Sleep and Cardiometabolic Disease: A Wearable-Electronic Health Record Study in the US All of Us Research Program (Circ CQO 2026)

10,362 All of Us adults with linked Fitbit and EHR data (median age 53, 70% female). Eleven sleep metrics from a 180-day baseline (at least 90 valid nights), outcomes counted after a 180-day washout, Cox models. Each 1-SD (29 min) higher night-to-night variability in sleep duration was associated with higher CVD-composite risk (HR 1.24, 95% CI 1.09-1.41), and bedtime, waketime and midpoint variability behaved similarly; mean sleep duration was not associated (0.96).

• Form factor: wrist (Fitbit, consumer)

Background: Sleep health spans duration, quality, regularity, and timing, yet Life's Essential 8 scores sleep by duration alone, and regularity studies used about 1 week of data. How these dimensions, measured over months, relate to incident cardiometabolic disease is unknown, as is whether nightly sleep-cycle structure adds information.

Methods: This cohort study included US adults in the All of Us Research Program with linked Fitbit and electronic health records (2009–2023). We derived 11 sleep metrics over a 180-day baseline (≥90 valid nights). Incident diagnoses, grouped into phecodes, were counted after a 180-day washout to limit reverse causation. Primary analyses related these metrics to prespecified endpoints, including a CVD composite (myocardial infarction, heart failure, cardiomyopathy, or ischemic stroke), heart failure, hypertension, and type 2 diabetes. Cox models adjusted for demographic, socioeconomic, lifestyle, and comorbidity factors. Exploratory analyses examined wearable-estimated sleep cycles (7,199 participants; 1,017,812 sleep-stage nights).

Results: Among 10,362 participants (median age, 53 years; 70% female), 230 developed the CVD composite (median follow-up, 3.7 years). Each 1-SD (29-minute) higher night-to-night variability in sleep duration was associated with higher CVD composite risk (hazard ratio, 1.24 [95% CI, 1.09–1.41]), as was higher variability in bedtime (1.23 [1.10–1.37]), waketime (1.20 [1.09–1.32]), and midpoint (1.21 [1.09–1.35]); mean sleep duration was not (0.96 [0.85–1.08]). Variability in sleep duration was also associated with heart failure (1.32 [1.13–1.55]), hypertension (1.16 [1.08–1.25]), and type 2 diabetes (1.20 [1.07–1.35]). More sleep cycles were associated with lower atrial arrhythmia risk (0.72 [0.58–0.89]), independent of sleep duration, regularity, and timing (0.58 [0.44–0.77]).

Conclusions: Night-to-night irregularity in sleep duration and timing was associated with the CVD composite, heart failure, hypertension, and type 2 diabetes; mean duration was not associated with the CVD composite. More wearable-estimated sleep cycles were associated with lower atrial arrhythmia risk beyond conventional sleep metrics, a novel finding requiring confirmation.

Verbatim from the publisher's DOI record (opens in a new tab) · Abstract as deposited by the publisher with Crossref; © the authors/publisher

No results from this paper on ChatPPG leaderboards yet. Browse the leaderboards.