Formatting modifications in GRADE evidence profiles improved guideline panelists comprehension and accessibility to information. A randomized trial

Objective: To determine the effects of formatting alternatives in Grading of Recommendations Assessment, Development, and Evaluation (GRADE) evidence profiles on guideline panelists' preferences, comprehension, and accessibility. Study Design and Setting: We randomized 116 antithrombotic therapy guideline panelists to review one of two table formats with four formatting alternatives. After answering relevant questions, panelists reviewed the other format and reported their preferences for specific formatting alternatives. Results: Panelists (88 of 116 invited [76%]) preferred presentation of study event rates over no study event rates (median 1 [interquartile range (IQR) 1] on 1-7 scale), absolute risk differences over absolute risks (median 2 [IQR 3]), and additional information in table cells over footnotes (median 1 [IQR 2]). Panelists presented with time frame information in the tables, and not only in footnotes, were more likely to correctly answer questions regarding time frame (58% vs. 11%, P < 0.0001), and those presented with risk differences and not absolute risks were more likely to correctly interpret confidence intervals for absolute effects (95% vs. 54%, P < 0.0001). Information was considered easy to find, easy to comprehend, and helpful in making recommendations regardless of table format (median 6, IQR 0-1). Conclusion: Panelists found information in GRADE evidence profiles accessible. Correct comprehension of some key information was improved by providing additional information in table and presenting risk differences.