2016 Theses Doctoral
The Impact of Built and Social Environment on Physical Activity among Older Adults
Physical activity, defined as bodily movement produced by skeletal muscles that results in energy expenditure, has many known mental and physical health benefits for older adults. However, as of 2008, only 22.6% of older adults in the United States reported meeting recommended physical activity guidelines. This dissertation examines the role of the built and social environment on physical activity among older adults, with particular focus on physical disorder, or the visual indications of neighborhood deterioration. All empirical analyses use data from the New York City Neighborhood and Mental Health in the Elderly Study (NYCNAMES-II), a three-wave longitudinal study of about 3,500 older adults living in New York City.
We first systematically review the existing literature concerning physical disorder as an influence on physical activity among adults of all ages. We find that most prior studies of disorder and activity have been cross-sectional and that disorder has not consistently been associated with less activity across all studies. However, we also find indications that older adults’ activity levels may be more negatively impacted by disorder than younger adults’ activity levels.
Next, we use a longitudinal analysis to estimate the association between neighborhood disorder and total physical activity among the NYCNAMES-II cohort. In multivariable mixed regression models accounting for individual and neighborhood factors, for missing data, and for loss to follow-up, we find that each standard deviation increase in neighborhood disorder was associated with an estimated 3.0 units (95% CI: 1.9, 4.2) lower PASE score at baseline, or the equivalent of about 10 minutes of walking per day. There was no significant interaction between physical disorder and changes in PASE score over two years of follow-up.
We next apply a latent transition analysis to identify patterns of types of physical activity the same cohort, identifying seven latent classes of activity. Of these seven classes, three pairs of classes were roughly equivalent except for participation in exercise. About three quarters of subjects remained within each latent class between waves; most transitions that did occur were between classes defined by exercise to the parallel class without exercise or vice-versa. More neighborhood disorder was modestly associated with moving out of a sports and recreation class (Relative Risk = 1.27, 95% CI = 1.00, 1.61 between waves 1 and 2, Relative Risk = 1.28, 95% CI = 0.85, 1.93 between waves 2 and 3).
Finally, we develop the Neighborhood Environment-Wide Association Study (NE-WAS), an agnostic approach to systematically explore the plethora of neighborhood measures available to modern researchers equipped with geographic information systems (GIS) software. We find that only neighborhood socioeconomic status and disorder measures were associated with total activity and gardening, whereas a broader range of measures was associated with walking.
Substantively, we conclude that more physical disorder was associated with less physical activity, potentially due to decreases in sports and recreation among those living amidst physical disorder, though latent transition analysis estimates were too imprecise to rule out chance. Future longitudinal research on physical disorder as an influence on physical activity would benefit from longer periods of follow-up in which more subjects moved between neighborhoods. Methodologically, the NE-WAS approach appears to be a promising way to systematize neighborhood research as the scale of available spatially located administrative data continues to grow. Future NE-WASes might profitably focus on comparing the spatial scale of neighborhood measures.
- Mooney_columbia_0054D_13496.pdf binary/octet-stream 2.6 MB Download File
More About This Work
- Academic Units
- Thesis Advisors
- Rundle, Andrew G.
- Ph.D., Columbia University
- Published Here
- August 5, 2016