Objective: To review recent economic analyses and determine if means of improving methodology used in modeling stroke management may exist.
Study design and setting: The Medline database was searched for pharmacoeconomic models of treatments or interventions in acute, non-transitory ischemic stroke. Search terms were: stroke, cost, cost-effectiveness, cost analysis, stroke management, model, modeling, and economic. All English-language articles published from January 1997 to January 2008 were reviewed.
Results: Ten Markov models and three decision analytical models were identified. All models had a societal perspective and all but one had lifetime horizons. They were all based on common patient states of disability, mortality and recurrence of stroke. Inputs used in the models were transparent and valid. Intracranial hemorrhage, cardiovascular events and data closely related to local settings were not systematically considered. One-way sensitivity analyses were the most common, but few parameters were tested and these varied between models. Consensus key drivers were therefore difficult to determine.
Conclusion: The overall structure of the models reviewed was sound. However, they should include more systematically cardiovascular events and intracranial hemorrhage, as well as local epidemiological data. Further multi-way sensitivity analyses would help to identify key cost drivers with greater precision and robustness.