P3‐398: Physical Activity, Potential Confounders and Cognitive Decline in the Elderly

Background: Physical activity has been related to lower risk of dementia, but the relationship of physical activity and the decline of global and domain-specific cognitive function remains unclear. Methods: This is a follow-up study. A total of 605 elders aged 65 years or older were recruited from health checkup program at National Taiwan University Hospital at baseline (2011-2013), and 483 of them had follow-up data (2013-2015). Physical activity was accessed by self-report International Physical Activity Questionnaire, which includes walking, moderate and vigorous activities. Global cognition was assessed by Montreal Cognitive Assessment. Domain-specific cognition was evaluated by Wechsler Memory Scale-Third Edition (logic memory and attention), Trail Making Test and verbal fluency test (executive function). The outcome of this study was change of global or domain-specific cognition from baseline to follow-up. Physical function was assessed by 8- feet gait speed test. Multivariable logistic regression models were used to explore the association between physical activity and cognitive decline over 2-year follow up. Stratification analysis was performed by physical function, sex, age groups, apolipoprotein E (APOE) e4 status for the association above. Results: Elders walked 150 mins/week or above (i.e., >495 MET-min, guideline of World Health Organization) showed protective effect on decline of global cognition (AOR=0.66, Ptrend Q1-Q5=0.045) and logic memory (thematic I: AOR=0.65, thematic II: AOR=0.55, recall II: AOR=0.62) than those not met the guideline. Associations for global cognition remained in elders with better physical function (gait speed at least 0.6 m/s : AOR=0.63), women (AOR=0.53), the young old (age 65- 74: AOR=0.60), APOE e4 carriers (AOR=0.22), and elders with lower education level (less than 6 years: AOR=0.10). For the decline of logic memory-thematic II, significant associations remained in elders with better physical function (AOR=0.63), women (AOR=0.51), and the young old (AOR=0.54). No association was observed for attention and executive domains. Physical function significantly modifies the association between regular walk on the change of logic memory-thematic I (Pinteraction=0.008). Conclusions: Regular walk protected against the decline of global cognition and memory domain. Significant associations between regular walk and cognitive decline were observed in some subgroups, which provide useful information for dementia prevention and reducing public health burden.