Abstracts

Rationale: In the US, 460,000 children have epilepsy. Many respond well to treatment in the outpatient setting. Quick access to pertinent data about a population of patients may help physicians improve the delivery of care to children with epilepsy. Readily available commercial software can visually communicate complex data to physicians. However, it is unclear what the core concepts are that should be displayed for a population of children with epilepsy. The present research aims to advance visualizations for pediatric epilepsy population management by interviewing physicians to uncover these core concepts. Methods: This is a qualitative methods research project. A convenience sample of pediatric neurology experts participated in a 20-minute interview designed to identify clinical concepts relevant to managing a population of children with epilepsy. The interviews were transcribed and reviewed. We performed thematic analysis on the transcripts to generate a preliminary list of core concepts for display. In parallel, as the interviews were conducted, we developed prototype clinical dashboards to explore potentially relevant visualizations. These prototypes were based on a sample of clinical data from the Weill Cornell Medicine pediatric epilepsy practice, and designed in Tableau, a commercial data visualization software package (http://www.tableau.com). Results: The interview phase of the research concluded after 14 interviews. Thematic saturation was estimated to occur at the 11th interview. We performed three additional interviews for confirmation. The clinical concepts extracted from interview analysis spanned 9 core topics, with 33 sub-categories. The progression of the prototype dashboards is shown in Figure 1. These potential dashboards illustrate the potential to display data related to geography, visit history, medications, imaging, and epilepsy diagnosis. Conclusions: There is a core set of clinical concepts relevant to identifying children with epilepsy with increased risk of emergency department visits. Of these clinical concepts, many are stored in the electronic health record, and could be readily extracted for use in clinical dashboards. Access to robust and relevant data visualizations, such as the prototype dashboards, could assist healthcare providers in identifying patients with increased likelihood of epilepsy-related emergency department visits. Funding: Internal Funding from the Department of Pediatrics, Weill Cornell Medicine