Alcohol use is a major contributor to global morbidity and mortality [1], and even low levels of consumption confer measurable health risk [2]. Despite growing awareness of this risk, alcohol remains among the most widely used drugs worldwide [3]. Although behavioral interventions can reduce intake, effects are often modest and short-lived [4], highlighting the need for scalable strategies supporting sustained, population-level reductions in alcohol use.

Wearable devices have become increasingly popular for continuous health monitoring, enabling real-time feedback on physiology and behavior. However, evidence linking wearable use to alcohol-related behaviors remains limited. A recent randomized controlled trial found that a wearable-supported behavioral program was associated with clinically meaningful reductions in alcohol use over 12 weeks [5]. To evaluate whether similar patterns emerge at scale and persist over longer durations, we examined longitudinal trajectories of self-reported alcohol consumption during the first 72 weeks of wearable membership in a large, real-world cohort.

The final dataset included 11,651,827 person-days (1,776,673 person-wk). Mean age was 34.4 (SD 10.5) yrs, BMI was 25.1 (SD 4.0) kg/m², and participants contributed a mean of 59.2 (SD 13.5) weeks (median 65 wk, IQR 18‐72 weeks), with an average of 6.6 (SD 0.9) alcohol entries per week (Table S1 in Multimedia Appendix 1).

The daily probability of alcohol use declined over time (Figure 1), from 23.0% (95% CI 22.7 to 23.3) in Q1 to 17.2% (95% CI 16.9 to 17.4) in Q6, an absolute reduction of 5.8 percentage points (95% CI −6.0 to −5.7). Results were robust in sensitivity analyses, including restriction to complete seven-day drink number reporting, where drinking volume decreased by 1.11 drinks per week (95% CI −1.15 to −1.07; Figure S2 in Multimedia Appendix 1).

The baseline daily drinking probability was higher among males than females (P<.001), but the relative decline from Q1 to Q6 was greater in females (P<.001; Figure 2). Baseline daily drinking probability differed across age groups (Ps <.001). Progressively larger relative and absolute declines in drinking were observed with increasing age (Ps <.001; Figure 3).

Both goal setters (n=13,298; Table S2 in Multimedia Appendix 1) and nongoal setters showed significant declines, though the relative decline was greater among non-goal setters (P<.001; Figure S5 in Multimedia Appendix 1).

In this large real-world cohort of 30,000 WHOOP members, self-reported alcohol use declined over the first 72 weeks of membership, with both drinking frequency and volume decreasing across age and sex. These findings extend recent experimental results [5] by describing sustained reductions in alcohol use over 72 weeks in a large real-world cohort.

The observed declines may reflect increased awareness of health behaviors, heightened sensitivity to physiological feedback, or broader lifestyle changes after wearable enrollment [7]. Importantly, participants engaged in active self-monitoring through journal entries, a behavior shown to promote reductions in drinking [8]. Additionally, large-scale wearable-based analyses demonstrate that even moderate alcohol consumption disrupts cardiac autonomic regulation, sleep, and next-day physical activity [9]. Awareness of these consequences may reinforce behavior change. Notably, declines in drinking frequency were accompanied by reductions in volume, suggesting decreased frequency was not offset by increased intake on drinking occasions. Although causal effects cannot be inferred, the persistence of these changes over more than one year highlights the potential for digital health platforms to support sustained alcohol-related behavior change at scale.

Several limitations should be considered. Alcohol use was self-reported without standardized drink definitions, and participants were subscription-based wearable users who may be more health-conscious than the general population. The requirement for sustained reporting engagement (≥18 wk) may further limit generalizability. Although reporting frequency remained relatively high across quarters and sensitivity analyses restricted to complete reporting weeks and members with full 72-week follow-up yielded nearly identical results, informative missingness cannot be excluded. Without a control group, observed changes cannot be causally attributed to wearable enrollment or distinguished from secular trends [10]. Although statistically significant, absolute changes were modest, and clinical significance at the individual level cannot be determined. However, dose-response evidence indicates that alcohol-related health risks increase continuously with consumption [11], with no established safe threshold [12], suggesting even modest population-level reductions may carry public health relevance, particularly when sustained over time and observed across demographic subgroups. Longer follow-up is needed to evaluate persistence beyond the study window. Despite these limitations, this study provides one of the largest longitudinal assessments of self-reported alcohol use to date, describing long-term trajectories of drinking frequency and volume over 72 weeks among new WHOOP members.