SUBSTANCE ABUSE, MARGINALIZATION, AND HOMELESSNESS: A BAYESIAN PERSPECTIVE

Washington University, Saint Louis MO

Investigators

Linked publications & trials

Abstract

[unreadable] DESCRIPTION (provided by applicant): This proposal is a request for a dissertation research grant (PAR-06-476) and provides a foundation for the applicant's long term research agenda of investigating the complex interactions between homeless adults, the environments they live in, and local service systems. The special focus of this study is on the cycles of substance abuse, incarceration, victimization, and social isolation that push certain groups of homeless adults beyond the reach of the very supports that could help them gain housing, employment, and recovery from substance abuse or mental illness. The concept of marginalization is used to describe how isolated an individual is from basic needs such as supportive social contact, a safe environment, and legal income options. This study will examine what role substance abuse plays in the marginalized status of homeless adults and how the combined effects of marginalization and substance abuse impact service use and housing status changes. Thus, the specific aims of this project are to 1) describe longitudinal patterns of substance abuse and marginalization in a homeless population, 2) examine the combined effects of substance abuse and marginalization on housing status changes over time, and 3) investigate how marginalization and substance abuse impact service use over time. This dissertation will use secondary data collected as part of a NIDA funded longitudinal study of urban homelessness called SUNCODA (DA 10713, PI-Carol North). Adults in the Saint Louis metropolitan area (n = 400) were recruited from shelters and street locations (1999-2001) and interviewed at baseline, then at 1 and 2 year follow ups. Subject-matched service data (including shelter use and contacts with health, substance abuse, and mental health sectors of care) were collected from regional providers over the same time period. This data provides a unique opportunity to explore longitudinal changes in substance abuse/dependence, mental health, social contacts, victimization, criminal activity, housing status, service use, and employment among homeless adults. Results from preliminary analysis using this data demonstrate the feasibility of modeling housing status changes using a multi-state Markov model and different types of service use using Poisson regression methods. This study will extend these preliminary analyses into a Bayesian framework whereby missing data, non-normal distributional assumptions, and prior information from the literature review will be explicitly modeled. This innovative approach to understanding service use and housing status transitions among homeless adults will provide important information toward developing effective targeted interventions and policies. This dissertation study is a critical step in the applicant's research trajectory focused on building more comprehensive and realistic ecological models of homelessness using a combination of survey data, Bayesian inference, systems dynamics, and agent-based simulations. PUBLIC HEALTH RELEVANCE: This dissertation study will look at some of the critical factors involved in how homeless adults get housing and help. Information about how substance abuse, social isolation, legal problems, and employment impact these important outcomes will help practitioners design more effective interventions and inform guide policy makers about where to target services and support for a very vulnerable and hard to reach population. [unreadable] [unreadable] [unreadable]