Abstract
Patient decision aids are tools intended to facilitate shared decision-making. Currently development of a patient decision aid is resource intensive: it requires a decision-specific review of the scientific literature by experts to ascertain the potential outcomes under different treatments. The goal of this project was to conduct a formative evaluation of a generalizable, scalable decision aid component we call Veterans Like Me (VLme). VLme mines EHR data to present the outcomes of individuals "like you" on different treatments to the user. These outcome are presented through a combination of an icon array and simulated narratives. Twenty-six patients participated in semi-structured interviews intended to elicit feedback on the tool's functional and interface design. The interview focused on the filters users desired with which to make cases similar to them, the kinds of outcomes they wanted presented, and their envisioned use of the tool. The interview also elicited participants information needs and salient factors related to the therapeutic decision. The interview transcripts were analyzed using an iteratively refined coding schema and content analysis. . Participants generally expressed enthusiasm for the tool's design and functionality. Our analysis identified desired filters for users to view patients like themselves, outcome types that should be included in future iterations of the tool (e.g. patient reported outcomes), and information needs that need to be addressed for patients to effectively participate in shared decision making. Implications for the integration of our findings into the design of patient decision aids are discussed.