Purpose: Understanding statistical information that describes the risks and benefits of treatment is difficult for many patients. Attempting to make this process easier and to improve patients' ability to make informed treatment decisions, many decision aids and health education materials now include graphical representations of the risks and benefits of treatment. However, little research has tested people's perceptions of graphs (e.g., helpfulness, effectiveness, trustworthiness) or from which type of graphs people prefer to receive their medical information.

Methods: 2450 subjects from a demographically balanced panel participated in an online survey. Participants were randomized to receive risk/benefit information presented in one of 5 formats: pie graph, bar graph, pictograph, sparkplug, or modified pie (clock) graph. Respondents' perceptions of the helpfulness, effectiveness (of describing risks and benefits), scientificness and trustworthiness were assessed through a rating exercise (0-6 Likert Scale) consisting of 8 total questions. After reviewing all 5 formats, respondents rank ordered them in terms of which format “they would prefer if they were trying to understand the risks and benefits of a treatment.”

Results: Bar graphs were most highly rated in terms of their perceived helpfulness (4.14) and effectiveness (4.16) of presenting risk/benefit information. However, pictographs were considered easiest for determining exact numbers, and were rated as the most preferred method for learning about the risks and benefits of treatment. While pictographs were rated as most trustworthy (3.62), bar graphs were rated as most scientific (3.61). People indicated that if they were reading an article, they would be more likely to look at an accompanying bar graph and least likely to look at a pictograph (4.08 vs. 3.72). In all other questions, pie graphs received the lowest ratings. When patients rank ordered their preference for the 5 graphs, bar graphs were an overwhelming favorite (44% of the first place votes), followed by pictograph (18%), clock (10%), sparkplug (14%), and pie (12%).

Conclusions: Although people show a strong preference for receiving their risk/benefit information in bar graphs, the less precise nature of this format may actually result in poorer comprehension of risk/benefit information. When deciding how to present statistical information, researchers and decision aid developers should consider both the preferences of the target audience as well as their ability to comprehend the specific material presented.