HEALTHTECH PROGRAMME

TAVI is primarily used in people who are at high risk for heart surgery or for whom surgery is inappropriate. But it is increasingly considered as a treatment option for people who are at low or intermediate surgical risk following a position statement by NHS England, 2023. In response to this statement, the Society for Cardiothoracic Surgery in Great Britain & Ireland and the Royal College of Surgeons submitted a letter stating that the policy was not clinically appropriate and could increase patient risks if subsequent surgery was needed.

Clinical experts advised that the decision about which type of transcatheter heart valve to use is usually made by an interventional cardiologist and largely depends on the clinical characteristics of the person with aortic stenosis. The decision may also be related to the clinician's experience with a particular transcatheter heart valve or the range of valves that are locally available. Most NHS trusts will have access to at least 1 self-expanding and 1 balloon-expanding valve. The clinical experts explained that the anatomy of the valve being replaced, the level and distribution of calcium and the person's surgical risk are particularly important and can be strong predictors of clinical outcomes. The committee heard that most people with aortic stenosis (that is, more than 50%) would not need a specific transcatheter heart valve and a wide range could be used.

The committee noted that the valves being assessed vary in their indications (see section 2). Clinical experts stated that most people having TAVI are at high surgical risk, with a tricuspid valve anatomy. Also, most TAVI procedures are done to replace a native aortic valve. The committee noted that all the valves in the assessment are indicated for this population. A clinical expert stated that transcatheter heart valves are sometimes used outside of their indication when this is considered clinically appropriate.

The committee acknowledged the wealth of evidence on the clinical performance of transcatheter heart valves and the relative treatment effectiveness of TAVI compared with surgery. But, it noted that there was little comparative evidence between different transcatheter heart valves and between companies. The EAG explained that it considered the UK TAVI registry (see section 3.8) the strongest source of clinical evidence. This is because it provided recent data from the UK, and allowed the assessment of multiple valves, while adjusting for recorded confounders. The EAG explained that 4 available network meta-analyses were unreliable because of differences in patient characteristics in the included studies. This can lead to a breach of the assumption of transitivity (that a patient could have been randomised to any of the study arms included in the analysis). The EAG also highlighted that the network meta-analyses included valves that had been withdrawn from market or were no longer available for purchase.

The committee and companies queried why randomised controlled trial (RCT) data was not considered and noted that it could provide important information, especially about long-term outcomes. The EAG noted that 1 non-inferiority RCT comparing multiple transcatheter heart valves and 4 network meta-analyses (that included RCTs) were included in the evidence summary. But it explained that most RCTs identified during the evidence review included surgery as a comparator, and often included older generation valves or valves no longer available in the NHS. The EAG explained that because of the recent changes in the populations having TAVI and surgery in the NHS, evidence from an RCT where surgery is a comparator may not reflect current care. The committee queried whether published evidence from countries other than the UK was generalisable to the NHS. An expert adviser said that international evidence is broadly generalisable to the NHS. But a specialist committee member noted that the level of TAVI use in the UK is lower than in many other higher-income countries and that the populations may be different in terms of the proportions of people at different surgical risks.

The UK TAVI registry is a mandatory registry that collects information for all TAVI procedures across England, Wales and Northern Ireland. The UK TAVI registry was created to define the characteristics and clinical outcomes in people having TAVI, regardless of technology or access route, in every centre doing TAVI in the UK. The registry is managed by the National Institute for Cardiovascular Outcomes Research with clinical direction and strategy provided by the British Cardiovascular Interventional Society and the Society for Cardiothoracic Surgeons. The committee agreed that the dataset reflects clinical practice in the NHS, but that it has limitations. The EAG was able to collate data from 7,409 procedures where the TAVI device could be identified. It explained that the registry only contains data on in-hospital outcomes and that the available data only included valves from 4 companies. Clinical experts stated that several clinically important patient characteristics (see section 3.3) are not recorded in the UK TAVI registry, and that it was not designed to make direct valve comparisons. The EAG also highlighted that many fields in the registry were poorly completed.

To address the lack of long-term data in the UK TAVI registry, the EAG linked the data to Hospital Episode Statistics (HES) based on the NHS trust, age and sex. The EAG explained that the linked dataset censored 381 procedures from Wales and Northern Ireland and that no match was found for 520 procedures. This resulted in 6,508 matches, of which 6,270 were procedures to replace a native aortic valve. The committee agreed that the linkage was robust. But it noted that the longest follow up within the linked dataset was 31 months, so the results could not be considered to fully represent long-term outcomes. Also, the EAG's decision to only use cases with no missing data markedly reduced the sample size (to 3,917 from 6,270 records in the UK TAVI registry).

The committee concluded that the UK TAVI registry data did not capture enough detail to provide reliable estimates of relative efficacy between valves. Multivariate analysis of the linked dataset showed statistically significant differences in the odds of experiencing in-hospital stroke, in-hospital aortic regurgitation and in-hospital permanent pacemaker implantation between some of the transcatheter heart valves. These differences were not seen in outcomes after discharge from hospital. The committee noted that the analysis of the linked dataset was limited because it was not possible to adjust for clinically important patient characteristics that are not recorded in the UK TAVI registry or HES (see section 3.3). So, it was not possible to conclude whether the observed outcomes in the analyses were because of features of the valves or the clinical characteristics of the people with aortic stenosis. The EAG explained that the results are also confounded by how much a valve has been used in the NHS during the study period. This leads to higher uncertainty for those valves that have been used less frequently. A specialist committee member explained that the most commonly used valves may be more likely to be used for people who can have a transcatheter heart valve from any company, and who are less likely to experience complications. But it is also possible that cardiologists may prefer to use the transcatheter heart valve they are most familiar with for people with more complex anatomy who are more likely to experience complications. The committee acknowledged that the differences in how much each valve is used in the NHS can have a significant impact on the validity of the results.

