Sometimes It's the Little Things: Choosing Row vs. Column Percentages (303202)

Keywords: Tables, Figures, Graphics

Virtually all clinical-research papers include one or more Tables that summarize patient characteristics within their study cohort. However, the presentation of these Tables often leaves something to be desired. In particular, two oft-bungled decisions are a) how to define the "columns" (i.e. treatment group? study outcome? etc) for a basic descriptive table and then b) the selection of whether to present the row percentages versus the column percentage for categorical variables. The clinical researcher often struggles with this decision due to lack of familiarity with study design and/or understanding of how to display what the most meaningful statistics are, and the pure statistician may struggle as well if they lack sufficient understanding of the clinical question for a given paper. Through a series of targeted examples provided on the poster, this presentation will provide guidelines for young applied statisticians in choosing the summary percentage most useful to the clinical reader in a host of possible study designs. Ideally, it will also spur a larger conversation about working well with non-quantitative researchers to produce clean and readable results.