Drug-eluting stents for treating coronary artery disease

To compare the clinical effectiveness between stents, the EAG did targeted literature searches to identify relevant published clinical evidence. The review focused on randomised clinical trials (RCTs) comparing outcomes between the stents in the scope of this assessment. For 8 of the 29 stents there was no randomised evidence that compared one stent with another in scope. The committee agreed that RCTs were the most suitable source of evidence for this assessment. But it acknowledged that there was a large volume of other types of evidence (14 non-randomised or observational comparative studies and 54 single-arm studies) related to the stents in the scope.

If evidence was not available for a stent in the scope, the EAG looked for evidence on clinically equivalent predecessors. Manufacturers provided information on whether evidence for a predecessor stent could be generalisable to a stent in the scope, but this information was not available for all stents. The manufacturers clarified that where equivalence was stated, the changes between stent generations were usually related to the deliverability of the stent, rather than the polymer or drug.

The EAG identified 22 key RCTs comparing 1 or more stents with another in scope. Of the 22 studies, 21 were non-inferiority studies that determined whether a stent works as well as its comparator. The committee noted that most of the 22 studies showed similar clinical outcomes (target lesion failure, major adverse cardiac events, stent thrombosis, repeat revascularisation and death from cardiac causes) between the different stents.

The EAG examined whether any of the 22 key RCTs provided outcome data for the subgroups in the scope. Some data on subgroups was available for women and for people with left main-stem lesions, bifurcation lesions, high risk of bleeding or diabetes. Some of the studies reported subgroup results, and some reported whether the subgroup characteristic had an effect on the clinical outcomes. 3 studies had 1 of these populations as the main population. The subgroup results were similar to the overall study results. None of the subgroup characteristics had a significant effect on the clinical outcomes. The committee heard from clinical experts that this was reflective of their experience.

The committee noted that none of the 22 key studies reported results by ethnicity or the effect of ethnicity on clinical outcomes, or included any information about the ethnicity of study participants.

To present the comparative effectiveness of multiple stents in a single analysis, the EAG did a network meta-analysis (NMA). There was sufficient evidence to include 18 of the 29 stents in the NMA. Of the 22 key studies, 14 studies contributed to the 1-year analysis and 12 studies to the long-term analysis of 2 clinical outcomes that were reported in all the included studies: target lesion revascularisation and target vessel-related myocardial infarction. The wide 95% confidence intervals around the effect estimates from the analyses indicated that the estimates were uncertain. But, as with the results of the individual primary studies, most of the NMA results suggested that the 2 clinical outcomes were similar between stents. The EAG explained that having only limited data for each comparison in the analysis, even less so for the long-term analysis, was a key reason for the uncertainty. The committee recalled the assumptions around clinical equivalence (see section 3.4). Using evidence from predecessor stents may have added uncertainty to the results.

Only 2 studies in the NMA were done partly in the UK. The clinical experts explained that there are some differences in clinical practice between countries. For example, intravascular imaging during PCI is more common in the UK than in some other countries. But this difference would mean that better clinical outcomes could be expected from the trials if they were done in the UK. Distributions of populations with stable angina and acute coronary syndrome are the same across the world. The committee had no concerns about the generalisability of evidence to the NHS.

The EAG developed a multi-state Markov model to estimate and compare the cost-effectiveness of the drug-eluting stents. The model included 2 clinical events: target vessel revascularisation and target vessel-related myocardial infarction. The committee agreed that, for the purpose of comparing different stents, the model was an appropriate representation of clinical practice in the NHS.

To calculate the probabilities of the 2 clinical events in the model (target vessel revascularisation and target vessel-related myocardial infarction), the EAG used the relative clinical effect estimates from the NMA. The economic model reported results for 18 stents because only 18 of the 29 stents were included in the NMA. The committee recalled that the amount of randomised evidence comparing effectiveness between the stents in the NMA was limited, so the treatment effects were uncertain (see section 3.8).

The model's base case estimated outcomes with a 1-year time horizon following the index (first) PCI. In the alternative scenario estimating 5-year outcomes, the clinical event rate after 1 year was assumed constant. The clinical experts noted that it is not correct to assume that the long-term outcome rate would stay the same, but added that the evidence did not suggest a difference in clinical outcomes. The EAG explained that this assumption about long-term outcomes was made because of the limited data available. It cautioned that the long-term cost-effectiveness analysis should be considered exploratory. The committee recalled that there was considerable uncertainty, especially around the long-term effectiveness estimates from the NMA (see section 3.8).

The model included the cost of the stents using NHS Supply Chain weighted average of 2023 purchase costs or framework price, other PCI procedure costs, treatment and care costs after PCI and repeat revascularisation and myocardial infarction-related costs. The committee concluded that the cost of the stents is a small part of the total procedure cost, and the price differences between most stents are generally relatively small. The committee noted that stents aimed for use in specific cases cost more, and these should be used only when they are clinically appropriate.

There was a limited amount of evidence comparing effectiveness between the stents, and subsequent uncertainty in the treatment effects from the NMA. So, there was considerable uncertainty in the model results. The EAG presented the results of the economic evaluation in terms of net monetary benefit, including the central value and the 95% confidence intervals. The 95% confidence intervals around the net monetary benefit average estimates for all 18 stents in the model were wide and largely overlapped. At the £20,000 threshold, there was a low (less than 30%) probability of any of the stents being the most cost-effective. The committee had no concerns about the overall cost-effectiveness of stents. It noted that for the 11 stents not included in the economic model, there was no evidence available to suggest significant differences in cost effectiveness. But it concluded that, based on the model, it is uncertain whether some drug-eluting stents are more cost effective than others.

The committee discussed 2 hypothetical scenarios that estimated the financial impact of shifting towards stents with a lower price. In 1 scenario the shift was between the same manufacturer's brands. In the other scenario the shift was between different suppliers. The scenarios did not consider potential clinical differences or volume-based pricing. The committee recalled that the cost of the stents is a small part of the total procedure cost, and the price differences between stents are generally relatively small (see section 3.13). It was uncertain whether, in the context of the total spend on stents, these shifts would result in substantial savings.

The committee concluded that to show additional value more evidence comparing clinical outcomes of different drug-eluting stents for people with coronary artery disease would be needed. This would ideally be from an RCT designed to show the clinical superiority of a stent. If a company introduces to the market a new drug-eluting stent or a new generation of the technology with minor improvements, it should show clinical non-inferiority. The committee noted that long-term data (up to 5 years) needs to be captured to help inform a cost-effectiveness analysis. But it recognised that other factors not related to the stent or target lesion may become more important after 1 year, and this could limit the validity of the conclusions from any long-term studies. The committee acknowledged that the lack of evidence for the additional value of a stent against its comparators in the evidence review does not necessarily mean that there is no difference in cost-effectiveness.