Presenting decision-relevant numerical information to Dutch women aged 50-70 with varying levels of health literacy: Case example of adjuvant systemic therapy for breast cancer

Background If communicated adequately, numerical decision-relevant information can support informed and shared decision making. Visual formats are recommended, but which format supports patients depending on their health literacy (HL) levels for specific decisions is unclear. Study aim The aim of this study is to investigate: 1) the effect of survival rates and side-effects presentation formats on comprehension and 'feeling informed'; 2) differential effects among women with higher/lower HL, with adjuvant systemic breast cancer therapy as case example. Methods Two online experiments among women from the Dutch population without a history of breast cancer were conducted. Experiment 1 had a 3 (survival rate format: text block-bar graph- icon array) x 2 (HL: low-high) between-subjects design. Experiment 2 had a 5 (side-effects format: no probability information-probability information in numbers with or without a visualisation- probability information in numbers with or without a visualisation accompanied by a description of the side-effects) x 2 (HL: low-high) design. Primary outcomes were comprehension and feeling informed (Experiment 2 only). Formats were previously designed in cocreation with patients. Results In Experiment 1, presentation format did not affect gist or verbatim comprehension. Higher HL was associated with higher gist comprehension. Experiment 2 showed an interaction between presentation format and HL on 'feeling informed'. When provided with visualised probability information without a description of the side-effects, women with lower HL felt better informed than women with higher HL. Conclusion Visual formats did not enhance comprehension of survival rate information beyond a welldesigned text block format. However, none of the formats could overcome HL differences. When designing decision-relevant information, visualisations might not necessarily provide an advantage over structured numerical information for both patients with lower and higher HL. However, a deeper understanding of presenting side-effect information is warranted.