DIA: Health economics
| Jens Grueger of Novartis spoke on handling uncertainty in cost-effectiveness analysis and the possibility of risk-sharing in reimbursement decisions. He is concerned with “material uncertainty” that could could switch the decision on whether or not to reimburse. Three of the main causes of this is the extrapolation of RCT results to a broader patient population and to longer timescales, and consideration of efficacy vs effectiveness. These sources of uncertainty are highlighted when approval is accelerated due to, eg, societal pressure for conditional approval. He stated that decisions have to be made and waiting for final outcomes is not an option, but that uncertainty will exist over false positives (ie, wasted money) and false negatives (ie, patients have forgone benefits that would eventually have been proven). Uncertainty is increased under conditional approval or exceptional circumstances (eg, via the Orphan medicines route). He suggested that one solution might be possible using conditional reimbursement in cases of conditional approval, with periodic reassessment of price based on continuing collection of evidence. Using this model, pricing could increase or decrease based on the evidence, and the “evidence corridor” for acceptance of the price point could itself be updated. Ultimately, when evidence has been collected to justify a final approval, financial information would also exist to set the final pricing. This approach could also be applied retrospectively to longer-term outcomes with other medicines. However, will this continuing collection of evidence actually reduce the uncertainty, or does the issue then become what happens if the therapy is withdrawn? The problems of disinvestment go far beyond this. Similarly, will granting conditional reimbursement prevent the use of an optimal study design? Jens proposed another solution for use in exceptional circumstances (eg, orphan medicines) where sufficient pricing data cannot be generated and a formal HTA assessment is not possible. In these cases, a price notification procedure could be used, considering price information about (reimbursed) analogues. As the condition is rare, this will have a low budget impact. Risk-sharing agreements are also becoming prevalent, mostly in the form of moeny-back guarantees where payment is only made for responders. One example of this is Velcade in the UK. This is particularly useful where there is uncertainty over whether a given patient will respond. However, the “next frontier” is performance based reimbursement, where the pricing is non-linear and value-based. For example therapies given as third-line might be reimbursed at a higher price than the same therapy used as first-line. Thus, the reimbursement depends on the specific patient population (eg, for different indications) or based on other risk factors (eg, age, treatment history etc.) This obviously creates significant data requirements, but may be possible in the context of registries. This would return the prescribing decision to being based solely on expected clinical impact. |
