Abstract
BACKGROUND: For over a decade, drug overdose has been the leading cause of injury and accidental death in the United States. Most fatal overdoses involve opioids and occur during solitary drug use events when no one is available to initiate lifesaving responses (eg, naloxone). While there is a growing interest in devices providing early overdose detection and automated responses, little research has engaged end users in a device design process. OBJECTIVE: This study aimed to describe user experience, perceived harms and benefits, and the acceptability of a shoulder-mounted wearable sensor among people who inject drugs who wore a device prototype for 7 days and to explore real-time responses that could be incorporated into a next-generation sensor. METHODS: Individuals aged ≥18 years reporting past week injection drug use were recruited from a walk-in substance use disorder clinic in Camden, New Jersey. Participants completed a brief survey assessing sociodemographics and recent drug use and were assigned a sensor prototype that they were asked to wear near-continuously for 7 days. At endline, they completed semistructured interviews exploring acceptability, usability, and form and function preferences for next-generation devices with integrated automated response options. Structured field notes and transcripts were analyzed to identify key themes and design considerations. RESULTS: Participants (n=23) had a median age of 41 years. Most were primarily non-Hispanic White (65%) males (61%), experiencing homelessness (57%) who reported daily injection drug use (74%) within 3 months and receipt of opioid use disorder medication within a month (74%). A total of 16 people completed an exit interview. They found the concept of a shoulder-mounted overdose detection device acceptable and unanimously endorsed the need for long-lasting discreet devices. They emphasized the importance of having multiple response options that wearers could tailor to their individual circumstances and preferences, noting some might prefer an automatic call to emergency services, while others might prefer to alert a peer. Participants expressed a preference for solutions that alert first responders over automated biomedical solutions, such as naloxone injection, because of concerns about device error (eg, false positives) leading to precipitated withdrawal. CONCLUSIONS: After wearing a shoulder-mounted prototype, this small group of participants found the idea acceptable and provided feedback to improve usability and design. Data suggest that a variety of devices with differing functions, sizes, and capacities will be needed to meet user needs and increase the likelihood of adoption once devices come to market.