6007.0: Thursday, October 25, 2001 - 9:30 AM

Abstract #27762

Disability Profiles in a Representative US Sample: Results Using Latent Class Analysis

Barbara M. Altman, PhD, Office of Analysis and Epidemiology, National Center for Health Statistics, 6525 Belcrest Road, Room 730, Hyattsville, MD 20782, 3014584654, baltman@cdc.gov and John A. Fleishman, PhD, CCFS, Agency for Healthcare Research and Quality, 2101 E. Jefferson St., Rockville, MD 20852.

Disability is a multifaceted construct, and different indicators have been routinely used to measure different aspects. Some studies use limitations in performing major social roles as an indicator of disability; others use ADLs or IADLs ; still others assess limitations in upper or lower body actions . Most research employs a variable-centered perspective, which examines relationships among different sets of disability indicators. In contrast, this study employs a person-centered perspective, which attempts to discern different types of individuals based on their profile of disability indicators. The research question is whether meaningful classes of persons with disabilities can be formulated on the basis of multiple disability indicators. We use nationally representative data from the 1996 Medical Expenditure Panel Study. We use latent class analysis to estimate models that specify a number of underlying types of persons with differing disability profiles using six indicators. Preliminary analysis using a four-class model suggests one class (86%) of persons without any disability, a second class likely to have only functional (upper or lower body) limitations (10%), a third class limited in both physical functioning and in major roles (2%), and a fourth class with limitations in all of the observed indicators (2%). The preferred model will be selected using criteria such as the Bayesian Information Criterion. Individuals will be assigned to their most probable latent class, and correlates of class membership will be examined. Specialized latent class models will examine the degree to which a Guttman-scale model, postulating a hierarchical ordering of indicators, fits the data.

Learning Objectives: Information will allow the participant to assess the value of a multifaceted measure of disability for improving the analysis of questions related to disability outcomes.

Keywords: Disability,

Presenting author's disclosure statement:
Organization/institution whose products or services will be discussed: None
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.

The 129th Annual Meeting of APHA