Saturday, 15 January 2005 - 2:00 PMThis presentation is part of: Social Factors in AgingDimensions of Functional Disability in Elderly: Analyzing Basic and Instrumental Activities of Daily LivingElena A. Erosheva, PhD, University of Washington.
Purpose. Common approaches to disability measurement include summed indices and hierarchical scales. Items are usually selected to form a scale on the basis of expert appraisal but it is often unclear to what extent statistical procedures are involved to help determine or validate the item hierarchy. This paper takes a fresh look on the concepts of hierarchy and dimensionality in disability manifestations via presenting a multivariate analysis of 16 ADL and IADL measures from the National Long Term Care Survey (NLTCS). Methods. The NLTCS, conducted in 1982, 1984, 1989, 1994, and 1999, was designed to assess chronic disability in the U.S. elderly Medicare-enrolled population. The screener identified chronically disabled persons aged 65 or older. We consider 16 ADL/IADL items for community dwelling elderly in the sample. For every item, individuals were classified as being disabled if any of the triggering conditions were met, i.e., active help, standby help, equipment use, or unable to perform activity due to disability or a health problem. We test whether a latent unidimensionality assumption is appropriate for the data. Rejecting the unidimensionality hypothesis, we then use a factor analysis for dichotomous variables to get a general idea about the underlying covariance structure and latent dimensionality. Finally, we fit a number of latent class models to determine whether the extracted functional disability data can be successfully described by a traditional latent class model with a modest number of classes. Results. The covariance structure in the data is qualitatively stable across the survey waves and can be explained by three factors: (1) an ADL disability factor, (2) a cognitive disability factor, and (3) a physical strength deficit factor. Consistent with these findings, the results of the latent class analysis indicate that the 16 ADL and IADL items do not constitute a hierarchy. The difficulty order of the items established by the probabilities depends on a latent class. Implications. Understanding disability manifestations is the first step in developing a reliable measure of disability that can be utilized by governmental policies which require selection of eligible disabled persons. Understanding disability in elderly is particularly important in the society where the percentage of aging people is rapidly increasing.
See more of Social Factors in Aging |