In this month’s Q&A, we sit down with Dr. Devin Incerti to discuss his work on decision modeling and value assessment. Hear more from Dr. Incerti at ISPOR New Orleans where he will be participating in an issue panel titled “Can We Make Global Value Assessments More Flexible and Comprehensive?” on Monday, May 20, 5:00 PM–6:00 PM ET.
What are some of the challenges that you have faced in value assessment?
In many countries, a single HTA agency assesses the value of a health technology–typically using cost-effectiveness analysis (CEA)–and recommends whether the technology should be adopted. In the United States, decision making is more complicated. For one, the general public tends to be wary of policies that might be construed as rationing care. Furthermore, decisions are made by various payer and provider organizations with different patient populations and time horizons.
The US setting–where we have focused our recent efforts–consequently creates a number of challenges. The decentralized environment means that it is necessary to estimate value tailored to local populations. We have also seen growing demand for newer approaches, such as multi-criteria decision analysis that can consider a more comprehensive set of costs and benefits than traditional CEA. Finally, different methodologies for computing the benefits, risks, and costs of a health technology can generate very different results, which impede efforts, such as value-based pricing and outcomes-based contracting, to reward value in the US health system.
What are some reasonable steps that can be taken to overcome these challenges?
One key consideration is the aim of value assessment. In my mind, its ultimate purpose is to optimize the allocation of treatments in a particular population given some budget. This means that it is critical to develop methods than can properly consider patient diversity. We have taken steps in this regard by developing a new software package, hesim, which integrates statistical models capable of incorporating effect modifiers and prognostic factors with decision modeling.
Of course, before even considering differences across patients, we need to ensure that the underlying decision model is sound. One problem that we have run into in oncology is that decision models do not consider explicit sequences of treatment where patients move from first-line treatment to second-line treatment, and so on. We have developed new multi-line models and have found that conclusions based on explicit treatment sequences can be quite different than those based on conventional approaches.
From your perspective, where is decision modeling headed?
Advances in statistical and computational methods have made it easier than ever to extract valuable insights from an ever-increasing supply of data. State-of-the-art techniques in statistics and machine learning are widely accessible through open-source libraries and web-based platform technologies. Yet, cost-effectiveness analyses are still typically performed in spreadsheet software (almost always Microsoft Excel), which results in models that are non-transparent, inefficient, error prone, and incapable of making the best use of the available data.
I believe that the field will move away from Excel-based solutions and embrace the adoption of new technology-driven approaches. This shift will make decision models more data-driven, better able to account for patient diversity, and ultimately foster an environment in which coverage and pricing decisions are made in a manner that confers the greatest possible benefit to society.
If you would like to schedule an in-person discussion with Dr. Incerti, please email him at Devin.Incerti@PrecisionHealthEconomics.com