Visualizing Risk: Risk Graphics’ Impact on Patient Understanding and Choices in Preference Studies

OBJECTIVES: Including preferences of patients in the development of and decisions about treatments can optimize health outcomes. In health, specifically risk preferences (i.e., adverse reaction) are of interest. Since patients often have difficulties understanding risks, researchers frequently include graphics as visual aids. However, distinct features of different graphics can alter the way respondents process, perceive, and judge the probability of risks, potentially influencing their choices. Empirical evidence on which graphic is most suited for Discrete Choice Experiments (DCEs) is lacking.

METHODS: A literature review and structured individual interviews (n=13) were conducted to identify and score risk graphics in terms of gist and verbatim understanding, their perceived usefulness and risk processing in choice tasks. Consequently, three risk graphics were assessed on understanding, internal validity and reliability, the use of heuristics and choice predictions in a DCE study on flu vaccination (N=1,500 Dutch population aged 60+). Data were analyzed using mixed-logit models.

RESULTS: The eight graphics tested during the interviews impacted patients’ gist and verbatim understanding of vaccine effectiveness, risk of side effects and choices. The icon array, column chart and stacked column chart performed best and were selected for the subsequent DCE study. All three graphics were used by respondents and considered helpful. Although differences were found with regards to respondents’ comprehension, internal validity, reliability, detected heuristics and risk preference estimates, these differences were not statistically significant.

CONCLUSIONS: Icon arrays, column charts and stacked column charts seem to be equally effective and useful for visualizing risk in DCEs. Based on the reviewed literature, icon arrays may be more suitable to visualize magnitudes of absolute risks, while (stacked) column charts may be preferred when visualizing relative risk magnitudes. Although existing literature proposes that column charts may be associated with denominator neglect, we did not find any evidence for such effects in the DCE context.