Objective: Chronic obstructive pulmonary disease (COPD) is a major health problem with high societal costs. The Global Initiative for Chronic Lung Disease (GOLD) has identified a need for health economics data for COPD. For chronic diseases, such as COPD, where the natural history of disease is lifetime, a modeling approach for economic evaluation may be more realistic than prospective, piggy-backed clinical trials or specific COPD cohort studies. Simulation models can be used to extrapolate clinical data beyond the limited time frame of clinical trials, to analyze subgroups of patients or to explore uncertainty regarding the results by using sensitivity analysis techniques. Our purpose has been to develop a flexible computer simulation model for COPD that will represent disease progression and GOLD recommendations, useful for economic evaluations of new medicines to meet the needs of various payer requirements for reimbursement and resource allocation.
Methods: This article describes a two-dimensional Markov model, which uses data from multiple sources about disease progression, exacerbation frequency and duration, mortality, costs, burden of illness, and the relationships between those variables. The model is evaluated using stochastic uncertainty analysis, it allows comparison of treatments affecting different disease mechanisms, and it uses primary data validated against published sources.
Results: We have evaluated two hypothetical interventions treating different features of the disease (lung function decline and acute exacerbations). These analyses show that reducing lung function decline must be a long-term strategy compared to reducing the number of exacerbations. It was necessary to have a long term like 30 years, with 10,000 patients and 20% increase in price, or 20 years with equal prices to show cost-effectiveness with statistical significance for a treatment that reduces lung function decline.
Conclusions: Our study shows the value of modeling as a tool for evaluating different scenarios and for combining several sources of data, to provide estimates that would otherwise be unavailable. Clinical trials of this size and duration would be unrealistic.