Tobacco Use Trajectories and Associated Changes in Biometrics and Sleep During the First 72 Weeks of Wearable Membership: Observational Cohort Study

BACKGROUND: Tobacco use remains a leading preventable cause of morbidity and mortality. Digital health tools and wearable technologies offer scalable opportunities for behavioral self-monitoring. However, real-world evidence characterizing long-term tobacco use trajectories and associated physiological changes during wearable adoption is limited. OBJECTIVE: This study aims to characterize longitudinal trajectories of self-reported tobacco use during the first 72 weeks of wearable adoption and to examine associations between tobacco use and wearable-derived cardiopulmonary and sleep measures. METHODS: We analyzed data from 12,678 new wearable members (18-79 years) who contributed up to 72 weeks of daily self-reported tobacco use and wearable-derived biometric data. Longitudinal trajectories of tobacco use were examined across prespecified 12-week quarters (Q1-Q6) using generalized linear mixed-effects models. Associations between tobacco use and wearable-derived nocturnal resting heart rate (RHR), heart rate variability (HRV), respiratory rate (RR), and sleep duration were evaluated using linear mixed-effects models that accounted for within- and between-person variation and adjusted for demographic and temporal covariates. RESULTS: Across 3,765,573 person-days, the estimated daily probability of tobacco use declined from 55.1% (95% CI 53.8-56.4) during Q1 to 27.2% (95% CI 26.1-28.3) during Q6, representing an absolute reduction of 27.9 percentage points (95% CI -28.4 to -27.4). Among tobacco users with end-of-follow-up data, over one-quarter (1404/4975, 28.22%) reported no tobacco use during Q6. Greater logging engagement was associated with larger reductions in tobacco use; each 10-percentage-point increase in engagement corresponded to a 0.92-percentage-point greater decline from Q1 to Q6 (95% CI -1.58 to -0.26). Following tobacco use days, RHR was 1.71 beats/minute higher (95% CI 1.70-1.73), HRV was 3.54 ms lower (95% CI -3.59 to -3.49), RR was 0.19 breaths/minute higher (95% CI 0.19-0.20), and sleep duration was 9.78 minutes shorter (95% CI -10.08 to -9.49) relative to nonuse days. Reductions in tobacco use over time were associated with directionally favorable changes in RHR (P=.001), RR (P=.002), and sleep duration (P=.02). CONCLUSIONS: In this large-scale, observational, real-world study of wearable users, the probability of tobacco use declined over the first 72 weeks of adoption. Among participants with end-of-follow-up data, more than one-quarter reported no tobacco use during Q6. Tobacco use was consistently associated with less favorable cardiopulmonary and sleep measures, while reductions in tobacco use over time co-occurred with directionally favorable changes in these measures, although such changes may also reflect broader lifestyle or health changes. Overall, this study provides large-scale, longitudinal evidence that sustained reductions in tobacco use co-occur with favorable changes in physiological markers within a digital self-monitoring environment. As the findings derive from a single commercial ecosystem, independent replication in noncommercial, multiplatform settings will be needed to establish generalizability.