Visual risk graphics in a patient decision aid for atrial fibrillation: The role of usability in shaping risk perception and comprehension

Background Atrial fibrillation (AF) is associated with a substantially increased risk of stroke, and anticoagulant therapy is recommended for eligible patients to reduce this risk. Patient decision aids (PDAs) have been developed to support shared decision-making by improving patients’ understanding of stroke and bleeding risks. Visual risk graphics, including static and animated icon arrays, are commonly used in PDAs; however, prior studies have reported mixed findings regarding the effects of different visual formats on risk perception and comprehension. Objective This study aimed to examine whether differences among static and animated visual risk graphics remain observable when a PDA achieves acceptable usability. In addition, the associations among usability, numeracy skills, risk perception, and risk comprehension were explored. Methods This randomized controlled usability study evaluated an Atrial Fibrillation Decision Support tool incorporating three visual risk graphic formats: (A) static arrangement, (B) grouped animation, and (C) random animation. A total of 90 patients with non-valvular AF were randomly assigned to one of the three groups. Usability was assessed using the System Usability Scale (SUS), while risk perception gap, risk comprehension, and numeracy skills were measured using validated and literature-based instruments. Quantitative analyses included one-way ANOVA, correlation analyses, and ordinal logistic regression, complemented by semi-structured interviews. Results Ninety participants (77.2% male; mean age 66.98 ± 10.5 years) were included. All three versions of the PDA demonstrated acceptable usability, with mean System Usability Scale (SUS) scores above the benchmark of 68.05 (static: 73.58 ± 20.48; grouped animation: 78.67 ± 14.08; random animation: 76.25 ± 17.12). No statistically significant differences were found among graphic formats in usability, risk perception gap, or risk comprehension. Numeracy-based subgroup analyses yielded consistent results. Risk comprehension was positively correlated with usability (r = 0.453, p < 0.001) and negatively correlated with risk perception gap (r = −0.262, p = 0.013), while numeracy was weakly associated with usability (r = 0.247, p = 0.019). Ordinal logistic regression showed that age was the only significant predictor of risk comprehension (OR = 0.926, 95% CI = 0.879–0.975), whereas graphic format, numeracy, and other demographics were not significant. Qualitative interviews indicated higher perceived anxiety and cognitive burden with animated formats, particularly random animation, despite comparable quantitative performance. Conclusions When implemented within a patient decision aid with acceptable usability, static and animated risk graphics did not show statistically significant differences. Usability may be an important contextual factor associated with how visual risk graphics are experienced and interpreted within patient decision aids. These findings underscore the importance of prioritizing usability in the design and evaluation of visual risk communication in patient decision aids.