Experimental studies support a role of physical activity in protecting brain structure and function. Compared with inactive adult mice, active adult mice have reduced cerebral accumulation of neurodegenerative byproducts, including ß-amyloid plaques (3) and reactive oxygen species (4) that have been implicated in the pathogenesis of dementia. Voluntary exercise upregulates proteins that stimulate neural growth, most prominently brain-derived neurotrophic factor in the hippocampus, an area that is central to learning and memory and is particularly affected in Alzheimer disease and other dementing illnesses (5).

A recent review of observational studies concluded that exercise in late life is inversely associated with risk for all-cause dementia and Alzheimer disease (6). Six of 9 studies reviewed showed a reduced incidence of dementia among persons in the highest category of exercise compared with those who were less active. Risk reductions ranged from 20% (7) to 50% (8) over follow-up periods of 5 to 7 years (6). The studies included in the review enrolled healthy adults who were free of dementia or mild cognitive impairment and evaluated the association of baseline physical activity with subsequent development of dementia. These studies had an important potential limitation: reverse causation bias, in which what seems to be an effect actually triggers its putative cause. Although the measurements of physical activity were done before clinically manifest dementia, these studies could not exclude the possibility that lower exercise levels at baseline were a consequence of early, subclinical cognitive deterioration. In the context of such diseases as dementia, in which changes in the brain may develop over decades, reverse causation is difficult to refute even in prospective cohort studies.

In this issue, Larson and colleagues' study (9) provides important evidence that more physical exercise is associated with lower rates of dementia, even after the authors took great pains to avoid reverse causation bias. The study investigators recruited adults 65 years of age and older from a health maintenance organization in Seattle, Washington. They excluded participants with dementia and those scoring below the 75th percentile for cognitive performance on the basis of the Cognitive Ability Screening Instrument (CASI), thus selecting a cohort of 1740 participants with high cognitive performance. Most important, early cognitive deterioration could not explain why members of this cohort varied in baseline levels of physical activity. After an average of 6.2 years of follow-up, participants exercising 3 or more times per week had a 38% lower incidence of dementia than those who exercised less consistently. This inverse association was evident despite the uniformity of the study sample—participants were predominantly white with high levels of formal education, and most (74%) engaged regularly in physical activity.

The study by Larson and colleagues significantly adds to several studies published recently on the association of physical activity with risk for dementia. In the Finnish Cardiovascular Risk Factors, Aging, and Dementia Study (CAIDE), leisure-time physical activity in middle age was inversely associated with risk for dementia in older age. The association with physical activity in middle age is a strong argument against reverse causation bias, because it is extremely unlikely that subclinical dementia was affecting physical activity in middle age (10). Other studies have assessed specific types of physical activity. In the Honolulu-Asia Aging Study, older men who walked more than 2 miles daily had a 40% lower risk for dementia than did those who walked shorter distances (11). In the large Cardiovascular Health Cognition Study, the variety of exercise activity in late life was a stronger inverse predictor of future dementia than energy expenditure in physical activity (12). The evidence connecting physical activity and dementia is not consistent. Other studies have found inverse associations between various cognitive activities and incident dementia but no association with physical activity (13, 14).

Although experimental and epidemiologic studies support the role of exercise in late life as a means to prevent or delay the onset of dementia, several important questions remain unanswered. It is still uncertain whether this association is causal or whether physical activity is just a proxy measure for "life engagement," for other cognitive activities, or for other lifestyle or sociodemographic characteristics associated with dementia. In all likelihood, the study by Larson and colleagues and other trials have taken us as far as observational data can toward excluding the explanations of reverse causation and confounding by "life engagement." We need randomized trials to examine the effect of physical activity on cognitive function measures, and we need researchers to include clinical dementia end points in ongoing trials of lifestyle modification. Another unresolved issue is the selectivity of the relationship between exercise and dementia. Does this association apply only to specific subgroups? Previous studies have raised the possibility that the association of exercise with cognitive decline or dementia risk depends on the individual's apoliprotein E 4 genotype, but the evidence is not consistent (10, 12, 15). Of interest, the study by Larson and colleagues raises the possibility that the potential benefit of exercise may be more pronounced in persons with lower levels of physical function. In this study, the assessment of physical function was based on 4 performance tests: 10-foot timed walk, time to stand from a seated position in a chair to a standing position 5 times, standing balance, and grip strength in the dominant hand. Exercise, particularly weight-bearing activity and strength training, improves physical function and helps maintain independence in older adults (16). Although Larson and colleagues are the first to report an interaction between level of physical function and physical activity and dementia risk, their findings highlight the importance of exercise for persons at all levels of ability but particularly those who already have some limitation in physical function. Finally, no one has defined the type, frequency, intensity, or duration of physical activity that is most beneficial in preventing cognitive deterioration.

We are edging closer to placing prevention of cognitive deterioration and of dementia on the long list of health benefits induced by physical activity. Current recommendations call for all adults, including the elderly, to perform 60 minutes of moderate-intensity physical activity daily to promote health and vigor and to decrease the risk for chronic illnesses and early death (17). On the basis of the epidemiologic literature to date, these recommendations may also help maximize cognitive health and attenuate neurodegenerative disease processes in older age. Because more than 40% of adults older than 50 years of age express heightened fear about Alzheimer disease (18), perhaps that perspective will serve as an additional impetus to be physically active in later life.