On Projection of Costs and Incremental Cost-Effectiveness Ratios: A Data Example

Cost and cost-effectiveness analysis has been more and more important to both policy makers and clinicians in assessing health care delivery. The incremental costeffectiveness ratio (ICER) is widely used inpractice to evaluate the relative health benefit of one treatment over another. Due to the short time span ofclinical trials, the cost estimated from observed data are often right censored. The ICERs obtained from observed data are thus limited to a short time window. But for some intervention it is life time costs and benefits that health decision makers need to calculate the true cost effectiveness. Thus it is necessary to project the ICERs to a longer time horizon based on available data. In order to project ICER, we need to first project both the cost and health effectsince ICER is a ratio betweenthe two. Two difficulties encountered in projecting costs using monthly cost data are the death cost, which has to be counted backward, and the last month cost which is always incomplete. Using data from a four-year multicenter clinical trial study, we describe a method of projecting the future medical costs, using a mixed effect model from observed data. Combined with a method of projecting the survival distribution for different hypothetical scenarios, we are able to project future ICERs.