Variation in sleep duration linked to cognitive decline – UW Medicine

Older adults whose sleep duration varied over the course of a two-decade-long study were much more likely to develop cognitive decline compared with those whose sleep duration did not vary significantly, new research shows.

“Our finding tells us that maintaining healthy, consistent sleep habits long-term may be important in optimizing brain health as you age. So making sure that good sleep is a regular part of your life —not just on weekends, and not just on vacations — is important,” said Dr. Jeffrey Iliff, a professor of psychiatry and behavioral sciences and neurology at the University of Washington School of Medicine.

Iliff, who also conducts research at the VA Puget Sound Health Care System, led the investigation. Samantha Keil, formerly of UW and now a senior research fellow at Weill Cornell Medicine, was the paper’s lead author. They and colleagues report their findings today in the journal JAMA Network Open. 

Previous research has suggested that sleep of short duration, less than six hours a night, increases the risk of cognitive decline. The new study also found this to be true. But the finding that variation of sleep duration appeared to affect cognitive decline is new.

The study followed 826 older adults, average age 76, who had no signs of cognitive decline at the time they enrolled. During the study, which lasted nearly two decades, participants filled out five questionnaires in which they reported how long they had slept in the previous seven days. They also took a series of neuropsychological tests to assess cognitive function, including standard tests of attention and memory used to detect cognitive impairment in aging populations.

The study tested whether participants’ sleep patterns were associated with how long they lived before they experienced clinically relevant levels of cognitive impairment.

In their analysis, the researchers grouped the participants into three groups: “short sleep,” if they slept less than seven hours; “medium sleep,” if they slept right about seven hours; and “long sleep,” if they slept more than seven hours.

They found that being a “short sleeper” was associated with a more than 3.6-fold increased risk of developing cognitive decline, a finding in line with previous research. But they also observed that people whose sleep changed the most year-to-year were associated with a more than threefold increased risk.

Iliff said that previous research probably missed this association between variable sleep duration and cognitive decline because most assessed participants’ sleep duration only at one or two points in time instead of following them over many years as they aged.

It is unknown why short, or disturbed, sleep and cognitive decline appear linked, and the connections between these two processes are complex. For one, the degenerative processes that cause cognitive decline also damage the brain’s sleep-regulating centers, leading to poorer sleep. It also appears that poor sleep quality itself harms the brain, leading to impaired cognitive function.  This may happen because toxic proteins that accumulate in the brain during our waking hours are cleared during sleep. Less sleep, then, would mean more waking hours for the proteins to accumulate and less time for them to be cleared.

“The data coming from several epidemiological studies suggest that it is the sleep that you get in midlife that determines your risk for cognitive decline later in life,” Iliff said. “So people should focus on good sleep now as a priority, the same way they might focus on maintaining a healthy diet or exercising regularly.”

Other key collaborators were Dr. Abigail Schindler at the VA Puget Sound Healthcare System, Dr. Sherry Willis of the UW Department of Psychiatry and Behavioral Sciences, and Dr. Miranda Lim from the Oregon Alzheimer’s Disease Research Center and VA Portland Health Care System.

 This work was supported by the U.S. National Institutes on Aging (P30AG066518, AG052354, AG055653-03, AG054456) and the U.S. Veterans Administration (RX002947).