits from reduction in days of parenteral support per week, However, clinical and patient experts agreed that a reduction of even 1 day of parenteral support per week can have a huge impact for people with SBS. This is because it allows for respite and gives time for normal activities for the person and their family and carers. The committee noted that using vignettes instead of trial data does not meet the NICE reference case. However, it agreed with the company and ERG that the use of vignette utilities was justified in this case.

The company's approach to estimating carer disutility adds uncertainty to the cost-effectiveness results

3.19 Both adults and children with SBS commonly need caregivers for help with day-to-day tasks, complex medical procedures and emotional support. Carer utilities are linked to days of parenteral support per week in the modelling. The company's adult base case assumes adults will have 1 carer, while its base case for children assumes 2 carers based on the assumption that the child's parents would act as carers. Patient experts emphasised the challenging experience of being a carer for a person with SBS. They highlighted the amount of time taken to provide caregiving duties as well as the impact of the high responsibility and emotional burden of keeping people with SBS alive and well. They confirmed that the carer role for somebody with SBS had a huge impact on the carer's quality of life. Clinical experts confirmed that the expectation of carers was high, and they were often formally trained to be able to undertake care that was usually only done in hospitals. The ERG accepted the company's approach to modelling carer disutility. But, it did specify that the carers utilities derived from the UK caregiver survey do not provide support for an association between days of parenteral support per week and carer health-related quality of life. Clinical and patient experts highlighted that reduction in days of parenteral support per week can have a huge impact on carers of people with SBS. The committee raised concerns that the company may have overestimated carer disutility by assuming children would have 2 carers rather than 1, which would favour teduglutide. It also questioned whether it was appropriate for all adults to have a caregiver. It concluded that the company's approach to estimating carer disutility added uncertainty to the cost-effectiveness results.

Using the mean price of home parenteral nutrition available on the NHS is appropriate for decision making

3.20 Parenteral support is provided by the NHS through the home parenteral nutrition (HPN) framework. The company estimated the resource use of HPN (consisting of parenteral support bags, catheter lock solution [Taurolock] and costs for delivery and nursing) in its submission based on resource use studies for both adults and children. It obtained prices using publicly available sources and expert input. There are several HPN providers in the HPN framework, each with different prices for the various components of HPN. These prices are confidential so cannot be disclosed here and were not available to the company. According to the NICE methods guide, the price used should be transparent to the NHS and nationally available. When commercial discounts are to be considered, the lowest nationally available tender price should be used. Feedback from NHS England was that choosing the lowest cost HPN provider was unlikely to reflect the price paid across the NHS. The ERG provided ICERs using the lowest cost HPN provider, highest cost HPN provider and the mean price of all HPN providers to explore uncertainties around the true price of HPN in the NHS. When doing this, the ICERs ranged from cost-saving to cost-ineffective. The committee would have preferred for a weighted average of the different provider costs to have been used based on market share data, but noted that this was not available. A patient expert highlighted that parenteral support provision and delivery is a very complex area and will differ according to individual needs. They also noted that supply issues add to the complexity. The committee considered the cost of parenteral support to be highly uncertain and noted the large impact on the cost-effectiveness results. It concluded that using the mean price of HPN was likely to be most appropriate for decision making because it is unlikely that the lowest HPN price would be accessed by the entire population with SBS. It also concluded that it would consider the highest and lowest cost HPN providers in scenario analyses.

Concomitant medication resource use and costs are overestimated in both the company's and ERG's base case

3.21 When on parenteral support, people with SBS often take numerous concomitant medications, including proton pump inhibitors, antimotility agents (such as loperamide and codeine), fragmin and ondansetron. The company estimated the resource use of these concomitant medications following expert discussion and took their costs from the BNF. The ERG also provided scenario analyses exploring different dosing regimens and formulations for the concomitant medications in response to feedback from clinical experts before the committee meeting. During the committee meeting, clinical experts provided clarification around the resource use of concomitant medications for people on parenteral support in clinical practice:

The clinical experts also confirmed that most concomitant medications are prescribed in primary care. Only intravenous proton pump inhibitors, ondansetron and Taurolock are available as secondary care prescriptions. The committee noted that there will be cost implications of this because of the different prices available to primary and secondary care providers. The company's base case used higher dosing frequencies and different drug formulations compared with clinical practice. The ERG's base case differed to the company's base case in terms of assumptions surrounding associated medications. However, while the ERG's scenario analyses explored uncertainties around concomitant medication resource use, it overestimated the use of concomitant medicines in clinical practice, and therefore the costs (which in the model offsets some of the costs of teduglutide). The ERG confirmed that ondansetron, intravenous proton pump inhibitors, codeine by injection and fragmin were major drivers of the cost-effectiveness results. The changes seen between clinical practice and the ERG's base case would likely have an impact on the ICER. The committee considered that neither the company's nor the ERG's base case accurately reflected the use of concomitant medications in the NHS, and accounting for this would substantially increase the ICER. It concluded that concomitant medication resource use and costs are overestimated in both the company's and ERG's base cases.

