Transforming the District of Columbia's Public Homeless
Assistance System
Martha R. Burt
Sam Hall
June 2, 2008
Urban Institute
2100 M Street, N.W.
Washington, D.C. 20037
www.urban.org
The contents of this report are the views of the authors and do not necessarily reflect the views or
policies of the Urban Institute, its trustees, or funders.
i
CONTENTS
Chapter 1: Introduction ............................................................................................................... 1
Methods and Data Sources.............................................................................................................. 1
Data from TCP............................................................................................................................ 1
Other Information ....................................................................................................................... 2
Chapter 2: Understanding How Homeless People Use the District’s Emergency Shelter
System ............................................................................................................................................ 5
Highlights........................................................................................................................................ 5
Introduction..................................................................................................................................... 6
D.C.’s Emergency Shelter Population ............................................................................................ 6
What Are the Differences among “FY06 Cohort,” “Active in FY07,” and “TCP Approach”? . 8
Single Adults Active in ES in FY07........................................................................................... 9
Gender, Race, and Age ........................................................................................................... 9
Veterans and Disabilities ...................................................................................................... 10
Living Arrangement on the Night Before Entering Shelter.................................................. 10
Families: AHAR and Virginia Williams Family Resource Center Data .................................. 10
Age, Race, and Gender ......................................................................................................... 11
Disability............................................................................................................................... 11
The Application and Placement Process............................................................................... 12
Living Arrangement at the Time of Applying for Shelter .................................................... 13
Length of Stay (LOS) and the FY06 Cohort Analyses ................................................................. 13
Single adults—FY06 Cohort......................................................................................................... 13
Total Length of Stay for the FY06 Single Adults Cohort......................................................... 14
Long-Stayers ......................................................................................................................... 15
Reentry.................................................................................................................................. 15
LOS Categories, Demographic Trends, and Missing Data................................................... 17
Age and Missing Values ....................................................................................................... 17
Race/Ethnicity....................................................................................................................... 18
Gender................................................................................................................................... 19
Disability............................................................................................................................... 20
Mental Illness........................................................................................................................ 21
Substance Abuse ................................................................................................................... 22
Physical Disabilities.............................................................................................................. 23
Families (FY06 Cohort)................................................................................................................ 23
Length of Stay (LOS) and Families (FY06 Cohort) ................................................................. 24
Long-Stayers ......................................................................................................................... 24
Those Who Exit and Return in Under 12 Months ................................................................ 25
Demographic Trends for Families in Different LOS Categories.............................................. 25
Race, Gender, Age, Family Size, and Missing Data............................................................. 26
Disabilities of Heads of Family Households, by Length of Stay.......................................... 26
Domestic Violence................................................................................................................ 27
ii
Social Supports ..................................................................................................................... 28
Income and Benefits ............................................................................................................. 28
Conclusions............................................................................................................................... 29
Benefits of LOS Analysis Using a Cohort Approach ............................................................... 29
Data Limitations and Data Management ...................................................................................... 30
The Closed Nature of the District’s HMIS Greatly Hampers Its Analytic Capabilities........... 31
Other Problems ......................................................................................................................... 33
Implications................................................................................................................................... 35
Chapter 3: Scope and Structure of the Local Homeless Assistance System for Single Adults
and Families................................................................................................................................. 37
Highlights...................................................................................................................................... 37
Emergency Shelter System Structure............................................................................................ 38
The Situation for Families ........................................................................................................ 38
Community Care Grants ....................................................................................................... 39
Emergency Rental Assistance Program ................................................................................ 39
Emergency Shelter ................................................................................................................ 39
The Situation for Single Adults ................................................................................................ 40
Beginnings of Specialization within Single Men’s Shelters..................................................... 41
Issues with Low-Barrier Shelters.............................................................................................. 42
Sources of Guidance ............................................................................................................. 43
Transitional Housing..................................................................................................................... 43
Permanent Supportive Housing .................................................................................................... 45
Supportive Services ...................................................................................................................... 47
Services for Homeless People Living on the Streets and in Emergency Shelters .................... 47
If Ending Homelessness Is the Goal, What Services Are Needed and How Should They Be
Configured?........................................................................................................................... 48
Services for People in Permanent Supportive Housing ............................................................ 49
Implications................................................................................................................................... 49
Chapter 4: District Government Agency Activities Related to Homelessness ...................... 51
Highlights...................................................................................................................................... 51
“Silo” Problems in the District ................................................................................................. 52
Interagency Communication and Coordination........................................................................ 54
Data Sharing ............................................................................................................................. 55
District Government Programs that Impact Homelessness ...................................................... 56
Department of Human Services (DHS) ................................................................................ 57
Department of Mental Health (DMH) .................................................................................. 59
The Department of Health’s (DOH) Addiction Prevention and Rehabilitation Agency
(APRA) ................................................................................................................................. 60
D.C. Housing Authority (DCHA)......................................................................................... 61
Department of Housing and Community Development (DHCD) ........................................ 62
Fire and Emergency Medical Services (FEMS) ................................................................... 63
Metropolitan Police Department (MPD) .............................................................................. 63
iii
Department of Corrections (DOC)........................................................................................ 64
Court Services and Offender Supervision Agency (CSOSA)............................................... 65
Pretrial Services Agency (PSA)............................................................................................ 66
Office of Property Management (OPM) ............................................................................... 67
Implications................................................................................................................................... 67
Chapter 5: Implications and Recommendations...................................................................... 69
1. Move Chronically Homeless People into PSH ......................................................................... 70
2. Create a Process to Prioritize Who Gets the NExt Available PSH Unit................................... 71
3. Reconfigure Emergency Shelter ............................................................................................... 73
4. Make the HMIS Work .............................................................................................................. 74
Overall Substantive System Changes ....................................................................................... 74
Technological Improvements ................................................................................................... 75
Performance Monitoring and Performance-Based Contracting................................................ 76
Conclusions................................................................................................................................... 77
References..................................................................................................................................... 79
Appendix A: List of People Interviewed ...................................................................................... 81
Appendix B: List of Acronyms..................................................................................................... 83
iv
ACKNOWLEDGMENTS
Many individuals have assisted us on this project; we would especially like to thank the
following:
•
Directors and staff at DHS who took the time out of their busy schedules to provide
insight and direction at all levels of this report; special thanks to Clarence Carter, Fred
Swan, Kate Jesberg, Sakina Thompson, and Deborah Carroll.
•
Other District agency staff and directors, with special thanks to Michael Kelly at DCHD,
Tori Whitney at APRA, Stephen Baron at DMH, Leila Edmonds and Guyton Harvey at
DHCD, Susan Schaffer and Spurgeon Kennedy at PSA, Calvin Johnson at CSOSA, Brian
Jordan at MPD, and Michael Williams at FEMS, for their willingness and energy in
helping us understand the breadth of homelessness in the District and how it involves city
agencies.
•
TCP Director Sue Marshall, and staff Cornell Chapelle, Amy McPherson, Michele
Salters, Clarence Stewart, Tamura Upchurch, Xiawei Zheng, and Mathew Winters, who
provided us significant access to the many resources at The Community Partnership.
Special thanks to Darlene Mathews and Tom Fredrickson for their help with providing
and analyzing HMIS data.
•
Mayoral staff at both the Office of the City Administrator and the Office of the Deputy
Mayor for Planning and Economic Development, particularly Laura Zeilinger, Julie
Hudman, and Leslie Steen, who helped shape our Workplan and the structure of our
research.
•
Council Member Tommy Wells and his staff, Adam Maier, Ram Uppuluri, and Yulondra
Barlow, who worked with us in defining our goals for this assessment.
•
Advocacy groups, especially Washington Legal Clinic for the Homeless staff Patricia
Mullahy Fugere, Mary Ann Luby, Amber Harding, Marcy J. Dunlap, and Andy Silver,
who informed many of the sections in this report.
•
The more than 60 District homeless shelter directors and providers whose insights, data,
and anecdotes greatly informed many of the sections presented here.
For a complete list of people who contributed to this assessment, see Appendix A.
We appreciate the significant contributions everyone made to our research, and those who
reviewed preliminary drafts of our reports. The findings and views herein are solely those of the
authors, who are responsible for any errors or omissions.
1
Chapter 1: Introduction
Chapter 1
Introduction
Homelessness has been a continuing presence in the District of Columbia for almost three
decades. It only became a high priority issue for public action, however, when the administration
of Mayor Adrian Fenty assumed control of District government in January 2007. As a City
Council member, the Mayor had been instrumental in passing the Homeless Services Reform Act
of 2005 (HSRA); he quickly made clear that ending homelessness in the District would be
among the most important goals of his administration.
Toward this end, the Department of Human Services was authorized to contract with the Urban
Institute to conduct an assessment of the District’s homeless assistance system, with the
expectation that the results of such an assessment could help guide efforts to transform the
system to make it more effective at reducing and ultimately ending homelessness. This
assessment began in July 2007. In addition to the present report, it has produced one lengthier
report, The Community Partnership and the District of Columbia’s Public Homeless System,
which looks at the roles of the Community Partnership and the ways that it has performed them
(Burt and Hall 2008a). A summary report, Major Recommendations: Summary Report of the
Urban Institute’s Assessment of the District of Columbia's Public Homeless Assistance System
(Burt and Hall 2008b) presents the assessments major recommendations, integrating findings
from the two longer reports.
This report describes the nature of homelessness in the District of Columbia and assesses the
structure of the District’s homeless services. It describes people who use homeless assistance
services; the network of providers that offer the services; the ways that D.C. government
agencies fund homeless services, provide direct services specifically targeted to homeless
people, and find themselves impacted by homelessness because homeless people are some of the
most frequent users of agency resources. After presenting “what is,” the report turns to “what
could be” and “what ought to be” if the District government is going to succeed at one of its
stated goals—to significantly reduce or even end homelessness within its boundaries by 2014.
For many of the issues addressed, and particularly when describing “what could be” and “what
ought to be,” we offer the experiences of some other communities that have faced the same
challenges as the District and are seeing their efforts pay off in reduced homelessness.
METHODS AND DATA SOURCES
Urban Institute staff pursued a number of approaches to gathering the information that we
describe in this report, which came from a variety of sources. These include:
DATA FROM TCP
• Interviews with most TCP staff, from the Executive Director to the accountants.
•
Examination of budget and contract documents between DHS and DHCD and TCP.
Chapter 1: Introduction
2
•
Examination of budget and contract documents for more than 100 contracts between TCP
and homeless service providers.
•
Examination of standard reports from the HMIS maintained by TCP.
•
Direct analysis of hundreds of thousands of raw HMIS records on emergency shelter
users, supplied to us by TCP.
•
Review of performance standards adopted by different contract agencies and comparisons
of the standards adopted to actual performance.
•
Examination of common standards for provider behavior and working with clients.
•
Examination of inspection and maintenance reports.
OTHER INFORMATION
• Interviews with more than 100 public agency staff, staff of elected officials, and staff of
homeless assistance and advocacy agencies ranging from executive directors to case
workers.
•
Analysis of Administrative Review and Fair Hearing records maintained by DHS,
covering the first six months of these processes under the HSRA.
•
Analysis of monitoring reports covering the first six months of operations of DHS’
Shelter Monitoring Unit.
•
Examination of critical documents including Homeless No More (the District’s ten-year
plan to end homelessness), the HSRA and its amendments, regulations, common
standards, earlier reports on the District’s homeless assistance system, the Inspector
General’s report on D.C. Village, among others.
•
The fist author’s knowledge of approaches used by many communities around the
country to address the issues facing the District.
Parts of this report build on information in our first report for this evaluation, The Community
Partnership and the District of Columbia's Public Homeless System (Burt and Hall, 2008a). In
particular, we incorporate into the present report parts of the earlier document that discuss the
structure of the District’s network of homeless assistance and the ability to conduct data analyses
that are timely and useful for policy making.
This report has four more chapters. Chapter 2 describes single adult and family emergency
shelter users, looking at length of stay and household characteristics. It also discusses at some
length the problems we encountered with the District’s homeless management information
system (HMIS) 1 and what needs to be done to make it a more useful tool for policy makers.
1
A full list of acronyms can be found in Appendix B.
Chapter 1: Introduction
3
Chapter 3 describes the District’s homeless assistance system as currently configured, including
the recent changes made to the family emergency shelter system as part of closing D.C. Village
and making strides toward transforming the homeless service system for families.
Chapter 4 describes activities and programs of District government agencies that specifically
serve homeless people or are impacted to a significant degree by homeless people. It also
describes the information resources these public agencies maintain for their clients or users that
might or could be used to identify people who are homeless.
Chapter 5 presents recommendations in several areas that we hope will help inform the homeless
system transformation promised by the Mayor. Implications from the information provided in
chapters 2, 3, and 4 are drawn together, along with relevant aspects of the findings reported in
Burt and Hall (2008a). We offer examples of approaches used in other jurisdictions for possible
adaptation by the District.
Chapter 1: Introduction
4
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
Chapter 2
Understanding How Homeless People Use the District’s
Emergency Shelter System
HIGHLIGHTS
2
•
From October 1st, 2006 to September 30th, 2007 (FY07) an estimated 12,768 different
single adults stayed in the District’s emergency shelter system.
•
On an average night in FY07, an estimated 1,988 single adults stayed in emergency
shelter, occupying about 88 percent of regular emergency shelter capacity.
•
Just under half (47 percent) of single adults entering emergency shelter in FY06 stayed
fewer than 7 days; 86 percent stayed fewer than six months.
•
Of single adults entering in FY06, 494 (4 percent) stayed 365 days or longer; another
1,167 (10 percent) stayed between 181 and 364 days. These 14 percent of single adults
account for over half of emergency shelter bed nights used by single adults in a year.
•
Moving all single adults who use more than 180 emergency shelter nights in a year into
permanent supportive housing could result in being able to close about 1,300 emergency
shelter beds for single adults, or around half of current emergency shelter capacity. The
District could then design and offer less crowded, more specialized, more effective
emergency shelter programs.
•
In FY07, an estimated 529 families, including 1,718 adults and 1,043 children, stayed
in emergency shelter.
•
On an average night in FY07, an estimated 136 families, including 711 persons in
families, used emergency shelter.
•
65 percent of families entering emergency shelter in FY06 left in fewer than six months.
•
Of families entering in FY06, 70 (19 percent) stayed 365 days or longer; another 60 (16
percent) stayed between 181 and 364 days. These 35 percent of families account for
just over half of emergency shelter unit nights used by families in a year. 2
•
Moving all families who use more than 180 emergency shelter nights in a year into
permanent supportive housing could result in being able to close about 114 emergency
This calculation uses family emergency shelter units accounted for in the 2007 Housing Inventory Chart, before
DC Village was closed.
5
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
shelter units for families, or just under half of the emergency shelter capacity at the time
of this analysis. This transition has already started with the closing of D.C. Village.
•
The current technology impedes TCP’s ability to manage the District’s HMIS because
staff are preoccupied with technical problems and data management issues that can and
should be handled by computer programs. Further problems exist within the HMIS
because of the closed nature of the system and inadequate intake procedures.
This chapter describes the people who use the District’s emergency shelters, focusing on those
who entered shelter in FY 2006 and following them for 12 months from the date of their first
FY06 shelter entry. Other data come from analyses conducted by TCP.
INTRODUCTION
Limited descriptions of the scope of homelessness in the District have been available for years.
Annual point-in-time counts have revealed how many people are homeless on a particular winter
day each year for the past seven years, and allowed us to estimate how many had been homeless
over the course of the whole previous year.
In addition, the HMIS database administered by TCP has for several years assembled data from
virtually all emergency shelters and from about 40 percent of transitional housing and permanent
supportive housing programs operating in the District. HMIS data have been used to report the
number of people using emergency shelters on any given day or over the course of a year and the
average number of days each person stays in emergency shelter.
More sophisticated analysis of HMIS data recently became possible when TCP acquired the
Advanced Reporting Tool (ART) from the vendor that supplies the HMIS software. This
improved technology, plus expanded data availability by including the population of families
applying for shelter through the Virginia Williams Family Resource Center (VWFRC), should
allow more detailed descriptions of the sheltered population. Data analysis conducted for this
report by Urban Institute researchers has moved beyond the type of reporting produced by HMIS
in the past, but data limitations still made it challenging for us to provide meaningful answers to
some of the important policy questions that this evaluation set out to answer.
This chapter first describes the single adults and families who use the District’s emergency
shelter (ES) system. In the process, we hope to provide D.C. policy makers with a better
understanding of who uses the emergency shelter system, their patterns of use, how long they
actually stay, and the implications of these patterns for policy decisions surrounding emergency
shelter structures and services. Thereafter, we describe the limitations of the current HMIS and
the difficulties we had in producing the analyses presented. We end with recommendations for
some much-needed upgrading of the District’s HMIS and technical capacity to produce useful
data in a timely manner.
D.C.’S EMERGENCY SHELTER POPULATION
The analyses reported in this chapter that describe everyone who was active in the Continuum of
Care during FY07 (October 1, 2006 through September 30, 2007) were done by TCP as part of
6
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
7
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
8
its contribution to the federal Annual Homeless Assessment Report (AHAR). 3 We call these the
“Active in FY07” analyses, which TCP conveyed to the Urban Institute as soon as they were
completed. Urban Institute researchers conducted the remaining analyses that tracked shelter use
for 12 months from date of entry for everyone who entered a Continuum of Care emergency
shelter in FY06. We call these the “FY06 Cohort” analyses. There is yet a third approach—the
TCP approach. Figure 2.1 graphically illustrates the differences in the three approaches—
differences that lead to very different answers to critical questions.
WHAT ARE THE DIFFERENCES AMONG “FY06 COHORT,” “ACTIVE IN FY07,” AND “TCP
APPROACH”?
Figure 2.1 illustrates the differences among the various approaches to analyzing HMIS data. The
top panel in figure 2.1 shows the Cohort approach, the middle panel shows the “active during a
specified time period” approach used for the AHAR, and the bottom panel shows the approach
used by TCP to calculate length of stay when it is doing its regular analyses rather than analyses
for the AHAR. Each horizontal bar in figure 2.1 represents one person’s use of emergency
shelter during FY05, FY06, and FY07. The same people—A through H—appear in each panel,
but their time in the system is not treated the same by the different analytic approaches. Take
person A, who enters shelter in June 2006 and stays 12 months. Each cell shaded red means the
person used shelter that month and that month is included in the length of stay analysis. Each cell
shaded green means the person used shelter that month but the month is not counted in the length
of stay analysis. Each cell shaded blue means the person was not in shelter that month.
A Cohort approach gives the most accurate picture of how people flow into and out of a system.
If you want to know how long people stay once they enter shelter, what proportion stay only a
short period of time and what proportion stay seemingly forever, a Cohort approach will give
you the right answers.
An “Active in FY07” approach tells you how many people used a shelter service in FY07, and
how many days or months of shelter they used in total during that year. It does not care when
people entered shelter or when they left, only whether they were present in FY07 and how much
shelter they used in that year.
The TCP approach answers the question “Of people who used the system in FY07, how much
time have they spent in shelter since they entered, whenever that was?” Table 2.1 summarizes
the approaches, the questions posed, and the answers one would get from the hypothetical set of
eight people shown in figure 2.1. 4
3
The AHAR is a report that Congress requires HUD to submit every year—the third AHAR is about to be released.
The AHAR is designed to be able to track progress toward ending homelessness. It presents homeless counts and
characteristics of homeless people based on homeless management information systems operated by Continuums of
Care and on bi-annual point-in-time counts of homeless people conducted by Continuums of Care. HUD contracts
with Abt Associates to produce the AHAR. In turn, Abt Associates works with a random sample of 80 jurisdictions
(usually counties) selected to be representative of the entire country. The District is one of those jurisdictions. All
jurisdictions contributing to the AHAR perform the same analyses, which are determined by Abt Associates.
4
Examples in figure 2.1 are illustrative only, and are not typical shelter use patterns in the District, where most
people stay only a few days.
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
9
Table 2.1: Different Approaches to Measuring Shelter Use and Length of Stay
Approach
FY06 Cohort
Active in FY07
(AHAR)
Regular TCP
Question answered
Once they enter shelter, how much of the next 12 months
do they use shelter? What is their average length of stay,
up to 12 months?
How much shelter was used in FY07? On average, how
much did each person use?
For everyone using shelter in FY07, how long had they
been in the system, counting from when they entered?
Results using
figure 2.1 examples
Total months used: 51
Average LOS: 8.5
Total months used: 37
Average LOS: 6.2
Total months used: 89
Average LOS: 14.8
Note: examples in this table are for illustrative purposes only, and are not typical of shelter use patterns in the District.
As the reader can see from figure 2.1 and table 2.1, the FY06 Cohort approach tells us that on
average each person entering shelter in FY06 used 8.5 months of shelter during the first 12
months after entering. The Active in FY07 approach tells us that on average a person using
shelter at all in FY07 used 6.2 months of shelter in FY07. And the Regular TCP approach tells us
that the average length of stay for all people using shelter in FY07, counting from the day they
entered shelter, is 14.8 months. As should be clear by now, the Active in FY07 approach
underestimates average length of stay for all people using shelter, while the Regular TCP
approach overestimates average length of stay because at any given point in time, long-term
stayers will be overrepresented among those present. This is the usual situation for all systems in
which there is flow into and out of a population, including welfare rolls, child welfare rolls, and
shelter users. The long-term stayers are overrepresented at any given point in time, leading
people to think that very high proportions of system users stay for very long stretches—an
impression that the data presented below clearly show is wrong.
SINGLE ADULTS ACTIVE IN ES IN FY07
The most recent data on single adults in ES, calculated using the Active in FY07 approach and
including both the actual counts from the HMIS and the estimated total population, reported:
•
From October 1st, 2006 to September 30th, 2007 (FY07) 11, 562 individuals stayed in
District ES that report to HMIS,
•
An estimated 12,768 individuals stayed in both HMIS reporting and non-HMIS reporting
ES, and
•
On an average night in FY07, 1,796 single adults stayed in ES that report to HMIS and
1,988 were estimated to stay in ES as a whole.
•
Per-person shelter use in FY07 averaged about 73 nights, or 2.4 months, for males and 33
nights, or 1 month, for females. A single ES bed in the District is used by about 5 men or
12 women each year.
Gender, Race, and Age
Of single adults in ES, 83 percent (9,648 in HMIS data, and 10,654 estimated overall) were male
(almost the exact opposite of what we will see in the families’ data); 57 percent were “Black or
African American” and 5 percent were “White, Non-Hispanic/Non-Latino.” For 34 percent, the
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
record did not provide information concerning race or ethnicity. Most single adults in the
emergency shelter (40 percent) were aged 31–50, with the next highest percentage (18 percent)
being aged 51–60; 31 percent did not provide information on date of birth or age. TCP reports
that the average age of homeless single adults in ES is approximately 49 years old. In general,
data paint the picture of single adults using ES as mostly middle-aged, African-American men.
Veterans and Disabilities
A total of 873 single adults in HMIS data mentioned veteran status—around 8 percent. For the
whole ES system there were an estimated 964 veterans. Information about veteran status was
missing, however, for 52 percent of the records. Similarly, 11 percent were recorded as having a
disability, but 60 percent of the records lacked information concerning a disability.
