Sleeping nine hours or more tracked with a 17% higher death rate over 30 years in Taiwan

Most worry about sleep runs one way: too little of it. A long Taiwanese study adds weight to the less familiar half of the story. Adults who said they usually slept nine hours or more died at a somewhat higher rate over the following three decades than those who slept seven or eight, while people who slept six hours or less showed no clear difference.

The study, published in September in the Journal of Sleep Research by Yu-Chien Lin, Kuo-Liong Chien, Jui Wang and colleagues at National Taiwan University, drew on the Chin-Shan Community Cardiovascular Cohort in Taiwan. Over a median of 30.8 years, long sleep was associated with increased risks of all-cause mortality, by about 17%. That follow-up is longer than any cohort in the best-known pooled analysis of sleep and death, and the result is smaller than the same cohort’s earlier results suggested, and seen only for long sleep, not short.

Why long sleep keeps turning up in death statistics

Researchers have been counting deaths by sleep length for decades, and long sleepers have been an awkward feature of the results. In 2010, Francesco Cappuccio and colleagues at the University of Warwick pooled 16 studies covering nearly 1.4 million people. Short sleepers had a 12% higher risk of dying during follow-up. Long sleepers fared worse, with a greater risk of death of about 30% in the pooled estimate.

Later reviews kept finding the same tilt toward the long end. A 2017 analysis led by Maki Jike at Nihon University, drawing on 137 prospective cohort studies and more than 5 million people, put the excess death risk for long sleepers at 39%. A 2018 dose-response analysis of 74 studies led by Chun Shing Kwok at Keele University found the risk climbing with every hour beyond eight, from about 14% higher at nine hours to 47% at eleven. For people sleeping under seven hours they found no difference that could be told apart from chance, and they concluded that longer duration of sleep may be more associated with adverse outcomes than short sleep. A 2020 review limited to older adults pooled 28 articles, involving a total of 95,259 people, and found a clear link for long sleep and only a marginal one for short sleep.

Most of that evidence rests on people’s own reports of their sleep, a weakness the new study shares: in Cappuccio’s review, sleep duration was assessed by questionnaire. The gap the Taiwan study fills is time. Follow-up in Cappuccio’s cohorts ranged from 4 to 25 years, so none of them watched people for three decades.

The Chin-Shan cohort had already weighed in once. In 2010, Chien’s group reported on 3,430 adults aged 35 years or older followed for a median of 15.9 years, during which 901 died. They found a U-shape association, with risk rising at both ends and the steepest rise among people sleeping nine hours or more, whose risk was 43% higher in an analysis adjusted for age and sex. Sleep quality counted too: in that report, sleep duration and insomnia severity were associated with all-cause death. People reporting insomnia nearly every day had a 70% higher death rate than those without it, and those who slept nine hours or more while also having frequent insomnia had about two and a half times the death rate of the comparison group. The new paper returns to the same community with almost twice the follow-up.

How the Chin-Shan sleep study followed 3,305 Taiwanese adults for three decades

The Chin-Shan Community Cardiovascular Cohort is a prospective cohort of participants from one community in Taiwan, enrolled in 1990-1991 to study heart disease and its causes. Each participant answered structured questionnaires about their usual sleep, and the researchers placed them in one of three groups: six hours or less, seven to eight hours, or nine hours or more.

For this analysis the team excluded individuals with incomplete data and anyone who already had heart disease or stroke, leaving 3,305 participants. Leaving out people who were already ill matters, because illness can change how long someone sleeps, and including them would tilt the comparison against long sleepers. The researchers then tracked who died, and from what, over a median of 30.8 years.

They compared death rates among the three sleep groups after adjusting for potential confounding factors, differences between the groups that could otherwise muddy the picture. They checked whether the pattern held across men and women, older and younger people, and people with and without obesity, high blood pressure or diabetes. And they ran a mediation analysis, a statistical way of asking how much of a link travels through a third factor, to see whether markers of hardened arteries or blood pressure might carry part of it.

How much higher the death rate was for long sleepers

Only long sleep stood out. Compared with people who slept seven to eight hours, those who slept nine hours or more had a death rate about 17% higher over the follow-up. The uncertainty around that figure was wide enough that the true difference could be as small as 1% or as large as 35%, so the result clears the usual statistical bar, but only just.

For deaths from heart disease and stroke, the estimate for long sleepers was similar, about 18% higher, but the authors report that long sleep was not significantly associated with cardiovascular mortality: the plausible range ran from somewhat lower risk to clearly higher. Short sleep, six hours or less, showed no significant association with either outcome, meaning no difference that could be told apart from chance.

The link looked similar in the sub-group analyses by age, sex, body weight, high blood pressure and diabetes, so it was not confined to one kind of person.

Set beside the cohort’s 2010 report, the long-sleep link has shrunk, from 43% to 17%, and the U-shape has lost its short-sleep arm. The two numbers are not strictly comparable. The earlier figure was adjusted only for age and sex and used seven hours as its reference, while the new one carries a fuller adjustment and compares against seven to eight hours. Even so, the direction of travel is toward a smaller effect.

Was blood pressure the link between long sleep and death?

The authors suggest one route. In their words, the extra risk “may be partially mediated by systolic blood pressure,” the top number in a blood pressure reading, which measures the pressure in the arteries as the heart beats. If long sleepers tended to have higher blood pressure, and higher blood pressure shortens life, part of the link would run through it.

That is a hypothesis about a path, not proof of one. Mediation analysis can show that numbers line up in a way that fits a path, but in an observational study it cannot rule out that something else raises both sleep time and blood pressure.

Is long sleep a cause of early death or a symptom?

The authors are confident in their reading. “This study found that longer sleep duration increases the risk of all-cause mortality,” they write. The design cannot carry that verb. Nobody’s sleep was changed, so the study shows that long sleepers died sooner on average, not that sleeping longer made them do so.

There are plenty of reasons a long sleeper might die sooner without the sleep being to blame. Kwok’s review lists depressive symptoms, low socioeconomic status, unemployment, and low physical activity among the things that travel with long sleep and could confound the link, and notes that long sleep may stem from comorbidities that lead to fatigue, such as chronic inflammation or anemia. It may also be an early sign of a disease that has not yet been diagnosed. Excluding people with heart disease at the start removes some of that, but not illnesses the study did not record.

Two more limits apply. Sleep was self-reported, and people are not precise judges of how long they sleep. And three decades is long enough for habits to change: someone sleeping nine hours at the start may have slept quite differently twenty years later. Both problems would blur the comparison, which may be part of why the effect is modest. The new analysis also leaves sleep quality aside, although the same cohort’s earlier report linked frequent insomnia with higher death rates.

The wider field is stuck at the same point. After pooling more than 5 million people, Jike’s team wrote that the next studies need to establish whether the relationship between long sleep and health outcomes is causal and modifiable, meaning whether changing sleep would change the outcome at all.

What the Taiwan sleep study adds to sleep advice

The study does not change what health agencies say. The US National Heart, Lung, and Blood Institute puts the expert recommendation for adults at 7 to 9 hours a night. The agency adds that more than 9 hours is not necessarily harmful, and may help young adults, people recovering from lost sleep and people who are sick. The Taiwan study’s long-sleep group started at nine hours, the top of that range, and it compared groups of people; it does not mark nine hours as a danger line for any one person.

What it adds is time. Thirty years of follow-up in a Taiwanese community is rare, and its message fits the pooled evidence: long sleep marks a group with somewhat higher mortality. Whether a change in someone’s sleep needs looking into is a question for their doctor.

For now, nine hours or more of sleep is better read as a signal worth understanding than as a habit shown to shorten life.