Trouble falling asleep was linked to faster memory decline in midlife, but not in old age

Poor sleep is often listed as a risk factor for dementia. But studies disagree: some find that people who sleep badly decline faster, others find no link, and a few find that older people who complain of poor sleep actually do better. A team led from the University of California, San Francisco, asked whether age explains the confusion. In about 17,000 adults, they found that the same sleep complaint meant different things at different ages.

This is an observational study: it shows links, not cause and effect. It is also a preprint, not yet peer-reviewed.

The idea behind it is that sleep and the brain affect each other in both directions. In midlife, poor sleep might harm the brain over the years. In old age, the diseases that cause dementia, which start long before memory problems show, also damage the parts of the brain that keep us awake. Someone whose brain is slowly losing that "wake-up" capacity may simply fall asleep easily, so having no trouble sleeping would then not be a good sign.

What they did

• Who: 16,884 adults without dementia from three studies. The main one was the English Longitudinal Study of Ageing: 9,473 people aged 50 to over 90, followed for about ten years. The findings were then checked in Whitehall II, a study of 5,580 British civil servants who were mostly in midlife (44 to 68), and in the Rush Memory and Aging Project, 1,831 mostly older people in the Chicago area (median age 80).
• Sleep: people reported how long they usually slept and how often in the past month they had trouble falling asleep or woke during the night. "Often" meant at least once a week; "short sleep" meant six hours or less.
• Thinking: at repeated visits, they did a word-list memory test (hear a list of words, then recall them) and a verbal fluency test (name as many animals as possible).
• Analysis: the team estimated how fast scores changed over five or ten years for people of each starting age, comparing those with and without each sleep complaint. They took into account sex, education, work, and in a second step health and lifestyle factors such as blood pressure, diabetes, depression, smoking and exercise.

What they found

• At 50, trouble falling asleep went with faster decline. People who had difficulty falling asleep at least once a week lost a little more on the memory test over five years. The difference was 0.12 standard deviations, a measure of spread; the authors say it is similar in size to the effects seen in recent prevention trials that combined several lifestyle changes.
• The link faded with age and then flipped. It shrank to about zero around age 70, and at older ages people with trouble falling asleep declined slightly less than those without.
• The same pattern showed up in the other two studies. In Whitehall II, trouble falling asleep at 45 went with more decline in fluency over 15 years. In the Chicago study, the link moved from unfavourable at 70 towards slightly favourable at 90.
• Short sleep showed a similar pattern, but weaker. It faded once health and lifestyle factors were taken into account, so part of it may be explained by them.
• Waking at night and long sleep (eight hours or more) showed no change with age.

What to keep in mind

• Self-reported sleep. Nobody's sleep was measured in a lab. People with early memory problems may report their sleep less accurately, although the results held after excluding 420 people who later developed dementia.
• Survivors and dropouts. The authors note that the reversal in old age could partly be because the people who stay in a study into their eighties and nineties are not typical. A statistical check for dropout pointed the same way, which is reassuring but not proof.
• No brain measures. Without scans or blood markers of brain disease, the explanation for the reversal (a brain losing its ability to stay awake) remains the authors' interpretation.
• Not designed to find causes, and the participants were mostly White. The team did not correct for testing many combinations; instead they looked for patterns that repeated across studies.
• Not yet peer-reviewed.

Why it matters

The study suggests that sleep complaints should be read in the light of age. In midlife, trouble falling asleep was linked to faster memory decline, which makes it a candidate for prevention trials: the authors call for studies testing whether treating insomnia in midlife slows decline. In very old age, the same link turned around, and the authors argue that an older person who no longer has trouble falling asleep is not necessarily sleeping better: it may be a sign of a brain that is less able to stay awake. Both ideas still need to be tested. This is not advice to start or stop any sleep treatment.