Measuring costs in cost-utility analyses - Variations in the literature

Objectives: Although cost-utility analysis (CUA) has been recommended by so me experts as the preferred technique for economic evaluation, there is con troversy regarding what costs should be included and how they should be mea sured. The purpose of this study was to: a) identify the cost components th at have been included in published CUAs; b) catalogue the sources of valuat ion used; c) examine the methods employed for estimating costs; and d) expl ore whether methods have changed over time, Methods: We conducted a comprehensive search of the published literature an d systematically collected data on the cost estimation of CUAs. We audited the cost estimates in 228 CUAs. Results: In most studies (99%), analysts included some direct healthcare co sts. However, the inclusion of direct non-healthcare and time costs (17%) w as generally lacking, as was productivity costs (8%). Only 6% of studies co nsidered future costs in added life-years. In general, we found little evid ence of change in methods over time. The most frequently used source for va luation of healthcare services was published estimates (73%), Few studies o btained utilization data from RCTs (10%) or relied on other primary data (2 3%). About two-thirds of studies conducted sensitivity analyses on cost est imates. Conclusions: We found wide variations in the estimation of costs in publish ed CUAs. The study underscores the need for more uniformity and transparenc y in the field, and continued vigilance over cost estimates in CUAs on the pari of analysts, reviewers, and journal editors.