Why we need a new Outcomes-based Value Attribution Framework for Combination Regimens in Oncology

Towse, A., Lothgren, M., Steuten, L. and Bruce, A. 

Consulting Report
February 2021

Sign Up for OHE Updates

Why not sign up for our email updates? Around once or twice a week we will send you a round-up of the latest news, events and publications from OHE.

I'm already signed up    No thanks

Using medicines in combination can deliver better outcomes for patients across different tumour types and disease stages. Yet many HTA agencies do not find that the expected additional benefits from adding a new medicine to a currently reimbursed medicine represents value for money to the health system. In markets that utilise cost-per-QALY approaches for assessing value, a clinically effective medicine might even be found to be “not cost-effective at zero price” when used as part of a regimen that increases treatment duration.

This publication contributes to the emerging debate as to why a new approach is required to address the value attribution problem in combination treatments. It sets out two key concepts required for an understanding of the relative performance of a combination regimen as compared to a monotherapy, considering (i) additive scale and (ii) the relative change in treatment duration as compared to the increase in overall survival.  It uses these to illustrate how current HTA approaches, lead in many cases to the add-on therapy not being cost-effective even at zero price.

The conceptual framework shows that two attribution approaches – the partial information incremental value approach, and the full information monotherapy ratio approach - are only comparable if an implicit assumption about additive scale is made, which the authors think implausible.  They argue that an approach is needed that is able to combine the best features of the two approaches recognising the contribution of each product as a monotherapy, and also recognises their respective contributions to the combination regimen.

General principles that might be expected from an attribution mechanism are set out.