There are outstanding uncertainties associated with the cost-effectiveness estimates

3.22 NICE's guide to the methods of technology appraisal notes that judgements about the acceptability of a technology as an effective use of NHS resources will take into account the degree of certainty around the ICERs. The committee will be more cautious about recommending a technology if it is less certain about the ICERs presented. The committee noted uncertainties with the model inputs, namely:

The committee considered that uncertainty relating to concomitant medication resource use and costs may be resolved through further clarification and analyses from the company. The committee also clarified that the uncertainty around the generalisability of clinical-effectiveness results from both the teduglutide and placebo arms of STEPS would mean the ICER would have to be comfortably within the range of cost effectiveness before recommending teduglutide. The other factors contributed to the level of uncertainty around the ICERs and the committee took this into account in its decision making.

Teduglutide is likely to be cost effective in children with SBS

3.23 The company's and ERG's cost-effectiveness estimates for teduglutide in children with SBS were well below what NICE normally considers an acceptable use of NHS resources. Because of confidential commercial arrangements for teduglutide and comparator treatments, the cost-effectiveness results cannot be reported here. The committee considered that the ICERs for children would increase if the assumptions relating to concomitant medication resource use and costs were updated to be aligned with NHS practice. However, even when taking this into account together with the other uncertainties (see section 3.22), the committee concluded that it was unlikely the ICER for children would be above £20,000 per QALY gained and highly unlikely to be above £30,000 per QALY gained. Therefore, teduglutide is recommended when treatment is started in children and young people aged 1 to 17, and this treatment can continue when they turn 18.

Teduglutide is not cost effective in adults with SBS when considering the current analysis

3.24 The cost-effectiveness estimates from the company and the ERG for teduglutide in adults with SBS were below what NICE normally considers an acceptable use of NHS resources. Because of confidential commercial arrangements for teduglutide and comparator treatments, the cost-effectiveness results cannot be reported here. However, the ICER was highly dependent on the costs related to the concomitant medications given alongside parenteral support. If these were correctly incorporated into the model, the ICER was likely to increase to levels above an acceptable use of NHS resources (see section 3.21). The other areas of uncertainty (see section 3.22) also have the potential to further increase the ICERs to an unknown degree. Therefore, teduglutide is not considered cost effective for adults with the current analyses.

For adults, no analyses reflect the committee's preferred assumptions

3.25 Because of confidential commercial arrangements, the cost-effectiveness results cannot be reported here. However, none of the company's nor the ERG's analyses reflected the committee's preferences. The committee considered that further analysis was needed to understand the impact of uncertainty on the economic analysis. It would prefer to see an updated base-case analysis for adults that aligns the concomitant medications with NHS practice. It would also like to see further scenario analyses considering different starting ages in the adult model, alongside justifications for the chosen starting age, and analyses that uses placebo arm data from STEPS rather than assuming a steady state for those not on teduglutide.

There may be additional benefits of teduglutide that are not captured in the cost-effectiveness analysis

3.27 The company considers teduglutide to be innovative because it represents a step change in the treatment of SBS and that existing treatment (parenteral support) only manages the symptoms of the disease. The ERG commented that the economic base case for teduglutide hinges on an evidence base with many uncertainties that cannot easily be resolved given the rarity and heterogeneity of SBS. The committee highlighted that there may be an uncaptured benefit to teduglutide in that it may prevent the need for intestinal transplant when parenteral support has not worked. The ERG noted that this was an important point to consider, but it was not possible to model this because of a lack of data on teduglutide's ability to reduce the need for intestinal transplant. The company stated that its base case for children is conservative because children may benefit more from teduglutide (see section 3.6). But the extent of this benefit is uncertain and may be countered by the fact that some children on standard care also have the potential to reduce their parenteral support needs (see section 3.8). The committee concluded that there may be additional benefits of teduglutide that are not captured in the cost-effectiveness analysis. But the extent of these benefits is unclear because of uncertainties in the evidence.

Teduglutide is recommended when treatment starts in children and young people aged 1 to 17

3.28 Teduglutide is recommended for use in the NHS for treating short bowel syndrome in children and young people aged 1 to 17, continuing into adulthood. The cost-effectiveness estimates for children were uncertain and did not reflect the committee's preferred assumptions about concomitant medications. But they were highly likely to remain below what is considered an acceptable use of NHS resources even when accounting for this and other uncertainties.

The committee is not able to make a recommendation for teduglutide in adults and requests further analysis

3.29 The committee was not able to make a recommendation for teduglutide for treating short bowel syndrome in adults. The base-case cost-effectiveness estimates for adults were much higher than those for children. They were also uncertain and highly dependent on the costs of concomitant medications, which were not accurately captured in either the company's or ERG's base case. Applying the committee's preferred assumptions was considered likely to result in cost-ineffective ICERs. The committee requests further analyses to explore this uncertainty.