Bringing Health Economic Modeling to the 21st Century
May 1, 2006, 00:00
10.1111/j.1524-4733.2006.00097.x
https://www.valueinhealthjournal.com/article/S1098-3015(10)64309-4/fulltext
Title :
Bringing Health Economic Modeling to the 21st Century
Citation :
https://www.valueinhealthjournal.com/action/showCitFormats?pii=S1098-3015(10)64309-4&doi=10.1111/j.1524-4733.2006.00097.x
First page :
Section Title :
Open access? :
No
Section Order :
8
This is a summary of the presentation given by David Eddy at the ISPOR 10th Annual International Meeting Second Plenary Session, May 17, 2005, Washington, DC, USA. Dr Eddy is Founder and Medical Director of Archimedes, Inc., Oakland, CA, USA.
At the core of every health economic analysis is a model. It is the model that converts data about diseases, interventions, and costs into projections of what will happen in the future. The importance of good data is emphasized by the expression “Garbage in, garbage out.” With a flawed model “Anything in, garbage out.” My purpose in this commentary is to argue that the increasing complexities of the problems we are trying to analyze are pushing our current models to their limits, and to offer a new type of model for our toolkit. Because both the need for the model and its design are rooted in my experiences, I will tell the story from a personal perspective.
Need for a New Type of Model
The need for a new type of model was gradually forced on me by the gap between the types of problems I wanted to analyze and the capabilities of the existing models. My own experiences began about 35 years ago with Markov models [1]. Given the continuing prominence of this type of model in health economic modeling, this is a reasonable place to start.
Every reader of Value in Health is familiar with the basic structure of a Markov model. The fundamental property that makes a random process a Markov process is that given the present, the future is conditionally independent of the past. Other ways to say this are that history does not matter, and the process has “no memory.” Typically a Markov process is defined in terms of discrete states; at any time the process will be in a particular state, and at discrete time intervals the process can make transitions between states. A textbook example is cars in a queue for a tollbooth. The state of the process is the number of cars in the queue at any time. The state changes every time a new car enters the queue or the front car is cleared through the tollbooth, and the trajectory of the process is determined by the probabilities of those events.
Categories :
- Cost-comparison, Effectiveness, Utility, Benefit Analysis
- Economic Evaluation
- Methodological & Statistical Research
- Modeling and simulation