Abstract:
|
Economic evaluation techniques such as cost-effectiveness analysis (CEA) are now used widely in the assessment of new medical technologies. The growing trend towards trial-based CEA affords the analyst patient-level data on costs and effectiveness, but this has raised statistical challenges--"some old, some new"--for both design and analysis. Several statistical challenges will be explored: (1) confidence interval estimation methods for the key summary statistic of the incremental cost-effectiveness ratio (ICER); and (2) incremental net-benefit and cost-effectiveness acceptability curves as a means of transforming ICER into a single metric, with interpretation conditional upon the monetary value of health gain (we will emphasize the estimation of cost-effectiveness rather than the testing of hypotheses relating either to costs, effects, or cost-effectiveness); (3) skewed cost data and potential solutions that maintain the economists' interest in the arithmetic mean rather than the median; (4) censoring adjustment in cost-effectiveness studies. Finally, we will describe the inherent tendency of health economists to adopt a Bayesian perspective.
|