Umbrella review finds ‘inconclusive evidence’ supporting cognitive benefits of exercise

Key takeaways:

• Positive associations between physical exercise and cognition may have been overestimated.
• Small exercise benefits became insignificant after accounting for publication bias.

After re-examining more than 20 meta-analyses of randomized controlled trials, researchers said they found “inconclusive evidence” supporting a positive link between regular physical exercise and cognition in healthy participants.

“The positive impact of physical exercise on cognition seems indisputable according to the scientific evidence accrued over the last 2 decades,” Luis F. Ciria, PhD, a postdoctoral fellow at the University of Granada in Spain, told Healio. “A wide range of narrative and meta-analytic reviews report cognitive gains associated with regular physical exercise in healthy populations across the lifespan.”

Such findings “are even driving current public health policies (eg, Physical Activity Guidelines Advisory Committee Scientific Report) aimed at fostering exercise adherence in consonance with the World Health Organization that currently recommends regular exercise as a means to maintain a healthy cognitive state,” he said. “The question we pose here is whether those claims, policies and recommendations are well backed up by scientific evidence.”

In their umbrella review, Ciria and colleagues compiled 24 meta-analyses of 109 randomized controlled trials (RCTs) involving 11,266 healthy participants. The researchers analyzed several RCT factors, including potential publication bias and moderating variables like control activity, age range and type of cognitive outcome.

Nearly all the meta-analyses found that physical exercise had a significant positive effect on cognitive function in participants (Cohen d = 0.29, 95% CI, 0.23-0.36). However, a re-estimate with only target interventions and healthy populations showed the effects “remained positive although dispersed,” Ciria and colleagues wrote in Nature Human Behavior, and added the results were likely overestimated.

The researchers also found small exercise benefits for all cognitive outcomes (Cohen d = 0.22; 95% CI, 0.16-0.28) that became smaller after adjusting for key moderators (Cohen d = 0.13; 95% CI, 0.7-0.2) and insignificant after adjusting for publication bias.

Ciria said that while the researchers knew most prior meta-analyses reported evidence of positive associations between physical activity and cognitive function, “we also knew that the quality of the literature, especially the primary sources they used, was not good, so we suspected that the quality of the evidence was not good.”

“Therefore, although the results may seem surprising, for researchers who are concerned with replicability and transparency in science, these results only corroborate what had already been pointed out previously,” he said.

Ciria explained that because all participants were classified as healthy, “this work does not allow drawing any conclusions or recommendations addressed to people with any pathology or disorder.”

“Regarding current public health policies and guidelines for the promotion of physical exercise, we strongly believe there is no need to appeal to the alleged, as yet uncertain, cognitive benefits of physical exercise, especially when the current meta-analytic evidence from RCTs suggests that, even if the effect exists, it is notably small to assert its practical,” he said. “The benefits of physical exercise on human well-being, especially with regard to physical health, are in themselves sufficient to justify evidence-based public health policies to promote its regular application in our daily lives.”

For future research, Ciria pointed to the development of “comprehensive theoretical models on the cognitive and neural mechanisms of the potential exercise-induced cognitive improvements,” along with larger study sample sizes to help increase statistical power.

“Finally, we encourage researchers in this field to adopt research transparency practices, include open-source datasets and codes, pre-register studies and their hypotheses, or engage in multi-laboratory initiatives,” he said. “All these avenues can definitely improve the field.”