Relative risk and odds ratio data are still portrayed with inappropriate scales in the medical literature

Abstract

Objective: To determine the frequency of inappropriate use of arithmetic scales to present relative risk (RR) and odds ratio (OR) data in figures among a group of high-profile general medical journals.

Study design and setting: Articles presenting RR or OR data in figures and published in one of five journals during 2002-2003 and 2007-2008 were evaluated to determine the type of scale used with the figures. Logarithmic and reciprocal (inverse) scales were identified as appropriate, whereas arithmetic scales were deemed inappropriate. If an article included one or more inappropriate figures, it was counted once as an incorrect publication.

Results: Fifty-two percent (53 of 101) of the articles with a graphical presentation of RR or OR data had used an inappropriate scale in 2002-2003 and 25% (58 of 231) in the time period 2007-2008. Although all five journals showed some improvement over the 5-year interval, there continued to be a problem with inappropriate presentations occurring with one in four articles.

Conclusion: Given the perpetuation of this problem, it may be time for the International Committee of Medical Journal Editors to address this issue by requiring that figures portraying ratio data use only logarithmic or reciprocal scales.