 | The 131st Annual Meeting (November 15-19, 2003) of APHA |
|
| The 131st Annual Meeting (November 15-19, 2003) of APHA |
Kevin Delucchi, PhD1, Helen Matzger, MPH2, and Constance Weisner, DrPH, MSW1. (1) Department of Psychiatry, University of California, San Francisco, 401 Parnassus Ave., Box 0984-TRC, San Francisco, CA 94143-0984, 415-476-4180, kdelucc@itsa.ucsf.edu, (2) Alcohol Research Group, University of California, San Francisco, 2000 Hearst St., Suite 300, Berkeley, CA 94709
An important concern for health services research in the alcohol field revolves around understanding the long-term course of problem drinking. While both hierarchical modeling and latent curve analysis are applicable methods, they do not provide a parsimonious description of common paths over time. Using data from assessments over five years of two surveyed samples—problem drinkers and dependent drinkers—we present results from a study adapting a relatively new approach to identify and characterize profiles of drinking behavior by clustering subjects’ changes in drinking into latent profiles. We test both variables which affect which profile a subject is classified to and which effect the shape of the profile. Simultaneously we account for the fact that profile membership is not known but estimated. This removes a potential bias in the more common “classify then analyze” approach. Results indicate that for both the problem and dependent drinkers a five profile model best fit the data. Tests indicate significant change over time. In both samples the largest group were subjects who continued to drink at an elevated level. Changes in drinking, however, are seen in the other latent profiles and differences are found between the two samples.
Learning Objectives:
Keywords: Statistics, Alcohol
Presenting author's disclosure statement:
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.