The Future of Health Economic Modeling- Have We Gone Too Far or Not Far Enough?
Abstract
This is a summary of the presentation given by A. Mark Fendrick at the ISPOR 10th Annual International Meeting Second Plenary Session, May 17, 2005, Washington, DC, USA. Dr Fendrick is Professor at the Department of Internal Medicine and Health Management & Policy, University of Michigan Medical Center, Ann Arbor, MI, USA. This presentation immediately followed the presentation, Bringing Health Economic Modeling to the 21st Century, by David Eddy.
As clinicians, we have to remind ourselves of the real motivation for health-care modeling. The development of models—like all clinical research methods—is a tool to help patients, to help the health-care systems act more efficiently, and ultimately, to achieve the goals of providing high quality and value in health-care services.
Have models had an impact? There is some evidence, although scant, that modeling does impact guideline development and reimbursement decisions. One large challenge, however, is that ultimately most “do we, don’t we?” decisions are still being made by clinicians at the bedside. There is virtually no evidence suggesting that models alone affect what clinicians do. Thus, we should not be overly optimistic that “better” models will ultimately change how we practice medicine. As we look into the future, the question to be asked is: Will models, no matter how complex or how simple, ever be accepted as a real time decision-making tool? Can we ever rely on the computer alone, as in 2001: A Space Odyssey, and put the trialists out of business except for the sole purpose of being able to validate an existing model?
In my view, the motivation for “incremental” modeling acceptance is quite straightforward: address the questions for which rigorous evidence is not currently available. On the success side of the equation, modellers have used a synthesis of data from various sources to extend the duration of trials, to examine populations not included in trials, or to address those questions for which trial data do not exist. Clearly, one advantage is the relative cost of simulations when compared with the real cost of collecting empiric data. As we acknowledge how expensive it is to do trials, it is important to remember that it may be quite complicated to obtain all this information for models, particularly for those clinical conditions that are less widely studied. Modeling in this context has been quite useful in defining or refining research questions. Many in the community are quick to recognize these exercises as hypothesis-generating when compared with hypothesis-testing.
As clinicians, we have to remind ourselves of the real motivation for health-care modeling. The development of models—like all clinical research methods—is a tool to help patients, to help the health-care systems act more efficiently, and ultimately, to achieve the goals of providing high quality and value in health-care services.
Have models had an impact? There is some evidence, although scant, that modeling does impact guideline development and reimbursement decisions. One large challenge, however, is that ultimately most “do we, don’t we?” decisions are still being made by clinicians at the bedside. There is virtually no evidence suggesting that models alone affect what clinicians do. Thus, we should not be overly optimistic that “better” models will ultimately change how we practice medicine. As we look into the future, the question to be asked is: Will models, no matter how complex or how simple, ever be accepted as a real time decision-making tool? Can we ever rely on the computer alone, as in 2001: A Space Odyssey, and put the trialists out of business except for the sole purpose of being able to validate an existing model?
In my view, the motivation for “incremental” modeling acceptance is quite straightforward: address the questions for which rigorous evidence is not currently available. On the success side of the equation, modellers have used a synthesis of data from various sources to extend the duration of trials, to examine populations not included in trials, or to address those questions for which trial data do not exist. Clearly, one advantage is the relative cost of simulations when compared with the real cost of collecting empiric data. As we acknowledge how expensive it is to do trials, it is important to remember that it may be quite complicated to obtain all this information for models, particularly for those clinical conditions that are less widely studied. Modeling in this context has been quite useful in defining or refining research questions. Many in the community are quick to recognize these exercises as hypothesis-generating when compared with hypothesis-testing.
Authors
A. Mark Fendrick