1071 — Ending Homelessness: Can We Rationally Allocate Social Benefits?
Lead/Presenter: Stefan Kertesz,
All Authors: Kertesz sg (Birmingham VA Medical center)
DeRussy A (Birmingham VA Medical Center)
Holmes S (Boston University)
Austin E (University of Alabama at Birmingham)
Pollio D (University of Alabama at Birmingham)
Kim Y (University of Alabama at Birmingham)
Permanent housing interventions such as Housing First (HF) have been shown to reduce hospital use by formerly homeless persons in some studies. These programs promise meaningful health service cost offsets when they target clients with particularly high levels of health need and health service utilization. Partly on this basis, in 2012, VA directed that veterans with greatest health and social vulnerability be prioritized for its HUD-VASH Housing program. We assessed whether this directive was associated with rising selectivity in clients entering HUD-VASH.
We compiled data from veterans seeking homeless assistance at 8 VA Medical Centers from 2011-2014 (n = 15,006). We identified 2 markers for excess hospital utilization (top 10% for emergency department visits and top 10% for hospital admissions). We flagged vulnerabilities in other domains: Homelessness (Chronically Homeless, Custody of Children, Literal Homelessness e.g. shelter, outdoors); clinical status (# of chronic medical disorders, # of chronic mental disorders, Alcohol abuse, Drug abuse); Service History (Iraq/Afghan Veteran, Year ofService); and demographics. We compared these vulnerabilities among HUD-VASH entrants (n = 5608,37%) and non-HUD-VASH entrants (n = 9393,63%), and modeled HUD-VASH Entry (yes/no) as a dependent variable in logistic regression, iterating models for VA's Pre-Housing First (2011-12) and Post-Housing First (2013-14) era.
Markedly increased selectivity was found for Chronic Homelessness (72% vs 40% of HUD-VASH Entrants vs non-Entrants in 2011-12, 81% vs 37% in 2013-14). Female gender and custody of children wereassociated with HUD-VASH entry in both eras. Other vulnerabilities, such as number of mental or medical disorders, had minimal or no association with entry into the HUD-VASH program. Contrary to expectation, frequent hospital admission was inversely associated with HUD-VASH entry (OR 0.6 and 0.7 for Pre-HF and Post-HF, respectively) and ED use was not associated with HUD-VASH entry.
VA policy changes shifted allocation of housing resources toward chronically homeless persons, especially women and parents. However, many characteristics demarcating increased vulnerability were not associated with HUD-VASH entry.
For VA and for other healthcare delivery systems that seek to advance population health, additional steps will be required to allocate clinical and social resources to especially vulnerable populations.