National dissemination of supported housing in the VA: model adherence versus model modification

The continuing development and dissemination of emerging evidence-based practices may be facilitated by the availability of descriptive information on the actual delivery of the service, and its variability, across sites. This paper presents data on the participation of 2,925 homeless veterans in the Housing and Urban Development-Veterans Affairs Supported Housing (HUD-VASH) program at 36 sites across the country, for up to five years. While most conceptual models emphasize rapid placement, sustained intensive case management, rehabilitation services, and "permanent" housing, no program has yet presented empirical data on the actual delivery of such services over an extended period of time.

Using extensive longitudinal data from the VA's national homeless outreach program, the Health Care for Homeless Veterans (HCHV) program, a quantitative portrait presents what happens in supported housing in a large real-world dissemination effort.

Program entry to HUD-VASH was generally slow with 108 days (sd = 92 days) on average passing between program entry and housing placement. Total program participation lasted 2.6 years on average (sd = 1.6 years)-just half of the possible 5 years. Service delivery became substantially less intensive over time by several measures, and three-fourths of the veterans terminated within five years, although the vast majority (82%) were housed at the time. Few veterans received rehabilitation services (6%) or employment assistance (17%) and most service delivery focused on obtaining housing.

These data suggest that real-world supported housing programs may not adhere to the prevalent model descriptions either because of implementation failure or because veteran needs and preferences differ from those suggested by that model.