Physical activity and cognition in the northern Manhattan study.

BACKGROUND To test the hypothesis that leisure time physical activity (PA) is associated with cognitive status. METHODS We assessed cognition using the Mini-Mental Status Examination (MMSE) at enrollment and using the modified Telephone Interview for Cognitive Status (TICS-m) administered annually since 2001 in the Northern Manhattan Study. Baseline measures of leisure time PA were collected via in-person questionnaires. Total PA was categorized into 3 groups based on the metabolic equivalent (MET) score, a composite of total reported intensity and time. We used linear regression models to examine the association of PA with MMSE, and generalized estimating equations for change in TICS-m over time. RESULTS There were 3,298 stroke-free participants with MMSE data (mean MMSE 26.0 ± 3.8) and 2,279 with TICS-m scores available. Compared to no PA, those with the upper quartile of MET scores had greater baseline MMSE scores (adjusted β = 0.4, p = 0.01) but no association with change in TICS-m over time. There were interactions (p < 0.05) between PA and both insurance and education; compared to no PA, those in the upper quartile of MET scores had a greater MMSE score only among those with Medicaid/no insurance (adjusted β = 0.83, p = 0.0005) and those who did not complete high school (adjusted β = 0.68, p = 0.001). CONCLUSIONS Increased levels of PA were associated with better baseline MMSE, particularly among those with socioeconomic disadvantages, but not with cognitive decline.