Higher REM fraction tracks with lower risk of 83 diseases in 95,559 wrist recordings

Eighty-three associated conditions out of more than 1,000 outcomes tested works out to at most about 8 percent of the scan ^[1]. Deep sleep's 7 conditions is under 1 percent ^[2]. Bonferroni is a strict correction at that test count, so the 156 surviving associations are not a multiple-comparisons artefact ^[6].

The increment matters as much as the hazard ratio. The REM estimates are scaled per interquartile range, which here is 47.6 minutes of REM ^[5]. A hazard ratio at that increment compares someone near the top of the middle half of the distribution with someone near the bottom of it. For Parkinson's disease the figure was 0.20 ^[8], an 80 percent lower hazard ^[3].

Exposure was measured once. Participants wore the accelerometer for seven consecutive days and nights, a deep-learning algorithm assigned the stages, and diagnoses were then counted for a median of 8.9 years ^[1][2][3]. The release gives the poor correlation between self-reported and objectively measured sleep as the reason real-world sleep patterns have been hard to study at all ^[16]. It also states that the study cannot establish that sleep patterns directly cause disease risk ^[9]. The design cannot say whether a diagnosis arriving in year seven had already been trimming REM during that one week in year one. The 0.20 for Parkinson's disease is the estimate most exposed to that question.

The two counts, 86 phenotypes with non-linear duration associations and 83 diseases associated with REM, sit close together, though they are not the same unit ^[5][11]. Night-to-night irregularity and wakefulness after sleep onset each tracked higher risk of several conditions, among them anxiety and substance use disorders ^[17].

"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 favorable distributions of sleep stages," the authors said ^[13]. The authors credit stage distribution as the explanation and issue the recommendation in hours. The same statement goes further, saying that maintaining 6 to 8 hours "can effectively reduce the risk of multiple diseases" ^[14]. That is causal phrasing on a design the release describes as unable to show causation ^[9].

One boundary on all of it. The cohort is British, drawn from UK Biobank ^[1], while the policy case set out in the release is Chinese: the 2026 China Sleep Health White Paper puts the average surveyed sleeper at 6.97 hours a night, with 23 percent under 6 hours and only 26 percent reporting a consistent schedule ^[15]. The hazard ratios were estimated in the British cohort. The paper comes from Shengzhi Sun of Capital Medical University and colleagues and was published in PLOS Medicine on September 17 ^[4].