<p>Tests of fixed effects on answer accuracy.</p>

Interactions between self-medicating consumers and over-the-counter medication (OTC) influence the quality of information processed and, hence, the appropriateness of medication-use decisions. Previous work has shown that information deemed necessary for the safe and effective use of an OTC, information that is required to be present in the Drug Facts Label (DFLs), is often overlooked, and that decision-making related to OTC use could be improved. We proposed the concept of “personalized labeling” to address these shortcomings. This strategy uses an augmented reality interface which presents users with recommendations related to an OTC’s appropriateness for that individual’s use They receive the recommendations via a smart phone app that allows them to point their camera at packages on the store shelf that they are considering purchasing. Specifically, a virtual green check mark or a red stop sign is imposed over the interrogated product after the theoretical app recognizes the OTC and compares product specific warnings with the individual’s health history and medication usage. To develop proof of concept evidence for this strategy, we utilized a computer-based absolute judgement task. Seventy-two participants were randomly assigned to either a concept-educated group (educated on how personalized labeling would work) or a control group (uninformed about the personalized labeling strategy). Both groups viewed stimulus which included both standard and personalized labels to make binary decisions (yes/ no) related to a drug’s appropriateness for use by a theoretical patient considering single-ingredient OTC products. Decision accuracy and decision time were measured as indicators of labeling effectiveness and efficiency, respectively. Results showed that within the concept-educated group, participants made significantly more accurate (ME = 0.977, SE = 0.007) and faster (ME = 9.584s, SE = 0.854) decisions with the personalized label when compared to the standard label (accuracy: ME = 0.933, SE = 0.017; p = 0.002; time: ME = 19.052s, SE = 2.322; p < 0.001). In contrast, the control group showed no significant difference in accuracy or decision time between the two label types. Additionally, participants across both groups took longer to correctly answer “Yes” (appropriate for use) compared to “No” (not appropriate for use), reflecting that decision time are faster for target-present conditions than target-absent conditions. While educating consumers about the tool seems important, it overlooks that they must install the app and share their health and medication information to use it. As such, we interpret this finding as evidence that personalized labeling could significantly improve consumer decision-making related to the identification of appropriate products. However, future studies are needed with a broader range of populations, package types, use contexts, and real-world conditions.