Could exercise worsen dementia? The idea runs against one of the fondest hopes of patients with Alzheimer disease, their caregivers, and physicians. It contradicts some early research and tentative recommendations. But it is a key finding of one of the largest studies yet to examine the question.

In the Dementia And Physical Activity (DAPA) trial, the mean score on the Alzheimer’s Disease Assessment Scale-Cognitive Subscale (ADAS-cog) worsened more for people with dementia who were assigned to a year of vigorous exercise than for people who kept to their usual routines.^[1]

The difference was small but statistically significant, says Bart Sheehan MRCPsych, MD, consultant liaison psychiatrist at the Coventry and Warwickshire Partnership Trust in Coventry, United Kingdom. “It does raise the possibility that, at this point, vigorous exercise might be damaging for people.”

The finding has experts in the field taking a harder look at what they thought they knew about the way physical activity affects a declining brain. It comes as a particular blow because no one has found a way to halt Alzheimer disease. “People are desperate for a treatment,” Sheehan said.

Until the DAPA results came out, exercise was looking like one of the most promising possibilities—if not to stop dementia, then at least to slow its progression. “Among patients with dementia or mild cognitive impairment, randomized controlled trials (RCTs) documented better cognitive scores after 6 to 12 months of exercise compared with sedentary controls,” wrote the authors of a 2011 meta-analysis.^[2]

Such results were enough to prompt the Mayo Clinic website, a health information website for consumers, to advise that “Exercising several times a week for 30 to 60 minutes may… improve memory, reasoning, judgment and thinking skills (cognitive function) for people with mild Alzheimer’s disease or mild cognitive impairment.”^[3]

But these findings were from relatively small trials. And negative results have also cropped up in the literature for years, including in other reviews of the literature.^[1] Funded by the British government, Sheehan and his colleagues set out to settle the question with the most authoritative trial possible.

They recruited 494 people with mild to moderate dementia according to the criteria in the Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM-IV). All lived in the community and were able to sit on a chair and walk 10 feet without assistance. The average age was 77. Sixty-one percent were men.

The researchers randomly assigned 329 to exercise and 165 to make no change in their physical activity. The exercisers attended group sessions in a gym twice a week for 4 months under the guidance of physical therapists. Each session lasted 60-90 minutes. The researchers asked them to work out for an additional hour each week at home during this period. The sessions included cycling in place for 25 minutes of moderate to hard intensity, as well as weight training such as biceps curls, shoulder forward raise, lateral raise, and sit-to-stand using a weighted vest or waist belt.

After the 4 months, the researchers prescribed a home-based program of unsupervised exercise of 150 minutes each week. They encouraged the participants to choose activities at home that they preferred and followed up with phone calls to encourage them. Eighty-eight percent reported continuing the exercises at home. Less than 1% of the participants reported doing structured exercise outside of the trial.

The people who evaluated the patients didn’t know which ones participated in the exercise programs and which ones did not.

After 12 months, the patients improved their fitness compared with the usual-care group. But when it came to cognitive function, the researchers recorded abysmal results. On the ADAS-cog, where a higher score means worsening function, the usual-care group went from 21.4 to 23.8, a worsening of 2.4 points, as might be expected with the progressive diseases that cause dementia.

But the exercisers fared even worse, going from a mean score of 21.2 to 25.2, a worsening of 4.0. For perspective, a normal score for someone who does not have dementia is 5, while the average score of someone diagnosed with probable Alzheimer’s or mild cognitive impairment is 31.2.

The difference was statistically significant (P =.03). It’s not clear whether it has clinical significance, Sheehan says. Still, it startled the researchers.

“It didn’t come as a surprise that physical exercise was not effective as a treatment for dementia, because dementia is notoriously difficult to treat,” he said. “I think what was a surprise is the very strong signal that it may make dementia worse.” They ran the statistics again and again but found no mistake.

And the finding held up regardless of the patients’ sex or mobility and regardless of whether they were diagnosed with Alzheimer’s versus other kinds of dementia, or whether they had mild versus severe cognitive impairment.

Despite their improved physical fitness, the exercisers did not improve in activities of daily living, behavior, or health-related quality of life.

The finding should influence what clinicians say to people with dementia and their caregivers, said Sheehan, who has treated many such patients. He now tells them that exercise won’t help with such core features of dementia as memory or the ability to organize oneself, and that it might actually do damage.

People who are already exercising and enjoying it shouldn’t necessarily stop, he added. But they must weigh the enjoyment and other health benefits—which are many—against the risk for harm.

Not everyone interprets the results of the DAPA trial as pessimistically as Sheehan. “We don’t have the evidence yet to be able to say that exercise is going to improve cognitive function,” said J. Carson Smith, PhD, an associate professor of public health at the University of Maryland. “But there is more evidence of a benefit in mild cognitive impairment and in people at increased risk for dementia.”

He is among the researchers whose small studies have suggested that exercise can improve cognitive ability in people with mild cognitive impairment. Epidemiologic studies measuring the benefits of long-term exercise for preventing dementia are even more impressive.

This includes a recent sample of 191 Swedish women who were 38-60 years of age in 1968 when they underwent an ergometer cycling test. Examinations of dementia were done six times up to 2010 and supplemented with information from medical records. Women with high physical fitness at middle age were nearly 90% less likely to develop dementia decades later, compared with women who were moderately fit.^[4]

Smith and others have found biological differences between more and less fit people that could explain a difference in dementia risk. Lower cardiovascular fitness is associated with a smaller brain volume two decades later, for example.^[5]

It’s hard to explain why exercise in healthy people might protect against cognitive decline, but exercise in people with dementia might make it worse. Sheehan theorized that already weakened brains might be too fragile to withstand the temporary loss of oxygen that comes with vigorous exercise. But there isn’t much information yet to support or refute such ideas.

Such studies can’t prove cause and effect. Not only physical activity but also genes affect physical fitness. And people who exercise may have other healthy behaviors.

But even Sheehan has not given up on the idea that physical activity can help people in their declining years. Some kinds of exercise can improve balance, for example. “People say, ‘I wish my father could recognize me,’ but they also say, ‘I wish my father didn’t fall over,'” he points out.