Applications of decision modeling to assess the value of clinical research
It is well recognized that decision analytic models are especially valuable in clinical research when it is not possible or prohibitively expensive to perform an experiment that would allow the outcomes of the medical intervention to be directly observed. Another instance in which decision models may be useful in deciding which experiments are likely to be most valuable to perform. Specifically, value of information techniques provide a powerful decision-analytic tool to prospectively assess the value of research. The approach can not only allow identification of the value of specific studies but also identification of optimal study designs that maximize the expected value of research. This paper examines several recent efforts to apply value of information techniques to identify the value of research. The effects of the research on health are modeled using a net health benefits framework that calculates the expected health benefits of an intervention in quality-adjusted life years net of the health benefits forgone by the spending needed to achieve those benefits. Techniques are also identified to bound estimates of the value of research using limited information. Examples are drawn from a variety of recent applications in the literature, including choice of atypical antipsychotics in schizophrenia, whether to use acetyl cholinesterase inhibitors in early Alzheimer’s disease, and alternative potential strategies to prevent or treat muscular dystrophy. The potential to use value of information methods in policy making is also examined.