Design Principles for Interactive Dashboards in Drug Safety Surveillance: Design Science Research

Adverse drug reactions pose a serious threat to health care, leading to patient harm and substantial economic burden. Dashboards for drug safety surveillance are a valuable tool to tackle it.

This qualitative study aims to develop a dashboard for drug safety tracking with active and iterative involvement of end users. To support dashboard development, we formulate and iteratively refine design principles (DPs) for drug safety dashboards using the affordance theory.

Following a design science research approach, we conducted 3 cycles of iterative design and evaluation involving the end users. Four professional end users (with expertise in drug screening, drug discovery, and data science) and 6 nonprofessional end users (drug consumers) were engaged in the requirements gathering sessions through co-design workshops, usability testing through think-aloud sessions, and heuristic evaluation.

The analysis resulted in a set of 8 DPs refined using the prototype's affordances (ie, actionable properties of the dashboard that guide user interaction and interpretation of data). Following themes emerged in the formulation and refinement of DPs: addressing the bootstrap problem through designing for immediate use (DP1a), allowing identification of patterns through visualizing causality while signalizing uncertainty (DP1b), tracking trends for relevant variables (DP1c), implementing user-controlled views (DP2a) and customizable levels of data granularity (DP2b), guiding user's visual attention through spatial layouts (DP2c), designing for higher public value (DP3a) and providing features to support decision-making for varied stakeholders' groups (DP3b). A high-fidelity dashboard prototype for drug safety surveillance was proposed as a result of applying the final set of DPs. Heuristic evaluation of the prototype revealed an overall usability score of 84%.

Applying DPs rooted in affordance theory led to a purposeful and user-relevant artifact that can improve understanding of drug safety data and potentially guide decision-making processes for professional users. Our theoretical contribution lies in providing refined DPs, while also demonstrating how affordances can aid dashboard development in pharmacovigilance. Our findings may be applicable to similar health information systems in related domains.