Clinical Applications of Wearable Devices for Individuals with At-Risk Substance Use and Substance Use Disorders: A Systematic Review

This PROSPERO-registered and PRISMA-compliant systematic review was conducted by a team from the University of Toronto and the Centre for Addiction and Mental Health, Toronto, Canada. The authors included English-language studies published between 2000 and July 2023 examining wearable devices (defined as portable, wireless, passively-measuring accessories or clothing) among adults with substance use disorders (SUDs) or predetermined at-risk use. Devices detecting drugs biochemically alone and standard polysomnography were excluded. From 2,743 titles, 33 studies met the inclusion criteria. Most used were wrist-worn devices (25 of 33) with the remainder chest bands or a clip-on sensor. Opioids were the most studied substance (9 studies), followed by alcohol and stimulants (5 each), cannabis and polysubstance use (4 each), and tobacco and general at-risk use (3 each). 23 studies were observational, 9 randomised controlled trials and one case study, with samples ranging from one to 119 participants. All were rated good quality on a modified NIH tool although few justified sample size and SUD ascertainment varied considerably.

The findings were organised into three domains. Short-term use was generally well tolerated, with willingness highest for discreet, comfortable devices. Participant acceptability of hypothetical overdose-detection devices was shaped by concerns over location tracking, manual activation burden and false positives, while participants without access to stable accommodation were notably less willing than housed participants to wear a device that might alert others or trigger police involvement. Sleep was the most thoroughly studied outcome, with actigraphy across alcohol, cannabis and opioid-using populations consistently showing reduced sleep efficiency, longer onset latency and greater night-to-night variability. Additional evidence showed that regularity improved with treatment and that early variability predicted higher relapse odds. Five further studies examined wearable-derived predictors of craving, stress and recurrence, and six examined direct detection of use or overdose events. Evidence was strongest for discrete behaviours such as cocaine use or smoking gestures, and weakest and least consistent for opioid-related events. Nevertheless, no study demonstrated that wearable monitoring improved long-term recovery outcomes and real-world overdose detection remained unverified.

Relapse rates following SUD treatment remain high and contemporary models frame relapse as a fluctuating, moment-to-moment phenomenon shaped by shifting affect, stress and craving. These dimensions are poorly captured by cross-sectional or retrospective assessment, and wearable devices are attractive to researchers and treatment professionals because they promise continuous, passive and real-world data that could complement or reduce the burden of assessment, which remains a major component of treatment expense. Sas et al have produced an admirable study that shines a light on the developing nature of the wearables field in addiction. The technology demonstrably picks up physiologically meaningful signals, but the literature remains fragmented, overwhelmingly short-term and has not yet asked the most pressing clinical question - whether monitoring changes outcomes.

The opioid overdose-detection literature illustrates where the field currently sits. Interest and enthusiasm was high among participants in “detection-and-reversal” devices with automated naloxone administration and tied to past overdose experience, particularly among those at elevated risk following incarceration or residential treatment. Overdose events, however, were rare or entirely absent during the monitoring windows of studies using real devices, leaving claims about detection accuracy essentially untested. Sas et al are appropriately careful to separate acceptability evidence produced under hypothetical testing conditions from performance evidence drawn from real-time trials, a distinction that is often not present in the broader literature. Their most uncomfortable finding is a paradox in compliance, where a passive monitor cannot mitigate risk if it is removed or left uncharged during the period it was deployed to observe.

For clinicians and researchers considering wearables as an adjunct to SUD care, this review offers a useful and sobering map of the evidence. Actigraphic sleep monitoring is the domain best supported for near-term use and the finding that sleep instability rather than poor sleep per se tracks relapse risk is a genuinely actionable insight worth building on. Elsewhere, the case for adoption is somewhat weaker than the technology's popularity suggests. It is striking that across a search window spanning twenty three years no study has tested whether wearable-informed monitoring translates into reduced relapse, improved engagement or safer overdose response. Until that gap closes, wearables should continue to be treated as a research tool rather than a validated clinical intervention.

The review also foregrounds an equity dimension deserving serious attention - that those most vulnerable to overdose are least willing to wear detection devices, and design factors such as device bulk, resale value, stigma and perceived lack of discretion meaningfully affect device uptake. Future implementation, particularly of overdose alert systems, will need to engage these preferences directly rather than assuming that technical capability drives adoption. Questions of data ownership, clinical responsibility for alerts and validated response pathways remain unresolved and will likely emerge as a pressing policy area as the field matures and such technology becomes more widely adopted. We should applaud Sas et al for their highly comprehensive approach to the subject of passive wearable devices for people with SUD, yet it remains imperative that we not lose focus on the bedrock of addiction treatment - an individualised, integrated, patient-centred and non-stigmatising social and medical approach personalised to the unique situation of the people who require it.