Purpose: The purpose of this article is to examine the effects of presenting treatment benefits in different formats on the decisions of both patients and health professionals. Three formats were investigated: relative risk reductions, absolute risk reductions, and number needed to treat or screen.
Methods: A systematic review of the published literature was conducted. Articles were retrieved by searching a variety of databases and screened for inclusion by 2 reviewers. Data were extracted on characteristics of the subjects and methodologies used. Log-odds ratios were calculated to estimate effect sizes.
Results: A total of 24 articles were retrieved that reported on 31 unique experiments. The meta-analysis showed that treatments were evaluated more favorably when the relative risk format was used rather than the absolute risk or number needed to treat format. However, a significant amount of heterogeneity was found between studies, the sources of which were explored using subgroup analyses and metaregression. Although the subgroup analyses revealed smaller effect sizes in the studies conducted on physicians, the metaregression showed that these differences were largely accounted for by other features of the study design. Most notably, variations in effect sizes were explained by the particular wordings that the studies had chosen to use for the relative risk and absolute risk reductions.
Conclusions: The published literature has consistently demonstrated that relative risk formats produce more favorable evaluations of treatments than absolute risk or number needed to treat formats. However, the effects are heterogeneous and seem to be moderated by key differences between the methodologies used.