Living Arrangement on the Night Before Entering Shelter
Information regarding residence on the night before coming into ES paints a picture of highly
unstable living situations preceding a shelter stay: 43 percent had been in their previous living
situation for one week or less; 69 percent had been in their pre-shelter living arrangement for less
than three months.
In terms of where homeless single adults are coming from, data from the Active in FY07
(AHAR) analyses suggest that the majority had lived within the District before entering the ES
system. Indeed, only 8 percent gave a zip code outside of the District’s jurisdiction, compared to
20 percent who gave a District zip code. Given available data, there does not appear to be much
movement into the District based on the attractiveness of the District’s homeless system, but
prior address zip codes are missing for 72 percent of single adults, so the conclusion may be
premature.
FAMILIES: AHAR AND VIRGINIA WILLIAMS FAMILY RESOURCE CENTER DATA
For families, HMIS data now include data from the Virginia Williams Family Resource Center
(VWFRC), the District’s central intake point for families entering the ES system, as well as data
on actual ES use by families. Combined, the Active in FY07 analyses and VWFRC data show
that during FY07:
•
529 families used ES that participate in the HMIS. These families included 1,661
homeless persons in families, of whom 1,008 were children.
•
The total number of families within the District’s ES, including HMIS-participating and
nonparticipating shelters, is estimated at 572, including 1,718 homeless persons of whom
1,043 were children.
•
On an average day in FY07, 697 homeless persons in families used ES that participated
in the HMIS, with an estimated 711 individuals for the entire Continuum of Care (CoC).
•
Per-family shelter use in FY07 averaged about 154 nights, or 5.2 months, for females
and 190 nights, or 6.1 months, for males. One ES family unit bed in the District is used
by about 2 people each year.
10
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
11
Age, Race, and Gender
The age and race of the head of household in homeless families active in FY07 in emergency
shelters were well documented. On average, homeless household heads active in FY07 were
around 34 years old—significantly younger than the 49 year average for homeless single adults.
Not surprisingly, given findings for homeless families nationally, homeless household heads in
the District are primarily women (89 percent), in sharp contrast to findings reported above on the
gender of homeless single adults (84 percent male), as shown in figure 2.2.
Figure 2.2: Gender Distribution of Single Adult and Family Emergency Shelter Users
Family Head of Household
in ES (FY07)
Singles in ES (FY07)
16%
11%
Male
Male
Female
Female
89%
84%
So urce: TCP analysis o f VWFRC data
So urce: TCP analysis o f VWFRC data
Disability
Figure 2.3 shows the types and numbers of disabilities for the household heads of the 529
families served in FY07 in the ES system, derived from the “Active in FY07” analyses.
Source: TCP’s Active in FY07 analysis for AHAR.
Type of Disability
Ab
us
e
Su
bs
ta
nc
e
M
ca
l
Ph
ys
i
ca
l
M
ed
ic
ob
il i t
y
al
Is
su
e
Ph
ys
i
en
ta
lH
ea
lth
Le
ar
ni
ng
M
DS
/A
I
HI
V
pm
De
ve
lo
ea
lt h
H
Ch
ro
ni
c
en
ta
l
100
90
80
70
60
50
40
30
20
10
0
Is
su
e
Number of Families
Figure 2.3: Disabilities Among Heads of Homeless
Family Households Active in FY07
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
Substance abuse is by far the most common type of disability recorded for heads of homeless
families in the District, followed by mental health issues. Unfortunately, the usefulness of these
data is weakened by high rates of missing data for questions on disabilities (we will improve on
this slightly in our cohort analysis below). The AHAR notes that information about the presence
or absence of a disability of any type was missing for 37 percent of family heads of households,
and far, far more lacked information about specific disability types. It is likely that disability
rates are much higher than the HMIS data can document. However, one can still glean something
about the distribution of disabilities from the information in figure 2.3.
The Application and Placement Process
The VWFRC is the starting place for families wanting to enter the District’s public shelter
system for families. It also serves as a focal point for families seeking a variety of assistance who
may be in precarious housing situations but are not literally homeless at the time they apply. The
VWFRC has the task of assessing each family’s situation and determining whether it is eligible
for shelter—that is, whether it is literally homeless and does not have any alternative housing
arrangement available to it. A good part of the VWFRC’s work involves helping not-quitehomeless families find alternative arrangements or the short-term assistance they may need to
remain in the housing they have.
Figure 2.4: Results of Applications for Family Emergency
Shelter
Entered emergency
shelter
10
Entered alternative
shelter
28
9
Disappeared without
completing application
Entered temporary
living situation
6
Withdrew application
3
Did not qualify for
emergency shelter
22
Missing information
Source: TCP analysis of VWFRC data
22
TCP’s analysis of information supplied by the 1,884 families applying for shelter through the
VWFRC in FY07, shown in figure 2.4, indicates that only 28 percent (529 families) were placed
in emergency shelter. Such placement could have occurred at the family’s first or subsequent
application. Another 28 percent simply withdrew their request or disappeared, finding housing or
shelter elsewhere or living in unsheltered homeless situations. the remainder did not qualify for
12
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
shelter (9 percent), entered into an alternative shelter (3 percent), or went into a temporary living
situation while awaiting shelter (22 percent).
The data in figure 2.4 show whether a family was accepted into shelter or had some other
outcome of its FY07 application(s). TCP was also able to ascertain from VWFRC data that 37
percent of FY07 applicants were requesting shelter for the first time. The remainder had made
earlier requests, including 13 percent that had made three or more requests.
Among families that enter shelter, many use the shelter system as it was intended to be used—as
a temporary place to stay while they get back on their feet. These are the families that apply
once, are considered homeless, enter shelter, and leave after a short stay. But a small proportion
of families (about 14 percent) cannot seem to get away from incipient or actual homelessness,
either applying for emergency shelter multiple times or having chronic or repeated shelter use,
For families that do actually enter shelter, our cohort analysis below will give a better sense of
just who these families are and how long they are truly staying.
Living Arrangement at the Time of Applying for Shelter
As with single adults, families applying for shelter had not spent much time in their most recent
living arrangements. Among the 51 percent of families with information on this issue, 37 percent
had been in their previous living situation less than one month. Many homeless families were
seeking shelter from temporary living situations that may have been the last of several they had
exhausted before applying for shelter.
As we saw with single adults, only 6 percent of homeless families Active in FY07 had a previous
zip code outside of the District (54 percent of family records did not contain this information).
The implication is that family shelters in the District do not appear to be acting as magnets for
homeless families in Virginia, Maryland, or other states.
Length of Stay (LOS) and the FY06 Cohort Analyses
As explained above, we believe that length of stay calculations are more appropriately done—
because they are more informative—using a cohort rather than an “active in the system”
approach. The FY06 Cohort approach used for all the analyses reported in this section gives the
most accurate possible picture of what happens to single adults and families once they enter a
District shelter and does not give undue weight to people who stay the longest.
SINGLE ADULTS—FY06 COHORT
During the initial interviews of this study, a number of people voiced fears that families and
single adults were entering the ES system and just staying put. Backed up by statistics estimating
the average length of stay at around 230 days for families and 80 days for single adults, people
were concluding that the District’s ES system was ineffective at moving people quickly back into
housing. Through our cohort analysis, however, we found a different story for both single adults
and families. In what follows, we present findings from our FY06 Cohort analysis, breaking
down LOS and illustrating the demographic makeup of those who stay for long as opposed to
short periods of time.
13
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
14
TOTAL LENGTH OF STAY FOR THE FY06 SINGLE ADULTS COHORT
Most single adults entering ES in the District do not stay very long at all. In fact,
•
•
Just under half (47 percent) of single adults that entered the emergency system in FY06
stayed one week or less;
86 percent stayed less than half of the 12 month follow-up period used in our FY06
Cohort analyses.
The average length of stay (ALOS) for the whole FY06 Cohort was 64 days out of the possible
365, with a median LOS of 11 days. This ALOS is 20 percent shorter than the 80 days reported
by TCP; the median LOS is dramatically shorter, strongly suggesting that the very short stayers
as a group should be treated very differently from the long-stayers.
Figure 2.5 graphically depicts the number of single adults within each length-of-stay group; table
2.2 gives the actual numbers. Of the 12,246 single adults in the FY06 Cohort, 5,733 (47 percent)
stayed fewer than 7 days and 10,585 (86 percent) stayed fewer than 6 months of the entire 365day follow-up period. Overall, only 4 percent of single adults spent all 365 days of the follow-up
period in the system, leading us to conclude that for the vast majority of single homeless
individuals in the District, ES is not a long-term choice. It should be obvious to policy makers
that the “very short stayers” and “very long-stayers” in the District’s ES system require very
different service interventions. For the 47 percent who stay fewer than 8 days, the ES system
might just be giving them what they need—a temporary respite. Policy attention would be better
directed at determining what to do for the remaining shelter users, leaving the very short stayers
to use shelter briefly and leave on their own.
47
50%
45%
40%
35%
30%
25%
20%
15%
10%
5%
0%
14
18
1
91
Total length of stay
or
ov
er
Da
ys
ay
s
D
36
5
th
ro
ug
h
18
0
36
4
D
ay
s
Da
ys
4
th
ro
ug
h
90
61
th
ro
ug
h
60
th
ro
ug
h
31
Source: Urban Institute analysis of HMIS data
(FY06 Cohort)
10
6
Da
ys
Da
ys
30
8
th
ro
ug
h
ew
er
or
f
7
10
9
da
ys
Percent
Figure 2.5: Total Length of Stay (FY06 Singles Cohort)
(n=12,246)
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
15
Table 2.2: Single Adults’ Length of Stay in Emergency Shelter (FY06 Cohort)
Number of days spent in ES (FY06
Cohort)
Total cohort
Percent of total FY06 population
(n=12,246)
100
Number of single adults within
LOS category
12,246
7 or fewer days
47
5,733
8 through 30 days
14
1,756
31 through 60 days
9
1,127
61 through 90 days
6
722
91 through 180 days
10
1,247
181 through 364 days
10
1,167
365 or more days
4
494
Source: Urban Institute analysis of HMIS data (FY06 Cohort)
Long-Stayers
While most homeless single adults are leaving shelter shortly after entry and not coming back,
the 14 percent of the FY06 Cohort that stays longer than six months (long-stayers) has a major
impact on shelter resources.
•
494 individuals (4 percent of the total single adults cohort) stayed 365 days of the 365
days we had data for; another 1,167 (10 percent) stayed 181–364 days.
On any given night for the FY06 Cohort, 21 percent of the 2,316 ES beds in the District CoC for
single adults (494 beds) were occupied by 4 percent of that population. In addition, those staying
for a total of 181–364 nights use on average 258 days, or 71 percent of the possible 365 nights in
our cohort analysis. Thus on an average night single adults in the LOS category of 181–364 days
occupy an additional 828 beds. In all, adding the bed usage of the two long-stayer groups
together shows us:
•
On any given night for the FY06 Cohort, 1,322 beds, or 57 percent of the District’s ES
beds for single adults, were occupied by long-stayers.
The policy implications of this shelter use pattern are enormous. Indeed, if the District were able
to move all 1,661 of those long-stayers (those who stayed over 180 days of the 365 days studied
here) into permanent supportive housing (PSH) per the Mayor’s plan, the District could reduce
the number of ES beds by 57 percent and still meet the needs for emergency shelter of short
stayers.
Reentry
In our initial interviews with District officials, some were concerned that homeless individuals
were continually entering and exiting ES, never breaking free of the cycle of homelessness. We
found this not to be the case for the majority of the population. We considered a person to have
left shelter if he or she went 30 days or more without spending a night in shelter. A reentry meant
that the person returned to shelter at any time after that 30-day period but still within the 365-day
follow-up period, even if only for one day. We found that:
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
16
•
4,949 single adults (40 percent of the entire FY06 cohort) left ES after an initial stay of
fewer than 7 days and did not reenter the system within 12 months of their initial entry
date. 5
•
7 percent of the FY06 Single Adults Cohort also stayed only briefly the first time they
used shelter but they did come back at least once within the 12 months following their
initial entry.
Overall, 34 percent of the entire FY06 Single Adults Cohort leave and reenter shelter
within 365 days of their initial entry.
•
Table 2.3 shows the number of single adults who leave and do not return within 365 days and
the number that leave and do return during that time period. Overall, the majority of single adults
(53 percent, or 6,506 individuals) will exit within two months and not return within 12 months of
their entry date.
Table 2.3: Single Adult Patterns of Reentry and Length of Stay (FY06 Cohort)
Number of days spent in ES
(FY06 Cohort)
Number that do not reenter
within 12 months of their
initial entry
8032 (66 percent)
4949
Number that do reenter within
12 months of their initial entry
8 through 30 days
1044
712
31 through 60 days
513
614
Total population (n = 12,246)
7 or fewer days
4214 (34 percent)
784
61 through 90 days
256
466
91 through 180 days
403
844
181 through 364 days
373
794
365 or more days
494
NA
Source: Urban Institute analysis of HMIS data (FY06 Cohort)
For 34 percent of our FY06 Single Adults Cohort, though, it is clear that they are having
significant trouble leaving the ES system and staying out. Of the 4,214 single adults who had
more than one shelter stay over the 12 month follow-up period, 28 percent reentered the system
three times, each after an absence of more than 30 days. It is likely that this group of homeless
single adults faces different problems than those who exit and do not return; it would probably be
useful to develop a specialized case management approach to help them retain housing.
5
In line with literature today, we use the 30 day exit criterion first established by Dennis Culhane. If a homeless
individual uses shelter for a few nights, then doesn’t use it, then comes back for a few more nights, he or she is not
considered to have left, and all nights of shelter use are treated as one contiguous stay. Only if the person does not
use shelter again for at least 30 days do we consider him or her to have exited the system. If he or she returns
thereafter but still within the 365-day follow-up period, we count it as a re-entry.
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
17
LOS Categories, Demographic Trends, and Missing Data
Policy makers would benefit greatly by knowing what differentiates people with longer periods
of stay and reentries from those who leave quickly and do not return. Here we look to answer
two questions:
•
Which single adults stay in ES longer than others?
•
Which single adults reenter ES after leaving?
Data in the HMIS to answer this question are extremely limited—we have only age,
race/ethnicity, gender, and disability status (yes/no) available to us. Our analysis of even these
demographic characteristics of each LOS category in our FY06 Single Adults Cohort is
hampered, however, by very large amounts of missing data for some variables. In the tables that
follow we provide results based on available data but also show the proportion in each category
for which the information is missing.
Age and Missing Values
The average age 6 of the entire FY06 Single Adults Cohort is 45 for both those who reenter and
for those who do not. People who reenter do not appear much different in age from those who
have only one stay. Longer stayers appear to have slightly higher average ages, but the level of
missing data precludes any firm conclusions (table 2.4).
Table 2.4: Age by Length of Stay for Single Adults (FY06 Cohort)
No Reentry
Number of days spent in
ES (FY06 Cohort)
Total population
Reentry
Average age
Percent missing
Average age
Percent missing
45
57
45
28
7 or fewer days
44
74
45
57
8 through 30 days
44
48
43
42
31 through 60 days
46
45
44
27
61 through 90 days
45
30
44
24
91 through 180 days
45
18
45
14
181 through 364 days
47
6
46
6
365 or more days
48
4
NA
NA
Source: Urban Institute analysis of HMIS data (FY06 Cohort)
Shorter lengths of stay are associated with high levels of missing data—74 percent missing for
those staying a week or less who do not reenter. It is extremely difficult to parse out what aspect
(missing data? age? other?) accounts for the changes by LOS category. Table 2.4 shows that as
LOS gets longer the amount of missing information shrinks and suggests that the average age
increases. The decrease in missing values associated with longer lengths of stay is almost surely
6
Age is age at entry, which is calculated by comparing date of birth to system entry date.
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
18
the result of the increased opportunity to get information from individuals entering shelter. But
we cannot tell if the increase in age is due to the relationship between age and LOS or if it is a
matter of more people giving date of birth.
It seems unacceptable that information is missing for so many people. The extent of missing data
on age is a problem for this analysis and also a problem for the District. Records for 45 percent
of those staying 31 to 60 days in shelter who do not reenter, for instance, do not contain
information on date of birth. One would think that by the time a person enters ES on 30 nights,
shelter operators would have ample opportunity to get this information. This problem cuts across
most descriptive variables in the HMIS. The result of these glaring gaps is, as noted, a serious
reduction in our ability to draw strong conclusions about the relationship between demographic
characteristics, LOS, and reentry in ES, and an inability for the District to use even this basic
information to help shape policy.
Race/Ethnicity
The demographic breakdown for the entire FY06 Single Adults Cohort was around 36 percent
African American, 4 percent Caucasian, and 60 percent missing for those who do not reenter. For
those who did reenter it was 62 percent African American, 5 percent Caucasian, and 26 missing. 7
As with age, we see in table 2.5 the same trend in the percent missing, with lower percentages of
unknowns coinciding with both longer LOS and reentry.
Table 2.5: Race/Ethnicity by Length of Stay for Single Adults (FY06 Cohort)
No Reentry
Number of days spent
in ES (FY06 Cohort)
Total population
Percent
African
American
36
Reentry
Percent
Caucasian
Percent
missing
4
60
Percent
African
American
62
Percent
Caucasian
Percent
missing
5
26
7 or fewer days
21
3
73
37
4
53
8 through 30 days
44
5
47
50
3
40
31 through 60 days
44
6
45
64
4
25
61 through 90 days
63
5
26
65
6
21
91 through 180 days
70
7
15
74
5
12
181 through 365 days
82
6
6
81
6
7
365 or more days
82
8
3
NA
NA
NA
Source: Urban Institute analysis of HMIS data (FY06 Cohort)
Given the level of missing data, only one observation can be made with reasonable certainty—
the racial/ethnic breakdown for long-stayers (staying more than 180 days in shelter) is at least 81
percent African American for both reentry and no reentry groups.
7
Race categories such as “other” made up for a small portion of the overall demographic breakdown, and are
therefore not mentioned here.
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
19
Gender
Because staff can usually determine a person’s gender by observation, this variable had a low
rate of missing data in the HMIS: 12 percent for those who do not reenter and only 3 percent for
those who do. As shown in Table 2.6a, around 73 percent of the FY06 Single Adults Cohort who
do not reenter were male; of those who do reenter, 84 percent were male.
We were able to specify a gender for all but 1.5 percent of single adults in the FY06 Cohort by
examining the shelters they stayed in. Since shelters are sex-specific—some serving only men
and others serving only women—we replaced missing values accordingly. Table 2.6b gives the
results. Comparing tables 2.6a and 2.6b, one can see that the percent male increases, and does so
more for the shorter length-of-stay categories, which originally had more missing data. The
proportion who are male increases even for people who reentered, fewer of whom had missing
data in the first place.
Table 2.6a: Gender by Length of Stay for Single Adults (FY06 Cohort)—
Original HMIS Data
No reentry
Number of days spent in ES
(FY06 Cohort)
Total population
Reentry
Percent male
Percent missing
Percent male
73
12
84
Percent
missing
3
7 or fewer days
72
15
84
6
8 through 30 days
68
10
80
5
31 through 60 days
72
7
84
2
61 through 90 days
79
2
82
2
91 through 180 days
79
3
86
1
181 through 364 days
75
2
87
<1
365 or more days
77
<1
NA
NA
Source: Urban Institute analysis of HMIS data (FY06 Cohort)
Table 2.6b: Gender by Length of Stay for Single Adults (FY06 Cohort)—After Using Shelter Location
to Determine Gender
No reentry
Number of days spent in ES
(FY06 Cohort)
Total population
Reentry
Percent male
Percent missing
Percent male
81
2
87
Percent
missing
<1
7 or fewer days
84
2.5
85
<1
8 through 30 days
74
3
85
<1
31 through 60 days
76
3
86
<1
61 through 90 days
81
<1
84
0
91 through 180 days
80
<1
87
<1
181 through 364 days
75
0
87
0
365 or more days
78
<1
NA
NA
Source: Urban Institute analysis of HMIS data (FY06 Cohort)
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
20
Because there are few cases of missing data, we are able to say with more confidence that for the
entire population at least 75 percent of single adults using the District’s emergency shelters are
male. It also seems that males are more likely to reenter the system. Percentage male is
consistently higher for those who do reenter than for those who do not. LOS and gender do not
appear to be related.
Disability
To get a general sense of the rate of disability in the FY06 cohort of single adults, we combined
data indicating the presence of mental illness, substance abuse, physical disabilities, and chronic
homelessness from whatever data fields such information was to be found. 8 We created one
variable that indicates whether or not an individual has any disabilities. Later we analyze each
disability separately, but first we illustrate, over LOS categories and reentry groups, the
percentage of single adults in the FY06 Cohort for which HMIS data indicate the presence of at
least one disabling condition, plus the level of missing data.
In the entire FY06 Single Adults Cohort, around 30 percent of those who do not reenter after
leaving the system had a disabling condition and 57 percent of those who do reenter were
identified as disabled (table 2.7). Conclusions on the broad population are again made difficult
by the amount of missing data—62 percent of values for those with one stay and 33 percent for
those with more than one stay were missing. This lack of data is particularly distressing
considering the “disability” variable is a composite of three other variables and only considered
lacking if all three are missing. The presence of disabilities is also one of the criteria for
establishing chronic homelessness, which is the focus of much policy interest.
Table 2.7: Disability and Length of Stay for Single Adults (FY06 Cohort)
No reentry
Number of days spent in
ES (FY06 Cohort)
Total population
Percent
yes
30
Percent
no
8
Reentry
Percent missing
62
Percent
yes
57
Percent
no
10
Percent
missing
33
7 or fewer days
14
6
80
30
7
63
8 through 30 days
35
10
55
41
10
49
31 through 60 days
37
13
50
56
12
32
61 through 90 days
57
13
30
62
10
28
91 through 180 days
68
15
17
70
13
17
181 through 364 days
79
15
6
83
8
9
365 or more days
83
14
3
NA
NA
NA
Source: Urban Institute analysis of HMIS data (FY06 Cohort)
8
Chronic homelessness was included in this variable creation because having a disability is one of the conditions
that HUD requires as part of its definition of chronic homelessness. So it seemed viable to assume that even if other
disability categories were not checked in a person’s record, if the record indicted “yes” on chronic homelessness, the
person had at least one disability.
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
21
For single adults who used shelter more than 180 days during the follow-up period, table 2.7
indicates data availability for at least 91 percent of records. At least 83 percent of the population
that stayed for all 365 days of the follow-up period had a disability, as did reentry people who
used 181–364 days of shelter within a year after first entering ES. We can thus firmly conclude
that the great majority of long-stayers were disabled in some way.
It also appears that disabilities are more prevalent in the reentry group. Consistently across all
LOS categories, the reentry group has a higher percentage of recorded disabilities and, often,
lower percentages of those recorded as not having a disability. Strong confidence in this
conclusion is impossible, however, given the amount of data missing.
