Abstract
BACKGROUND: N-of-1 trials have emerged as a personalized approach to patient-centered care, where patients can compare evidence-based treatments using their own data. However, little is known about optimal methods to present individual-level data from medication-related N-of-1 trials to patients to promote decision-making. OBJECTIVES: We conducted qualitative interviews with patients with heart failure with preserved ejection fraction undergoing N-of-1 trials to iterate, refine, and optimize a patient-facing data visualization tool for displaying the results of N-of-1 medication trials. The goal of optimizing this tool was to promote patients' understanding of their individual health information and to ultimately facilitate shared decision-making about continuing or discontinuing their medication. METHODS: We conducted 32 semistructured qualitative interviews with 9 participants over the course of their participation in N-of-1 trials. The N-of-1 trials were conducted to facilitate a comparison of continuing versus discontinuing a β-blocker. Interviews were conducted in person or over the phone after each treatment period to evaluate participant perspectives on a data visualization tool prototype. Data were coded using directed content analysis by two independent reviewers and included a third reviewer to reach a consensus when needed. Major themes were extracted and iteratively incorporated into the patient-facing data visualization tool. RESULTS: Nine participants provided feedback on how their data were displayed in the visualization tool. After qualitative analysis, three major themes emerged that informed our final interface. Participants preferred: (1) clearly stated individual symptom scores, (2) a reference image with labels to guide their interpretation of symptom information, and (3) qualitative language over numbers alone conveying the meaning of changes in their scores (e.g., better, worse). CONCLUSION: Feedback informed the design of a patient-facing data visualization tool for medication-related N-of-1 trials. Future work should include usability and comprehension testing of this interface on a larger scale.