Model-based cost-effectiveness analyses for prostate cancer chemoprevention : a review and summary of challenges.

BACKGROUND AND
PURPOSE: Decision-analytic modelling is often used to examine the economics associated with using a specific treatment. As a result, it is important to understand structural and methodological approaches used in published decision-analytic models for examining the cost effectiveness of 5α-reductase inhibitors (5ARIs) for prostate cancer (PCa) chemoprevention. This understanding allows us to provide recommendations for using decision modelling in future economic evaluations of chemoprevention for PCa.
METHODS: A review of the published literature was performed using MEDLINE and the Cochrane Library to identify studies involving mathematical decision models that evaluated 5ARIs for PCa chemoprevention. Published articles were reviewed and key modelling components were extracted and summarized. Recommendations for developing future decision models to examine the economic consequences of PCa chemoprevention were presented.
RESULTS: We identified seven published models of PCa chemoprevention. All the models identified used a Markov framework with time horizons ranging from 4 years to lifetime. Due to the wide range of patient risk groups examined, PCa risk data were taken from the Surveillance, Epidemiology, and End Results (SEER) and other databases or estimates published in relevant clinical trials. Treatment effects included change in the incidence of high- and low-grade PCa and impacts on benign prostate hyperplasia. Adverse events were considered to affect compliance, discontinuation and quality of life. Quality-of-life impacts were similar among studies. Examination of modelling parameter sensitivities was comprehensive.
CONCLUSIONS: Published models have examined the cost effectiveness of PCa chemoprevention; however, limitations exist. Decision models should take into account the full PCa clinical pathway when compiling health states. The time horizon should be long enough to consider the full benefit of chemoprevention while allowing actual time receiving the drug to occur from the start of the model until a man's life expectancy is less than 10 years. Baseline PCa risk should be specific to the population of concern. Models should examine the impact on both low- and high-grade tumours and account for the impact of 5ARIs on benign prostatic hyperplasia. Because chemoprevention has an upfront effect, the structure of the model should be constructed so that the downstream effect of avoiding or delaying recurrence can be considered. Adverse events due to chemoprevention should be considered through compliance, discontinuation or quality-of-life impact, and understanding the impact of avoiding PCa and benign prostatic hyperplasia events are important model properties.