Mental Illness
Instances of mental illness among single adults in the FY06 Cohort are given in table 2.8. For the
total population of single adults who exit without reentering the system within 12 months of their
entry date, 6 percent of records show a “yes” for mental illness, 24 percent show a “no,” and 70
percent are missing any information. Those with multiple stays are less likely to have missing
data (42 percent) and more likely to show a “yes” for mental illness (15 percent).
Again, as LOS increases, so does the percent of the population that can be identified as having a
mental illness; unfortunately, missing values once more make any conclusions on the
relationship between LOS and mental illness impossible. The percentage of single adults
recorded as having a mental illness goes up as LOS increases, but so does the percentage
recorded as not having one. We therefore cannot draw any conclusions about the relationship
between LOS and having a mental illness.
One might conclude that those who reenter ES during the follow-up period have slightly higher
rates of mental illness across LOS categories. For those who stay for a total of three to six
months, for example, 20 percent of those who reenter are considered mentally ill while 53
percent are not and 27 percent have missing data. For those with only one continuous stay, only
15 percent are identified as having a mental illness, 53 percent are identified as not having a
mental illness, and 31 percent are missing the relevant data. Without the missing data, this
conclusion must remain weak.
Table 2.8: Mental Illness and Length of Stay for Single Adults (FY06 Cohort)
No reentry
Number of days spent in
ES (FY06 Cohort)
Total Population
Percent
yes
6
Percent
no
24
Reentry
Percent missing
70
Percent
yes
15
Percent
no
43
Percent
missing
42
7 or fewer days
3
13
84
7
23
70
8 through 30 days
8
26
66
12
33
55
31 through 60 days
8
29
63
15
43
42
61 through 90 days
13
44
43
20
43
37
91 through 180 days
15
53
31
20
53
27
181 through 364 days
18
57
25
21
58
21
365 or more days
17
66
17
NA
NA
NA
Source: Urban Institute analysis of HMIS data (FY06 Cohort)
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
22
For LOS categories of more than 180 days, 75 percent or more of records contain the relevant
information. Between one-fourth and one-third of those staying for longer than 180 days for
which information existed had a mental illness.
Substance Abuse
Overall, for those who do not reenter, around 18 percent of single adults were noted as having a
substance abuse problem, 15 percent were recorded as not having a problem, and data were
missing for 67 percent. Those who did reenter had an overall distribution of 40 percent identified
as having a substance abuse problem, 20 percent as not, and 40 percent with no data.
As shown in table 2.9, substance abuse is the only disability for which the percentages of those
recorded as having a problem were higher than those recorded as not having a problem, across
all but one LOS category. It also seems that the recorded levels of substance abuse were higher
among those who reentered than among those who did not. Those who reenter had twice the rate
of identified substance abuse (40 percent to 20 percent) as to those identified as not having a
problem. For those who do not reenter, that ratio was close to even.
As a further illustration, look at the LOS category of 181–364 days. For both entry groups the
percentage of missing data is the same (making the comparison somewhat cleaner). In this LOS
category, 10 percent more single adults who reentered had a substance abuse problem (57
percent) than was true for those who did not reenter (47 percent).
Table 2.9: Substance Abuse and Length of Stay for Single Adults (FY06 Cohort)
No reentry
Number of days spent
in ES (FY06 Cohort)
Total population
Percent
yes
18
Reentry
Percent
no
15
Percent
missing
67
Percent
yes
40
Percent
no
20
Percent
missing
40
7 or fewer days
9
8
83
18
13
69
8 through 30 days
17
20
63
29
17
54
31 through 60 days
21
17
61
37
21
41
61 through 90 days
38
22
40
45
19
36
91 through 180 days
44
28
28
48
27
25
181 through 364 days
47
35
18
57
25
18
365 or more days
54
34
11
NA
NA
NA
Source: Urban Institute analysis of HMIS data (FY06 Cohort)
Just as the percentage with a confirmed substance abuse problem rises with LOS, so too does the
proportion on whom this information is available. Therefore, we cannot draw any firm
conclusions about the association between substance abuse and LOS. We can, however, say
something about substance abuse among long-stayers. With only 11 percent of the data missing
for those staying a year or over, we know that at least 54 percent have a substance abuse
problem. Confirmed substance abuse reached as high as 57 percent for those who reenter and
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
23
stay a cumulative 181–364 days, even with 18 percent of the data missing. It is clear that
substance abuse in the long-stayer population is one of the biggest problems among users of ES.
Physical Disabilities
Table 2.10 shows the percentage of single adults in the FY06 Cohort reported as having a
physical disability. Records for about 7 percent of those who did not reenter and 16 percent of
those who did reenter indicate the presence of a physical disability. As has been true through
these analyses, however, missing values for physical disabilities muddy any conclusions we
might draw from the data. Around 66 percent of the values for those who do not reenter and 37
percent of those who do reenter were missing.
With respect to the effect of a disability on reentry, the differences between those who reenter
and those who do not were even less than for other disability types. While those who reenter did
tend to have higher rates of physical disability across LOS categories, this was not always the
case: for those who stay 61–90 days, the rate of disability was three percentage points higher for
those who do not reenter, even with comparable percentages of missing values. As with mental
illnesses and substance abuse, the amount of missing data for physical disabilities makes it
impossible for us to tell the effects of having a physical disability on LOS.
Table 2.10: Physical Disability and Length of Stay for Single Adults (FY06 Cohort)
No reentry
Number of Days Spent
in ES (FY06 Cohort)
Total population
Percent
yes
7
Percent
no
27
Reentry
Percent
missing
66
Percent
yes
16
Percent
no
47
Percent
missing
37
7 or fewer days
3
14
83
7
25
68
8 through 30 days
7
31
62
10
37
53
31 through 60 days
8
36
56
15
45
40
61 through 90 days
17
47
35
14
54
32
91 through 180 days
18
60
21
20
55
25
181 through 364 days
23
61
16
25
62
13
365 or more days
23
69
8
NA
NA
NA
Source: Urban Institute analysis of HMIS data (FY06 Cohort)
FAMILIES (FY06 COHORT)
Policy makers we talked to before we designed this study were concerned that families as well as
single adults were entering ES and not leaving. While families do tend to stay longer than single
adults, it is clear that the majority are not staying “forever.” We first discuss the LOS for
families entering ES in FY06 (FY06 Families Cohort) and then look at the differences among the
families within each LOS category on various demographic characteristics. Because all families
placed in shelter receive an intake assessment through VWFRC, the family records contain fewer
missing values than was true for single adults.
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
24
LENGTH OF STAY (LOS) AND FAMILIES (FY06 COHORT)
LOS for families followed trends similar to those for single adults, with most families entering
the system leaving in less than six months. Our FY06 family cohort had an average LOS of 153
days, with a median of 107 days. As shown in figure 2.6, 21 percent of families entering ES in
FY06 left within one month, another 22 percent left in two to three months, and an additional 22
percent left in three to six months; none of these reentered the system within 12 months of their
initial entry. Thus, 65 percent of those entering the District’s ES system for families leave in
fewer than 180 days and do not reenter.
Figure 2.6: How Long Do Families Stay in ES? (FY06 Families Cohort)
25%
22
21
19
Percent
20%
16
15%
12
10
10%
5%
D
ay
s
ay
s
or
36
5
18
136
4
Length of stay
ov
er
D
ay
s
D
91
-1
80
D
ay
s
61
-9
0
D
ay
s
31
-6
0
<3
1
D
ay
s
0%
Source: Urban Institute Analysis of HMIS data (FY06)
C h t)
Patterns of family stays in emergency shelter are considerably different than patterns for single
adults; nevertheless, they still show that most families do not stay “forever.” TCP publishes an
average length of stay for families based on its approach to counting time in shelter (see figure
2.1 and accompanying discussion). But our Cohort approach, which as we explained earlier is
the right way to assess length of stay patterns, reveals that the true ALOS for families is 153
days—one-third shorter than the calculation done using the TCP approach. The median LOS is
even shorter—107 days.
Long-Stayers
The number of families staying in ES for a long period of time is not insignificant, however. As
shown in figure 2.6, around 19 percent of the FY06 Families Cohort (70 families) stayed at least
one year. Thus on any given night during the covered time period, 70 of the 261 family ES units
in the CoC—or, 27 percent—are occupied by people who entered shelter in FY06 and stayed for
(at least) 12 months. Another 16 percent of families who entered shelter in FY06 stayed between
180–364 days, with an average LOS of 260 nights—71 percent of the 365 days studied in our
cohort. In all, on any given night, 54 percent of units are occupied by families who stay 181 days
or more. 9 A targeted policy for moving these families into PSH would mean a minimum
9
This calculation uses family emergency shelter units accounted for in the 2007 Housing Inventory Chart, before
DC Village was closed.
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
reduction of 112 ES units. This transition has already started with the closing of D.C. Village.
Those Who Exit and Return in Under 12 Months
Unlike single adults, among whom 34 percent leave and return within the 12-month follow-up
period used for our FY06 Cohort analysis, fewer than 3 percent of families exit shelter for at
least 30 days and then return within 12 months after first entry. Table 2.11 shows that only 8 of
the 380 families in our FY06 Families Cohort left and returned to the emergency shelter system
within 12 months of their initial entry. These families represent very small percentages of each
LOS category, so analytic results barely change if they are left out. 10 Therefore all analyses of
family LOS are done and presented only for first entries in FY06.
At any rate, it seems evident that families in the ES system either leave or stay—rarely do they
reenter within the year. If they do exit and return, it is likely that their leave is for a short period
of time, under 30 days, in which they either move to a different shelter or enter into an extremely
unstable living situation that lasts for fewer than 4 weeks. 11
Table 2.11: Length of Stay and Reentries (FY06 Family Cohort)
Number of days spent in ES
(FY06 Cohort)
Fewer than 30 days
31 through 60 days
61 through 90 days
91 through 180 days
181 through 364 days
Percent of those who leave that re-enter the system
within 12 months
4 (3 families)
0 (0 families)
5 (2 families)
2 (2 families)
1 (1 family)
Source: Urban Institute analysis of HMIS data (FY06 Cohort)
DEMOGRAPHIC TRENDS FOR FAMILIES IN DIFFERENT LOS CATEGORIES
Knowing what the population of homeless families looks like, particularly the differences
between LOS groupings, is important for creating targeted policy and effective case management
in the District. This section addresses two basic questions:
10
Families that return are not significantly different from families who leave and remain out of shelter on gender
and race/ethnicity. The average age of returners was four years younger than that of the rest of the FY06 cohort (29
vs. 33), but as age is missing for about 1 in 4 families, particularly in the shorter LOS categories, this result should
be interpreted with caution. The 8 returning families stayed initially for an average of 33 days and left for an average
of 118 days before returning. In all, because of the small number of returning families, we could not draw any firm
conclusions about factors affecting shelter return.
11
It must be noted here that due to the manual nature of the concatenation process, there is ample room for error on
this point. In other words, it is quite possible that families leaving for more than 30 days and returning to the system
were on occasion given their initial entry date by those concatenating the data. There is no way to tell if this has
happened, but we believe our conclusion still stands given that the number of mistakes made would have to be
significantly large in order for our interpretation to change.
25
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
•
What do homeless families entering in FY06 look like?
•
Why do some families stay longer than others in ES?
26
As with the single adults data, missing information made it difficult to draw firm conclusions
about factors influencing length of stay. First we note the data problems and then look at basic
demographic characteristics of age, race, and gender. We also look at issues of disability,
domestic violence, social supports, and income, drawing what conclusions we may.
Race, Gender, Age, Family Size, and Missing Data
In general, the race, gender, age, and family size of the heads of household (HoH) showed little
relationship to LOS. As Table 2.12 shows, the entire FY06 Cohort of family heads of household
was on average 33 years old (significantly lower than the 45 year average age for single adults).
Around 93 percent were African American and 92 percent were female. Average family size was
three people, with an average of two children. Consistently across all LOS categories, the
average family in the FY06 Families Cohort was a single African American mother, aged 33,
with 2 children.
Table 2.12: Demographic Characteristics and Length of Stay for Families (FY06 Cohort)
Race
Gender
Age
Percent
African
American
93
Percent
other
Percent
missing
Percent
female
Percent
missing
Average
age
Percent
missing
4
3
92
1
33
23
Fewer than 30 days
90
6
4
96
1
33
56
31 through 60 days
100
0
0
98
0
31
21
Number of days spent in
ES (FY06 Cohort)
Total Population (n=372)
61 through 90 days
97
3
0
85
0
31
30
91 through 180 days
92
<1
7
92
3
33
22
181 through 364 days
95
3
2
83
2
33
4
365 or more days
97
3
0
96
0
35
0
Source: Urban Institute analysis of HMIS data (FY06 Cohort)
There were no obvious trends in any of these basic demographic characteristics as LOS
increased, nor did an obvious link exist between LOS and missing values. In general, race and
gender data were almost totally complete, while age availability varied widely across LOS
categories. There is really no excuse for the level of data missing from these family records. The
intake assessment that each family receives at the VWFRC should get all this information. If it
does not, case workers in family shelters should be updating information about families as they
get to know them. Very few families stay such a short period of time in shelter that it would be
impossible for case workers to learn more about them.
Disabilities of Heads of Family Households, by Length of Stay
To determine disability status we used a number of different questions from intake forms. While
individual disability categories had extremely high rates of missing data (80 percent for a
question regarding “Type of Disability,” for example) we were able to get at a more general
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
27
indicator of disabilities by combining questions on substance abuse, disability type, and length of
disability, to get a response rate of 48 percent overall indicating whether or not a HoH was
disabled in any way—be it a physical disability, a mental disability, or substance abuse. Again,
this level of missing data about something as important as disabilities poses a real problem for
planning how to help families leave shelter and be able to sustain housing.
Table 2.13: Disability and Length of Stay for Families (FY06 Cohort)
Number of days spent in
ES (FY06 Cohort)
Total Population (n=372)
Percent yes
Percent no
Percent missing
29
19
52
30 or Fewer days
24
34
43
31 through 60 days
21
14
64
61 through 90 days
25
18
57
91 through 180 days
34
15
51
181 through 364 days
42
14
44
365 or more days
26
14
60
Source: Urban Institute analysis of HMIS data (FY06 Cohort)
The level of missing values showed no relationship to LOS (table 2.13). Because of the large
amount of missing data, it is difficult for us to draw any conclusions about the relationship
between LOS and disabilities. Of the entire FY06 Families Cohort, 29 percent of HoHs were
recorded as disabled, and there appears to be a positive relationship between LOS and disability,
with a peak at just under a year. It is likely that the proportion of HoHs with a disability would
be higher for those staying a year or more, but because we only have information on 40 percent
of the population, we cannot say this with any certainty.
Domestic Violence
Domestic violence as a factor in homelessness is highest for those staying a year or longer, at 19
percent (table 2.14). It seems clear that for many families staying in shelters for 365 or more
days, violence has at least something to do with the length of their stay.
Table 2.14: Victims of Domestic Violence and Length of Stay for Families (FY06 Cohort)
Number of days spent in ES
(FY06 Cohort)
30 or Fewer days
31-60 days
60-90 days
90-180 days
180-365 days
365 or more days
Percent Yes
Percent Missing
6.5
2.5
11.0
2.5
5.0
19.0
43
2
24
17
5
0
Source: Urban Institute analysis of HMIS data (FY06 Cohort)
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
28
Social Supports
It appears that longer periods of stay were associated with a reduced network of family and
friends who could offer social support (table 2.15), but we are again blocked from drawing any
strong conclusions by the level of missing information. With that said, the fact that 66 percent of
those staying less than a month had been staying with family or friends prior to coming to shelter
is not insignificant, nor is the fact that this proportion was only 10 percent for those staying a
year or longer. If all 24 percent of those with missing data who stayed a year or longer were to
report staying with family or friends, that would still only bring their percentage to 34 percent—
well short of the situation for families staying the shortest time in shelter. It appears that shorter
stays are associated with the availability of a social support network while longer stays are
associated with their absence.
Table 2.15: Social Supports and Length of Stay for Families (FY06 Cohort)
Number of days spent in ES
(FY06 Cohort)
Percent living with family or friends
before coming to ES
Percent missing
66
34
51
33
14
10
9
31
27
41
59
24
30 or Fewer Days
31-60 Days
60-90 Days
90-180 Days
180-365 Days
365 or more days
Source: Urban Institute analysis of HMIS data (FY06 Cohort)
Income and Benefits
Families’ receipt of income and benefits was also telling. Records for those who stayed longer
indicated higher rates of both earned income and TANF benefits. Table 2.16 shows this
relationship, with 36 percent of those leaving in less than 30 days having TANF benefits as
opposed to 71 percent of those staying for more than a year. Those whose records show any
earned income went from 12.5 percent for those staying under a month, to 27.5 percent for those
staying over a year.
Table 2.16: Earned Income and TANF (FY06 Cohort)
Number of days spent
in ES (FY06 Cohort)
30 or Fewer Days
31-60 Days
60-90 Days
90-180 Days
180-365 Days
365 or more days
Percent claiming TANF
benefits
Percent
missing
Percent claiming
earned income
Percent
missing
36.0
41.0
56.0
57.0
59.0
71.0
20
38
28
28
22
14
12.5
15.0
16.0
15.0
23.0
27.5
20
38
28
28
22
14
Source: Urban Institute analysis of HMIS data (FY06 Cohort)
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
The information in table 2.16 raises the question of whether case workers in family shelters are
helping the longer stayers get benefits or connect to work, or whether families arrived in shelter
with these resources. If the former, the level of contact with families that would facilitate these
connections again raises the question of why the shelter records for these families contain so
much missing data. For both earned income and TANF, missing values still present an obstacle
to drawing conclusions in all except the longest LOS categories.
Looking at the lengthiest LOS categories, the high rate of TANF receipt might suggest a link
between longer periods of stay and increased willingness to use public funds. It also suggests that
those with longer periods in ES who claim an earned income do not seem to be able to afford
housing even when they are working—thus implying additional problems, such as a disability.
CONCLUSIONS
While missing values make conclusions on the relationship between LOS, reentry, and
demographic characteristics extremely difficult, we are able to get a sense of the makeup of
long-stayers for both single adults and families. In general, the long-stayers among single adults
have high rates of disability with serious levels of addiction, are predominantly male, mostly
African American, and middle-aged. Families staying for long periods of time are mostly headed
by single, African American females, on average about 33 years old, with two children
accompanying them in shelter. These long-staying families have high rates of TANF receipt,
high rates of domestic violence, seemingly less ability to rely on family and friends, and some
disabilities (although not as high as for single adults). Experience suggests that these
characteristics are particular to long-stayers and are probably less severe for people in shorter
LOS categories, but we are not able to support or refute that notion with the available data.
In the future, we encourage shelter staff to make a greater effort to fill out client information, and
shelter managers and supervisors to make sure they do. While “low-barrier” shelters (which
make up a large portion of single adults ES) are not required to gather extensive information
from clients, it is clear that practices should be revised to assure better information, and that at
the very least intake staff should be more effective in getting and recording information for
clients who stay longer periods of time. It is unacceptable that the percent missing for the entire
cohort of single adults reached as high as 70 percent for some very basic types of information.
Family shelters are never considered low-barrier and therefore the amount of missing data for
head of households is even less acceptable. Intake procedures and subsequent data gathering and
data entry need to be enhanced for both groups to ensure that the District has the information it
needs to manage the system to the maximum advantage. Otherwise, the HMIS will remain far
below its true potential to inform effective and thoughtful policy.
BENEFITS OF LOS ANALYSIS USING A COHORT APPROACH
Results of a Cohort approach may be used to triage people using emergency shelter and structure
the services offered to suit different circumstances. For example,
•
The characteristics of people who leave quickly (say, within one month) after entering an
emergency shelter and do not come back are very different from those who live in
emergency shelters for years. Knowing what characterizes the former group and how it
differs from the latter group would help system managers to construct an intake
29
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
instrument that helps triage clients into those who are likely to become long stayers and
those who are likely to leave on their own. For instance:
o People staying fewer than seven days probably do not need much more than
they are currently getting in shelter—a bed for a few nights.
o People in the middle categories of length of stay would benefit from case
work intervention early on to help them connect to the things they need and
leave shelter faster.
o People in the probable-long-stayer group are candidates for permanent
supportive housing offered quickly through a Housing First model.
Alternatively, some people in this last group may have few disabilities and
may work, but do not earn much, so they choose to stay in shelter rather than
try to get a place of their own. A housing subsidy—either shallow or deep—
may be all they need to leave shelter for good.
•
The characteristics of people in different LOS groups would also indicate which public
agencies would need to offer services to address identified problems, either before people
leave shelter or after, to prevent them returning to homelessness.
•
Knowing the characteristics of people in different groups would also help in decisions
about what to require or expect of shelter users by way of exerting themselves to work
with case managers and access other resources (e.g., job training or placement) to help
them leave homelessness.
DATA LIMITATIONS AND DATA MANAGEMENT
As noted many times in the above presentation of results, we encountered several problems with
the information on emergency shelter users involved in the District’s HMIS. Many of the
problems involved the inflexibility of the software being used and the resultant inability of TCP
or analysts such as ourselves to answer a wide variety of questions in a clear, correct, and timely
fashion. Other problems had to do with data quality and the completeness of basic data fields that
are expected to be filled in during intake or shortly thereafter. Before we address these concerns,
we want to note that some of the delays we encountered in getting data were compounded by the
timing of our study—the holiday season and the beginning of hypothermia season generally
increase the demands on the HMIS for reporting. On top of this, TCP had to conduct the Annual
Point in Time Survey and analyze and report its “Active in FY07” data for the Annual Homeless
Assessment Report that HUD must send to Congress every year. The data analysis for AHAR is
time and staff intensive and the software supplied for doing it turned out to have its own
problems – problems that surfaced as TCP staff worked to develop an adequate interface with the
District’s HMIS. In addition, the data sets requested by the Urban Institute were often extremely
large and complex. All in all, it was a taxing time for both the HMIS system and its managers
and consequently for Urban Institute researchers.
Beyond these circumstantial factors, however, the system has problems that need to be addressed
if HMIS is to be an effective policy resource. The District’s HMIS is bedeviled by both
technological and data collection problems. Bowman Internet Systems—the software and related
services vendor for the District’s HMIS and for about 80 percent of all Continuums of Care in
30
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
the country—experienced technical difficulties on multiple occasions during our assessment,
halting TCP’s ability to pull data and run queries through the new Advanced Reporting Tool
(ART). 12 Bowman was going through a system upgrade that exacerbated the problems and
slowed the AHAR process and consequently our access to data; we therefore expect these
problems to lessen in the near future, but not to disappear entirely.
Although it was specifically designed to add flexibility, speed, and ease of use to the querymaking process, the ART system also had separate technical difficulties and did not prove to be
as flexible or useful as TCP (and we) had originally hoped. ART could not readily produce the
answers to our specific questions or produce data sets that were readily susceptible to analysis.
We had to do significant amounts of data cleaning before we could even begin to approach the
research questions this study is designed to answer. Maybe some time in the future, ART will be
able to produce the answers to needed policy questions in a timely enough manner, but as of now
it cannot do so.
Technical problems with ServicePoint also slowed the process. Whenever TCP tried pulling
large data sets from the HMIS, the data transferred so slowly that the ServicePoint session timed
out (after only 20 minutes). Data queries then had to be rebuilt, split into multiple sections, and
then merged after they were dumped into Excel–yet another time consuming process. Further,
ServicePoint only extracts data into MS Excel. The data sets requested by the Urban Institute
often produced more rows of data than are available in an Excel Spreadsheet and thus special
queries had to be created and mapped out. TCP has made a request to Bowman to add the option
of pulling data into other software, but currently Bowman does not have this capability. TCP has
brought these issues to Bowman’s attention, but Bowman remains unresponsive.
What Bowman should have said is that it can write TCP a program (in XML) that would allow
TCP to pull data many times the size of its current files and convert it into any format desired,
whether Excel, Access, STATA, SAS, SPSS, or some other program. Bowman has done this for
Michigan’s statewide system, and it can do it for the District. Obtaining such an XML program
and learning how to use it should be a top priority for TCP.
THE CLOSED NATURE OF THE DISTRICT’S HMIS GREATLY HAMPERS ITS ANALYTIC
CAPABILITIES
The District’s HMIS is completely “closed,” meaning that staff at Program A cannot see any
information entered by staff at Program B—not even when Agency Q runs both Program A and
Program B. Each program assigns each person served a “unique ID,” but that ID is only
“unique” within the program. The same person may have several “unique IDs” if he or she uses
several programs, rendering the whole concept ludicrous as practiced. Before it can conduct any
analysis that wants to know about people rather than bed nights, TCP, the HMIS administrator,
has to jump through many programming hoops to concatenate all the information about one
person into one record.
12
TCP mentioned that these technical failures happen multiple times a month.
31
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
TCP does not even have its system set up to scan incoming data from program intakes to see if
characteristics match someone already in the system. HMIS servers in many other communities,
even if the systems are “closed,” are set up to do this. If they find that the person is truly new to
the system, these HMIS assign a truly unique ID number and give it back to the entering program
to use thereafter for that person. If another program enters data for the same person, the system
recognizes that the person is already in the system and tells the second program what to use as
the unique ID. Having a truly unique ID for each person in the system greatly simplifies the task
of describing system users for policy makers.
Because the District’s HMIS is closed, homeless data can be extremely unwieldy to use.
Individual providers at emergency shelters currently cannot access the database to add to a
client’s intake information or to see if the client stayed in another emergency shelter the previous
night. Every time a homeless individual checks in—which is each night for single adults—new
“intake” data must be recorded. As a result, TCP receives as many as 365 data points for a single
homeless adult in the course of a year (the family system works differently). These data must in
turn be concatenated into one continuous record by TCP, using first and last names, birthdays,
and other demographic information to match records.
Currently, TCP concatenates data sets manually, sifting through almost 200,000 observations,
merging any data entries together if they represent contiguous dates for the same person, and
creating a single complete record for each family or single adult. This year, for everyone active
in the system in FY07, this process took several TCP staff several weeks! Further, as soon as
new data come in, which is every day, the matching process would have to be done again to
incorporate the new data before any analyses could be run.
The closed nature of the District’s HMIS and its lack of an automated procedure to assign a
systemwide unique ID makes this merging process necessary, but there is no reason why it
should have to be done manually. A system-derived unique ID would help tremendously, even in
a closed system; so would using a statistical analysis package such as STATA or SPSS to
concatenate and analyze the data. These packages are not very expensive and quite easy to use. 13
The cost of these programs would easily be offset by savings in labor hours and more accurate
and in-depth data analysis. For example, Urban Institute researchers were able to write the
programming to concatenate about 192,000 intake records for single adults in about nine hours.
Once written, STATA took a little more than an hour to tell us that these records pertained to
about 12,700 separate individuals and to assemble the 192,000 intake records into 12,700 new
person-based records. 14 Not only does this beat spending weeks doing a manual concatenation,
but the program can be re-run at any time to answer questions about what happened to people
using shelters as of yesterday, rather than being done so painfully once a year. Changing from a
completely closed system to some greater level of openness would make the process even easier.
13
Both programs can be purchased online. STATA can cost as little as $1,400 a year for a single user or it can be
leased. Both programs offer discounts to nonprofit organizations.
14
TCP would only have to create the programming for this analysis once, after which it could be used for any
number of future concatenations. If TCP were using STATA, it could even take the code we have already written
and adapt it for its own use.
32
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
These three changes—system-derived unique ID, statistical software package, and greater system
openness—would almost immediately fix many problems that prevent TCP from being able to
respond to policy makers quickly and accurately with answers that could help shape policy.
OTHER PROBLEMS
Even for questions pertaining to shelter use that require only intake records, not person-based
records, to analyze even seemingly simple queries takes TCP inordinate amounts of time and
effort. We believe this is again due mainly to inflexible and inadequate technology. For instance,
when asked for “the average daily number of families and single adults using emergency shelters
in 2007,” TCP responded that answering such a question required running 730 separate
Dashboard reports (365 for single adults and 365 for families) and then taking their average—
literally a day’s worth of work. The answer to such a question should be readily available to
policy makers and advocates virtually at the push of a button.
To reiterate, the HMIS as a whole struggled to provide the data requested for this assessment. In
the end, the data needed was delivered, but the process took longer than acceptable. Again, this is
in some part due to the timing of our assessment and the overflow of requests coming into TCP
and to the inflexible software TCP must use, but it also stems from a need for more training on
dealing with data requests and data in general. (This problem is of course prevalent in
organizations throughout the country.) Most requesters will not have the time or the expertise to
sift through spreadsheets as we did. Ideally, the HMIS management should be able to look at
research questions and provide in a timely manner either the answer as a result of internal
analysis or the complete relevant data set for analysis by researchers. Over the course of our
assessment, a significant amount of time was spent sifting through spreadsheets and making
additional requests, corrections, and clarifications. This does not have to be the case.
Once received, the data on families had problems with completeness and overall quality. For the
most part, single adults data were clean and more readily available, although less informative
because fewer data fields are required. However, they were by no means complete, as we have
seen. The real issues occurred with families, where there is supposed to be a thorough intake
assessment that should result in relatively few missing values. Services data in general are
known for incomplete entries and incorrect data points, but a few problems in the data sets we
received were particularly egregious. Within the FY06 families cohort, for certain sections—
education level, for instance—over 85 percent of the records did not contain the relevant data. To
make matters worse, there was serious ambiguity as to what the missing data panels meant, as
some at TCP interpreted them as a “no” and others as “missing”— two very different things with
different policy implications. What is more, we could not use some variables in the data set we
received for certain analyses because TCP had already merged some data before sending it to the
Urban Institute. 15 Most of these ambiguities were cleared up, after significant analysis on the
part of UI and TCP, but some remain and are noted in previous sections.
15
As an example, the concatenation process that TCP undertook for family data merged different shelter stays into
one stay if a family switched shelters and thus the different shelter names would not show up in the data we
received. Rather, the variable describing where the family was sheltered would show only one location rather than
the two they truly stayed in. This limited our ability to say anything about the differences in stays for certain
shelters.
33
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
Urban Institute researchers are relatively sophisticated about data. We know what we are looking
for, we know when we are not getting it, and we keep asking until we get the data we need to
answer the questions we want answers to—all assuming that the data exist. Given all that, it took
nine requests and transfers of nine data sets with different characteristics before we received the
data set that we needed and that would yield the answers we sought.
Our experience raises the questions: “What do people who are not knowledgeable about data
analysis do?” and “How can policy makers get the information they need to make decisions?”
Part of the job of managing the HMIS is disseminating information to people who need answers
but are not in the position to do analysis themselves, and who also may be phrasing questions in
a way that does not adequately convey the information needed to an HMIS analyst. This is not to
say TCP staff managing the HMIS are doing a bad job—they certainly are not. Throughout this
assessment, TCP staff were responsive, helpful, and knowledgeable about the ins and outs of the
data system. We were continually impressed by their energy and willingness to put in long hours
to help our analysis. Here we are saying two things:
•
The current technology impedes TCP’s ability to give attention to important areas
because they are preoccupied with technical problems and data management issues that
can and should be dealt with by computer programs.
•
Clear communication is essential if TCP staff is going to be able to give policy makers
the answers they need. Policy makers need to be very careful in phrasing their requests,
and HMIS analysts need to discuss with policy makers the various ways they can
structure their data system queries, some of which will get closer than others to the
answers that policy makers need. It is very easy for everyone to end up frustrated and
disappointed if a clear process is not followed.
More training on all sides, including the people requesting information as well as those
attempting to supply it, will be needed to insure that data requests are responded to correctly and
that policy makers are getting answers to the questions they are asking. In addition, continuing
interaction among policy people and data people is vital to assure that the technical computer
language of a data request will actually extract the data that the policy people intended. Our
FY06 Cohort analysis of lengths of stay presented earlier in this chapter is a perfect example of a
seemingly simple policy question that can be analyzed in three very different ways to produce
three different answers. People must be clear what they are asking for, and analysts must be able
to translate that clarity into a data query. 16
16
An example of how this process can go awry even for very sophisticated data analysts and researchers happened
recently when Urban Institute researchers were looking at how long people with serious mental illness placed in
permanent supportive housing remained in housing, whether in the original location or elsewhere. Initial results had
us “losing” half the people—something the mental health agency that commissioned the study was sure had not
happened. As it turned out, in writing the code to extract the data for analysis, the programmer had limited the
search to housing programs run by the same agencies in which clients were first placed—not because we asked for
that, but because we had not been absolutely clear about what we wanted and it did not occur to us that anyone
would limit the search in that way. As these programs represented only half the units available in the city, it is not
surprising that we lost a lot of people. Luckily we were working closely with mental health agency staff who did not
believe the results and were able to find the “lost” people and determine their housing status. This example
34
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
IMPLICATIONS
1. Open the HMIS, at least enough to make it possible to give each shelter user a truly
unique ID. Even better would be a system in which shelter staff were really trying to use
the data in the system. They would therefore see what a problem it was to have so much
missing data, and they could do something about it.
2. Get TCP a statistical analysis package such as STATA or SPSS.
3. Revisit data collection requirements at intake for every type of shelter in the District.
Think of what needs to be known about people for the purpose of helping them leave
shelter. That should be the goal and it is important to ask why it does not appear to be the
goal. There is no reason why the HMIS should have so much missing data on families, or
why we so little information is forthcoming about single adults in shelters.
4. Part of the problem of missing data for single adults using emergency shelter is the
practice of treating each night of shelter use as a new entry, combined with the closed
nature of the HMIS system. To address this problem, the District might consider moving
to a “bed management” approach that does one complete intake the first time a person
appears and thereafter just checks off every time the person uses the shelter.
5. Develop a clear communications structure whereby policy makers can request and obtain
needed information for their decisions.
6. Train everyone, not just on procedures but on the reasons for procedures.
7. Moving all single adults who use more than 180 ES nights in a year into permanent
supportive housing would allow the District to close about 1,300 ES beds for single
adults, or around half of current ES capacity. The District could then design and offer less
crowded, more specialized, more effective emergency shelter programs.
a. Some of these would be brief shelter programs with few services for the 47
percent of single adult shelter users who stay one week or less.
b. Others could focus on specific issues such as promoting employment, addressing
disabilities, and establishing connections to housing.
illustrates the importance of policy makers and analysts working closely together to be sure that the answers being
given actually make sense in relation to the questions that were asked.
35
Chapter 2: Understanding How Homeless People Use the Emergency Shelter System
36
Chapter 3: Scope and Structure of the Local Homeless Assistance System for Single Adults and Families
37
Chapter 3
Scope and Structure of the Local Homeless Assistance
System for Single Adults and Families
HIGHLIGHTS
•
A Districtwide structure—in the sense of there being clear rationales and pathways for
placing system users in one program or another and helping them move through whatever
stages and programs they might need—exists only for emergency shelter for families.
•
Before the Emergency Rental Assistance Program (ERAP), VWFRC was seeing about
200 families a month; after ERAP it has been seeing 130 to 150 families a month—a
reduction of 25 to 35 percent.
•
The HSRA specifies that homeless families must be placed in apartment-style shelters; 93
units of such shelter are available at Spring Road, Park Road, and Girard Street. Other
than these locations, VWFRC has only the 10 units offered by Capital Hill Group
Ministries in congregation-based settings and the hypothermia and overflow space
available at D.C. General to which to refer homeless families.
•
Only 216 ES beds out of a total 2,426 for single adults (9 percent) have any type of
specialized identity, approach, or expectations. Only 15 beds are specifically for
substance abuse recovery, in a system where 55 percent of single adults whose disability
status is known have substance abuse problems. The rest are basically warehousing.
•
One area for expanded offerings within District shelters is employment-related skills
development and supports for job search and retention.
•
44 agencies in the District offer 89 transitional housing projects that can accommodate
2,265 people at any one time, including the children in projects serving families.
•
The District already has more than 3,200 beds of permanent supportive housing, many of
them of long standing.
•
According to monitoring reports from TCP and DHS, at least two-thirds of the people
using low barrier shelters never talk with a case worker.
•
For some things it does not matter so much where caseworkers are placed, whether under
the rubric of a shelter provider or in an independent agency, as long as (1) there are
enough of them for the population being served and (2) they have something major to
offer. That is, they really need to be able to help people get services and benefits,
therapeutic options and employment options, and above all, housing. Explicit in this
second condition is that the community has to offer enough services, of the right type,
Chapter 3: Scope and Structure of the Local Homeless Assistance System for Single Adults and Families
38
that homeless people can qualify for. The issue is not simply where the caseworkers sit; it
is whether they really have access to resources that clients need.
This chapter describes the nature, number, capacity, and focus of the District’s residential
programs offering emergency shelter, transitional housing, and permanent supportive housing.
We also briefly examine supportive services available in the District to homeless people and the
provider network that makes it all happen. It describes where TCP fits in the structure, what parts
are organized together and what parts are not. We use this information, together with information
provided in the first report of this assessment, covering providers’ past experiences dealing with
District government agencies and their sources of program revenue, to draw a comprehensive
picture of the scope and structure of the District’s homeless assistance system. Along with
findings reported in other chapters, this information helps us form the recommendations for
system change that we reserve for chapter 5.
EMERGENCY SHELTER SYSTEM STRUCTURE
On the whole, the heading above is a misnomer because for the most part the array of homeless
assistance programs in the District does not have a structure—not in the sense of there being
clear rationales and pathways for placing system users in one program or another and helping
them move through whatever stages and programs they might need, including directly into
housing. Individual nonprofit homeless assistance agencies may have their own rationales and
pathways through which their clients move, but a Districtwide structure exists only for
emergency shelter for families, as shown in figure 3.1.
THE SITUATION FOR FAMILIES
Virtually all families
seeking emergency
shelter in this system
enter through the
Virginia Williams
Family Resource
Center (VWFRC).
The VWFRC has a
number of resources
it can call on to help
families requesting
assistance, shown in
the boxes in figure
3.1. The solid arrows
in figure 3.1 indicate
pathways for those
who are actually
homeless; the dashed
arrow indicates
uncertainty about what will happen to those who are actually homeless once the system is again
at capacity. The dotted arrows indicate resources available to VWFRC for those who are not yet
actually homeless.
Chapter 3: Scope and Structure of the Local Homeless Assistance System for Single Adults and Families
39
Only 15 to 20 percent of families who come to VWFRC are actually homeless when they come
in. Most of the rest are doubled up and in some type of housing-related crisis, which may be
interpersonal, economic, or both. VWFRC staff do an intake interview with all families to
determine why their housing is at risk, their needs, potential eligibility for services, and what
types of help might create positive change in their immediate family situation that would let
them stay housed. At VWFRC a family can consult employment and housing specialists, access
public benefits for which they might be eligible, and work with a case manager to try to resolve
its housing crisis without losing housing. In addition to the connections to mainstream resources
that can be made at VWFRC, the Center has two resources designed specifically to keep people
in housing—Community Care Grants and the resources of the Emergency Rental Assistance
Program (shown in figure 3.1).
Community Care Grants
If the family is not yet actually homeless and appears likely to be able to benefit from a
Community Care Grant, through which it will work with a nonprofit agency for up to a year to
stabilize its housing and financial situation, VWFRC will refer the family to that resource.
Community Care Grants give the nonprofit agency a certain amount of money it can spend to do
“whatever it takes” to help the family stabilize, be it move-in money, child care, car repairs, or
new teeth.
Emergency Rental Assistance Program
If the family’s situation is clearly economic and a short-term housing subsidy appears to be the
solution, since early 2007 VWFRC has had the resource of the District’s Emergency Rental
Assistance Program (ERAP) to help with housing costs. Before ERAP, VWFRC was seeing
about 200 families a month; after ERAP it has been seeing 130 to 150 families a month—a
reduction of 25 to 35 percent. But ERAP is only available to people who are already in a unit or
have identified a housing unit and are approved for it, so it helps people stay where they are or
get into a new place they have already found. Further, resources for each family are only
sufficient to cover a few months’ rent or deposits. If a family is already in housing and has rent
arrearages within ERAP financial limits, ERAP may be used to pay the arrearages. Alternatively,
ERAP may pay for the security deposit and first month’s rent (up to $900) to allow the family to
move to a new residence. To qualify, families have to show that they can continue to pay the rent
when their housing emergency is over—that is, in the time frame that ERAP can cover. In
reality, family expectations of being able to pay the rent themselves in a few months fall through
fairly often. Families may only apply to ERAP once a year.
Emergency Shelter
Family shelter resources are shown in table 3.1. If a family is one of the few arriving at VWFRC
literally homeless, the Center will first attempt to find the family a place to stay temporarily with
relatives or friends; if these attempts fail a shelter referral will be made. VWFRC refers to only
four of the eight agencies in the District providing emergency family shelter, which together
offer 79 percent of year-round family emergency shelter units in the District. The remaining
units of family emergency shelter, none of which report to HMIS, are supplied by Dorothy Day
House (5 family units), Covenant House (6 family units, but counted with beds for youth in table
3.1), and two shelters serving victims of domestic violence that together offer 26 units.
Chapter 3: Scope and Structure of the Local Homeless Assistance System for Single Adults and Families
40
If eligible for shelter or a Community Care Grant (shelter diversion), families are placed in those
settings, to the extent that the programs have a unit available. The HSRA specifies that homeless
families must be placed in apartment-style shelters; 93 units of such shelter (341 beds) are
available at Spring Road, Park Road, and Girard Street. Other than these locations, VWFRC has
only the 10 units (33 beds) offered by Capital Hill Group Ministries in congregation-based
settings and the hypothermia and overflow space available at D.C. General to which to refer
homeless families. With the closing of D.C. Village, the District lost capacity to serve 68
families (42 percent of its then-capacity), albeit in a congregate setting that violated HSRA
provisions and left a great deal to be desired. The question now is, what will happen when
demand for family shelter exceeds supply? The dashed line from VWFRC to the box labeled
“Overflow” represents the still-evolving plans for what to do with homeless families when it is
not hypothermia season and no unit is available for an eligible family.
THE SITUATION FOR SINGLE ADULTS
Four agencies offer eight year-round programs for single women, comprising in total 396 regular
beds. A ninth agency offers 6 beds for single women along with 30 family units and is included
with the family projects in table 3.1 rather than those for single women. One agency offering 31
beds does not participate in HMIS; the others do. Access to most of these programs is achieved
by standing in line—if one is not in line, one does not get shelter for the night. Even those who
are in line may not get shelter if the line is too long. Two programs, with 43 beds, are
subcomponents of larger programs that offer guests 24-hour access to their beds as a privilege
for making progress toward goals agreed-to by guests and program staff.
Table 3.1: Agency and Project Participation
In the CoC/HMIS—Emergency Shelters
Population served
Agencies with yearround projects
Year-round bedsa
Year-round
projects
#
Offering
# in
HMIS
#
Offering
# in
HMIS
# Offered
# in HMIS
% in
HMIS
Families
8
4
8
4
472 (140)
374 (104)
79
Single women
4
3
8
7
396
365
92
Single men or both
men and women
8
3
14
9
2,195
2,061
94
Youth
3
0
3
0
37
0
0
Totals
19
8
33
20
3,100
2,800
90
a
For families, the number of beds is shown, with the number of units in parentheses. Numbers reflect the situation
as January 31, 2008, after closing D.C. Village, which could shelter 238 people at a time (68 families).
Chapter 3: Scope and Structure of the Local Homeless Assistance System for Single Adults and Families
41
Eight agencies offer 14 year-round projects for single men or single men or women; together
they have 2,195 year-round beds. Two of these agencies also offer seasonal beds in five
locations. In addition, three other agencies offer only seasonal/overflow beds. Access to most of
these projects is achieved by standing in line, as is true for the single women’s projects. Three
youth projects offer a total of 37 beds; one serves young mothers with children, but is counted
with the youth projects rather than those for families because access is restricted to youth.
BEGINNINGS OF SPECIALIZATION WITHIN SINGLE MEN’S SHELTERS
As already mentioned in chapter 2, we expect that the District can move the 14 percent of shelter
users who stay more than 180 days out of shelters and into permanent supportive housing
through a combination of housing development—which the PSH Work Group of the Interagency
Council on Homelessness’s Strategic Planning Committee is already working on—and working
with shelter residents. Then shelter beds could be reduced by half and the remaining beds
restructured to address specific needs and promote more rapid exit from shelter for the 86
percent of shelter users who use fewer than 180 days. The 47 percent who use 7 or fewer days
can be accommodated in true emergency or intake shelters. The 39 percent who use between 8
and 180 days and who have the potential to return to housing on their own need the attention of
case workers for assessment, triage, and assignment to specialized shelters that will help them
address the issues that keep them homeless.
Some District shelters have the rudiments of this specialization. Emery has been designated a
working men’s shelter. Every resident has a job, every resident’s bed is guaranteed for the night,
storage is available for clothes and equipment needed for work, and shelter hours and curfews
are modified to accommodate people’s work schedules. Emery has space for 110 men at a time.
The Housing Assistance Center at New York Avenue (48 beds) works the same way—all
residents have a job and the idea is that they are saving money to get into housing. Plans were to
have a second Housing Assistance Center, at 801 East, but resources have not been sufficient for
that to happen. Thus around 7 percent of the emergency shelter beds in the system are designed
to support people who are both homeless and working.
During our interviews with homeless assistance providers, many commented on the fact that no
shelters have specialized services to help people GET work. If shelter residents get a job by their
own efforts they can move into the specialty shelters, but there are no available steps to help
someone along the way if he or she does not already have a job. Certainly one area for expanded
offerings within District shelters is employment-related skills development and supports for job
search and retention.
Another very small proportion of District shelter beds are designated as “TRP,” for Transitional
Recovery Program, or as “24-hour” in the case of beds for single women. People occupying
these beds know they have a bed for the night, somewhat more room, a bit of storage space, and
more attention from caseworkers. They are expected to be “in transition,” working on whatever
the issues are that keep them homeless, whether lack of a job, substance abuse, health problems,
or other issues. People in these beds still do not have an imposed maximum length of stay, but by
the nature of the program they are in, they are expected to be working toward moving out of
shelter within some reasonable period of time.
Finally, 15 beds at 801 East are set aside specifically for substance abuse recovery, in a program
run by APRA. That’s it—among the 2,426 emergency shelter beds in the CoC, only 216 or 9
Chapter 3: Scope and Structure of the Local Homeless Assistance System for Single Adults and Families
42
percent, all for single adults, have any type of specialized identity, approach, or expectations.
Only 15 beds are specifically for substance abuse recovery, in a system where 55 percent of
single adults whose disability status is known have substance abuse problems. The rest are
basically warehousing, as everyone interviewed who actually runs these shelters agrees.
ISSUES WITH LOW-BARRIER SHELTERS
Of the 370 beds for single women that report to the HMIS, 312 (84 percent) are low barrier,
meaning that nothing is required of guests to enter or to return for as many nights and years as
they can get in, and virtually nothing is asked of them even by way of information, if the missing
data shown throughout chapter 2 is to be believed.
Of the 2,061 beds for single men that report to the HMIS, most are low barrier. As it does for
single women’s shelters, this means that nothing is required of guests to enter or to return.
Further, the practically nonexistent entry requirements for low barrier shelters seem to have been
interpreted to mean that shelter users are being asked for the bare minimum of information about
themselves at entry—and they are giving even less.
It is certainly true that the less one knows about a person the less one can help them. In
consequence, the general practice of keeping “hands off” the people who use low barrier shelters,
even if they return night after night for months or years, means that staff may know as little, and
be as little able to help, after a person has used a shelter bed for months as was true on the day
the person first entered the program.
It should be clear to any reader that this is not the way to help people leave shelter. The paucity
of information about the vast majority of men and women using low barrier shelters that we
discovered in the work done for this report, coupled with what we learned about casework
practice in shelters that we described in the first report for this assessment, clearly leads to the
conclusion that all these shelters accomplish is warehousing people one night at a time. Frequent
users of low barrier shelters are allowed to opt out of speaking to a case worker, and once they
do, caseworkers do not “bother” them again, being overwhelmed with people who do want help.
Thus, according to monitoring reports from TCP and DHS, at least two-thirds of the people using
low barrier shelters never talk with a case worker.
Our first report for this assessment discussed this issue with respect to contract size and RFP
type, but the issue is also pertinent to the task of the present report, which is to assess how the
District could transform its homeless assistance system into one that works. In the first report we
wrote (Burt and Hall 2007, p. 30):
Some maintain that emergency shelter operators are not obligated to provide more
than a single case manager simply because clients are not required to participate.
But this low service level is clearly not the intention of the Homeless Services
Reform Act, nor is it good practice. In fact, we disagree with the notion that lowbarrier shelters do not need as much staff time to work with clients—actually, they
need much more, because they should be working to engage residents in efforts to
leave the shelters rather than just ignoring everyone who does not seek out their
services. Of the providers we spoke to, almost all felt they could use more case
management, and most put the optimum ratio at around 15–20 clients per case
manager. Indeed, the low case management level is more a product of dwindling
Chapter 3: Scope and Structure of the Local Homeless Assistance System for Single Adults and Families
43
resources, unwieldy client loads, and numerous contract deliverables. The many
downsides of this dilution of services will be discussed in more depth in our second
report, but we have seen that the results are significant and negative.
We make no claim that our documentation of how little is known about people using low barrier
shelters provides any new insight—many people in the District knowledgeable about the local
shelter system have been saying the same thing for years. Our findings simply add weight to
those perceptions, and highlight the difficulties that District policy makers will have in
developing any coherent plan to change the emergency shelter system—which certainly needs to
happen—based on the information currently available to them to describe current users.
Sources of Guidance
Many communities around the country have experience with the type of transformation of
emergency shelter that we strongly recommend for the District. New York City is probably the
most immediately comparable, since its “right to shelter” legal environment means the city pays
for almost all emergency shelter, as does the District, and people can use shelter as long or as
often as they choose, as happens here. 17 For the past four years or so, New York has had an
initiative to reduce long-term use of shelters by working with long-term shelter stayers while also
developing permanent supportive housing as an alternative to the shelters. This is a combined
effort of the Department of Homeless Services, the Continuum of Care organizing structure with
its broad community membership, and shelter staff and contracted agencies such as the Center
for Urban Community Services that do the actual work. Common Ground in New York is also
very experienced in engaging homeless people living for years on the streets and helping them
move into permanent housing, having transformed the Times Square area with their outreach
coupled with permanent supportive housing development and access in the immediate area.
TRANSITIONAL HOUSING
The intent of transitional housing projects is to serve people who are expected to be able to
return to housing on their own, but for whom emergency shelter does not offer enough time or
services to help them get there. “On their own” does not imply that they will be able to afford
housing based solely on earnings; it does mean that with a housing subsidy and after a short
stabilization period they will be able to keep themselves in housing without needing ongoing
supportive services. That is, candidates for transitional housing should not be those whose
disabilities preclude their ever living completely independently.
Little is known nationally about how transitional housing fits into a community’s network of
homeless assistance programs, in the sense that no communities to our knowledge have any
central control over who enters transitional housing projects. Most people get into transitional
housing upon referral from other service agencies. Most commonly these are emergency shelters,
but they may also include community action agencies, welfare agencies, child welfare agencies,
17
Homeless people in the District do not have a legal right to shelter at all times, although they do have a right to
shelter for five months of the year during hypothermia season and District-supported emergency shelters may not
place length-of-stay limits on guests. The only limit on people using emergency shelter in the District is the
availability of shelter beds.
Chapter 3: Scope and Structure of the Local Homeless Assistance System for Single Adults and Families
44
agencies offering emergency food and shelter assistance and homelessness prevention, social
service agencies, and clergy, among others.
We know as little, or as much, in the District as is known nationally about how transitional
housing projects fit into the overall scheme of homeless assistance programming. Table 3.2
presents the basics, telling us that 50 agencies in the District offer 88 transitional housing
programs that can accommodate 2,925 people at any one time, including the children in
programs serving families. Transitional housing programs exist in the District to serve all types
of homeless people. Some serve only families, only single women, or only single men; some
serve any single adult; some that focus on domestic violence serve single women and women
with children. A few serve youth. Some focus on substance abuse recovery, some on helping
people who have a mental illness, some on developing work skills and getting a job, some on
parenting and household management, and some on different combinations of these and other
issues. Some allow people to stay for two years or even more, while others have maximum
lengths of stay as short as three or six months.
Table 3.2: Agency and Project Participation
In the CoC/HMIS—Transitional Housing Projects
Population served
Agencies with
year-round
projects
Year-round bedsa
Year-round
Projects
#
Offering
# in
HMIS
#
Offering
# in
HMIS
# Offered
# in HMIS
% in
HMISb
Families
28
18
35
26
1,746 (535)
1,373 (124)
79
Single women
12
9
16
12
296
197
68
Single men
11
5
16
10
443
290
65
Singles, men or women
14
4
18
6
401
115
29
Youth
3
1
4
2
39
22
56
Totals
50
29
88
54
2,925
1,997
68
a
Reflecting the situation as of January 31, 2008; the number of beds available for families is shown; the number of units
appears in parentheses. Family beds include 600 (200 units) that were added in six projects as part of the System
Transformation Initiative that accompanied the closing of D.C. Village, and another 60 beds (20 units) in a seventh project
that opened in 2007. .
b
One agency participates for one project but not for another.
Some agencies offering transitional housing are large multi-service agencies that run the
equivalent of a mini-continuum of care within their own walls, offering many other types of
Chapter 3: Scope and Structure of the Local Homeless Assistance System for Single Adults and Families
45
assistance, shelter, and housing to homeless and formerly homeless people. Are most of the
people that their transitional housing programs accept coming from other programs within the
same agencies, making it difficult for people from other agencies to access the programs? We do
not know. Are programs with short maximum lengths of stay oriented toward people with
simpler problems and issues who do not need as long to move on, or are other things than client
needs driving length of stay requirements? Again, we do not know. Are some better than others
at helping similar people leave homelessness and achieve stable housing? Performance
standards and reporting that would allow anyone to make these judgments are in their infancy, as
we discussed in the first report for this assessment (pp. 78ff.).
It would take a more intensive and detailed study than the current one to discover all these things
about transitional housing programs in the District (the same is true for permanent supportive
housing programs, as we will see below). At some time, the District will probably want to
undertake such a study and bring to the table the issue of how transitional housing resources are
being and should be used. These are important resources and it would be useful to develop a
strategic plan for them. Doing so, however, will probably rank behind resolving system issues at
both ends of the service system that are linked together—emergency shelter and permanent
supportive housing.
PERMANENT SUPPORTIVE HOUSING
Permanent supportive housing (PSH) provides safe housing, made affordable to people with
extremely low incomes through housing subsidies or other means, with supportive services to
help people retain their tenancy for as long as tenants need or want to remain (i.e., there is no
time limit on residency). What distinguishes PSH from regular affordable housing is that (1) all
tenants have been homeless, (2) all tenants have a disability that threatens their continued
housing tenure if supportive services are not available, and (3) supportive services are available
and have the primary purpose of helping people retain housing. Otherwise, PSH resembles
regular affordable housing in that tenancy is secure as long as the tenant abides by the terms of
the leasing arrangement, just as is true for any other tenant. What distinguishes PSH from
residential living situations for people with serious mental illness, developmental disabilities, or
other physical, mental, or emotional disabling conditions is that (1) all tenants have been
homeless (in units designated as PSH, not necessarily in every unit in a building), (2) tenancy is
secure as long as the tenant abides by the terms of the leasing arrangement, just as any other
tenant would have to do, and (3) supportive services are geared primarily to helping people retain
housing, not to therapeutic ideas of “improvement” or control.
PSH is configured in many ways—a program may use a single building that contains all its units
and tenants, it may use apartments scattered throughout a neighborhood or community and
connect with tenants either in their homes or at a program site, it may occupy a few units in a
building otherwise providing affordable housing to income-eligible households, or any of several
other variations and combinations. It may be dedicated to a specific subgroup of formerly
homeless people identified by household type (single men, single women, families, mixed), or
disability (people recovering from substance abuse, mental illness, living with chronic health
conditions, or any combination). It may be “clean and sober” or not. It may be “low barrier”
(come as you are, without requirements for having either addictions or mental illnesses “under
Chapter 3: Scope and Structure of the Local Homeless Assistance System for Single Adults and Families
46
control”) or “housing ready” (usually meaning that addictions or mental illnesses are “under
control”).
A community needs all types of PSH, because people who need PSH have different needs and
preferences for where and how they want to live. Some want, need, and seek clean and sober
living, either to start with or at some point along the way to help them maintain sobriety. Others
will not even begin to leave homelessness if they must first control or stop their addictions or go
(back) on psychiatric medications. Housing First models of PSH were developed for the latter,
and have proven highly effective at helping the most severely disabled, hardest to serve,
treatment resistant street homeless people and long-time shelter stayers to leave homelessness.
Since these long-time homeless people, usually with multiple disabilities, are the people the
District has determined to serve with its commitment of 2,000 net new units of PSH for single
adults by 2014, these units should be structured along the basic outlines of Housing First models
until the demand is met. Current estimates of about 1,600 long-term shelter users and 300 to 400
street dwellers will just about take all these units. The remaining 500 units of PSH to which the
District has committed itself are for families with histories of long or repeated homelessness.
Table 3.3: Agency and Program Participation
In the CoC/HMIS—Permanent Supportive Housing Programs
Population served
Agencies with
year-round
projects
Year-round bedsa
Year-round
projects
#
Offering
# in
HMIS
#
Offering
# in
HMIS
# Offered
# in HMIS
% in
HMISb
Families
Single women
9
4
3
3
11
4
6
3
1171 (369)
54
721 (229)
33
62c
61
Single men
7
4
11
3
106
49
46
Singles, men or
women
Totals
18
9
33
15
1,894
526
28
30
13
59
27
3,225
1,329
41
a
Reflecting the situation as of January 2008; the number of beds available for families are shown; the number of
units appear in parentheses.
b
One agency, DMH, is responsible for 606 PSH beds for single adults and 30 PSH units for families (with 88 beds),
and does not report to the HMIS.
c
One agency, TCP, is responsible for 62 percent of the family PSH beds that are reported in the January 31, 2008
Housing Inventory Chart, all of which report to HMIS.
The District already has more than 3,200 PSH beds, many of them of long standing. Except for
PSH serving families, relatively little of it reports to the HMIS, as table 3.3 shows. By examining
provider names and target population information in TCP’s Housing Inventory Chart, we can get
an idea of the scope of PSH for some specialized populations. For instance, six agencies offer
213 PSH beds (including 27 that are in 12 family units) serving households living with
Chapter 3: Scope and Structure of the Local Homeless Assistance System for Single Adults and Families
47
HIV/AIDS, accounting for about 7 percent of all PSH beds. At least 850 beds (including 88 that
are in 30 family units) are offered by agencies with a mission to serve people with severe and
persistent mental illness, accounting for 26 percent of PSH beds. More than 400 beds have a
major focus on serving people with substance abuse issues, including 123 that are in 33 family
units, and another 100 beds (at least) that are explicitly for people with both serious mental
illness and a co-occurring substance disorder.
One can also see the role of some District government agencies in these resources. Using funding
from the District’s budget, DMH provides resources to support 606 beds for homeless single
men and women with serious mental illness, as well as 30 units for families with a seriously
mentally ill parent. The Office of HIV/AIDS in the Department of Health has obtained federal
resources (Shelter Plus Care certificates) to provide housing subsidies for 36 units that house
formerly homeless people living with HIV/AIDS. We assume, but cannot tell from the
information currently available to us, that these two public agencies also fund supportive services
for these housing units.
SUPPORTIVE SERVICES
One would have to examine project budgets and also look at the patterns of referral and
partnering with other agencies to get a really firm grasp on the supportive services that help
people leave homelessness and stay housed. 18 In its annual application to HUD for funding from
the Supportive Housing Program, TCP includes six pages listing more than 100 organizations
that provide nonresidential services that may help someone avoid homelessness (prevention) or
leave it (outreach and supportive services). Prevention services include rent, mortgage, or utility
assistance, counseling, and legal assistance. Outreach includes street outreach, mobile clinics,
and law enforcement activities. Supportive services include case management, life skills, alcohol
and drug abuse services, mental health counseling and treatment services, health care
HIV/AIDS-related care, education, employment, child care, and transportation. Most agencies in
the District offering one or more of these services to low-income households are listed.
SERVICES FOR HOMELESS PEOPLE LIVING ON THE STREETS AND IN EMERGENCY SHELTERS
Relatively few of the 100+ agencies listed in the most recent HUD application are explicitly
focused on serving homeless people and the likelihood that many homeless persons would
approach or receive services from most of them are fairly low. Some of them are family support
collaboratives, which were developed with a focus on helping families involved with child
welfare to keep their children at home and develop better family relations. Many have helped
families avoid homelessness, but only recently, with the closing of D.C. Village, have they been
drawn into the front lines of responsibility for stabilizing formerly homeless families in housing.
Several other agencies are community action or community development organizations, whose
mission is broadly to serve low-income populations. Some are public agencies with housing,
18
This level of examination is well beyond the scope of this assessment. The first author has gathered similar data in
several studies and several communities and is familiar with methodologies for performing such analysis. The Urban
Institute recently received foundation funding to support a study of this type in the District, to look at permanent
supportive housing programs. Such a study will be very useful to the PSH Work Group as it moves into the phase of
deciding which people to serve and what they will need.
Chapter 3: Scope and Structure of the Local Homeless Assistance System for Single Adults and Families
48
health, mental health, or substance abuse missions and services, but without an explicit
orientation toward serving homeless people.
Study after study of homeless people on the streets, using emergency shelter, or staying in other
places not meant for human habitation have found that most have difficulty accessing services
from these types of agencies, and the agencies have trouble serving homeless people. So people
do not go to those agencies when they are homeless. This is a primary reason why homeless
assistance agencies in the District and elsewhere have felt it necessary to incorporate many types
of specialized services into their own repertory and staffing patterns. Often, too, they find
themselves having to raise money through fund drives or soliciting philanthropic resources to
pay for specialized staff such as psychiatrists that they believe the relevant public agencies
should be supporting.
The historic weakness of funding from public agencies in the District for case management and
supportive services that providers told us about and we described in the first report for this
assessment (Burt and Hall 2008a) is partly responsible for the number and strength of agencies
whose array of homeless-related services is relatively complete and self-sufficient—the nonprofit
mini-continuums that we spoke of above.
If Ending Homelessness Is the Goal, What Services Are Needed and How Should They Be
Configured?
In the District, a large majority of homeless people use shelters at some time or other—many use
them often. Others are known to street outreach workers. We saw in Burt and Hall (2008a) that
case management levels in the District’s shelters were for the most part abysmally low, and
sometimes even nonexistent. There can be no question that significantly more trained casework
staff are needed, to reach staffing levels of no less than one caseworker for every 30 or 40 people
in shelter. To make headway with long-time homeless people, even that ratio is too high—15 to
20 clients with multiple problems and issues would be the most that one person could reasonably
support.
There has been some discussion in the District of separating case management from the
sheltering function and locating the case management in an independent agency (i.e., not with
the same nonprofit agency that runs the shelters). For some things it does not matter so much
where these caseworkers are placed, under the rubric of a shelter provider or in an independent
agency, as long as (1) there are enough of them for the population being served and (2) they have
something major to offer. That is, they really need to be able to help people get services and
benefits, therapeutic options and employment options, and above all, housing. Explicit in this
second condition is that the community has to offer enough services, of the right type, that
homeless people can qualify for. The issue is not just a matter of where the caseworkers sit; it is
whether they really have access to resources that clients need.
An advantage to placing caseworkers in an agency external to all shelters is that the workers
could continue to support people as they move to different programs within the homeless
assistance network and as they move out of homelessness. This continuity of casework would be
a decided advantage over having to establish relations completely anew in each program. On the
other hand, if a person moves to a new program rather than into housing in the community,
teamwork and coordination would need to be developed between the original caseworker and the
staff of the new program.
Chapter 3: Scope and Structure of the Local Homeless Assistance System for Single Adults and Families
49
Under one condition, an external casework agency would be ideal. That condition would be the
availability of “whatever it takes” funding, or a combination of funding streams such that no
client would ever be denied a needed service because he or she was not eligible for a particular
funding source. The agency would make every effort to help clients qualify for the benefits and
services they need, but if someone needed something and was not eligible for the program that
pays for it, “fill-the-gaps” funding from the District would come into play. This type of
“whatever it takes” funding has been used in California (the AB 2034 program) to great
advantage; several other communities have experimented with similar concepts, including
Hennepin County, Minnesota in its Rapid Exit from Shelter approach for families. The District’s
own Community Care Grants are a form of this type of funding. The advantage of locating this
funding-casework combination in one place, external to specific shelters or programs, is that
functionally it becomes a model of money following the client, rather than the client having to
apply to each program separately.
Outreach is another type of service that is more or less useful for ending homelessness depending
on what outreach workers are able to deliver. Outreach teams that can connect people to housing
are very successful at helping people move out of homelessness. Outreach that can only offer
coffee and blankets is the equivalent of overflow shelter accommodations—it does what it does,
which is good in itself, but it is not likely to reduce homelessness. Therefore, ideally outreach
teams will have real resources to which they can link clients, including housing.
SERVICES FOR PEOPLE IN PERMANENT SUPPORTIVE HOUSING
In theory, all tenants in permanent supportive housing have adequate access to casework staff
and resources. In reality, staffing levels and configurations in PSH vary greatly. Although PSH is
supposed to be “permanent” housing, or at least “as long as you need it” housing, nationally the
average length of stay in PSH is somewhere between 1 and 1.5 years, with a significant number
of people who leave doing so within the first three to six months. On the other hand, some PSH
projects have succeeded in keeping people in housing for years. 19 The difference lies in the
quality and structure of services.
In most communities, funding for the services component of PSH is the hardest to get and the
hardest to keep, due to the diverse nature of the variety of help that people need and the very
fragmented nature of service funding streams and programs. District investment in “whatever it
takes” funding could really help at the front end, reducing the time a person remains homeless to
the bare minimum, and at the back end, making sure that people are able to stay in housing once
they get it.
IMPLICATIONS
1. Move long-stayers in shelter and long-time street people into PSH—the task of the PSH
Work Group. Doing this will require creating the housing, assembling the services,
19
Average length of stay in scattered-site housing that uses Shelter Plus Care certificates through the Santa Monica
(California) Housing Authority is 4.5 years. In the District each year, only about 11 percent of people using Shelter
Plus Care to subsidize rent leave their units. Of those who leave, about half go to other permanent housing (a good
outcome) and about half leave for unknown destinations without communicating with the program..
Chapter 3: Scope and Structure of the Local Homeless Assistance System for Single Adults and Families
50
figuring out how to prioritize who gets the next unit, and getting providers to work with
accepting the priorities.
2. Reconfigure emergency shelters for single adults—give the job to a new subcommittee of
the ICH’s Strategic Planning Committee, to resemble the PSH Work Group
3. Develop a service structure and funding sources to help people move out of homelessness
quickly; for those in PSH, make sure services are adequate to assure housing retention.
Figure out the advantages and disadvantages of creating an external source of case
management. Create a “whatever it takes” funding stream similar to Community Care
Grants that are currently available for families, and make it available to case managers
for single adults.
Chapter 4: District Government Agency Activities Related to Homelessness
51
Chapter 4
District Government Agency Activities Related to
Homelessness
HIGHLIGHTS
•
There are positive examples of interagency communication on homelessness, but
nearly all of the public agency staff we interviewed noted a serious need for a more
comprehensive, structured and regular forum.
•
In both day-to-day communication regarding services and long-term planning of
targeted programs, more interagency communication and awareness is needed.
•
The lack of interagency communication can:
o Lead to clients receiving the same services from a number of different providers,
also known as “doubling-up.”
o Leave gaps in services when different public agencies assumed others were
providing a given service.
o Lead to inefficient allocation of resources when public agencies provided services
outside of their specialty.
o Perhaps most importantly, lead to a lack of coordination of individual care.
•
Some public agencies give each other access to administrative data about specific
individuals to facilitate targeted and coordinated services (e.g., DMH and APRA case
managers working in the D.C. jail can access jail information as well as their own
agency data. However, no systematic cross-agency data matching has occurred that
would show policy makers the full extent and patterning of multiple system use.
•
Most public agencies want to move forward on a data sharing system and have been
eager to work with the Urban Institute on a data matching analysis as a first step.
Throughout our interviews, public agency representatives expressed a general frustration over
the inefficiencies in the District’s homeless services system. While few interviewees had a firm
understanding of what public agencies other than their own were doing in regard to
homelessness, many thought that homeless individuals were either getting services from more
than one agency or were being forgotten. To a large extent, each District agency acts inside its
respective “silo” when it comes to homeless services and information on clients, and few
participate in regular and effective interagency communications regarding homelessness, let
alone actual comparisons to see if several public agencies are serving the same people. 20
20
DMH and APRA staff who work with their own clients and eligible inmates in the D.C. jail do have access to jail
information on a person-by-person basis, but no attempt has been made to compare across agencies to determine the
percentage of jail inmates who are DMH or APRA clients, or vice versa. This latter type of agency-to-agency
Chapter 4: District Government Agency Activities Related to Homelessness
52
Communications that do exist tend to be among service agencies, not between service agencies
and housing agencies. The prevailing sentiment among District housing and service agencies
then becomes one of “the other guy is not doing enough.” Service agencies cry out for more
housing and housing agencies complain about a lack of services; all the while, communication
remains minimal.
This is not to say that there is no communication or coordination occurring in the District: the
Interagency Council on Homelessness holds bi-monthly meetings and two work groups are hard
at work. The Criminal Justice Coordinating Council’s Substance Abuse Treatment and Mental
Health Service Integration Taskforce (SATMHSI) is currently working on a comprehensive plan
for individuals with disabilities in the criminal justice system. Many public agencies have
collaborated for specific projects. There are other positive examples to draw from, as outlined
below.
Nor is the lack of communication necessarily the fault of the District agencies. Part of the
problem lies in legal barriers to data sharing that make it impossible at present for agencies to see
exactly where homeless clients are going. 21 Still, in both day-to-day communication regarding
services and long-term planning of targeted programs, more interagency communication and
awareness is needed. Until this happens, the District will not have a truly efficient or effective
system of homeless assistance and prevention.
This chapter briefly describes the problems associated with uncoordinated service systems in
general, and then examines public agencies individually to see how each is impacted by
homelessness, outlining what programs and resources District agencies currently have in place.
Suggestions for improvements and specific examples of successes around the country are given
in chapter 5.
“SILO” PROBLEMS IN THE DISTRICT
Homelessness is a difficult and often expensive problem, one that cannot be neatly tucked into a
single public agency such as DHS. Ending homelessness for many people requires coordinated
services and information from widely differing areas of responsibility—from affordable housing,
to employment, substance abuse, domestic violence, physical disabilities, and mental health. As
is evident in Continuums of Care throughout the country, a successful system puts a high
premium on efficiency and coordination across public agencies and providers.
The “silo” problems are fairly intuitive. The lack of interagency communication can:
•
Lead to a client receiving the same services from a number of different providers, or
“doubling-up.”
comparison of caseloads has just begun between FEMS and DMH, in the context of the Criminal Justice
Coordinating Committee's SATMHSI effort to address the needs of jail inmates with mental illness and/or substance
abuse disorders.
21
Other communities have overcome these barriers while continuing to assure the privacy and confidentiality of
client information. The District could do it also, if it chooses to do so.
Chapter 4: District Government Agency Activities Related to Homelessness
•
•
•
53
Leave gaps in services when different public agencies assume others are providing a
given service.
Lead to inefficient allocation of resources when public agencies provide services outside
of their specialty.
Perhaps most importantly, lead to a lack of coordination of individual care.
All four problems have, to a notable extent, played out in D.C.’s homeless service system and
this is in no small part due to the fragmented nature of public agencies. First, a number of
interviewees mentioned homeless individuals bouncing around from emergency shelters, to
clinics, to hospitals, to detox centers, to jail—all unbeknownst to public agencies and their
subcontractors (private housing and service providers). Some refer to the process as the
“homeless shuffle.” 22 At each stop, a provider gives a service without taking into account what
the client has already received –for instance, without knowing that client A already has case
management at provider B and has gotten treatment for TB at hospital C. Some public agency
subcontractors mentioned that they could not even tell if a client had been to another one of the
shelters their own organization runs. This is not only inefficient and ineffective, it is also
potentially dangerous.
Just as a lack of information can lead to excessive and untargeted services, as just noted, it can
also leave a homeless client with significant gaps in care. Public agency directors frequently
complained that other service providers and public agencies incorrectly assumed that they were
providing a given service. As an example, one District agency director noted, “Sometimes
homeless people get moved into the housing but the support services do not follow. They talk
about providing case management, but people end up just taking them off their list and assuming
[we] will take care of it—something we do not have the internal resources [for].” This lack of
support service surrounding housing for formerly homeless people with disabilities was
mentioned numerous times.
Similarly, some District agencies were inefficiently crossing service boundaries, and not
necessarily by their own volition. For example, DMH is the primary agency running a Sobering
Station dominated by alcoholic homeless individuals—clearly the jurisdiction of APRA (Griffin
and Mead 2007). While sharing responsibilities is a good thing, especially in regard to
homelessness, it is important to recognize expertise, responsibility, and leadership in certain
areas and to allocate funds accordingly. As one District program director put it, “Why am I
providing services for someone who should be getting it from, say, DMH?” Just as we would
not want a dentist to fix a broken arm, it is important for District agencies to own their
responsibilities and to bring specific areas of expertise to the table.
In all, these problems highlight something larger in homeless services: a need for informed
coordination at an individual level. In the face of dual disorders, skyrocketing housing prices,
and transient clients, ending homelessness requires highly coordinated and individually targeted
22
Griffin, Patricia and Martha Johnson Mead (2007), “Strategic Plan for Persons with Serious and Persistent Mental
Health and Substance Abuse Disorders Involved in the Criminal Justice System in the District of Columbia.” The
Criminal Justice Coordinating Council Substance Abuse Treatment and Mental Health Services Integration
Taskforce. Working Draft.
Chapter 4: District Government Agency Activities Related to Homelessness
54
services from a number of different specialists. As one public program director put it, “I should
be able to know what other agencies are dealing with the same client. No agency can solve this
problem alone….” It is well known, for instance, that people with a mental illness are
overrepresented in the criminal justice system and have longer incarceration periods and higher
rates of recidivism; add homelessness to the mix and the problems compound. Suddenly, DMH’s
problem is DHS’s, DOC’s, CSOSA’s, and PSA’s. One agency’s client becomes another’s very
quickly, and the policy choices should change the more overlap that occurs. For the safety of
other homeless individuals in a given program, the well-being of the specific client, and the
general efficient use of public monies, public agencies and providers must have as much
pertinent information as possible.
INTERAGENCY COMMUNICATION AND COORDINATION
There are positive examples of interagency communication on homelessness, but nearly all of the
public agency staff we initially interviewed noted a serious need for a more comprehensive,
structured and regular forum. One public agency program manager told us:
Just in terms of literally coordinating a lot of our efforts internally, we need to
find out what projects are compatible with other agencies’ activities, we need the
identification of units, the development of units, man hours to stay in the loop in
terms of what’s going on in city hall, [what’s] going on in activist meetings,
staying in contact with TCP, and really just staying on top of the programs we
currently manage.
It is true that most of the public agency staff interviewed had established one or more channels of
communication between their agency and another, but these connections were almost always
narrow, concerning the use of new funds or existing funds in a jointly run program. In general,
interagency communications that occur on a project basis are effective, but until very recently
there has been no forum for communication related to system planning or resource development
and allocation.
We reserve to chapter 5 the recent developments, of which the activation of the Interagency
Council on Homelessness stipulated in the Homeless Services Reform Act is the most
comprehensive of four mechanisms with the potential to develop and implement a new vision for
the District’s response to homelessness. In this chapter we focus instead on the evidence that key
areas of communication among public housing and public service agencies were improving.
Examples of interagency coordination include, among many:
•
•
•
DMH currently awards temporary rental subsidies to mentally ill individuals on the
DCHA waiting list for housing vouchers (which has a homeless preference, see
below).
DMH’s Homeless Outreach Team (HOT) works closely with FEMS and MPD to
provide transfers and referrals to the Comprehensive Psychiatric Evaluation Program
(CPEP) for homeless individuals.
DHCD and DMH are working together to create service-enriched housing for
mentally ill homeless people, with DHCD orchestrating unit production for DMH
clients and DMH using $14 million from its budget to cover the housing plus services
once the housing is occupied.
Chapter 4: District Government Agency Activities Related to Homelessness
•
•
•
•
•
•
•
55
DHS works with DCHA to provide third party verification of homelessness to qualify
people for homeless set-aside Section 8 certificates.
DHCD works with TCP, and thus DHS, to determine funding allocations for PSH
providers.
OPM works with DHS and TCP to do shelter maintenance and major repairs.
In the Options Program, PSA and DMH provide case management and housing,
through Green Door, for mentally ill defendants who are homeless. DMH subsidizes
10 such housing units.
CSOSA works with DCHA to help offenders who held Section 8 vouchers before
they were charged to regain their subsidized unit.
DHS, APRA, and DMH have a Memorandum of Agreement to use a detoxification
facility as a hypothermia shelter for alcoholics.
DMH, DHS, and the D.C. Superior Court work together through Community
Connections to provide intensive clinical services and housing for 50 homeless
persons who have been legally charged with a crime and have a mental illness
(Griffin and Mead 2007).
Some public agencies were able to build on communications by establishing forums, panels, and
brownbag sessions to brainstorm new and coordinated efforts with a more systemic focus. For
example, DHCD recently conducted a brown-bag to discuss “what was needed and how much”
in the homeless service system and what to include in the RFPs that it was about to issue for its
$12.5 million of earmarked funding. The meeting, attended by local advocacy networks,
providers, and District agencies (DCHA, DMH, and DHS) had a panel discussion and a
question/answer session. Other interviewees mentioned attending TCP meetings to discuss
further funding options and use of funds.
DATA SHARING
Far fewer public agencies shared information on specific individuals so that targeted and
coordinated services might be applied, but data sharing does exist. FEMS and DMH have begun
to compare numbers in an attempt to identify frequent users, but what policy implications will
come from the exchange are still ambiguous, at best. DMH and DOC treatment staff also share
data to ensure DMH clients maintain treatment. DOC, PSA, and CSOSA share information to the
extent that each acts as a step in the same judicial path, but it is clear that more open dialogue
and policy brainstorming is needed between DOC and the two supervision agencies. APRA
currently has access to CPEP records through the Safe Passage Information System, but again,
we did not see strategic use of the data for the advancement of homeless services. In sum, there
is reason to believe that more strategic thinking is needed on how to share data while maintaining
client privacy, as well as on how to use shared data to target policies for the homeless. However,
these examples do show that data sharing is indeed possible.
Fortunately, our interviews and research revealed that most public agencies wanted to move
forward on a data sharing system and were eager to work with the Urban Institute on a data
matching analysis as a first step. The majority of interviewees in public agencies expressed a real
desire to work with others in a more coordinated, client specific way, and some, including
CSOSA, MPD, and FEMS, have already made additional commitments to open up their data
systems to detect the overlap with people who are clients of DMH. The SATMHSI taskforce,
Chapter 4: District Government Agency Activities Related to Homelessness
56
moreover, has drawn up preliminary plans for a multi-agency data base to track people with
mental illness throughout the corrections system; there is no reason to believe these plans cannot
include the HMIS and homeless service providers.
DISTRICT GOVERNMENT PROGRAMS THAT IMPACT HOMELESSNESS
Part of our charge in this assessment was to determine which District government agencies are
affected by homelessness and whether they have programs and activities that explicitly serve
homeless people. Many public officials we interviewed were genuinely interested in knowing
this information. One public agency program director stated, “If literally I knew what each
agency did in a nutshell, that would speak volumes.”
We hope the information presented here will serve as a reference point, to be updated and
maintained as a public resource in the future. It should also be a tool for further action—effective
services require interagency coordination. First we look at the facts of public agency activities,
and then address the issues of coordination, collaboration, and implementation.
For each District agency, we briefly describe (a) how it impacts, or might be impacted by,
homelessness, (b) what relevant programs it currently offers, and (c) what data it produces or
controls that might be useful, first in understanding homelessness in the District and how people
are involved with different District agencies and then in taking steps to end people’s
homelessness. We do not cover all public agency activities that may reach homeless people. 23
In the interest of time and project resources, we focus only on programs and activities
specifically aimed at homeless populations or that are obviously highly impacted by
homelessness.
The District agencies surveyed were:
Service agencies:
• Department of Human Services (DHS)
• Department of Mental Health (DMH)
• Department of Health’s Addiction Prevention and Recovery Agency (DOH/APRA)
Housing agencies:
• D.C. Housing Authority (DCHA)
• Department of Community Development (DHCD)
Public safety agencies:
• Fire and Emergency Medical Services (FEMS)
• Metropolitan Police Department (MPD)
• Department of Corrections (DOC)
• Court Services and Offender Supervision Agency (CSOSA)
23
As an example, the Addiction Prevention and Rehabilitation Agency (APRA) has many specialized service
programs for specific populations such as seniors or Latinos. Within each subgroup there will almost certainly be
homeless individuals, but because the programs target people by other characteristics and APRA does not maintain
data on the housing status of people served, we cannot tell how many homeless people might be affected. Therefore
we do not include them here.
Chapter 4: District Government Agency Activities Related to Homelessness
57
• Pretrial Services Agency (PSA)
Other agencies:
• Office of Property Management (OPM)
Department of Human Services (DHS)
The Department of Human Services pays for the lion’s share of homeless services in the District,
mainly through the contract it maintains with the Community Partnership (TCP) to transfer
funding from its own coffers to the agencies that perform the actual services. Currently TCP
manages 90 subgrants funded by DHS dollars, which pay for ES, TH, PSH, and a variety of case
management and support services. Within DHS, the Family Services Administration (FSA),
directed by Fred Swan, is directly responsible for providing services and aid to homeless
individuals and families. DHS is perhaps the one District agency that is specifically charged with
reducing homelessness rather than providing services to address certain problems facing
homeless people, such as mental illness, substance abuse, or an inability to afford housing.
Programs Contracted through TCP. TCP distributes DHS funds through subcontracts
to mostly nonprofit and some for-profit agencies to support essential activities of the homeless
assistance network. Contract numbers and types vary from year to year, depending on what is
needed and on what new activities DHS has been instructed to focus on by City Council or the
Mayor’s office and has received money for. In our previous report, we described the way these
new activities may arise and how they are worked through the system. If policy makers decide to
continue the new activities, TCP’s contract for the next year is modified to reflect the increased
responsibilities and resources. Under the contract in place in 2007, TCP issued subcontracts to
various agencies offering homeless assistance programs to run:
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
9 hypothermia/severe weather shelters,
8 low barrier shelters,
13 temporary emergency shelters,
2 Housing Assistance Centers in emergency shelters (but only one is really
functioning as a HAC),
19 transitional housing programs,
16 permanent supportive housing programs, of which 7 use Shelter Plus Care
certificates,
10 outreach programs,
6 supportive services programs, which include the Virginia Williams Family
Resource Center, other case management programs, an employment services
program, an on-site clinic for families (now gone, since D.C. Village closed), and
community voicemail,
3 programs of rental assistance for foster care families, reunification services, and the
Emergency Rental Assistance Program,
Several contracts to cover operating expenses such as shelter maintenance and food
service, vans, training, property management, and attorney’s fees, and
Contracts for activities identified during the year, especially those related to closing
D.C. Village.
Chapter 4: District Government Agency Activities Related to Homelessness
58
Data. All subcontracts that TCP manages for DHS require providers to participate in HMIS,
including virtually all emergency shelter beds/units and high proportions of transitional and
permanent supportive housing programs serving families (those serving single adults are less
well represented). Requirements are simplest for emergency shelter programs, which are only
asked to submit 16 data elements for each person:
1. Name
2. Social Security number
3. Date of birth
4. Race and ethnicity
5. Gender
6. Veteran status
7. Residence prior to program
8. Zip code of last permanent address
9. Month and year person left last permanent address
10. Program entry date
11. Program exit date
12. Unique person identification number
13. Program identification number
14. Program event number (is this the 1st, 2nd, 3rd, etc. time the person has used the
program? For 12-hour emergency shelters where a bed is not guaranteed every night,
each night a person uses shelter is treated as another event)
15. Unique household identification number
16. Children questions (do you have children with you? If yes, items 1–15 are asked for
each one.)
As we saw in chapter 2, however, many of these data elements are missing for substantial
numbers of homeless service users. The fault may lie with what providers of specific types of
service (e.g., hypothermia, low barrier shelter) are asked to collect, with the diligence applied by
shelter workers to the task of getting good information from shelter users, with staff turnover and
the perpetual need to train new data gatherers, with the way shelter intake is set up, or some
combination of these. But whatever the reasons, the HMIS is missing a lot of very basic
information for a lot of shelter users.
In addition to the universal data elements just mentioned, transitional and permanent supportive
housing programs and some family emergency shelters require other data points, including:
•
•
•
•
•
•
•
•
•
Income and sources
Noncash benefits
Physical disability
Developmental disability
General health status
Pregnancy status
HIV/AIDS status
Behavioral health status (mental illness and substance abuse)
Domestic violence
Chapter 4: District Government Agency Activities Related to Homelessness
•
•
•
•
•
•
•
•
59
Education
Employment
Veteran status
Services received
Destination (varieties of permanent housing, or other venues)
Follow-up after program exit
Children’s education
Other children questions
We spent our time for this assessment working with the HMIS data from the emergency shelter
system rather than the longer-term and more specialized programs, because the biggest issues
had to do with understanding how emergency shelters were being used, how many of the clients
were long stayers, what difference it would make to the need for emergency shelter if the District
developed the promised 2,500 units of PSH and moved the longest-term homeless people into
them, and similar issues. So it remains an open question what the data from people in transitional
and permanent supportive housing programs that report to the HMIS would tell us about who
uses these programs and how well the programs are performing.
Department of Mental Health (DMH)
The Department of Mental Health is an integral part of the District’s response to homelessness.
As shown in chapter 2, mental disabilities are very common among homeless individuals,
especially those staying in shelter (or on the streets) for long periods of time. What is more,
mentally ill clients were frequently cited by housing providers as the most difficult homeless
population to serve, requiring more support services to keep them in stable housing. Without
programs directly addressing mental illness, the District’s hardest to house people will remain on
the streets, in emergency shelters, or in any number of unfit living situations. It is clear that the
link between housing and services could be improved for this population.
Programs within the Homeless Outreach Team. DMH has a number of programs
specifically oriented toward helping people in the District who are homeless. Most services
for homeless people are run through its Homeless Outreach Team (HOT)—a field-based
crisis intervention team used as a contact for shelter providers, emergency dispatchers, or
District citizens. In addition, HOT runs, or helps run, a number of homeless service
programs, including:
•
•
•
•
Sobering Station—Detox facility open for homeless men and women during
hypothermia seasons (run with APRA, at D.C. General).
DMH service rotation at St. Elizabeth’s Hospital—A required program for
psychiatrists in residency at St. Elizabeth’s, where residents train with HOT staff at
shelters for the homeless.
Mental health trainings—HOT runs trainings for shelter providers and police
officers on a regular basis.
Police Service Area 101 (PSA101) —A pilot project to see if HOT helps avoid
arrests in a high service area for the mentally ill living on the streets. HOT works with
the Metropolitan Police Department to help out with area responses to nonemergency situations with mentally ill homeless individuals.
Chapter 4: District Government Agency Activities Related to Homelessness
60
Programs within the broader DMH include:
• Case management at Franklin School Shelter—Funded by DMH and administered
through Anchor Mental Health (4 case managers)—new program.
• The Day Service at Hermano Pedro Shelter—Funded by DMH, provides case
management services for mentally ill shelter users.
• Urgent Care Outpatient Clinic—Outpatient clinic at D.C. Superior Court, deals
with people with mental illnesses, many of whom are homeless, in court for traffic
and petty misdemeanor offenses.
Data. HOT keeps data on Excel spreadsheets for everyone the team encounters on its rounds.
DMH also has extensive management information systems for all of its clients, whether
homeless or not, to record Mental Health Rehabilitation Services (MHRS data). These data
reside in two large databases—ANASAZI and eCURA. Between the two, DMH clients can be
identified as homeless or not.
The Department of Health’s (DOH) Addiction Prevention and Rehabilitation Agency (APRA)
As we saw in chapter 2, substance abuse affects a high proportion of people using emergency
shelter in the District. Lack of a stable living environment poses unique difficulties for people
desiring to end their addictions; from what we have been told, the District’s low barrier shelters
offer a fertile environment for continued abuse of drugs and alcohol. Unfortunately, APRA
barely focuses on homeless people as homeless, and cannot even tell what proportion of
individuals it serves is homeless. There is reason to believe APRA’s interest is increasing and
data collection is improving, so the agency may be in a better position to address the associated
problems of homelessness and addictions in the near future. At this point, it is clear that the
necessary programs are not in place and that more focus is needed.
Programs. APRA does have programs that specifically address both homelessness and
substance abuse. The following is a list of each program, their general purpose, and the
homeless population they serve.
•
•
•
•
Project RISE—Housing for individuals in substance abuse treatment, fully funded
by APRA and run through Catholic Charities at its 801 East shelter (15 beds).
Hypothermia shelter—APRA runs a severe weather shelter (also referred to as a
sobering center) located at its detoxification facility on the D.C. General campus.
During severe weather homeless individuals may use the facility to sober up.
Satellite intake centers for Drug of Choice Treatment Vouchers—Staff of some
District homeless shelters and clinics (La Casa, Whitman-Walker Clinic, Neighbor’s
Consejo, and La Clinicia de Pueblo) are authorized to issue treatment vouchers. Most
vouchers, however, are distributed by APRA at 1300 First St. NE.
Project Orion—Working together, APRA and Unity Health Care provide addiction
prevention services for individuals in hard to serve communities. Primary care
services are provided by mobile units that go throughout the city. Homeless
individuals receive free services, including “education, counseling, and testing for
Chapter 4: District Government Agency Activities Related to Homelessness
61
HIV/AIDS, sexually transmitted diseases, hepatitis B and C and tuberculosis, and
medical and case management services.” 24
Data. Project RISE client data are kept in HMIS. However, with regard to its general
services, APRA reports it is unable to provide data related to homelessness, not so much
because it does not usually ask about housing status, even though it does not, but because,
staff said, the agency has no reliable data or data systems on any of its clients or services.
D.C. Housing Authority (DCHA)
The D.C. Housing Authority (DCHA) is deeply involved with the issue of homelessness in the
District. Committed to providing housing and expanding opportunities for rental assistance for
low and moderate income individuals and families, DCHA deals primarily with the affordable
housing side of homelessness, although its clients occasionally have services given through local
service providers. DCHA has a “homeless preference,” as well as a specific proportion of its
housing subsidies set aside for homeless people. To date, DCHA has housed over 5,000
previously homeless people, many of whom, DCHA staff report, require inordinate amounts of
housing resources. Because homeless clients tend to come into housing with lower incomes than
even most other low-income households, DCHA is forced to pay proportionally more of the rent
for any unit. (Vouchers pay the difference between an apartment’s rent or the Fair Market Rent,
whichever is less, and 30 percent of a client’s income. The lower the income, the more money
DCHA must use per unit to compensate.) DCHA says the consequence is that it cannot to do as
much for those families who need vouchers but are not homeless.
Homeless clients, especially those with a mental illness or substance abuse problem, can also
pose other unique problems for DCHA. Many people interviewed, including DHCA staff
themselves, felt the agency did not or was not able to coordinate enough with providers who
might supply support services. A common complaint was that a large number of previously
homeless clients who needed services were forgotten once they were put in housing, leading to
problems—especially with mentally ill clients—that strain relations between DCHA and
landlords. This is clearly a place where coordination among and between public agencies and
private service providers is sorely needed.
Programs. In addition to the large preventive role that DCHA plays through various housing
assistance programs, the agency has two programs/funding streams that have a specific
homeless focus. They are:
•
24
Section 8 Vouchers with a homeless preference—The Section 8 Voucher waiting
list gives preference to those who claim to be homeless. During eligibility
assessments for the vouchers, a third party must confirm a family’s homelessness
before it is approved.
Goode, T., Sockalingam, S., Lopez-Snyder, L. (2003). Bridging the Cultural Divide in Health Care Settings: The
Essential Role of Cultural Broker Programs. Washington, D.C.: National Center for Cultural Competence,
Georgetown University Center for Child and Human Development.
Chapter 4: District Government Agency Activities Related to Homelessness
•
62
Local Rent Supplement Program (LRSP)—Local housing dollars that run though
DCHA give homeless families preference in much the same way that preferences
work for Section 8 vouchers.
Data. DCHA does not have any databases specific to homelessness. It does maintain
extensive files on Section 8 voucher applicants and recipients, along with third party
verification of homelessness if relevant. Files include tenant income, household
composition, and other information from a tenant’s application, plus any information that
changes during annual reassessments.
Department of Housing and Community Development (DHCD)
The Department of Housing and Community Development (DHCD) serves as both a provider
and monitor of homeless services in the District and is also a key agency for the development of
future housing opportunities, especially for those with special needs.
Programs. Along with DHS, DHCD funnels federal and district money for homeless
services. Specifically, DHCD has four funding streams set up to fund District homeless
programs directly. They are:
•
•
•
•
•
Shelter Plus Care (federal funding) —124 units with rental assistance and service
programs for individuals or families who are seriously mentally ill, chronically
addicted, or HIV positive. Currently, all Shelter Plus Care units that flow through
DHCD are tenant based rental assistance units in scattered site housing. For these
units, DHCD monitors the program but TCP contracts out the funds to providers or
landlords.
Emergency Shelter Grant (ESG—federal funding) —Gives rent funds to
individuals and families who face eviction. Currently DHCD monitors the program
but TCP contracts out the funds to providers.
$12.5 million earmark—Provides funds for PSH and mixed income housing with
wraparound services.
$25 million Federal Housing Production Trust Fund—An RFP is currently
outstanding for a number of housing services, including PSH for the homeless.
$5 million DMH funding—DMH has transferred these funds to DHCD so DHCH
can issue a targeted RFP for developers to create PSH units for DMH clients who are
homeless.
In addition, DHCD conducts housing development training sessions to attract and educate
developers for the creation of additional affordable housing units for the homeless.
Data. Both the ESG and the Shelter Plus Care units are in HMIS; however, the $12.5 and $5
million earmark funds allocated through DHCD RFPs will not be. For the latter, it is likely
that basic demographic information on program clients will be kept. It would definitely be in
the interest of the PSH Work Group to assure that information about occupants of these other
PSH units is more extensive than that and also that it becomes part of the HMIS—perhaps
through the work of whichever service provider is doing the supportive services that ought to
be available to unit occupants.
Chapter 4: District Government Agency Activities Related to Homelessness
63
Fire and Emergency Medical Services (FEMS)
D.C.’s Fire and Emergency Medical Services (FEMS) deals with homeless individuals on a daily
basis through transports and emergency response calls. Over a given year, FEMS conducts
around 80,000 transports and 120,000 response calls. Homeless individuals account for a large
number of the repeat users of the system; repeat use is especially common for those homeless
individuals with mental illness or substance abuse problems. FEMS Assistant Chief Michael
Williams estimates that around 1 percent of all calls (1,200 calls) were from the top 20 frequent
users—the majority of whom were mentally ill or substance abusers and homeless.
Programs. FEMS fields emergency response units currently consisting of 8 paramedics in
radio equipped minivans. The units can do the following:
•
•
•
Transports—Transports include rides to CPEP or District hospital emergency
rooms.
Emergency Response—Paramedics are trained to do physical and mental health
assessments.
FD12 first responders—EMS is working to get FD12 authority so its staff can be
first responders along with MPS and DMH. FD12 authority would allow FEMS to
treat and transport against a person’s will, once appropriate criteria were met.
Data. Dr. Michael Williams has, in the last eight months, started a database of transports and
responses that keeps:
•
•
•
•
•
•
Electronic Patient Care Reports (ePCRs),
Every call received since the project began,
Type of call,
Diagnoses, issues (substance abuse, metal illness, disability, etc.),
Where transported (CPEP, hospitals, etc.), and
Client address (or, “no fixed address”).
Metropolitan Police Department (MPD)
Because homeless individuals are involved in a disproportionate share of public disturbances in
the parts of town they frequent, the Metropolitan Police Department has a direct interest in
alleviating homelessness in the District. Interviewees mentioned a need for more information to
deal with homeless people, especially those with a mental illness or substance abuse problems.
Indeed, often times it is more effective to refer homeless individuals who are creating a
disturbance to a shelter, case manager, CPEP, or HOT than making an arrest, which the police
see as unproductive for people who have been homeless a long time. Interviewees mentioned
that more information on a person’s mental health history could lead officers to make more
appropriate referrals. In the words of an MPD staff, “We don’t have the information to not make
an arrest. If we did, we could act more appropriately.” 25
25
If a crime has been committed, an officer is obligated to make an arrest, whereupon PSA will take over with
psychiatric evaluations and referrals.
Chapter 4: District Government Agency Activities Related to Homelessness
64
Programs
•
•
•
Police Service Area 101 (PSA101)—A pilot project to see if the HOT helps avoid
arrests in a high service area for homeless mentally ill living on the streets. MPD
works with HOT to patrol the area and refer homeless individuals to their appropriate
services or shelters.
Transports—MPD officers give transport to homeless individuals to hospitals,
CPEP, hypothermia shelters, or to jail.
FD12—All officers are FD12 certified and thus can take an individual to CPEP if he
or she is a danger to themselves or others.
Data
•
•
•
•
•
Police Identification (PDID) numbers – a permanent identification number used for
all arrestees
Whether and where transported (none, CPEP, jail)
PSA101 response and arrest data
FD12 information (details of client characteristics and disposition)
Arrest records
Department of Corrections (DOC)
The link between homelessness and incarceration has been well documented. The Department of
Corrections (DOC), through PSA and CSOSA, has a vested interest in finding stable housing
situations for offenders. PSA and CSOSA, however, act as the medium for individual case
management. Homelessness affects the DOC directly in that homeless people have a high risk of
arrest and incarceration for petty crimes, and indirectly in that it increases chances of recidivism.
Along those lines, housing options that are not associated with a high risk of repeat offending are
needed to make sure those offenders with no other options will, in fact, have an alternative.
Interviewees from the criminal justice system, it should be noted, largely did not view the ES
system as a viable option, generally seeing the shelter system as contributing to recidivism.
The DOC director was highly cognizant of the potential effects on the jail of being able to
identify frequent jail users who were homeless (and possibly also heavy users of other public
systems) and put them on a priority list for accessing PSH. The reduction in jail use could be
considerable. He, along with the DMH and FEMS directors, participates in the CJCC’s
Substance Abuse Treatment and Mental Health Service Integration Taskforce, which has been
grappling for quite a while with the need to share data to identify clients in common and with the
legal barriers that have thus far prevented this from happening. All are very interested in
participating in the data matching project that the Urban Institute has proposed.
Programs. No programs specifically addressing homelessness, although DOC is involved in
a number of activities around prisoner reentry that focus on preventing recidivism.
Data. DOC maintains extensive records on inmates, covering dates of entry and release,
crimes/charges, basic demographic characteristics, other descriptive information, and all
aspects of health, mental health, substance abuse, and other types of care received by inmates
while they are incarcerated. Housing status/homelessness is not a variable in these data, but
Chapter 4: District Government Agency Activities Related to Homelessness
65
the data could be matched with HMIS information and possibly information maintained by
other agencies (e.g., DMH’s homeless outreach team) to determine homeless status.
Court Services and Offender Supervision Agency (CSOSA)
Through its focus on recidivism, prisoner reentry, domestic violence, and other offender
supervision programs and support services, the Court Services and Offender Supervision Agency
(CSOSA) deals with homelessness both directly and in a preventive role. One CSOSA program
manager estimated that as many as 30 percent of the offender population he deals with faced
homelessness. Prisoner reentry into the community has long been associated with a high risk of
homelessness, which in turn is associated with a heightened risk of recidivism. CSOSA therefore
places homeless offenders in housing when it can, linking them to services and support networks.
The agency’s mission ties directly to finding stable living arrangements for its clients.
Finding a stable living situation often proves difficult, however. Agency staff said that case
managers were hesitant to place individuals in ES because it was not conducive to supervision
and follow-ups. Along the same lines, housing has to be found for individuals living in high
crime environments where the chances of recidivism are increased, and not just for those who are
homeless. Sex offenders and violent crime offenders were mentioned as particularly hard to
house because providers were hesitant to accept them, so they faced homelessness as their only
available option. In all, agency staff mentioned a real need for more housing options for the
homeless offender population, deeming the housing search a “constant struggle.”
Programs. CSOSA is a supervision agency without any “in-house” programs specifically
geared toward homeless individuals. It does not serve people who have served their full jail
sentence, because the law has no further basis to supervise them. Nor does CSOSA supervise
“petty misdemeanants,” and that probably includes a lot of homeless people arrested for
loitering, creating a public nuisance, and the like. For people “in between,” whose sentences
run from 1 to 12 months and who are being released before their sentence ends, CSOSA does
provide case management and directs homeless offenders to needed services. The following
are some referrals CSOSA uses specifically for homeless clients:
•
•
•
•
Public Law Placements—Placement in halfway houses as a condition of parole.
Transitional housing—East of the River transitional housing.
Emergency Shelters—As a last resort, offenders are sent to District shelters.
Section 8 Vouchers—CSOSA works with DCHA to get offenders who previously
held Section 8 Vouchers, pre-booking, back into a unit.
Data. CSOSA does keep a lot of information on its supervisees. Its data systems do not cover
petty misdemeanants or people who have served their full term, but for former jail inmates
who do come within its purview, CSOSA data include:
•
•
•
•
Offender addresses.
Public Law placements.
Housing stability and housing type for all offenders as they come into CSOSA (this is
kept in a computer data base called “Screening Tool”).
How many times the offender moved in the last month (Screening Tool).
Chapter 4: District Government Agency Activities Related to Homelessness
•
•
•
66
Type of shelter/housing the offender stayed in within the last month (Screening Tool).
Offender housing history (Auto Screener).
Offender mental health history (Auto Screener).
Pretrial Services Agency (PSA)
D.C.’s Pretrial Services Agency serves the Superior Court for the District of Columbia and the
United States District Court for the District of Columbia in pretrial recommendations and
supervision of defendants, many of whom are homeless. Established in 1997, PSA acts
independently under CSOSA as a federal agency, providing conditional releases and services to
defendants awaiting trial. One of PSA’s guiding principles is that “non-financial conditional
release, based on the history, characteristics, and reliability of the defendant, is more effective
than financial release conditions.” As a result, defendants go through pretrial interviews to assess
their ability to safely reenter the community, to return for trial, and to not engage in criminal
activity. PSA is thus a potentially important check point for homeless individuals who are
entering the justice system—one that could serve as an intake point for homeless services and
one that would supply invaluable data. To a large extent, PSA is aware of this opportunity.
Unfortunately, PSA does not work with D.C. traffic court defendants or petty misdemeanants,
who are tried in a different court. The crimes for which many homeless people are arrested
would put them in the category of petty misdemeanants, meaning that PSA clients would not
include a lot of the frequent jail users we would want to identify.
Undeniably, a lack of a fixed address—often designated as homelessness—presents unique
problems for PSA and its ability to help defendants effectively. Most significantly, a lack of
address or contact information negatively affects case management and the option for conditional
release. An electronic monitoring release, for example, might be the best option for a particular
person, but one can only get electronic monitoring if he or she has a permanent place of
residence. Homeless defendants often cannot be reached for follow through, court notifications,
and general case management, which makes PSA less likely to release them and more likely to
keep them in jail. It is clear that PSA sees homelessness as an impediment to its services and
recognizes that it is in its best interest to approach homelessness directly.
Programs. As with CSOSA, PSA serves as a supervision agency and therefore does not
have any programs specifically targeting homelessness. It does, however, supervise homeless
defendants, provide case management, and make specific referrals for treatment and housing.
Such referrals include:
ƒ
ƒ
Emergency shelter referrals—PSA’s Social Services and Assessment Center refers
clients to emergency shelters when they have no other housing in the community.
Options Program referrals—PSA supports 10 housing slots at the Green Door for 5
male and 5 female homeless mentally ill defendants.
Data. Since PSA does fairly in-depth assessments of the people in its charge, as it is
responsible for advising judges on possible dispositions for defendants, its files contain lots
of data. For our purposes some of the information recorded includes defendant addresses and
address types (several variables – homeless, shelter, and halfway house) and indicators of
disability, such as the mental health and substance abuse information collected by the
Alcohol Severity Index (ASI) instrument.
Chapter 4: District Government Agency Activities Related to Homelessness
67
Office of Property Management (OPM)
Since January 2007, the Office of Property Management (OPM) has dealt directly with homeless
services by helping DHS’s Facilities Management Operations Division (FMOD) with its
homeless shelter repair responsibilities. Today, OPM has completely taken over structural
maintenance under a Memorandum of Agreement with DHS. OPM also contributes to reducing
homelessness in other ways, by searching out housing units that the district might buy and
convert into homeless assistance programs.
Programs
•
•
Structural maintenance for District Emergency Shelters
Finding new units for ES, TH, and PSH
Data
•
•
Maintenance requests
Maintenance contracts
IMPLICATIONS
1. District agencies impacted by homelessness already know they have clients in common
and already want to share information on individual clients, with the goal of coordinating
services. To date they have met numerous roadblocks in their efforts to share data. The
District needs to consider the legal bases that may exist to allow data sharing, including
how to establish appropriate safeguards and privacy protections but still enable crossagency data matching that would optimize use of public resources at the same time that it
increases the odds that homeless people get the services they need to leave homelessness.
2. Pay specific attention to increasing communication among public agencies concerned
with housing and those concerned with services.
3. Identify frequent users and common clients by comparing public agency data bases—see
Chapter 5.
Chapter 5: Implications and Recommendations
68
68
Chapter 5: Implications and Recommendations
69
Chapter 5
Implications and Recommendations
Rarely in the past decade or so has there been an opportunity such as the present for broad
system planning and the development of a common “vision” for the District’s CoC. Even under
the D.C. Initiative (1994–1999), District government agencies were not involved in the changes
that moved the system beyond emergency shelter. Since the D.C. Initiative ended, and until very
recently, no “table” has existed to which all stakeholders might come to work together to shape a
new sense of how the District should respond to homelessness. Even passage of the Homeless
Services Reform Act in late 2005 and its creation of an Interagency Council on Homelessness
did not stimulate such activity until a new administration took office in January 2007.
The Mayor has pledged to replace homeless shelters with housing. He is committed to producing
2,500 new units of permanent housing with supportive services for the most severely disabled
and longest-term homeless people in the city. He has activated the Interagency Council on
Homelessness (ICH) and given it work to do. He has closed D.C. Village and moved many longterm homeless families into apartments where they will receive the services they need to help
them transition out of homelessness. These steps are important in themselves, but will be even
more important if they are just the first steps in a true transformation of the District’s approach to
homelessness.
For many of the recommendations growing out of this report to come to fruition, the District
needs a consistent forum where commitments to big new goals can be taken, ideas for how to
reach those goals can be exchanged and polished, stakeholders can be held accountable for
fulfilling their parts of the plans, and the leadership is present and empowered to make sure plans
are carried out, bottlenecks are resolved, progress is tracked, and the ultimate goals are reached.
Throughout the two reports produced for this assessment, we have drawn conclusions from our
findings about the directions in which the District might want to move the current system and
what it might take to reach a new and more effective set of programs and activities. Despite the
many topics we were asked to address, the most important recommendations converge on four,
and they all interrelate. The Strategic Planning Committee of the ICH is the obvious body to take
up these recommendations, and indeed has already started on some of them. We state them
simply first, and then take up each in turn:
1. Move chronically homeless people from shelters and streets into permanent supportive
housing. Make sure these PSH units have attached to them the supportive services that
people need to help them retain housing through the active involvement and financial
support of District government agencies.
2. Create a process to prioritize who gets the 2,500 new PSH units based on disability plus
frequent use of emergency rooms/psychiatric evaluation center/emergency medical
services/jail or long tenure on the streets or in shelters. Providers getting the resources to
offer the new units must commit to taking people based on this process.
69
Chapter 5: Implications and Recommendations
70
3. As long stayers in shelter, who use between 50 and 60 percent of all shelter bed nights,
move into PSH, cut the number of shelter beds in half and reconfigure the whole system
to be more customized, more geared to facilitating rapid exit, more resource-full, and
better linked with mainstream resources.
4. Make the HMIS work. Make it work for providers assisting individual clients, for
program directors trying to track performance, for TCP to answer simple questions
simply, and for policy makers so they can get a straight answer quickly and accurately.
Make it capable of being used to assess program performance. A key element in
achieving this goal is to open the system to a greater or lesser degree.
1. MOVE CHRONICALLY HOMELESS PEOPLE INTO PSH
The Mayor is already committed to producing the 2,500 units of PSH needed to accomplish this
goal. The PSH Work Group of the ICH is already established and working on how to implement
this commitment. But the work is just beginning. Stakeholders representing different agencies
and interests have to learn each other’s
Writing RFPs to Get What You Want
languages and reach agreement on what they
are doing, separately and together. To reach the At a recent PSH Work Group meeting, DHCD staff said
“We can put out an RFP asking for developers to build
scale of 2,500 new PSH units will probably
for homeless people, among other populations. But we
require bringing in new developers and service
can’t control what comes in, so if we don’t get any
providers and thinking in new ways. The
proposals for PSH, we have to fund what we do get.”
example to the right illustrates one new
Not true.
thought; many others are needed.
In another community, the housing and community
development agency realized it would never reach its
After thinking about definitions of PSH,
PSH goal if it didn’t change its RFPs. Two things it tried,
among others, were (1) giving proposals many extra
chronic homelessness, and similar terms, the
points for including PSH, making it more likely they
PSH Work Group issued its draft plan on April
would get funded, and (2) every project was required to
2, 2008. Pretty soon the whole group is going
devote a minimum percentage of its new units to PSH,
regardless of who else would live in the other units.
to have to focus on the three sides of the action
Both worked - the number of PSH units expanded, and
that produce an occupied PSH unit—who is
new developers were brought into the “PSH business.”
going to build it, run it, support its tenants with
services; how are the right people going to get
into the units; and who is going to pay for it. With respect to the first issue, some communities
have launched very successful “matchmaking” activities to create partnerships between
developers/housing operators and supportive service providers, helping both partners to learn to
work together. These can be done in different ways—Connecticut and Los Angeles have done it
using training seminars; Portland, Oregon has done it more informally, but all have brought in
new players.
With respect to the second issue, different communities have used different approaches to
identify the group of long-term homeless people who should be the priority population for new
units. The most successful ones we know of have used data from public systems to identify
frequent users—frequent jail users, frequent detox users, frequent emergency room users,
frequent shelter users—and developed pathways to assure that these were the people next in line
for housing. Communities that have combined these strategies—creating PSH and targeting the
frequent users—have seen their street counts go down and are closing emergency shelters.
70
Chapter 5: Implications and Recommendations
71
On the third issue, the PSH Work Group may want to take a look at Seattle/King County’s
Funders Group, which has had phenomenal success in generating new PSH through joint RFPs.
In the process, the participating public agencies (and some foundations) have been able to target
their own funding to maximum effect, knowing that the matching pieces (capital, or operations,
or services funding) will be forthcoming when needed.
Site visits to two or three communities that have successfully mobilized resources for PSH—that
is, have housed chronically homeless people resulting in reduced homeless counts—would be in
order. Representatives of public agencies that would contribute each type of funding—capital,
operations/rent, and services—should participate. Possible locations, other than the “usual
suspects” of San Francisco and New York, include Denver, Portland/Multnomah County,
Oregon, Seattle/King County, and the states of Connecticut, and Minnesota.
2. CREATE A PROCESS TO PRIORITIZE WHO GETS THE NEXT
AVAILABLE PSH UNIT
Once new PSH units exist, there is the question of who will be invited to occupy them, and how
those people will be chosen, contacted, and assisted to take possession. The Mayor has explicitly
committed his administration to filling
Prioritizing Who Gets the New PSH Units
the 2,500 net new units with chronically
homeless people—that is, people with
Communities have used a variety of mechanisms to assure
the longest histories of homeless and
that the most vulnerable homeless people get new PSH units.
Examples include:
with at least one disability. In other
communities pursuing the goal of ending
• San Francisco—(1) Solicit participation from street
chronic homelessness, new PSH tenants
homeless people, give everyone a number, pick randomly
often have multiple serious disabilities.
from numbers for each new unit; (2) Use focal points for
services to chronically homeless people—community
The challenge is to develop one or more
clinics, emergency rooms, mobile unit—two to three months
mechanisms to identify the relevant
before new building/units will be ready, solicit candidates
people, contact them, and offer them
from these focal points. Offer housing.
housing.
• Portland, OR—From jail records, identify most frequent
users of county jail. Offer housing.
Different communities have used
different approaches to identify the group
• Seattle—From hospital and detox center records, identify
of long-term homeless people who
most frequent users. Offer housing.
should be the priority population for new
• Santa Monica, CA—Identify long-term street homeless
units. Some examples are shown in the
candidates from (1) police, EMS, outreach team, and now
box to the right. The most successful
local resident experience; (2) assessment of vulnerability to
communities we know of do two things.
dying on the streets. Offer housing.
They use HMIS data to identify long• New York—Identify (1) long-term stayers and (2) those with
term shelter stayers, and they use data
high vulnerability to dying on the streets. Offer housing.
from public systems to identify frequent
users of jails, detox, emergency rooms,
• Portland, ME—From knowledge of shelter operators and
police, identify long-term homeless people. Offer housing
and hospitals, developing pathways to
assure that these are the people next in
line for housing. Because frequent users of public crisis services are most likely to be the
chronically homeless people who have lived for years on the streets rather than in shelters, the
combination of these approaches tends to capture both street homeless people and long-term
71
Chapter 5: Implications and Recommendations
72
shelter stayers. Communities using these combined strategies have seen their street counts go
down and are closing emergency shelters. 26
On the subject of creating a prioritizing process through analysis of shelter and public agency
records, the PSH Work Group has a very interested ally in the Criminal Justice Coordinating
Council’s Substance Abuse Treatment and Mental Health Services Integration Task Force
(SATMHSI). SATMHSI membership closely parallels the public agency members of the ICH, so
working together should be easy. SATMHSI members have already done a significant amount of
work concerning homelessness and its relationship to mental health, substance abuse, and the
criminal justice system, and have written a strategic plan outlining how they hope to reach their
goals of more appropriate treatment of the people within their area of responsibility. When they
began their work, homelessness was not explicitly on their agenda, but it soon became clear that
homelessness was a common element in the histories of many of the arrestees, defendants, and
jail inmates who posed the greatest challenges to the system.
