Daily Physical Activity And Mortality Risk In The Very Old

Abstract

PURPOSE: The lack of longitudinal studies has left open the question as to whether greater accelerometry-based assessment of physical activity would be shown to be protective even in very old persons. We used data from the Age, Gene, Environment (AGES)-Reykjavik Study to assess the association of mortality with total counts, measures of activity type (sedentary time, count intensity), and distribution of counts over the day. METHODS: Data were collected from a hip worn Actigraph GT3X accelerometer; 589 subjects had 4 or more days of data. Covariates included age, sex, reported health status, coronary calcium, smoking status, depression, health conditions, cognitive score, C-reactive protein, BMI, walking speed, and leg strength. RESULTS: The sample (age=80±4.8) consisted of 226 men and 363 women followed for 5 years for mortality (116 deaths). Multiple measures of activity including total counts (per standard deviation, OR=1.74, 95% CI 1.23, 2.46), proportion of time sedentary, and volume of moderate or vigorous activity were all associated with mortality risk in adjusted models. However, once other measures were adjusted for total counts, none of the other measures remained statistically significant while the odds ratio for total counts remained relatively unchanged. Over the day, those who died had proportionately more counts from noon to 4 PM (P<.005) but fewer counts from 8 to 12 PM (P=.01); after adjustment for total counts, only the relationship from noon to 4 PM was significant. CONCLUSIONS: These longitudinal data support the importance of total physical activity to health risk in older persons. Further, they suggest that not only are there differences in the pattern of energy expenditure over the day, but that these patterns may be associated with increased risk.