Does Active Substance Use at Housing Entry Impair Outcomes in Supported Housing for Chronically Homeless Persons?

The Collaborative Initiative on Chronic Homelessness (CICH) was a multisite federally funded demonstration program begun in 2004 to provide permanent supported housing and supportive primary and mental health services to persons experiencing chronic homelessness (8). Through a competitive process, 11 communities were selected for funding (8). They were charged with providing comprehensive health services and linking them to housing and to use of service models that have been proven effective (most notably, intensive case management [9] or a “housing first” housing model [6]). Funding was provided by the U.S. Department of Housing and Urban Development, the Health Resources and Services Administration, the Substance Abuse and Mental Health Services Administration, and the U.S. Department of Veterans Affairs (VA).

The federal agencies sponsoring the initiative invited the VA Northeast Program Evaluation Center to conduct a national evaluation of CICH client outcomes in order to monitor the initiative's implementation and effectiveness (8). During the 4.5-year data collection period (from February 2004 to September 2008), staff at each site administered assessment interviews to voluntary participants at program entry and quarterly thereafter for up to four years. Because CICH was a service initiative and not a formal research study, there was no systematic inclusion of a comparison group not receiving CICH housing or services (10).

Altogether, 756 individuals out of 1,242 program participants (61%) consented to participate in the national program evaluation and were assessed at program entry. Participation was voluntary and did not influence housing placement or other services. Institutional review boards at each participating site approved the informed consent procedures. After complete study description, written informed consent was obtained. Compared with those who did not participate in the national evaluation (N=508, 41%), participants were more likely to be older, male, black, and have medical or mental health problems but were less likely to have substance use problems or be recruited from a community location (streets, parks, or other outdoor areas) (8). Of the 756 evaluation participants, the study's analytic sample was limited to 410 (54%) participants who were housed within 12 months of program entry and completed an assessment interview just after housing placement and at least one additional follow-up assessment. They reported either high (>15 days of use) or no substance use in the month they were housed. Of 62 assessed but not included in the study, 58 (8%) were never placed into housing and four (<1%) were placed after 12 months. Those never placed did not differ significantly from the study group. [An appendix showing participant selection and comparison of baseline characteristics is available as an online supplement to this article at ps.psychiatryonline.org.] The remaining sample (N=694) was restricted to high-frequency substance users (N=120, 16%) and abstainers (N=290, 38%) in order to ensure comparison between an actively using and a low-use group. Intermediate-frequency substance users (one to 15 days of use) (N=284, 38%) did not differ significantly from either high-frequency users or abstainers or fell intermediate to each group (see online appendix).

Full-time evaluation assistants were trained at project initiation and regularly throughout data collection. Baseline and follow-up interviews were administered in person, and self-report items were recorded during the interview. Interviewer observations and disabling-condition assessments were then confirmed with program staff. Follow-up interviews were typically administered in person or by phone as needed—for example, if participants had relocated. Thus follow-up assessments continued even if participants had discontinued CICH services.

Because participant substance use at the precise time of entry into supported housing was not measured, we used data from the ASI (taken during the first assessment after housing entry) to identify high-frequency substance users, defined as those reporting more than 15 days of alcohol, marijuana, or other substance use during the month in which they were housed (N=120, 29%). Abstainers reported no substance use and constituted the comparison group (N=290, 71%).

Because high-frequency and abstinent participants may have differed in ways that could confound outcome comparisons, significant baseline group differences were identified by using chi square and independent-sample t tests. Excepting substance use variables, these differences were included as covariates in outcome comparisons.

Mixed linear regression models were used to test differences between the high-frequency and abstinent groups over time, with controls for the baseline value of the dependent variable, site, and previously identified baseline covariates. The significance of main effects for time, group, and group × time interaction was evaluated. Least-squares means of the outcome measures for each group averaged over all time points were calculated, as were least-squares means at each time point. Cost and expenditure variables were log-transformed to better normalize the data for statistical analysis. All analyses were carried out in SPSS version 16.0 (18).

No significant group difference in follow-up rates was found. The mean duration of follow-up was 17.9 months, with over 50% of participants actively participating in the evaluation two years after baseline.

Despite similar age at the onset of homelessness among abstainers and users, high-frequency users reported more lifetime years homeless (Table 1). They also reported more years in jail and were more likely to have received public support (through SSI, SSDI, or GA) and to have a diagnosis of posttraumatic stress disorder (PTSD), developmental disability, a medical problem, an alcohol or drug problem, or co-occurring mental and substance use problems. All of these issues suggest poorer overall adaptation or greater addiction severity. Having never been married was less common among high-frequency users. There were no other significant differences on baseline demographic characteristics, eligibility criteria, or health services use (Table 1).

At housing entry, neither number of days housed nor days homeless in the previous 90 days differed significantly between groups (Table 1). High-frequency users reported fewer days institutionalized—with half as many residential treatment days. Because this study's baseline assessment occurred within three months of receiving housing, the number of days housed in the past 90 days was higher than one might otherwise expect for a chronically homeless population. Baseline clinical status measures not related to addiction did not differ significantly between groups, including for the SF-12, BSI, observed psychosis scale, subjective quality of life, and percentage reporting past month employment (Table 1).

Consistent with group definitions, high-frequency users averaged drinking to intoxication on six days of the past 30, and 76% reported illicit drug use in the month before the baseline assessment at time of housing placement (Table 1). Within this high-use group, 40% (N=48) met the >15 day threshold for alcohol use only, 14% (N=17) for marijuana only, 31% (N=37) for other illicit drugs only, and 15% (N=18) for any combination. Indeed, 40% reported using illicit drugs other than marijuana on most days in the past month, indicating significant severity (not shown in table). The ASI-alcohol and ASI-drug scores were substantially greater among the high-frequency users than the abstainers, as were past 30-day alcohol and drug expenditures, reflecting recent active use (Table 1).

There were no significant differences in outpatient medical, mental health, substance use, or Alcoholics Anonymous or Narcotics Anonymous service use (Table 1). At baseline, high-frequency users had accrued higher past-90-day medical costs but lower mental health costs (Table 1).