It's Not Just How Long You Sleep. It's What Kind.

A study published Thursday in the journal PLOS Medicine finds that how long a person sleeps may matter less than what kind of sleep they get. Drawing on nearly 96,000 adults in the UK Biobank who wore wrist accelerometers around the clock, researchers report that the amount of rapid-eye-movement and deep sleep a person logs is linked to future risk across dozens of diseases, independent of total sleep time.

The research, led by Jingsong Luo and Shengzhi Sun of Capital Medical University and Peking University in Beijing, is among the largest efforts yet to connect objectively measured sleep architecture, rather than a person's own estimate of how many hours they slept, to long-term health outcomes. Total sleep time has dominated public health messaging for decades. This analysis argues that the composition of that sleep carries independent weight.

The numbers

Researchers tracked 95,559 participants for a median of 8.9 years after they wore accelerometers for seven consecutive days, then cross-referenced sleep patterns against more than 1,000 subsequent disease diagnoses. After statistical correction for multiple comparisons, the team identified 156 significant associations between sleep measures and incident disease.

• Greater REM sleep was linked to lower risk of 83 conditions spanning cardiovascular, neurological, metabolic and psychiatric disease, including heart failure, dementia and Parkinson's disease.
• Greater deep sleep, also called slow-wave sleep, was linked to lower risk of seven conditions, among them type 2 diabetes and major depressive disorder.
• Across 86 disease categories with a measurable dose-response pattern, 69 showed the lowest disease risk at six to eight hours of total sleep.
• Sleeping less than five hours a night accounted for 37 of the 41 associations found between short sleep and elevated disease risk.

Sleep irregularity from one night to the next, and more time spent awake after initially falling asleep, were separately linked to higher risk of anxiety and substance use disorders.

How the sleep stages were measured

Large cohort studies have traditionally relied on self-reported sleep duration, a method known to be imprecise. This study instead used a deep-learning model, described by the authors as validated against 1,113 nights of laboratory polysomnography, to infer sleep stages from raw wrist-movement data collected by the accelerometers. That approach let researchers estimate REM, light and deep sleep at population scale without requiring participants to sleep in a lab.

It is also the method's central limitation. Independent sleep scientists who reviewed the paper for the Science Media Centre noted that wrist movement is a proxy, not a direct physiological reading. Dr. Nina Rzechorzek, a clinician scientist studying circadian rhythms at the University of Cambridge, said the sleep stages "were inferred by a deep-learning model from wrist movement, rather than measured directly using gold-standard polysomnography," and pointed out that when the researchers excluded disease diagnoses made within two years of the sleep measurement, to reduce the chance that early undiagnosed illness was itself disturbing sleep, only 95 of the original 156 associations held up.

"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 study authors wrote.

Who this affects

The cohort skews toward middle-aged and older adults in the United Kingdom, but the underlying question, whether short and fragmented sleep is a symptom or a driver of disease, is a live one in American public health as well. According to a 2024 data brief from the CDC's National Center for Health Statistics, 30.5% of American adults reported sleeping less than seven hours a night. Separate CDC survey data on insufficient sleep shows the burden falls unevenly, with the highest rates among Native Hawaiian and Pacific Islander adults, adults aged 45 to 64, and residents of some Southern and Appalachian states.

People managing chronic conditions already linked to poor sleep, including diabetes, depression and cardiovascular disease, are the population most likely to be affected by any eventual clinical application of this research, since it raises the possibility that treating fragmented or shallow sleep, not just insufficient sleep, could become a target for prevention.

Reaction and what happens next

Reaction from outside researchers has been notably measured, reflecting the gap between a striking correlation and a clinically actionable finding. Dr. Greg Elder, an assistant professor of psychology and associate director of the Northumbria Centre for Sleep Research, cautioned that consumer-grade tools remain a step removed from clinical measurement. "Movement-based watches can't yet accurately determine what sleep stage we are actually in overnight," he said, adding that the study can show association but not that increasing REM or deep sleep would itself prevent disease. He noted the seven-day monitoring window is generally considered representative of a person's habitual sleep pattern, which he called a strength of the design, even as he questioned the granularity of the stage classifications derived from it.

Coverage of the study by outlets including Medical Xpress and New Atlas emphasized that the paper's authors themselves describe the work as observational, not proof that altering sleep stages changes disease outcomes. The researchers write that short, fragmented sleep could just as easily be an early symptom of disease already brewing as a cause of it.

The authors have called for prospective studies that pair wearable-derived sleep staging with clinical polysomnography to confirm the accuracy of the deep-learning classifications at scale, and for research that tracks whether interventions to improve sleep depth or REM proportion change disease incidence, rather than simply correlating with it. Until then, sleep researchers on both sides of the debate converge on a narrower, less headline-grabbing point: the data reinforce existing guidance that adults aim for six to eight hours of nightly sleep, a target that public health agencies have promoted for years on the strength of total-duration studies alone. What this analysis adds is a first large-scale suggestion that the internal structure of those hours, not just their number, may be doing more of the work than previously understood.