Comparative Analysis of Expert, Clinician, and Health Care User Interactions With Summary of Findings Tables: Usability Study.

Background: of findings (SoF) tables are widely used in systematic reviews and clinical practice guidelines to present evidence about health care interventions in a concise and transparent format. Although developed to improve accessibility and interpretation of evidence, previous studies have shown that users often experience difficulties understanding statistical information, certainty ratings, and the relationships between outcomes and treatment effects. Limited research has explored how different groups of users cognitively interact with SoF tables while solving evidence interpretation tasks, particularly when table complexity increases. Objective: This study aimed to investigate how different types of users-Grading of Recommendations Assessment, Development, and Evaluation (GRADE) or Cochrane experts, practicing clinicians, and health care users-interact with SoF tables of varying complexity while answering intervention-related questions, and to examine differences in search behavior, navigation patterns, and task performance. Methods: We used the Read&Learn tool (ERI-Lectura and LAIA-UV) in an online, single-session study to evaluate participants' interactions with SoF tables. Participants (n=120; 40 per group) accessed preselected SoF tables via a secure link and unique login. Participants completed tasks involving 4 SoF tables with increasing complexity. Specific table cells were blurred and had to be clicked to reveal information. Outcomes included the number of correct answers, total time spent on tasks, number of table cells visited, number of target and nontarget cells, and question-reading behavior. Results: Simpler SoF tables with a small number of outcomes and single target cells were correctly interpreted by most participants, regardless of their expertise. As table complexity and task demand increased, all participant groups demonstrated reduced performance and less efficient navigation patterns. Experts generally performed better than clinicians and health care users, particularly by spending less time reading nontarget cells and visiting fewer irrelevant table elements. Nevertheless, even experts experienced difficulties with tasks requiring synthesis and interpretation across multiple table cells. Questions requiring comparison and integration of information across outcomes resulted in the highest rates of incorrect responses in all groups. Heatmaps of the number of clicks and time spent on tables demonstrated that experts used more targeted search strategies, whereas clinicians and health care users explored a larger number of nontarget cells and spent more time navigating the tables. The most complex SoF tables produced the highest cognitive demands for all groups, suggesting that increasing element interactivity and information density substantially affect usability. Conclusions: SoF tables remain cognitively demanding even for experienced users of evidence synthesis. Increasing table complexity appears to reduce users' ability to identify, interpret, and synthesize relevant information. These findings suggest that current SoF formats may impose substantial intrinsic and extraneous cognitive load, particularly for nonexpert audiences. Future development of SoF tables should prioritize clearer presentation of clearer outcomes, inclusion of absolute alongside relative effects, and interactive or user-centered formats that support evidence navigation and interpretation.