Drug development in the pharmaceutical industry is increasingly influenced by measures of cost-effectiveness, such as cost per life-year gained, and some governments make decisions about whether to pay for drugs based on cost-effectiveness considerations. While cost per life-year gained is a key measure of cost-effectiveness, costs associated with the intermediate outcome of improving a biomarker, such as cholesterol level or blood pressure, provide important supplementary information, particularly where mortality data may be limited. In this case, cost-effectiveness can be interpreted as the additional cost per unit time of achieving an additional beneficial biomarker response to treatment. A problem in this context is that biomarker assessment is typically subject to measurement error which leads to bias in assessing the benefit of a drug, and hence in the assessment of its cost-effectiveness. We discuss the adjustment of cost-effectiveness analyses for measurement error and consider the potential magnitude of bias that can arise. Using example calculations in the context of cholesterol-lowering therapy, it is demonstrated that such biases can be significant, leading to costs being overestimated by in excess of 25%. Ignoring measurement error in cost-effectiveness analyses can, therefore, have a substantial effect on the interpretation of such analyses.
|Number of pages||10|
|Journal||Applied Stochastic Models in Business and Industry|
|Publication status||Published - Sep 2006|