The committee considered evidence on device-specific short and long-term outcomes from a number of peer-reviewed studies identified by the EAG. This included 4 network meta-analyses comparing multiple valves, 4 studies comparing multiple valves while adjusting for confounders, as well as a number of additional observational, non-randomised, single-arm and retrospective studies. The committee noted that the published evidence assessed by the EAG was not identified by a systematic search. The EAG acknowledged that this approach can lead to bias, but explained that this was a pragmatic choice given the abundance of published evidence, intended to address gaps in the real-world evidence.

Five transcatheter heart valves (Allegra, Evolut FX, Hydra, Myval Octacor and Trilogy) had no data in the UK TAVI registry because they were new to the NHS Supply Chain framework at the time of assessment. The committee noted that the published evidence identified by the EAG presented the best available evidence for these valves, but it acknowledged that it was sparse and subject to bias and limitations.

It was not clear in the clinical evidence whether there are differences in clinical effectiveness between different companies' transcatheter heart valves due to incremental innovations between the valves. But, the committee acknowledged that clinical equivalence between companies' valves could not be assumed. The committee recalled that for most people with aortic stenosis, many of the available valves could be used (see section 3.3). So, it is likely that for those people the valves are clinically comparable.

The committee queried whether it is appropriate to assume clinical equivalence between generations of a valve from the same company. Clinical experts commented that it was inappropriate to present results of the registry analysis separately for different generations of valves from the same company, because they considered these largely equivalent. A specialist committee member and company representatives explained that usually newer generations make incremental improvements and that these are often small changes which would not affect outcomes, such as durability. The EAG highlighted that clinical studies between generations typically have short follow up and do not provide long-term data, with the longest follow up being 1 year. It stated that, since differences in clinical outcomes between generations have been seen in the literature, long-term equivalence could not be assumed. A specialist committee member stated that it should not be assumed that a newer valve is non-inferior if the differences between valves are substantial (for example, changes in the leaflet tissue). This was based on the committee member's experience with surgical heart valves. The committee concluded that it is likely that newer generations of valves work as well as previous generations, but that this cannot be assumed.

The EAG adapted the economic model used in the economic evaluation for NICE's guideline on heart valve disease presenting in adults: investigation and management (from now, NG208), to allow for direct comparisons of different transcatheter heart valves. The committee considered the structure and assumptions of the EAG's economic model and agreed that it was an appropriate representation of clinical practice in the NHS.

The committee concluded that the clinical inputs to the economic model had limitations, because they relied on the results of the multivariate analysis of the UK TAVI registry, which were highly uncertain (see section 3.10). The transition probabilities between health states in the model were calculated from the event rates in the linked dataset. The committee recalled the bias and limitations associated with this dataset and agreed that this leads to significant bias in the results of the economic model.

Expert advisers and the companies suggested that data from RCTs could be used to inform the economic model, especially for long-term outcomes. The EAG explained that using data from different sources for different outcomes is likely to give biased results, because they will not account for all clinically important characteristics. The EAG also noted that although longer-term data is available from RCTs, it is restricted to comparisons of older generation valves, often with surgery as a comparator. The EAG highlighted that simultaneously sourcing all clinical inputs was a significant methodological advantage of using the UK TAVI registry data. It also noted that using different sources for clinical inputs was cited by stakeholders as a limitation of the economic evaluation in NG208.

Some companies have 'added value' arrangements with NHS Supply Chain, in which part of the cost of the valve is returned to be spent on related items or staff, based on the number of valves purchased. The committee concluded that it was appropriate to account for these 'added value' arrangements in the valve cost, but acknowledged that changes in the volume of use could affect the effective price of some valves. It highlighted that the price variation between the valves after the 'added value' was accounted for was smaller than the variation between the list prices. It also noted that the resources returned through 'added value' agreements can only be spent on structural heart-related products or services at the NHS trust level. The committee heard that analyses from the economic evaluation done for NG208 indicated that a transcatheter heart valve would have to cost £14,800 or less for the procedure to be cost effective for all surgical risks. Most of the transcatheter heart valves currently available in the NHS are above this price at both their list price and after 'added value' agreements have been accounted for.

The committee concluded that the model results were too uncertain to determine whether there were differences in the cost effectiveness of the transcatheter heart valves. The EAG presented the results of the economic evaluation in terms of net monetary benefit including the central value and the 95% confidence interval. The committee noted that although there were differences in the net monetary benefit of the different valves, the confidence intervals overlapped significantly. The committee agreed that it is not possible to establish whether the differences in net monetary benefit were because of differences in valve performance or because of confounding in the clinical data used to inform parameters in the economic model (see section 3.10).

The committee considered a hypothetical scenario that modelled a conservative estimate of a 10% market shift towards less expensive valves without considering potential clinical differences. It concluded that switching to less expensive valves priced below the cost-effectiveness threshold that covers all surgical risk groups (see section 3.18) could result in a cost saving for the NHS, which could fund additional TAVI procedures if reinvested into the service.