Cost-effectiveness Analysis of Supported Housing and Supportive Community: a Sub-study for the SAMHSA CMHS Housing Initiatives II Cooperative Agreement
Updated June 2002

Principal Investigators: Kristine Jones, Ph.D., Carole Siegel, Ph.D. with Judith Samuels, Ph.D.
Field Researchers: Kim Kennard, Ph.D., Kim Prchal, MSW.

This project aims to compare the cost-effectiveness of Supported Housing (SH) with Supportive Communities (SC). This is a site-specific sub-study of the SAMHSA funded study of Comparison of Housing Alternatives for SMI.

Methods: Interviews are conducted on an ongoing basis with new tenants in SH and CR housing re: utilization of general health, mental health and substance abuse services, other service use, medication use, supportive and competitive employment earnings, other income, housing expenses, legal and safety concerns, as well as actual residence for each night covered by the interviews. 

Per unit costs will be estimated for each resource category. The costs of an individual for a specific resource category will be constructed, as the product of that individual’s use of the resource and its respective unit cost. Total cost for the individual is the aggregation over all the various resource categories costs associated with that individual. The total cost for a housing model will be the sum of the costs associated with all persons initially tenants in that housing model.

The "effect" is the sum of all nights spent in residential categories considered ‘appropriate housing.’ The total effect for a housing model is the sum of all ‘appropriately housed’ nights for tenants initially placed in the housing model.

Measures will be constructed monthly and will be aggregated to yield four three month cost and effect variables as well as annual cost and effect variables. Similar exercises will derive costs and effect measures for sub groups of tenants.

Univariate analysis on both the cost and effect variables will be carried out. Each cost component will be compared and tested for a significant difference between the two housing models and between sub groups within housing models. An analysis of differences in cost patterns across the 12 months and a comparison of costs between those appropriately housed will also be conducted. ANCOVAs will be used and include covariates expected to have prognostic value.

Two different cost effectiveness ratios will be considered: 1) the standard expected costs to expected effects ratio, and 2) costs and effects measures averaged up to the housing model level. These average values are used in computing the ratios for the respective models

Confidence intervals will be obtained for the first ratio form using Fieller’s theorem. Testing for differences in these measures will be done using a log-likelihood test of the equality of two ratios. as developed by Siegel, et al (1996). The second ratio form is the expected costs to effect ratio. Cost effectiveness ratios are constructed on each individual then averaged up to the housing model level. For these ratios, standard ANCOVAs, including repeated measures models, and including covariates expected to have prognostic value will be used to compare costs of SH to those of SC.

Results:  A cost effectiveness measure – defined as social cost per ‘days appropriately housed’ has been constructed on a sample of 157 new tenants in either SH or CR programs.   To date, information on social costs as well as ‘days appropriately housed’ have been obtained through 1700 interviews with these new tenants.  Interviews have been ongoing since 1988. The interviews have been entered into a data management system developed for the project.  

INCLUSION OF GENDER AND MINORITY SUBJECTS

The gender and ethnic distribution of new tenants is expected to be similar to that found in the New York City’s Housing Initiative I’s study of new tenants moving into SH and SC residences. The gender distribution was 67% male and 33% female. The ethnicity distribution was 28% White, 51% Black/African American, 17% Hispanic/Latino and 4% other.

The findings of this project are expected to have prognostic value in comparing the costs of supported housing with supportive communities.

PLANS

Tasks to be completed include estimating unit costs for each resource type, linking unit costs to resource volume and accumulating costs for each sample member.  Once these tasks are completed cost effectiveness analysis on incremental net health benefits will be conducted. 

Reference: Siegel C, Laska E, Meisner M (1996). Statistical methods for cost-effectiveness analyses. Controlled Clinical Trials, 17:387-406.

Presentation: 
Jones, K. (Nov. 1999): Measuring costs for a cost-effectiveness study: case of supported housing vs. community residences. Paper presented at American Public Health Association conference, Chicago IL. 

[Top]

HOME  

Cores: Negotiating Lives in Communities | Methods for MH Services Research | Systems Integration |  Promoting Recovery  |
Topics: Service Delivery Systems | Homelessness | Instrument Development | Managed Care | Mental Illness and Chemical Abuse | Multi-cultural Issues |  Improving Services Research | Recovery | Treatment Innovations |
Statistical Methods and Computer Programs