A PAREXEL® Consulting analysis of the reimbursement decisions of three national reimbursement agencies from 2005 to 2014 found that just over a third (39 percent) of submissions to payers were approved.
One reason for this low acceptance rate of already-approved treatments is that pharmaceutical manufacturers are not presenting the data that payers want, demonstrating that a new treatment is more cost-effective than those already on the market.
Drug developers design their trials primarily for regulatory approval. This is reasonable; without regulatory approval their therapies would never reach payers for reimbursement consideration. This leads manufacturers to focus Phase III trials on their new treatment’s efficacy versus placebo or established standards of care. However, this is a less demanding hurdle than comparing them with the market-leading therapies commonly used by clinicians and against which payers are likely to compare. Consequently, developers are left trying to assemble the data payers demand after Phase III trials have finished, using the data they have. This often means constructing an indirect case against the payers’ preferred comparators. Not surprisingly, this can cause payers to become skeptical about the reliability of the developers’ arguments and the value of their treatments.
For example, in the UK, NICE issued a negative final appraisal for Benlysta, a lupus treatment, despite the fact that the drug’s developer, GlaxoSmithKline, sought approval for Benlysta only as an add-on therapy for patients with active lupus who were receiving other medicines, and even offered a discount on the list price. But NICE expressed uncertainties over Benlysta’s cost-effectiveness, noting the lack of data comparing the drug with Roche’s MabThera, the prevailing treatment.
Even if developers could predict accurately the data they would need to ensure success in every submission, it would be too overwhelming and expensive to collect it all. Therefore, they must collect and model better data, incorporating information from prior sub-missions for similar therapies and indications, and use it to design better-targeted trials. This is an achievable goal.
DON’T REPEAT MISTAKES OF OTHERS
The search for better data and economic models begins with avoiding mistakes others have already made. Developers should look at submissions for similar compounds to see where the gaps in the evidence were, factoring those into their trial design.
By reviewing previous reimbursement decisions, developers can identify the sources of payer uncertainty, which can help them to understand what data might be important and to prioritize accordingly. This may result either in the commissioning of further studies or in a change to the trial protocol. It can sometimes be relatively easy to collect the required data from a single trial that will meet the needs of multiple agencies.
PAREXEL, for instance, has an oncology client planning trials to satisfy two quality-of-life (QoL) measures: one disease-specific focused on pain and one generic based (the EQ-5D). NICE prefers EQ-5D data, which is not very specific concerning pain and therefore not very good at distinguishing severity in this illness. The specific measure, better at distinguishing between levels of illness, cannot be compared across different illnesses. If both measures are collected in the same trial, one could map from the specific to the general (for example, a given range of pain scores mapped to a corresponding range of EQ-5D values). This type of equation could be used to estimate EQ-5D values indirectly if the disease-specific measure is the only one collected in a clinical trial. NICE prefers EQ-5D data to be collected directly, rather than mapped, and other agencies prefer other measures, such as the Short Form Health Survey, or SF-36. Clearly, what is collected and what is mapped could have a pivotal role in what an agency might approve.
It is therefore important to review prior reimbursement agency submissions for the specific therapeutic area and determine for each whether mapping has been criticized or accepted, or whether the agency requires directly measured EQ-5D (or SF-36) data. On that basis, a developer can decide what to collect directly and what to map, and put in a plan to explain its strategy to payers and gain their trust at an earlier stage.
For one client, PAREXEL analyzed the relationship of hospitalization for heart failure to mortality, demonstrating a strong correlation between rehospitalization and life span. PAREXEL quantified the number of rehospitalizations that needed to be prevented to have a significant impact on mortality by an intervention. This allows the possibility of using readmissions as a surrogate endpoint in clinical trials, potentially saving time and money in development.
As in the oncology example, patient-reported outcomes (PROs) are becoming increasingly important to securing regulatory approval. They can strengthen the case for a product’s therapeutic value, particularly when establishing a label claim. They have also become relevant in overcoming many market-access hurdles, including reimbursement submissions, and in price setting. Therefore, it is important to factor PROs into early-stage planning.
The range of existing and validated PRO instruments is extensive and growing, and if a developer wants to compare symptom outcomes with a competitor’s product for the same indication, it may make sense to use the same instrument. However, if a new claim is being made for a treatment compared with a competitor’s, it may be wise to develop a new instrument. In either case, collecting this data is time consuming and expensive (especially when developing a new instrument), and should be done as early as possible. This enables a developer to determine what data various stakeholders and decision-makers will likely most value and what study design will produce the quality and quantity of data desired. This can all shorten time to market.
Along with improving trial design by taking into account different sources of data, it is also valuable to develop an economic model before Phase III. This makes it possible to determine the potential cost-effectiveness of any compound early enough to address key issues, or fill data gaps, before investing resources in a treatment that has no realistic chance of achieving commercial success.
In a recent PAREXEL engagement, an early-stage economic model illustrated that changes in patient utility were a key driver of cost-effectiveness, and that evidence in the literature for utility for patients with the target disease was subject to uncertainty. Following this modeling, the client, in addition to collecting EQ-5D in its Phase III trial, implemented a separate study designed to collect utility data for patients treated with the current standard of care. In this case, better information and better economic modeling reduced uncertainty and strengthened the argument for cost-effectiveness.