Probabilistic Markov model to assess the cost-effectiveness of bronchodilator therapy in COPD patients in different countries.

Objectives: The development of a probabilistic Markov model with a time horizon of 1 year to compare the cost-effectiveness of three bronchodilators: 1) the new long-acting anticholinergic tiotropium; 2) the short-acting anticholinergic ipratropium; and 3) the long-acting beta 2-agonist salmeterol, for the treatment of patients with chronic obstructive pulmonary disease (COPD) in different countries. In this article we compare The Netherlands and Canada.
Methods: The Markov model was structured along disease severity states and exacerbations. Transition probabilities between disease states and exacerbation probabilities were derived from patient-level data from six randomized controlled trials assessing the efficacy and safety of tiotropium. Resource utilization during exacerbations and maintenance treatment for The Netherlands were derived from two clinical trials, whereas for Canada these data were obtained from a countrywide observational study that used similar inclusion criteria as the trials. Second-order Monte Carlo simulations were undertaken in which values were randomly drawn from distributions of these parameters. Outcomes of the model are yearly treatment costs, exacerbations, and quality-adjusted life months.
Results: The mean difference in the number of exacerbations was 0.17 (95% uncertainty interval: -0.02-0.37) in favor of tiotropium when compared with salmeterol and the difference between salmeterol and ipratropium was 0.12 (-0.17-0.44) in favor of salmeterol. The number of quality-adjusted life months did not substantially differ between treatment groups and varied from 8.42 (SE 0.41) in the tiotropium group to 8.17 (0.46) in the salmeterol group and 8.11 (0.50) in the ipratropium group. In The Netherlands, costs in the tiotropium group were 42 Euros (-484-353) lower than in the salmeterol group, whereas costs in the salmeterol group were 128 Euros (-795-457) lower than in the ipratropium group. In Canada, costs were consistently lower than in The Netherlands and nearly the same in all treatment groups. Differences between the two countries were primarily a result of a higher length of hospital stay in case of an exacerbation in The Netherlands. The cost-effectiveness acceptability frontier of exacerbations showed that tiotropium was associated with the maximum expected net benefit for all values of the ceiling ratio above 0 Euros (The Netherlands) and 10 Euros (Canada) in the base case analysis.
Conclusions: This probabilistic model-based economic evaluation demonstrates how clinical trial data can be combined and integrated with country-specific information about resource utilization and unit cost to assess the cost-effectiveness of bronchodilators in COPD patients. Quality-adjusted life months did not substantially differ between treatment groups. In terms of exacerbations, tiotropium was associated with maximum expected net benefit for plausible values of the ceiling ratio. In sensitivity analyses, this outcome was most sensitive to changes in exacerbation rates.