On Wednesday 14th February 2018 the Health Economics team from Work Package 5 convened a workshop in Paris to coincide with the IPECAD (International Pharmaco-Economic Conference on Alzheimer’s Disease) conference taking place the following day. The workshop entitled “Modelling the economic value of Alzheimer’s disease interventions: How far have we come, and what next”, was a great opportunity to discuss progress and to consult invited experts in health economics about how to proceed from here.
The workshop kicked off with a short general introduction from Work Package 5 co-lead Alastair Gray (University of Oxford), who welcomed the group, providing an update and contextual overview for the day. Following this, the first session featured presentations from advisers on current and planned research on Alzheimer’s disease (AD):
- Anders Wimo (Karolinska Institute), presented some analysis of the cost-effectiveness of treating AD compared to the cost-effectiveness of preventing AD. He emphasised they key challenge of convincing payers of the importance of pre-dementia disease modifying drugs, given that the benefits of using such drugs would only become visible many years after starting treatment before the occurrence of significant symptoms.
- Next, Colin Green (University of Exeter) presented an outline of his prior work and ongoing research in health economic modelling as part of the IPECAD modelling group. In this modelling work the central issue was the use of a multi-domain approach involving cognition, behaviour and function. He also stressed that modelling in an HTA context was not specifically about describing in detail the experience of people with AD, with the aim being to inform decisions through addressing the key question of ‘whether this intervention is cost-effective for these people in this particular context?’
- The final speaker in the first session was Peter Neumann (Tufts Medical Center and Tufts University). He presented his work on disease progression delay in relation to modifiable lifestyle risk factors, guided by the question of what the impact such delay will be on disease progression and medical costs later on of making lifestyle changes earlier in life. In addition, the presentation focused on the importance of considering ‘spillover effects’ of AD to capture not only consequences of interventions on the person affected but also on their families and carers as well.
The second session was led by ROADMAP members Michele Potashman (Biogen) and Ron Handels (Maastricht University). It focused on the proposed structure of a future economic model as set out in the model specification which is being developed in Work Package 5.
- Ron Handels began the session by emphasising the importance of ensuring that the heterogeneity of people affected by AD needed to be captured in economic models. He also set out the arguments in favour of using broader measures of wellbeing, rather than relying exclusively on QALYs (quality-adjusted life years), in the ideal health economics model. He suggested that one approach here would be the use of numerous indicators covering QALYs, DALYs (disability-adjusted life years), and as many other relevant factors as possible, as this better reflects ‘real world’ decision-making. He concluded that a multi-domain score in which these factors remain distinct, and not simply one composite outcome measure in which all domains are aggregated, may be the most appropriate for health economic modelling in AD.
- Michele Potashman followed, first of all providing a summary of approaches that can be used in health economic modelling for AD. She stressed the diversity of situations in which models may be relevant, and highlighted the importance of ensuring that any modelling approach is consistently applicable and accurate across numerous contexts cutting across different countries, patient groups, disease stages, and health care settings as well as systems. She initiated a discussion on the extent to which disease progression should modelled on the basis of cognition, function, or should also try to incorporate behaviour.
Session three was also opened by Michele Potashman, who explored the question of how to extrapolate pre-dementia trial outcomes to later in life. One aspect of this is the difficulty of defining unambiguously to what ‘pre-dementia’ refers or when it begins, given that it necessarily deals in the absence of symptoms rather than their presence. Michele described the lack of clarity on whether extrapolation should be based on age, cognitive status, or biomarker status, and it was agreed that this is one area where better data are required.
- Following this, Filipa Landeiro (University of Oxford) described the progress that had been made in conducting the systematic reviews in WP5 of ROADMAP. In particular, she showed some initial results from the reviews of quality of life in people with dementia, and of resource use and costs. The large discrepancies between quality of life as self-rated by people with dementia – with little clear sign of decline as the disease progresses – and as rated by their carers – indicating a steep decline with disease progression – initiated a lively discussion on the difficulties this posed for HTA appraisals, and illustrated well that some of these difficulties are philosophical rather than technical matters. The systematic reviews are also a good way of identifying gaps in the literature and evidence base.
- The last speaker in this session was Pascal Lecomte (Novartis), whose presentation focused on the need to fully capture uncertainty in health economic models in this area. He led the delegates through the different types of uncertainty that would have to be captured, including assumptions concerning the benefits or drawbacks of continuing treatment; the basis on which extrapolated treatment effects can be said to have clinical validity; issues of compliance and discontinuation by patients when modelling; and whether there are particular sub-populations of interest that should be considered in modelling. Pascal also reminded the group that model validation will be a key requirement of any model, and will have to follow standard best practice guidelines as laid out by, for example, ISPOR:
- whether the model corresponds to current science and evidence;
- whether it behaves as intended;
- whether it compares well with other models;
- whether it compares well with actual event data from randomised controlled trials and real-world data;
- and whether it compares well with prospectively observed events.
There was general agreement on these points, and that the strongest forms of validation will be blind validation both externally and longitudinally.
The final session of the day was opened by Anders Gustavsson (Quantify Research on behalf of ROCHE), with a presentation considering the significance of mortality for economic modelling in AD. He began by providing some statistics regarding the relation between mortality and ageing, the relation between the two once an AD diagnosis has been factored in, and mortality independent of each disease stage. He then presented some data on the impact of disease staging on mortality. This outlined the increased risk of death at each stage of the disease, the impact of age upon the risk of being at each particular stage, and the possible impact on mortality of a disease modifying treatment. Anders closed with a discussion of four key questions:
- whether age and stage effects need to be considered in AD models;
- the challenges of identifying sufficient data on which to base models when considering mortality factors;
- how to incorporate the pre-clinical state in discussions regarding mortality;
- and how to incorporate the effect of nursing home care on mortality by providing better care.
Finally, the importance of identifying data limitations, and of prospectively planning ways to collect the data required, was emphasised by John Gallacher (ROADMAP Project Coordinator), who suggested that many of the issues that had arisen in the course of the day came back to these data deficiencies.
In conclusion, it was agreed that the proceedings of the day would be written up and circulated, and that all slides would be made available to participants.
We asked Colin Green, Professor of Health Economics, based in the University of Exeter Medical School, University of Exeter, UK to comment. Dr Green currently leads health economics across the Medical School, and is Head of the Health Economics Group.
“The ROADMAP Team – WP5 Health Economic Modelling – are taking a thorough approach in their assessment of model scope, structure and parameter inputs, with the aim of setting out a model specification. Together with the detailed set of reviews underway on resource use, costs and health-related quality-of-life, the WP5 efforts will set out a great foundation for a Phase II model development programme.”
Find out more about Dr Green here: https://medicine.exeter.ac.uk/people/profile/index.php?web_id=Colin_Green
Alastair Gray, Professor of Health Economics and Director of the Health Economics Research Centre (HERC), Nuffield Department of Population Health also provided us with a comment on the Workshop:
“The ROADMAP work package 5 workshop in Paris was an excellent day which more than met our expectations, with a very high level of engagement and active contributions over the whole day from the external experts and the ROADMAP participants. It helped to identify some key questions to be addressed when continuing preparation of our economic model specification, and some important evidence gaps, but also major areas of agreement and consensus.“