The advent of the direct acting anti-virals (DAAs) for hepatitis C, the increasing use of high cost biologics, and the escalating launch prices of oncology medicines in particular, have raised concerns that pharmaceutical expenditure growth could become increasingly difficult to sustain. In order to ensure adequate resource mobilisation, and to manage the entry of major new therapies, many countries see value in trying to anticipate changes in market dynamics and by doing so, to attempt to forecast future pharmaceutical expenditure.
Prepared with support from the European Commission, this OECD report explores countries' approaches to tracking pharmaceutical utilisation and expenditure and anticipating changes in pharmaceutical markets. It examines how these are used to inform the setting of budgets and spending caps, and as inputs to modelling future expenditure. Identifying key data needed to inform future projections, the report also highlights best practices across OECD countries, and proposes recommendations for countries currently undertaking, or planning to introduce pharmaceutical expenditure projections to inform future policy making.
A number of countries project pharmaceutical expenditures in the short term to support resource planning and budget setting. These estimates generally take into account supply-side movements and market dynamics, but even with sophisticated modelling techniques and detailed market intelligence, such projections are challenging.
Horizon scanning is used by several countries to identify late stage products in global industry pipelines, but only a few use the outputs of this activity in modelled projections. This resource-intensive undertaking would benefit from cooperation between countries to avoid duplication of effort. However, in order to inform expenditure projections effectively, it would need to be augmented by systematic monitoring and collation of country-specific data on timing of market entry.
Beyond predicting the arrival of new products in the market, modelling their impact on expenditures remains very challenging, since it requires anticipating their therapeutic value and likely price. Therapeutic value is likely to influence various drivers of uptake and diffusion: e.g. the place in therapy, whether use is likely to be additive or to displace older treatments, or whether it extends treatment to a previously untreated patient cohort. These are in turn influenced by the population to be treated, requiring a knowledge of the underlying epidemiology and burden of disease; the indications likely to be approved for marketing, and those likely to be accepted for coverage or reimbursement.
Determining the nature and timing of loss of exclusivity (LoE) of a product, as a proxy for the timing of generic or biosimilar market entry, is essential for modelling the impact of generic and biosimilar competition. The establishment of accessible and comprehensive public databases could improve access to the necessary intelligence both for generics/biosimilar manufacturers and analysts. An alternative could be for countries/payers to require the provision of comprehensive information on all forms of applicable IP protection in applications for coverage/reimbursement.
Data on past trends in generic uptake and their impact on markets (both volumes and prices) are useful to predict future effects of generic market entry, but less so for biosimilars. Biosimilar data remain sparse, not only because of the shorter history and smaller number of ‘follow-on’ products, but also because the uptake, acceptance and pricing effects of biosimilars appear to be more idiosyncratic, and seemingly dependent on drug class and location of use (i.e. hospital vs community), as well as on national/payer policies on pricing and substitution.
For effective short-term projections, many model parameters that cannot be populated empirically must be driven by assumptions, thus highlighting the need for testing multiple scenarios and performing extensive multivariate sensitivity analyses.
Comparing actual spending trends with projected estimates is important for adjusting assumptions and improving both the confidence in, and the predictive value of these heavily parameter-driven models, particularly if they are to be used to estimate the potential effects of proposals for policy reform. This also informs an assessment of the trade-offs between resource intensity and forecasting precision.