Purpose: The plight of homeless adults with co-occurring mental illness and substance abuse remains one of the most challenging issues for service providers. A recent innovation in services has emerged (‘housing first') that places consumer choice as a top priority. Compared to standard care (‘treatment first'), this approach reverses the ‘standard' care sequence by providing independent scatter-site housing and case management services immediately without first requiring treatment adherence and sobriety/abstinence. This study reports findings from the only randomized experiment testing these two distinct approaches—the New York Housing Study (NYHS). Research questions for this report were centered on outcomes across groups and over time (48 months) related to substance use and service utilization for psychiatric problems and substance abuse.

Methods: A total of 212 homeless mentally ill adults were randomly assigned to housing vs. treatment first conditions (the former represented by Pathways to Housing, Inc. and the latter represented by transitional housing and residential treatment programs). Pathways to Housing, Inc. was viewed as the experimental condition. Subjects were interviewed at 6-month intervals for 4 years with a 78% retention rate. Measures of outcome variables were the Drug and Alcohol Follow-Back Calendar (Sobell et al., 1988) and the Treatment Services Inventory (McLellan et al. 1991). Group and person-level differences were analyzed graphically and with a growth curve model following hierarchical linear modeling procedures using SAS Proc Mixed software.

Findings: There were no statistically significant group differences in reported levels of alcohol and drug use over time, with a slight trend toward lower alcohol use in the experimental group. Control participants were significantly more likely to use psychiatric and substance abuse treatment services.

Implications for Practice and Research. While control participants might be expected to use more services (since their housing was contingent on treatment adherence), the lack of group differences in substance use was noteworthy. Previously published NYHS findings on other outcomes showed greater housing stability and lower costs for the ‘housing first' group, but our ‘no group difference' findings are consistent with those for other outcomes (e.g., psychiatric status and quality of life) in which group differences are minimal. Although this raises questions about whether ‘housing first' is more effective in non-housing aspects of psychiatric rehabilitation, measurement issues must be considered. Despite the rigorous study design and low attrition rate, we note the likelihood of differential under-reporting of substance use and over-reporting of treatment utilization due to social desirability effects. A fuller understanding of the ‘housing vs. treatment first' debate requires pursuit of qualitative and mixed methods approaches to capture the complex interplay of mental illness, substance abuse and extreme deprivation in this population.