There is certainly an industry- wide need for more transparent, targeted and robust benefit risk assessments. In this blog we talk with Yannis about the article and why the Bayesian framework is particularly well suited to these efforts.
Could you give us a summary of the objectives of the publication? Yannis Jemiai (YJ) As you know benefit risk assessment has become a hot topic in recent years. While there are frameworks available to do the assessment in a qualitative way, it’s been argued recently that this should be quantified to ensure more transparency and consistency in how decisions are made across different regulatory and payer organizations. There have been a number of discussions around how to tackle this issue- this article specifically examines how a Bayesian framework can help.
So what are the advantages of a Bayesian approach to benefit risk assessment? (YJ) There are many advantages. In particular, the Bayesian statistical framework is very well geared towards decision making since it allows you to weigh different factors according to different prior beliefs that an individual stakeholder may have. One of the interesting features of this topic is that the factors a regulator would take into account as opposed to a payer or a patient could be quite different. These differences can be incorporated in to a quantitative assessment of benefit risk much more easily using a Bayesian framework.
It was somewhat surprising, when reading around the topic how qualitative benefit risk assessment has historically been, particularly when considered in the very quantitative context of drug development as a whole. Can you comment on that? (YJ) Yes, when you look at the literature, there are various reasons for this. In fact, sometimes the picture is not particularly complex, and there may not be many factors which need to be accounted for. For example, if you have one clear measure of efficacy, such as survival, and one clear measure of safety then it can be relatively straightforward for a committee to come together, discuss it and make a decision. But nowadays many drugs have much more complex benefit risk profiles. We may be looking at not just survival, but quality of life and other benefits which aren’t specifically treatment benefits- for example cost to society or reduced hospitalizations. These benefits may be more difficult to quantify.
What personally did you find most interesting working on this publication?
(YJ) It was a great learning experience to discover more about the proposals that are out there and the role that statisticians can play in bringing rigor and a quantitative approach to the discipline of benefit risk assessment and help make better decisions for patients and for society.
What are the next steps in taking this research forward? (YJ) The objective of this particular article is to advocate for a Bayesian approach to benefit risk assessment aimed at a broad audience. It doesn’t drill down into the statistical detail or advocate any particular method. The plan is to follow this up with a more technical article aimed at a statistical audience which provides more granularity on specific approaches and how to apply them, as well as some case studies to share experiences of using them in practice.
To read the abstract and obtain access details for the publication click the button below. The work was carried out as part of the DIA Bayesian Scientific Working Group, Benefit-Risk subteam.
1) The Case for a Bayesian Approach to Benefit-Risk Assessment Maria J. Costa, PhD, Weili He, PhD, Yannis Jemiai, PhD, Yueqin Zhao, PhD, Carl Di Casoli, PhD
Yannis Jemiai, Ph.D. is Senior Vice President at Cytel, where he leads the software products, strategic consulting, and marketing groups. Dr Jemiai received his Ph.D. in biostatistics from Harvard University