In a recent study published in JAMA Network Open, researchers examined the relationship between physical activity and cognitive decline. They looked at domain-specific and global cognition, dose-response relationships, and potential moderators.
Physical exercise is associated with a lower risk of cognitive decline; however, most evidence comes from short-term studies, potentially due to reverse causation bias. Research suggests that physical activity improves cognitive outcomes and reduces the risk of dementia, but the evidence is conflicting.
Recent intervention studies highlight the need to link cognition with physical activity, and most existing evidence comes from observational studies with short follow-ups and without previous cognitive levels. Modeling cognition as a continuous variable would enhance the statistical power.
About the Study
In the present meta-analysis, researchers examined the relationship between physical activity and cognitive decline and assessed the influence of factors such as follow-up duration, baseline age, exercise volume, and the quality of included studies on the association.
The team searched databases including PubMed, Scopus, PsycINFO, Web of Science, CINAHL, and SPORTDiscus for relevant English language full-text records until November 2, 2022. For the meta-analysis, the team included prospective cohort or case-control observational studies with individuals aged ≥ 20 years, follow-up durations of one year or more, and estimates of the associations between physical activity and cognition. The included studies reported validated cognitive outcome measures in later life (average and maximum age of 55 and 65 years, respectively).
The team included studies assessing physical activity using devices, questionnaires, or interviews. The primary outcome was the association between physical exercise and global cognition or specific cognitive domains (such as episodic memory, executive function, verbal abilities, verbal naming and fluency, processing speed, visual-spatial abilities, and working memory) at follow-up.
The team excluded studies that measured retrospective physical activity levels, cardiorespiratory fitness levels, phases of physical activity, physical activity beyond follow-up, and statistical adjustments for physical activity. They also excluded studies that included subjective cognitive measurements or rated cognition based on disability registers. Studies that included cohorts with dementia, specific diseases, or cognitive impairments at baseline were also excluded.
Two reviewers independently assessed the title, abstracts, and full texts of the identified records, evaluated the quality of evidence, and extracted data, with a third reviewer resolving disagreements in case of no consensus. The team conducted random-effects modeling and used logistic regression to determine the risk ratios (RRs) and regression coefficients for physical activity using moderators, scatterplots, and funnel plots. They analyzed data between January and August 2023, with final analyses in December of the same year.
Initially, the team identified 18,669 records and excluded 17,861 at the title-abstract screening and 703 at the full-text screening. As a result, 104 records, including 341,471 individuals, were analyzed. The binary outcome analysis comprised 45 records and 102,452 individuals; the subsequent cognition analysis comprised 14 records and 41,045 individuals; and the global cognitive change analysis comprised 25 records and 67,463 individuals.
Among the 45 studies with binary outcomes, one study, 13 studies, and 31 studies were of high, moderate, and low quality, respectively. Among the 14 studies examining global cognition at follow-up, none, four, and ten studies were of high, moderate, and low quality, respectively. For the 25 studies assessing global cognitive changes, none, five, and 20 were of high, moderate, and low quality, respectively. The included studies for all outcome measures were highly heterogeneous, with I2 values of 70% for binary outcomes, 76% for follow-up cognition, and 67% for changes in global cognition.
Physical activity, after funnel plot corrections, was associated with lower cognitive decline or impairment (pooled RR: 0.97); however, there were no statistically significant associations in subsequent analyses over more than ten years. Physical activity was associated with follow-up global cognition domains and their changes with standardized regression coefficients of 0.030 and 0.010, respectively, as determined by trim-and-fill assessment, without moderation or dose-response relationships based on participant age and follow-up duration, baseline cognitive adjustments, or study quality.
Specifically, verbal fluency and episodic memory were associated with physical activity with standardized regression coefficients of 0.05 and 0.03, respectively. There was a stronger inverse association between the extent of physical activity and cognitive decline or impairment up to 5,000 metabolic equivalent task minutes per week (i.e., moderate to vigorous physical activity for 16 hours per week).
Overall, the study’s results show a weak but positive association between physical activity and cognitive decline that remained consistent regardless of cognitive level or cohort age. This finding is crucial for public health to delay multifactorial diseases that cause dementia.
Follow-up duration, frequency, type, and quality of physical activity measurement influenced the association reported in studies with binary cognition outcomes. However, funnel plots revealed potential bias. The study also found weak associations between verbal fluency and episodic memory, with mixed results for executive function in follow-up and change analyses.