OBJECTIVES

To analyze perspective and costing approaches used in published cost-effectiveness analyses (CEA) and their impact on reported incremental cost-effectiveness ratios (ICERs).

METHODS

We analyzed Tufts Medical Center’s CEA and Global Health CEA Registries, containing 6,907 cost-per-quality-adjusted-life-year (QALY) and 709 cost-per-disability-adjusted-life-year (DALY) studies, published from 1974 to 2018. Using collected information on analytic perspective, costing methodology, and reported ICERs, we examined how often published CEAs include cost components beyond those traditionally included and investigated how the inclusion of different cost components influences cost-effectiveness ratios.

RESULTS :

A healthcare sector perspective was most prevalent (75% of cost-per-QALY and 66% of cost-per-DALY literature). Some CEAs did not clearly state the perspective used (13% of cost-per-QALY and 20% of cost-per-DALY literature). Although most studies included intervention-related medical costs, CEAs seldom included unrelated medical costs. Cost-per-DALY studies were more likely to include at least one non-healthcare cost component than cost-per-QALY studies (34% vs. 21%). The most common non-healthcare cost component was productivity in the cost-per-QALY literature (11%) and patient transportation costs in the cost-per-DALY literature (21%). CEAs rarely assessed impacts on other non-healthcare sectors, such as education or criminal justice system. Of 18,206 cost-per-QALY ratios, the median ICER was $26,000/QALY (inter-quartile range, IQR:3,100-100,000), while 18% of ratios were cost-saving. Of 4,994 cost-per-DALY ratios, the median ICER was $440/DALY (IQR:75-3,400), while 7% of ratios were cost-saving. Among cost-per-QALY ratios calculated from a healthcare sector perspective, the median ICER ($25,000/QALY, IQR:3,100-100,000), was lower than the corresponding median from a societal perspective ($30,000/QALY, IQR:2,300-150,000). Non-industry sponsored studies reported higher ICERs than industry-sponsored studies from both perspectives.

CONCLUSIONS :

Key study attributes, including perspective and cost components, vary substantially in the CEA literature. Researchers and analysts should be transparent about their choices. Revised guidelines by the Second Panel, such as the inclusion of an Impact Inventory, help to reduce inconsistencies across studies and confusion.