Abstract
OBJECTIVE: This study quantifies the effects of leisure-time physical activity (LTPA) on walking speed independently of body composition in an elderly cohort, and in those elderly with metabolic derangements due to age, diabetes, and cardiovascular disease (CVD). METHODS: 1655 community-dwelling women and men >55 years were measured for body composition (lean mass : fat mass ratio, LNFAT) , based on estimated bioelectric impedance by using population-specific prediction equations derived from dual-energy x-ray absorptiometry. In addition to LNFAT, LTPA, diabetes, CVD, walking speed, and other covariates were measured biannually over an 8-year period. LTPA was categorized as <22.5 Mets/week, ≥ 22.5 Mets/week, based on public-health recommended guidelines, and LNFAT was dichotomized based on its sex-specific median. Direct effects of high vs. low LTPA on walking speed were estimated for fixed levels of LNFAT, which represented an intermediary variable in the analysis. Stratified estimates of effects were obtained using subject status (e.g., age≥75 years, diabetes, CVD) at each visit. RESULTS: Walking speed was significantly greater (0.74, 0.75 m/s in women and men, respectively) if subjects experienced LTPA ≥22.5 Mets/week and > median LNFAT, compared with <22.5 Mets/week and ≤ median LNFAT (0.68, 0.69 m/s). While direct effects of LTPA contributed to higher walking speed, none were significant in the overall, nor the stratified groups of subjects, of either sex. CONCLUSIONS: Walking speed increases with greater LTPA and LNFAT in the elderly, but there was no evidence to indicate that walking speed increases from LTPA independently of body composition and the metabolic processes it represents.