Abstract
BACKGROUND AND OBJECTIVES: Determining whether a patient should discontinue antiseizure medication after a period of seizure-freedom is often challenging. Risk prediction tools can support shared decision-making. End-user stakeholder engagement throughout the tool development process is critical. We examined clinician views regarding potential novel graphically based seizure risk prediction tools. METHODS: We conducted 6 qualitative 1-hour focus groups of 3-5 providers who see patients with epilepsy (total N = 25). We used purposive sampling as informed by previous literature to include both epileptologists and general adult neurologists from geographically diverse settings. We obtained feedback from respondents regarding several possible novel seizure risk prediction tools and asked for any other suggested risk prediction formats. Interviews were transcribed verbatim and analyzed by independent readers using both deductive (researcher-driven) and inductive (response-based) reasoning. RESULTS: Although some respondents favored numerical-only calculator outputs, others favored graphical outputs, such as cumulative incidence functions (displaying risk across time) for themselves and pictographs (static patient icons colored to represent how many patients might have a seizure) to share with patients, including a comparison between continuation vs discontinuation (not available in existing calculators). Although responses were mixed, some respondents typically felt that CIs surrounding predictions would be cluttering and not improve their gist graph understanding. Other suggested output formats could include "stoplight" summary displays or decision charts. DISCUSSION: Our results provide vital feedback to direct future risk prediction tool development efforts. Although some preferred numerical-only presentations for their simplicity, others cited that graphical displays have many advantages regarding clarity and amount of information. Future studies will also require patient input.