Does One Size Fit All? What We Can and Can't Learn From a Meta-analysis of Housing Models for Persons With Mental Illness

Abstract

Numerous studies have evaluated the impacts of community housing models on outcomes of persons with severe mental illness. The authors conducted a meta-analysis of 44 unique housing alternatives described in 30 studies, which they categorized as residential care and treatment, residential continuum, permanent supported housing, and nonmodel housing. Outcomes examined included housing stability, symptoms, hospitalization, and satisfaction. Outcome scores were converted to effect size measures appropriate to the data. Effect sizes were combined to estimate random effects for housing models, which were then compared. All models achieved significantly greater housing stability than nonmodel housing. This effect was greatest for permanent supported housing (effect size=.63, p<.05). No differences between housing models were significant. For reduction of psychiatric symptoms, only residential care and treatment differed from nonmodel housing (effect size=.65, p<.05). For hospitalization reduction, both residential care and treatment and permanent supported housing differed from nonmodel housing (p<.05). Permanent supported housing achieved the highest effect size (.73) for satisfaction and differed from nonmodel housing and residential care and treatment (p<.001 and p<.05, respectively). The meta-analysis provides quantitative evidence that compared with nonmodel housing, housing models contribute to stable housing and other favorable outcomes. The findings also support the theory that different housing models achieve different outcomes for different subgroups. Data were not sufficient to fully answer questions designed to enable program planners and providers to better meet consumers' needs. It is important to answer these questions with research that uses common measures and adheres to scientific conventions.