This workshop was held as a collaboration between Arthritis Research UK and the MRC Network of Hubs for Trials Methodology Research.
This workshop was attended by rheumatologist clinicians, statisticians, health economists, representatives from NICE, and Hub members. Presentations and discussions focussed on the health economic models used to model the outcome of treatment of inflammatory joint diseases with the newer biologic drugs for rheumatoid arthritis (RA) and related disorders. One of the major drivers for this workshop from the rheumatological community was the lack of randomised ‘head to head’ trial data between the licensed agents. There was discussion of the use of indirect comparison approaches to existing data and the use of new methodologies such as Value of Information analysis, in the prioritisation and design of further research.
The group acknowledged that Value of Information (VOI) analysis has the potential to inform decisions on research priorities, and on the design of future studies. For example, VOI could be used to assess the value of developing an Early Rheumatoid Arthritis National Database to capture data on treatment of patients. It could also be used to inform investigators about the value of head-to-head trials of biologics, and the respective value of small, non-inferiority trial compared to larger trials intended to demonstrate superiority.
The group appreciated, however, that in RA there is no consensus on how the natural history of inflammatory joint diseases should be modelled. What are the key outcomes? And how are they inter-related both in terms of state and change? Without reliable cost-effectiveness models which clinicians understood and could sign up to, research priorities and study designs suggested by VOI analysis would be unlikely to command wide support in the academic arthritis community. Specific areas that need to be addressed are the incorporation of drug safety in the models, the changes during treatment in one of the key current outcomes in cost effectiveness analyses: the Health Assessment Questionnaire (HAQ) for disability, and the impact of withdrawing a treatment. Modellers agreed that cost-effectiveness results were sensitive to the way these particular aspects were modelled. There was also uncertainty about the best way of defining response to treatment, and considerable lack of agreement on the need for individual patient models.
As there is currently no consensus view on model assumptions, structure or outcomes chosen, the group agreed that it would be reasonable to form a Clinical Studies Group that could work towards finding consensus on the model inputs and methodology. This should be a collaborative exercise involving modellers and clinicians, with involvement of NICE.
Click on the links below to view the presentations from the workshop: