Slide 1:

The Dynamics of Homelessness and the Impact of Supportive
Housing on Services Use and Costs

Carol Wilkins
Director of Intergovernmental Policy
December 2003
www.csh.org

Slide 2:

Our Mission

CSH helps communities create affordable housing with services to prevent and end homelessness.

Slide 3:

Acknowledgements

Much of the data and many of the slides included in this presentation were prepared by

Dennis Culhane, Ph.D.
Center For Mental Health Policy & Services Research
University of Pennsylvania

The Corporation for Supportive Housing works in partnership with Dr. Culhane and other researchers to increase our knowledge about the dynamics of homelessness and the impact of supportive housing.

Slide 4:

Point and period prevalence of Public Shelter Utilization: New York and Philadelphia

• Percentage of individuals entering shelter on the day data was collected was .25% for New York and .4% for Philadelphia.
• Using historical data from 1992 – 1% of total number of homeless had been in the shelter for one continuous year in New York City compared to slightly more in Philadelphia.
• Using historical data from 1990 – 1992 – 2% of the homeless in New York had been in shelters continuously for 2 years compared to 2.6% in Philadelphia.
• Using historical data from 1990- 1995 3% of the homeless in New York had been homeless for 5 years. Philadelphia did not indicate individuals homeless for five years.

Slide 5:

Annual Rates of Shelter Utilization for Selected Populations

• General Population < 1.3%
• Poor Persons 4.5%
• Poor Children 9.36%
• Poor Black Children (<5 years) 16.12%
• Poor Black Women (18-29) 12.28%
• Poor Black Men (30-49) 19.57%

Slide 6:

Cluster Distributions: Persons and Shelter Days Consumed (Single Adults in New York)

• Three types of homelessness are graphed.
  • Of the homeless nearly 80% are considered transitionally homeless with an average of 1.36 stays a year. They stay an average of 57.8 days and use 30% of the total bed/days available in a given year.
  • Approximately 11% of the persons homeless in New York are episodically homeless with an average of 4.85 stays a single year with a cumulative total of 263 days. The episodically homeless use 18% of the available bed/days.
  • The chronically homeless represent approximately 9% of the total number of homeless with only 2.27 stays but the duration of those stays is 637.8 days. These extended stays use 50% of the total available bed/days.

Slide 7:

Disability Condition and Veteran Status by Cluster (Single Adults in Philadelphia)

• Mental Illness: 8% for transitional homeless, 12% episodic homeless, 20% chronic homeless
• Medical Condition: 12% transitional homeless, 20% episodic homeless, 25% chronic homeless
• Substance Abuse: 28% transitional homeless, 40% episodic homeless, 38% chronic homeless
• Veterans: 8% transitional homeless, 11% episodic, 15% chronic

Slide 8:

Implications for Designing Solutions to Homelessness

• Chronically homeless people need permanent supportive housing
• Transitionally homeless need:
  • Prevention
  • Help for a quick return to housing
• Episodically homeless people need:
  • “ Low Demand” or harm reduction models including safe havens
  • Residential treatment/transitional housing

Slide 9:

The Impact of Supportive Housing on Services Use for Homeless Persons with Mental Illness in New York City

Dennis Culhane, Ph.D. 
Stephen Metraux, M.A.
Trevor Hadley, Ph.D.

Center For Mental Health Policy & Services Research
University of Pennsylvania

Slide 10:

Funding Sources:

Fannie Mae Foundation

The United Hospital Fund of New York

The Conrad N. Hilton Foundation

The Rhodebeck Charitable Trust

The Corporation for Supportive Housing

Slide 11:

NY/NY: Background

• Agreement between NY State and NY City
• Funds capital, operating, and service costs for 3,600 supportive housing units in NYC
• Placement recipients must have an SMI diagnosis & a record of homelessness
• Data available on 4,679 NY/NY placement records between 1989-97

Slide 12:

Research Question

How do NY/NY housing placements affect the use of:

• City shelters
• State psychiatric hospitals
• State Medicaid services
• City hospitals (HHC)
• Veterans Administration hospitals
• State prisons
• City jails

Slide 13:

Data Sources

• NY/NY Housing Placements: 1989-97
• Single Shelter Users and Stays: 1987-99 
• State Hospital Users & Stays: 1990-96
• Municipal Hospital Users & Stays (non-Medicaid): 1989-96 
• Medicaid-Reimbursed Inpatient Hospital Stays: 1993-97
• Medicaid-Reimbursed Outpatient Visits: 1993-97 
• Veterans Hospital Stays: 1992-99 
• State Criminal Justice Prison Use & Convictions: 1987-97
• City Jail Use: 1987-99

Slide 14:

Research Method #1

Pre-Post Test Analysis

• From a single point in time, data was collected two years before the NYC shelter placement and two years after the placement.

Slide 15:

Research Method #2

Matched Pair Case-Control Design

• A total of 4,679 persons with NYC placement were matched with a control pool of users from the services system and matched on race, sex, age, substance abuse usage and mental health issues. The resulting pairs were matched on the pre-intervention services used.

Slide 16:

The Cost of Homelessness

Service Provider	Mean Days Used (2-year pre-NY/NY)	Per Diem Cost	Annualized Cost
NYC DHS - Shelter	137	$68	$4,658
NYC OMH - Hospital	57.3	$437	$12,520
NYC HHC - Hospital	16.5	$755	$6,229
Medicaid - Hospital	35.3	$657	$11,596
Medicaid - Outpatient	62.2 (visits)	$84	$2,612
VA - Hospital	7.8	$467	$1,821
NYS - DCJS - Prison	9.3	$79	$367
NYC DOC - Jail	10	$129	$645
Total			$40, 449

Slide 17:

Cost of Homelessness #2

This slide shows a pie chart with the following information:

• DHS Shelter $4658
• Mental Hospital $12,520
• NY Health and Hospitals Corporation $6229
• Medicaid Inpatient $11,596
• Medicaid Outpatient $2612
• VA Hospital $1821
• State Prison $367
• City Jail $645

Slide 18:

Reductions in Shelter Use
Source: the NYC Department of Homeless Services

• Persons 49.8%
• Days 60.5%

Slide 19:

Reductions in Inpatient Hospital Use
Source: the NYS Office of Mental Health

• Persons 25.4%
• Days 60.8%

Slide 20:

Reductions in Public Hospital Use
Source: the NYC Health and Hospitals Corporation (non-Medicaid)

• Persons 19.2%
• Days 21.2%

Slide 21:

Reductions in Medicaid-Reimbursed Inpatient Hospital Use
Source: the NYS Department of Health

• Persons 8 %
• Days 24.4%

Slide 22:

Increases in the Use of Medicaid-Reimbursed Outpatient Services
Source: NYS Department of Health

• Persons – 13.1%
• Days -75.9%

Slide 23:

Reductions in VA Inpatient Hospital Use
Source: US Departmentt of Veterans Affairs

• Persons 12%
• Days 24.4%

Slide 24:

Reductions in Incarceration in State Prison
Source: NYS Department of Criminal Justice Services

• Persons 64.1%
• Days 84.8%

Slide 25:

Reductions in Incarceration in City Jails
Source: NYC Department of Corrections

• Persons 26.9%
• Days 38.0%

Slide 26:

NY/NY Savings: Per Housing Unit Per Year

Serivces	Annualized Savings per NY/NY Unit
DHS Shelter	$3,779
OMH Hospital	$8,260
HHC Hospital	$1,771
Medicaid - Inpatient	$3,787
Medicaid - Outpatient	($2,657)
VA Hospital	$595
NYS Prison	$418
NYC Jail	$328
Total	$16,282

Slide 27:

Cost Savings #2

• DHHS Shelter $2819
• Mental Hospital $6162
• HHC Hospital $1321
• Medicaid Inpatient $2825
• Medicaid Outpatient – (negative) $2657
• VA Hospital $444
• NY Prison $312
• NY Jail $245

Slide 28:

NY/NY Housing Costs

Housing Type	Number of Units	Unit Cost	Net Cost per Housing Unit (% of total)	Net Cost of Housing Initiative
Community Residence	1,384	$20,534	$4,252 (20.7%)	$5,884,768
Supportive Housing	2,231	$17,276	$994 (5.8%)	$2,217,614
All NY/NY (weighted mean)	3,615	$18,523	$2,241 (12.1%)	$8,101,215

Slide 29:

NY/NY Housing Costs and Savings

• Savings per unit from reduction = $16,282
• Annual Per Unit Cost of Housing for Community Residence = $19,662
• Supportive Housing = $17,277
• Weighted Mean = $18,190

Slide 30:

Summary of Findings

• Homeless mentally ill are heavy service users (37% of last 2 years spent in institutional settings)
• Providing services for homeless mentally ill is expensive ($40,449 per person per year)
• Providing NY/NY housing for homeless mentally ill reduced costs by 30% ($16,272 in savings per unit)

Slide 31:

NY/NY Cost Study Conclusions

• 95% of supportive housing costs offset by savings from service reductions attributable to housing placements
• Study underestimated savings associated with program-funded services and crime
• Study did not quantify benefits to consumers or neighbors
• NY/NY was a sound public investment

Slide 32:

Looking Beyond New York City

• CSH has supported efforts to measure the impact of supportive housing on services use and costs in other States, including:
  • Connecticut
  • Minnesota
  • California
• New results available from a study of VA supported housing program (Rosenheck)
• State and local governments can partner with supportive housing providers and tenants to measure the impact of a range of models

Slide 33:

State and Local Government Strategies

• Measure impact of supportive housing on use of services in a few service systems where
  • Data is available and consistent from year to year
  • Reductions will have greatest fiscal or policy impact
  • Program design and logic model predict outcomes
• Measure service use 12 – 24 months before and after homeless people move into supportive housing
• Use supportive housing wait lists as control group if selection minimizes differences

Slide 34:

Consistent Findings: Housing + Services Make a Difference

• More than 80% of supportive housing tenants are able to maintain housing for at least 12 months
• Most supportive housing tenants engage in services, even when participation is not a condition of tenancy
• Use of the most costly (and restrictive) services in homeless, health care, and criminal justice systems declines
• Nearly any combination of housing + services is more effective than services alone

Slide 35:

San Francisco Health Housing and Integrated Services Network

• Two permanent supportive housing projects: 236 study participants entered housing (1994 – 1998)
• Shelter + Care program targeted homeless adults with co-occurring mental illness, substance abuse and/or HIV/AIDS
• Wait list with lottery to select for housing + few applicants denied housing (no clean & sober requirements) = random assignment to housing
• Hospital and treatment system data provided by SF Department of Public Health
• Analysis by Tia Martinez, UC Berkeley

Slide 36:

Decline in total Emergency Department Visits (N= 236)

Graph shows emergency department visit data for 13 –24 months prior to receipt of housing through 13 – 24 months after receipt of housing. There is a marked decline in medical emergency visits and a more moderate decline in psychiatric visits. Some clients indicated more than one emergency visit.

Over the span of 4 years the number of visits is as follows:

• Overall emergency visits:
  • 430 – 457 (before housing)
  • 202 – 228 (after housing)
• Medical emergency visits:
  • 351 – 378 (before housing)
  • 154 – 186 (after housing)
• Psychiatric emergency visits:
  • 79 – 79 (before housing)
  • 48 – 42 (after housing)

Slide 37:

Decline in total Hospitalizations (n=236)

Graph shows decline in hospitalizations for 13 – 24 months prior to receipt of housing through 13 – 24 months after receipt of housing. There is a less clear relationship between housing and medical hospitalizations on this graph. However, housing reduces the overall number of psychiatric hospitalizaitons.

Over the span of 4 years the number of hospitalizations is as follows:

• Overall hospitalizations:
  • 71 – 80 (before housing)
  • 44 – 52 (after housing)
• Medical hospitalizations:
  • 46 – 45 (before housing)
  • 26 – 40 (after housing)
• Psychiatric hospitalizations:
  • 25 – 35 (before housing)
  • 18 – 12 (after housing)

Slide 38:

Changes in Emergency Department Use From Year 1 to Year 2 Among Cases and Controls

Bar graphs comparing cases and controls for both years indicate a decline in emergency department usage as follows:

• Probability
  • year one
    • case =51
    • control =44
  • year two
    • case = 34
    • control = 44
• Mean usage
  • year one
    • case = 1.56
    • control = 1.68
  • year two
    • case = .66
    • control = 1.32

Slide 39:

Supportive Housing Reduces Use of and Costs for:

• Hospital inpatient care for medical and psychiatric conditions
• Hospital emergency room visits – especially for the most frequent users of ER
• Psychiatric emergency and institutional care
• Residential mental health & substance abuse treatment – especially detox
• Jails and prisons
• Emergency shelters

Slide 40:

Supportive Housing May Increase Use of and Costs for:

• Outpatient primary and specialty medical care
• Some mental health services (e.g. case management, pharmacy)
• Methadone (more consistent participation)
• Services to address substance abuse problems, including services delivered outside of traditional treatment programs
• Vocational and employment services
• Probation

Slide 41:

Policy Implications

• Costs and savings are often in different systems of care – and sometimes at different levels of government
• Coordinated investments are needed for housing and services
• Opportunities to maximize savings may be greatest when focusing on chronic homelessness and/or homeless frequent users of emergency and inpatient services

Slide 42:

The Bottom Line

• “Standard care” emergency responses to chronic homelessness are very costly
• Investments in supportive housing will significantly reduce services use and public costs in some systems of care
• The net cost of achieving much better outcomes is relatively small – if savings can be re-invested

Slide 43:

The Dynamics of Homelessness and the Impact of Supportive
Housing on Services Use and Costs

Carol Wilkins
Director of Intergovernmental Policy
December 2003
www.csh.org