Research on homeless populations with substance abuse and/or mental illness demonstrated the inter-relation ships among interpersonal-, organizational-, and societal-level variables in predicting service use and housing outcomes (Rosenheck, Morrisey, Lam, et al., 1998; Wenzel, Burnam, Koegel, et al., 2001; Pollio, Eyrich, North, et al., 2001). Although these complex analyses significantly advance knowledge of the diverse factors necessary to understanding service use, none have the capacity to examine the impact of geographic variables, such as distance traveled and convenience of access, on services and outcomes. This limitation seems critical, given qualitative research has repeatedly reported that homeless individuals use services based on accessibility, rather than on need factors or provision by organizational agreements (Leibow, 1965; Anderson, 1978; Snow & Anderson, 1987; Pollio & Kasden, 1996). Integration of this information into a multi-level service analysis can uniquely contribute to our understanding of service use and guide more effective development of interventions in this difficult-to-treat population. Testing the impact of geographic and spatial features on service use and outcomes requires a longitudinal data set collected specifically with provisions to assess agency level service use in small time increments and outcomes at established intervals. Additionally, individual demographic and organizational data must be obtained for all persons and providers. The proposed R03 builds on the Service Use, Needs, Costs, Consequence, of Drug Abuse in homelessness (SUNCODA: NIDA ROI-DA 10713) project providing a unique opportunity to integrate spatial information into a data set meeting all of the above criteria to examine the impact of geographic and spatial variables on service use and outcomes. The result of this process would be a Geographic Information Systems (GIS) database (a) mapping the locations of all homelessness, drug abuse, mental health, and outpatient health service providers in the St. Louis City geographic area, (b) specifying distances and transportation options across all these agencies, (c) locating these agencies on relevant established geographic constructs (e.g., census tracts, communities); and (d) linking this GIS database with already-collected monthly service use data for a representative sample of 396 homeless drug abusers. The resulting database would then be used to mount a spatial exploration of service use, and model the complex interaction of individual, organizational, and spatial factors to comprehensively explain service use patterns and associated outcomes.
