DISCRETE OUTCOME MODELS FOR EPIDEMIOLOGICAL SURVEYS
Biography
Overview
This application proposes to use data from the Longitudinal Study on Aging (LSOA) to assess the impact of physical activity and comorbid conditions on changes in functional status. Use of definitions of functional status encompassing both disability and dependency are proposed. Results from these analyses will allow a comparison of how physical activity and comorbid conditions may impact on disability and/or dependency, possibly leading to a better understanding of the role that physical activity may play in prolonging the time until functional decline and the need for long term care services. To achieve these aims, it will be necessary to develop statistical methods and computer programs permitting either marginal, transitional, subject-specific or latent variable modelling of discrete outcomes obtained from epidemiological surveys. Such methods must requisitely deal with issues such as: survey weights, stratified selection of clusters of participants, correlation of outcomes both within clusters and between repeated measurements on participants, and incomplete data resulting from institutionalization and death. The statistical methods resulting from this project will have broad applicability to other epidemiologic studies faced with any of the issues previously listed. An additional byproduct of this research project will be the availability of computer software useful for fitting the newly developed statistical methods.
Time