Abstract
OBJECTIVE: Although cost-effectiveness analyses (CEAs) have been advocated as a tool to critically appraise the value of health expenditures, it has been widely hoped that they might also help contain health care costs. To determine how often they discourage additional expenditures, we reviewed the conclusions of recently published CEAs. DATA SOURCES: A search of the Abridged Index Medicus (a subset of MEDLINE designed to afford rapid access to the literature of "immediate interest" to the practicing physician) between 1990 and 1996. STUDY SELECTION: We only included articles that reported an explicit cost-effectiveness (CE) ratio (a cost for some given health effect) in the abstract. DATA ABSTRACTION: From each abstract, we collected the value for the incremental CE ratio and the measure of health effect (life-years, quality-adjusted life-years [QALYs], other). We then categorized the authors' conclusion into one of three categories: supports strategy requiring additional expenditure, no firm conclusion, and supports low-cost alternative. Finally, we obtained the article and collected information on funding source. DATA SYNTHESIS: Among the 109 eligible articles, the authors' conclusion supported strategies requiring additional expenditure in 58 (53%) and supported the low-cost alternative in 28 (26%). We then focused on the 65 articles reporting either life-years or QALYs. Cost-effectiveness ratios ranged from $400 to $166,000 (per life-year or QALY) in the 39 articles (60%) in which authors supported additional expenditure, and ranged from $61,500 to $11,600,000 in the 13 articles (20%) in which authors supported the low-cost alternative. Despite identifying similar CE ratios, authors arrived at different conclusions in the overlapping range ($61,500 to $166,000). Of the 10 articles acknowledging industry funding, 9 supported a strategy requiring additional expenditure (p = .01 as compared with those without such funding). CONCLUSIONS: Authors of CEAs are more likely to support strategies requiring additional expenditure than the low-cost alternative. There is no obvious consensus about how small the CE ratio should be to warrant additional expenditure. Finally, concerns about funding source seem to be warranted.