Communicating Probabilities of Cervical Cancer Screening Results with Icon Arrays or Tree Diagrams: A Longitudinal Experiment.

Abstract

Simple graphical displays such as icon arrays and tree diagrams have been proposed for communicating health risks and supporting informed decisions. The UK's National Health Service (NHS) uses tree diagrams to communicate the chances of different cervical cancer screening results, but their effectiveness has not been compared to icon arrays. We conducted a well-powered longitudinal experiment involving 3,100 UK women eligible for cervical screening (25-64 years) to examine the effectiveness of icon arrays to communicate possible cervical screening results, relative to the UK NHS's tree diagram and to a numerical-only format. We also examined whether the presence (vs. absence) of explanatory text referring to different types of results (i.e. distinguishing between HPV positive results with vs. without abnormal cervical cells) moderated effects of presentation format. Presentation format did not affect verbatim or gist knowledge of probabilities at initial assessment (i.e. immediately after participants viewed the displays), but icon arrays were associated with better gist knowledge of absolute magnitudes than tree diagrams and numerical-only formats at 1-month follow-up. Participants exposed to icon arrays also perceived lower likelihood of adverse screening results and reported stronger screening intentions at initial assessment. For displays without explanatory text, icon arrays were also associated with more positive user evaluations and less negative affective reactions than tree diagrams at initial assessment. Overall, our findings suggest that icon arrays support enduring knowledge of approximate magnitudes of probabilities and are better suited than tree diagrams for communicating possible screening results.