At present, SATMHSI is somewhat stymied by each member agency’s inability to see whether
people it serves are also served by other member agencies. Not only is resolving this bottleneck
an essential step in targeting resources to the right people as well as getting each person all the
elements of support that will make a difference, but it will also be necessary if the agencies are to
track the impacts of their interventions with the most frequent users of multiple systems to see
whether they actually reduce service use across the board. The agendas of the PSH Work Group
and SATMHSI have many points in common; together, and with proper legal counsel to protect
individual privacy, they should be able to figure out how to share data in the interest of greater
system efficiency and effectiveness.
The people who are the focus of both groups face many challenges. Most get only piecemeal
help, if they get any help at all. They are “known” to these crisis public systems, but the evidence
for the failure of any one public system to resolve their issues is evident in their repeated use of
the same systems, and at the most expensive levels. Because the person’s issues interact and
cannot be picked apart and addressed one at a time, only when the resources of all the systems
26
At this time (Spring 2008), the Urban Institute is in a position to obtain data from FEMS, DOC, and DMH to
match with each other and with homeless people in the HMIS. We have cleared these procedures through the Urban
Institute’s Institutional Review Board for the protection of human subjects (IRB) and have been in communication
with the relevant agencies, all of which are very interested in cooperating and have begun discussions with their own
IRBs. We will be starting the process of data selection, transfer, and matching as soon as needed agreements are in
place. Thus we will be able to do much of the important work of cross-agency data matching and will be able to
report the extent and nature of the populations that use these four District systems. This in itself is vital information
that will make it possible to estimate cost savings to each department under various scenarios, such as “if 1,000 of
the highest users become stabilized in housing.”
However, we will not be able to identify to each agency the specific people who ought to be the first ones targeted
because to do so would violate the privacy commitments we made in obtaining the data. There are provisions within
the HMIS standards and procedures for identifying a specific person “for the purposes of offering services” the
person would not otherwise receive. Once the scope of frequent use and system overlaps are known from the Urban
Institute analysis, District agencies will have to develop appropriate procedures and safeguards for taking any next
steps, and the Urban Institute IRB will have to approve before names can be released.
72
Chapter 5: Implications and Recommendations
73
work together in an integrated way are they likely to be effective. The first point of effectiveness
is often getting the person into a stable housing situation.
3. RECONFIGURE EMERGENCY SHELTER
Data presented in chapter 2 indicate that almost half of single adults using emergency shelter in
the District use seven or fewer days and do not return. At the other extreme, 4 percent stay for at
least a year and another 10 percent stay at least 181 days; together these 14 percent of shelter
stayers use almost 60 percent of shelter resources. The expectation is that the new 2,500 units of
PSH will succeed in moving this 14 percent out of shelter and into housing, along with
chronically homeless street dwellers who do not use shelter.
The result should be to cut demand for emergency shelter beds in half for single adults. The
District could close half of its emergency shelter capacity and reconfigure the rest. Shelters could
be smaller, more focused, better staffed, more specialized. Intake could be more organized, more
designed to assess what would help people leave homelessness quickly and help them do so.
Public and nonprofit resources to help people get jobs and find housing should be in place. All of
this could begin to work once a person had spent 8 to 14 days in shelter, cumulatively, giving the
very short stayers time to leave on their own, as they are doing now, and focusing shelter
resources on the approximately 2 in 5 shelter users who will neither leave on their own within 7
days nor end up staying forever.
Once PSH development is under way and procedures are in place to identify the long-term
homeless people and frequent system users who are the target population for that PSH, the ICH’s
Strategic Planning Committee should set up a work group on “Emergency Shelter
Transformation” to parallel the PSH Work Group. The job will be no less challenging than that
of the PSH Work Group, but it needs to be done. Based on our conversations with them, current
providers of emergency shelter will welcome this challenge, as they are not happy with their
inability to offer effective help to the people they currently shelter. At the same time, different
service providers may be able to join in the work once shelters are smaller and more focused.
Reconfiguring the emergency shelter system also leaves the way open to set up some
performance expectations for the various types of shelter that will emerge. It is important to
remember that if the District expects to set performance standards for providers and to make
future funding contingent on meeting or exceeding standards, it must also come to terms with the
need to expect something from shelter users. We discussed this issue in our first report for this
assessment. The District will have to revisit the whole concept of “low barrier”; shelters could
still continue to admit people without demand or restriction, but what happens later would have
to be reconsidered. Even, or especially, in Housing First permanent supportive housing, which is
low barrier for entry, staff are very active in their work to engage tenants in activities that will
help them stay in housing. Staff in low barrier shelters should be no less persistent and
persuasive, but for them to be so, there has to be a far higher staff to client ratio than exists now.
For a more complex emergency shelter system to work, there would also have to be assessment,
triage, and significant levels of casework. Likewise, the service resources would have to be
available and accessible to address issues of employment, substance abuse, mental illness, and
violence, at least. The Emergency Shelter Transformation Work Group would have the task of
73
Chapter 5: Implications and Recommendations
74
deciding whether casework resources should be attached to particular programs, as they are now
(to the extent that they exist) or whether they should be placed in an independent casework
agency that could continue to serve individuals and families as they move between one program
and another, and provide continuity for some time after they return to housing. Setting up case
management for all emergency shelters in a separate program has appeal for several reasons. It
would be especially useful if the case management program had at its disposal a reasonable
amount of flexible “do whatever it takes” funding, as well as having the connections to help
people access mainstream benefits and services. It could use the “do whatever it takes” resources
to help people while they are waiting to qualify for public benefits, when they need services but
do not qualify for regular public benefits, or when public benefits do not cover particular services
the client needs.
4. MAKE THE HMIS WORK
For everything to work—all the changes recommended for finding and housing chronically
homeless adults and families, plus those for changing the emergency shelter system—the District
will have to have a more flexible, useful, and open homeless management information system
than now exists.
Throughout reports for this evaluation, there have questions about homeless people or homeless
assistance programs that we could not answer because the District’s homeless management
information system either did not have the relevant data or could not externalize the data it did
have in a way that made analysis possible. We trust the reader to remember the various data
problems we encountered, and will limit ourselves here to listing suggestions for improvement.
OVERALL SUBSTANTIVE SYSTEM CHANGES
1. Require shelters to collect more information on anyone using any shelter other than for
hypothermia. At a minimum, assure that the records contain name, race/ethnicity,
age/date of birth, gender, and the presence or absence of a disability. These are HUD
minimum standards for emergency shelter data.
2. For people using shelter more than X days (pick a period between 7 and 14 days), require
contact with a case worker and a completed assessment, followed by efforts to link
people to resources, including for employment, vocational rehabilitation, health, mental
health, and substance abuse needs. Put the information into the HMIS.
3. Incorporate into the HMIS information from outreach programs about the people they
work with on the streets.
4. Bring providers and advocates together to work through the issues involved in opening
the HMIS. Different communities have resolved these issues in different ways, from
“barely open,” which would be enough to establish a unique system ID number but not
much else, to “completely open,” which means that staff of one homeless assistance
agency can see whether someone has already used one or more resources in the system
and some basic information about the person. Completely open systems still only share
their information within the homeless assistance network, not with public agencies. It is
74
Chapter 5: Implications and Recommendations
75
time to have this discussion in the District, with the intent of moving from current
practice to one or more levels of openness.
Many CoCs have open systems (e.g., Cincinnati, many communities in Michigan),
including several of the District’s immediate neighbors (the Montgomery County, Prince
George’s County, and Southern Maryland CoCs). District providers could easily see how
these work in practice, what safeguards are in place to protect privacy, advantages of
providers’ open access to information to individuals and families seeking help, and other
aspects of how these systems work. At least one large provider agency with programs in
the District and also in surrounding counties with more open HMISs is already
sufficiently convinced of the advantages of an open system that it has asked TCP to
change the District’s HMIS for its own programs.
A more open HMIS should be able, at a minimum, to do the following: Program A enters
a person’s name and basic information into the HMIS for someone who is using Program
A for the first time. The HMIS searches existing records for a match. If it does not find a
match, it assigns the new person a unique system ID number. If it finds a match, it
conveys the already-assigned system ID number to Program A, which then uses that
unique system ID for the person just entering Program A.
5. Look into the benefits of using a bed management approach to emergency shelter use,
rather than the current practice in emergency shelters for single adults of entering and
exiting each person every night he or she uses a shelter. In a bed management approach,
the person requests shelter, completes an intake process, and is entered into the HMIS.
Every night thereafter that the person comes back to the shelter, the shelter operator notes
a night of shelter use in the person’s record. If the person fails to return anywhere in the
system for at least 30 days, the system automatically exits the person as of the last date of
use.
6. Develop a mechanism that gives adult homeless program users (single adults and those in
families) the option of participating in a multi-agency service team, which would involve
making their personal records accessible to specific service providers on their personal
team from which they would then receive services to help them leave homelessness. The
mechanism should meet HIPAA and other privacy requirements, which involve informed
consent for each specific agency to be involved in sharing data for that individual or
family.
TECHNOLOGICAL IMPROVEMENTS
7. TCP desperately needs software or programming improvements that let it “pull” data
from the underlying HMIS database easier and faster. The District is probably stuck with
ServicePoint, the software program that runs the local HMIS and the HMISs of about 80
percent of all CoCs in the country. But people elsewhere have figured out how to make it
work better, including using XML to extract data for analysis. A major impediment to
analysis for the District is the absence of a unique system ID, which is a consequence of
how TCP and local providers have implemented the concept of a “closed” HMIS. Even
with the more open system that we recommend, however, there will still be analysis
problems until and unless TCP’s analytic capacity is strengthened.
75
Chapter 5: Implications and Recommendations
76
8. TCP should obtain and use a statistical analysis software packages such as SPSS or
STATA (see chapter 2). These programs will greatly facilitate analysis once the data to
be analyzed have been extracted from the ServicePoint database. As just noted, however,
major improvements are needed in that extraction process—getting Bowman Systems to
write an XML program for the District is probably a good part of the answer.
PERFORMANCE MONITORING AND PERFORMANCE-BASED CONTRACTING
9. Four things are needed before the District will be able to move in the direction of
performance-based contracting:
a. There is a very big philosophical issue that will need to be addressed before it will
be worthwhile to work on designing a set of expectations for what providers
should do and secondarily for how data on performance will be recorded and
reported. That philosophical issue is, “What does the District expect of the
homeless people who use its homeless assistance programs?” If providers are not
permitted to place expectations on the people they serve, the District cannot place
expectations for performance on the providers. At present District government
seems content to expect nothing of people who use its emergency shelters (people
using transitional and permanent supportive housing programs must comply with
the rules of their specific program, which can vary widely). Clearly if the District
is going to transform its emergency shelter system as suggested above, it will
need to modify its current low expectations for making an effort to leave
homelessness.
b. Agreement among providers and other stakeholders on serious performance
expectations for programs of different types. Currently performance standards are
set by each provider for itself, they are set far too low (based on preliminary
performance data in the HMIS), and they are not developed by providers offering
similar programs that know the issues involved in meeting a particular level of
performance. As more is expected of emergency shelter programs (per
recommendation 3), realistic expectations for performance need to be in place.
i. As part of this work, District officials and homeless service providers will
have to agree on the performance domains to be measured. Obvious
domains are housing and employment. Other domains may be reduced
substance abuse, reduced psychiatric symptomatology, improved physical
functioning, reconnecting with family, getting children back from foster
care, and the like.
ii. They will also have to agree on what will count as an indicator of
improvement/progress/positive outcome for each domain and how they
will measure and record relevant indicators.
iii. Data fields for recording status on relevant indicators will have to be
incorporated into the HMIS.
76
Chapter 5: Implications and Recommendations
77
iv. Providers will have to assess progress periodically and record the results
in the HMIS.
c. Better data collection about program user circumstances at program entry, and
much better data collection about circumstances at program exit, especially for
emergency shelter programs (transitional and permanent supportive housing
programs that report to the HMIS already report exit status for most of their
clients).
d. Agreement that it is okay, after a certain period of time, for providers to expect
program users to work with program staff toward the end of leaving
homelessness. Some shelters may still remain “low barrier,” meaning nothing is
required at entry. But after a person has used a program for a while, staff should
be making a concerted effort to involve that person in services (see number 2,
above) and there should be sufficient staff to do so. Given that the person will still
be able to use emergency shelter even if he or she refuses services, it takes a
special approach to induce some people to accept services, but that approach must
be made, and made persistently. Staff should support each other in thinking
through different approaches and in analyzing what has and has not worked. PSH
programs using a Housing First/voluntary services approach have developed
considerable skill in bringing people into services; emergency shelter workers can
learn from these examples.
CONCLUSIONS
The recommendations offered here will require a serious organizational effort on the part of the
District government and stakeholders in the District’s homeless assistance network. The
organizational structure for implementing Recommendations 1 and 2 is the PSH Work Group of
the ICH’s Strategic Planning Committee, which is already in place and working. It concentrated
first on establishing the ground rules for PSH development and the mechanisms that will carry
through until 2,500 units are in place. Along the way it will have to address the issues raised by
our Recommendations 1 and 2, which identify the people who should be the occupants of those
units and the ways those prospective occupants should be recruited.
Recommendation 3 calls for a second organizing Work Group to address the issues that will be
involved in reconfiguring the emergency shelter system. As already noted, the ICH’s Strategic
Planning Committee is probably the proper home for this committee, working in parallel with the
PSH Work Group. Implementing Recommendation 4 will require a Work Group of its own, the
HMIS Work Group, possibly under the aegis of the ICH’s Operations Committee.
To a very real extent, the successes of the PSH and Emergency Shelter Transformation Work
Groups will depend on how well and how quickly the HMIS Work Group does its job.
•
The PSH Work Group will need the HMIS to be able to identify long-term shelter
stayers, and to integrate its data with DMH, FEMS, and DOC data to find the homeless
people who are frequent users of other public crisis services.
77
Chapter 5: Implications and Recommendations
78
•
Ideally, the HMIS would also be the vehicle for tracking PSH project performance in
retaining people in housing, and for recording the services they receive and determining
what seems to work best.
•
The Emergency Shelter Transformation Work Group will need good HMIS data to help it
understand the subgroups of people using shelter who stay more than 7 but fewer than
180 days so it can design appropriate specialized shelters.
•
A reconfigured emergency shelter system will need the HMIS to maintain records of
shelter stays, assessment information, case management support, service referrals, and
services received, so it can track the performance of different shelters and assess
outcomes (e.g., entered employment, entered treatment, returned to housing) both for
themselves and in relation to services received.
•
Therefore the HMIS Work Group will have to work closely with the PSH and Emergency
Shelter Transformation Work Groups to be sure the changes it will be implementing for
the HMIS meet the needs of those two work groups.
•
The HMIS Work Group will have to orchestrate discussions of “opening” the system to at
least some degree, with the goal of achieving a system that is useful for many purposes.
•
The HMIS Work Group will also have to orchestrate discussions of performance
expectations for providers and homeless people, which outcomes to hold which programs
accountable for, how to measure those outcomes, and how to report them. These
discussions will have to begin with data showing baseline performance, which TCP is
just beginning to be able to produce. The result of these discussions should be a coherent
set of shared expectations for program performance similar to those used in Columbus,
Ohio and included as Appendix A of our first report.
78
References
79
REFERENCES
Bass and Howes, Inc. 2000. “Assessment of the Community Partnership for the Prevention of
Homelessness.” Washington, DC: Report to the Deputy Mayor for Children, Youth, and
Families, November 8, 2000.
Burt, M.R. 2002. “Homelessness and the Homeless Assistance Network in the District of
Columbia and the Role of the Fannie Mae Foundation’s Help-the-Homeless Program.”
Washington, DC: Urban Institute. Prepared for the Fannie Mae Foundation, May 2002.
Burt, M.R. 1995. “Assessment of the D.C. Initiative’s First Year. Prepared for the Community
Partnership for the Prevention of Homelessness.” Washington, DC: Urban Institute.
Burt, M.R. and S. Hall. 2008a. The Community Partnership and the District of Columbia’s
Public Homeless Assistance System. Washington, DC: The Urban Institute. Available at
www.urban.org.
Burt, M.R. and S. Hall. 2008b. Major Recommendations: Summary Report of the Urban
Institute’s Assessment of the District of Columbia's Public Homeless Assistance
Assistance System. Washington, DC: The Urban Institute. Available at www.urban.org.
Enterprise Foundation. 1998. “A Report on the Existing Conditions and Opportunities for the
District of Columbia’s Homeless Housing and Service Systems.” Columbia, MD:
Enterprise Foundation. Prepared for the District of Columbia Financial Responsibility and
Management Assistance Authority (“the Control Board”).
Griffin, Patricia and Martha Johnson Mead. 2007. Strategic Plan for Persons with Serious and
Persistent Mental Health and Substance Abuse Disorders Involved in the Criminal Justice
System in the District of Columbia. The Criminal Justice Coordinating Council Substance
Abuse Treatment and Mental Health Services Integration Taskforce. Working Draft.
79
References
80
80
Appendices
81
APPENDIX A: LIST OF PEOPLE INTERVIEWED
D.C. Government
•
Office of the Mayor—Julie Hudman, Laura Zeilinger, Janice Ferebee, Leslie Steen,
Melissa Hook, and Oscar Rodriguez
•
Office of Councilmember Tommy Wells—Adam Maier, Ram Uppuluri, and Yolundra
Barlow
•
Service agencies:
• Department of Human Services (DHS)—Clarence Carter, Kate Jesberg, Fred
Swan, Sakina Thompson, Ricardo Lyles, Jean Wright, Deborah Carroll, Susie King,
Lisa Franklin-Kelly, and George Shepard.
• Department of Mental Health (DMH)—Stephen Baron, Barbara Bazron,
Michele May, and Eric Strassman
• Department of Health’s Addiction Prevention and Recovery Agency
(DOH/APRA)—Tori Whitney
•
Housing agencies:
• D.C. Housing Authority (DCHA)—Michael Kelly and Adrianne Todman
• Department of Community Development (DHCD)—Leila Edmonds and Guyton
Harvey
•
Public safety agencies:
• Fire and Emergency Medical Services (FEMS)—Michael Williams, John Dudte,
Mytonia Newman, and Patricia White
• Metropolitan Police Department (MPD)—Brian Jordan
• Department of Corrections (DOC)—Devon Brown, Reena Chakraborty, and
Henry Lesansky
• Court Services and Offender Supervision Agency (CSOSA)—Calvin Johnson,
Claire Johnson, and Joyce McGinnis
Pretrial Services Agency (PSA)—Susan Schaffer and Virgin Kennedy
•
Other agencies:
• Office of Property Management (OPM—Rick Gersten, Regina Payton, and
Spencer Davis
Providers
•
•
•
•
•
Anchor Mental Health—Peggy Lawrence
Calvary Women’s Services—Kristine Thompson
Catholic Charities—Chapman Todd
Center for the Study of Social Policy—Jim Gibson
Central Union Mission—David Treadwell
81
Appendices
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
82
Coalition for Nonprofit and Economic Development—Robert Pohlman
Coalition for the Homeless—Michael Ferrell and Omega Butler
Community Connections—Helen Bergman
Community Council for the Homeless at Friendship Place—Jean-Michel Giraud
Community of Hope—Kelly Sweeney McShane
Covenant House Washington—Nicole Lee
D.C. Central Kitchen—Robert Egger
D.C. Parent Training and Information Center—Danielle Greene
Downtown Business Improvement District—Chet Grey
Edgewood Brookland Family Support Collaborative—Louvenia Williams
Families Forward—Ruby Gregory and Joi Buford
House of Ruth—Crystal Nichols
Latin American Youth Center—Steve Chaplain
Latino Transitional Housing Partnership—Jarrod Elwell
Local Initiatives Support Corporation—Oramenta Newsome
My Sister’s Place—Karen Fletcher
My Sister’s Place—Nichelle Mitchem and Inga James
N Street Village—Schroeder Stribling
National Alliance to End Homelessness—Richard Hooks Wayman
New Endeavors for Women—Wanda Steptoe
Pathways to Housing—Linda Kaufman
Street Sense—Laura Thompson and Jesse Smith
Urban Living Institute—Urla Barrow
Washington Legal Clinic for the Homeless—Patricia Mullahy Fugere, Mary Ann
Luby, Scott McNeilly, Amber Harding, Marcy Dunlap, and Andy Silver
Wesley Seminary—Ann Michel
Women Empowered Against Violence—Heather Powers
Informal conversations with case managers and shelter managers at:
- Franklin School
- New York Avenue
- Adam’s Place
- 801 East
- New Endeavors by Women
- Blair
- Madison
- D.C. General (Harriet Tubman and the hypothermia shelter)
The Community Partnership—Sue Marshall, Cornell Chapelle, Tom Fredericksen,
Darlene Mathews, Amy McPherson, Michele Salters, Clarence Stewart, Tamura
Upchurch, Mathew Winters, and Xiaowei Zheng
82
Appendices
83
APPENDIX B: LIST OF ACRONYMS
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Advanced Reporting Tool (ART)
Administrative Review (AR)
Americans with Disabilities Act (ADA)
Annual Homeless Assessment Report (AHAR)
Average Length of Stay (ALOS)
Certificates of Occupancy (CO)
Community for Creative Non-Violence (CCNV)
Comprehensive Psychiatric Evaluation Program (CPEP)
Continuum of Care (CoC)
Court Services and Offender Supervision Agency (CSOSA)
Criminal Justice Coordinating Council (CJCC)
D.C. Housing Authority (DCHA)
Department of Housing and Community Development (DHCD)
Department of Corrections (DOC)
Department of Health, Addiction Prevention and Recovery Agency (DOH/APRA)
Department of Housing and Urban Development (HUD)
Department of Human Services (DHS)
Department of Mental Health (DMH)
Emergency Rental Assistance Program (ERAP)
Emergency Shelter (ES)
Emergency Shelter Grant (ESG)
Facilities Management Operations Division (FMOD)
Fiscal Year (FY)
Fire and Emergency Medical Services (FEMS)
Head of Household (HoH)
Homeless Management Information System (HMIS)
Homeless Services Reform Act (HSRA)
Housing Assistance Center (HAC)
Inspector General (IG)
Interagency Council on Homelessness (ICH)
Length of stay (LOS)
Metropolitan Police Department (MPD)
Office of Administrative Hearings (OAH)
Office of Property Management (OPM)
Permanent Supportive Housing (PSH)
Police Services Area 101 (PSA101)
Pretrial Services Agency (PSA)
Request for Proposal (RFP)
Shelter Monitoring Unit (SMU)
Shelter Plus Care (S+C)
83
Appendices
•
•
•
•
84
Substance Abuse Treatment and Mental Health Service Integration Taskforce
(SATMHSI)
Transitional Housing (TH)
Virginia William’s Family Resource Center (VWFRC)
Washington Legal Clinic for the Homeless (WLCH)
84