A large observational study tracking more than 95,000 people has found that REM sleep disease risk spans at least 83 conditions, with greater time in rapid-eye-movement sleep associated with lower rates of dementia, heart failure, and Parkinson’s disease. Published on 17 September in PLOS Medicine, the study also found that sleeping fewer than five hours per night was linked to elevated risk across 37 separate disease outcomes.
The research used data from 95,559 participants in the UK Biobank who wore wrist accelerometers for seven consecutive days and nights. Rather than relying on self-reported sleep, the team applied a deep-learning algorithm to estimate sleep stages (REM, deep, and light sleep) alongside total sleep duration, wakefulness after sleep onset (WASO), and night-to-night sleep irregularity. Participants were followed for a median of 8.9 years, with health records tested against more than 1,000 disease outcomes.
What the REM sleep disease risk data actually show
Greater REM sleep, measured per interquartile range of 47.6 minutes, was associated with lower risk of 83 diseases. The hazard ratios for individual conditions make the scale of the association concrete: heart failure came in at 0.74, dementia at 0.54, and Parkinson’s disease at 0.20. Greater deep sleep told a different story, linked to lower risk of seven conditions including type 2 diabetes and major depressive disorder rather than the broader sweep seen for REM.
The paper, authored by Jingsong Luo, Ruiyi Liu, Jie Yin, Wangnan Cao, Shengzhi Sun of Capital Medical University, China, and Rui Chen, identified 156 significant associations between sleep patterns and incident diseases overall, after Bonferroni correction, a statistical adjustment designed to reduce false positives when testing a large number of outcomes simultaneously.
Total sleep duration showed a non-linear relationship with disease risk. The lowest risk was concentrated in a six-to-eight-hour window, and that window held for 69 of the 86 disease phenotypes where significant non-linear associations were found. Below five hours, the picture worsens considerably, with people in that group facing elevated risk across 37 conditions.
Sleep irregularity and wakefulness: the numbers the headline missed
Beyond sleep stage composition and total duration, the study examined two further dimensions of sleep quality. According to the PLOS Medicine paper, greater sleep irregularity was linked to elevated risk of 3 diseases, while increased WASO (time spent awake after initially falling asleep) was associated with elevated risk of 6 diseases, including anxiety and substance use disorders. These are modest counts compared with the 83 linked to REM, but they add weight to the picture: it is not only how long or how deeply people sleep, but how consistently and continuously, that appears to matter for health outcomes.
Self-reported sleep data, the authors note, often correlates poorly with objective measures, which makes studies relying on questionnaires difficult to interpret. The accelerometer-based approach here sidesteps that problem to a degree, though it introduces its own assumptions, chiefly that a wrist device worn during normal daily life captures sleep stages with clinical accuracy.
The study draws its participants from the UK Biobank but frames its findings partly in a Chinese public health context. The 2026 China Sleep Health White Paper, cited in the paper, reports that the average surveyed Chinese person sleeps 6.97 hours per night, with 23 per cent sleeping fewer than 6 hours, and only 26 per cent maintaining a consistent sleep schedule. Sleep quantity and quality is a stated priority of China’s National Health Commission.
The authors are measured about what the data can support. ‘The findings provide additional evidence supporting the role of a 6-8 hours’ sleep duration as a health safeguard for middle-aged and older adults, likely attributable to more favourable distributions of sleep stages,’ they write. They also note directly that, as an observational study, the research cannot establish that sleep patterns directly cause changes in disease risk. Association and causation remain two different things, and no amount of careful accelerometry closes that gap on its own.
The study was supported by the National Science Fund for Distinguished Young Scholars, the Key Programme of the National Natural Science Foundation of China, and the Beijing Outstanding Young Scientist Programme, among other funders, who the authors state had no role in study design, data collection, analysis, or manuscript preparation. The paper is published as open access by PLOS.
