HOMELESSNESS, DRUG ABUSE, AND SERVICE USE

1
HOMELESSNESS, DRUG ABUSE, AND SERVICE USE
CHALLENGES FOR PRACTITIONERS, PROGRAM PLANNERS,
AND POLICY MAKERS
David E. Pollio, PhD George Warren Brown School
of Social Work
2
CHANGING POPULATION

• ? Early 20th century
• ?Hoboes, bums, winos
• ?Middle aged alcoholic men

? 1980s ?Deinstitutionalized mentally ill
? Famous descriptions ?75 schizophrenic
(Torrey 1986) ?50 schizophrenic, 100 mentally
ill (Lipton et al 1983) ?One-third mentally
ill
3
(No Transcript)
4
(No Transcript)
5
(No Transcript)
6
(No Transcript)
7
(No Transcript)
8
OVERVIEW OF PRESENTATION
A. ST. PATRICK STUDY B. SUNCODA METHODS
SUBTUDIES CHANGE OVER DECADES MODELING SERVICE
ACCESS LONGITUDINAL SERVICE USE C. TOWARDS THE
FUTURE
9
SERVICE USE OVER TIME AND ACHIEVING STABLE
HOUSING IN A MENTALLY ILL HOMELESS POPULATION
10
(No Transcript)
11
SERVICE READINESS THEORY
? Individuals initially obtain services to meet
survival needs

• ? Favorable outcomes occur after
• 1) readiness to change
• 2) services are available

? Service use increases before change and
decreases afterward
12
SERVICE READINESS THEORY

• ? Change occurs at strategic moment

? If change fails to occur, service use returns
to its previous level and purpose (survival)
? Service use, like homelessness, tends to be
cyclical
13
THEORETICAL MODEL FOR SERVICE USE AND OUTCOMES
14
TESTING THE MODEL
Data collected at St. Patrick Center comparing
service use by two groups
? Housed group (n58) Individuals
maintaining housing for 24 consecutive months

• - versus -
• ? Unhoused group (n55)
• Next unhoused service user selected from
  the agency

15
SERVICE USE

• ? Monthly service use data collected for both
  groups starting the month the housed group
  achieved housing, proceeding through next 24
  months

? Utilization of drop-in center, counseling,
health, and general services compared between the
two groups
? Comparison on total post-housing period
separate analysis of pre-housing and immediate
post-housing periods -- different patterns
hypothesized
16
COMPARISON OF TOTAL SERVICES USED BY GROUPS
ATTAINING AND NOT ATTAINING HOUSING
17
2-PHASE MODEL HOUSED VS. UNHOUSED
18
2-PHASE MODEL FOR DROP-IN CENTER USE, COUNSELING,
AND HEALTH SERVICES HOUSED ONLY
19
RESULTS

• ? Data support model for general service use,
  drop-in center, and counseling use

but not for health services
? Findings suggest benefit may be gained by
facilitating broad service use among homeless
populations
20
SERVICE IMPLICATIONS

• Findings suggest need for
• ? Multi-phase intervention with changing
  intensity
• ?Lower in initial stages when relationships
  established
• ?High at strategic moment
• ?Lower during consolidation of gains
• ? Relation-building services (e.g., counseling or
  drop-in center) in services package
• ? Repeated opportunities for service engagement

21
Service Use, Needs, Costs, and Outcomes for the
Drug Abusing Homeless Population
SUNCODA STUDY 1998-2003
FUNDED BY NIDA GRANT 10713
22
RATIONALE

• ? Research has not examined complex simultaneous
  interactions among
• ? demographics
• ? homelessness
• ? mental illness/drug use disorders
• ? services

? Understanding multiple interrelationships in a
more comprehensive model will ultimately improve
our ability to deliver effective services to the
homeless population
23
SUNCODA HOMELESS CONCEPTUAL MODEL
24
S ervice U se, N eeds, C osts, and O utcomes
for D rug A buse in homelessness
? Following a homeless sample (N397) for 24
months
? Random recruitment from shelters and street
locations
? Interviewed annually for substance use, mental
illness, service utilization, and other key
outcomes
? Interviewed every 3 months for intermediate
outcomes
25
Service Use, Needs, Costs, and Outcomes for Drug
Abuse in homelessness
? Collecting service use data from 35 participant
agencies using MIS and other standardized
collection methods
? Collecting agency organizational data
? Costing services for each participant agency
26
(No Transcript)
27
WHERE WE ARE NOW
? Baseline interviews completed (N397)
Service Use, Needs, Costs, and Outcomes for Drug
Abuse in homelessness
? 1-year follow-up interviews ongoing (N275)
? First year service data collection completed
(N385)
? Cost methods finalized
? Agency cost data collection ongoing
28
SAMPLE CHARACTERISTICS
Demographics Male 76 Nonwhite
82 Age (median years) 42
Education (median years) 11.7 Ever
married 46

• Homelessness
• shelters in last year 1.9
• shelter nights in last year 119
• Median age first shelter 34
• Median lifetime years homeless 2.5

.
29
PSYCHIATRIC DISORDERS
60
50
40
Percentage
30
20
10
0
30
SUBSTANCE USE DISORDERS
80
70
60
50
Percentage
40
30
20
10
0
31
METHODS SUBSTUDIES
32
TEST-RETEST ANALYSES IN DRUG USERS AND NONUSERS
HOMELESS SUPPLEMENT TO THE DIAGNOSTIC INTERVIEW
SCHEDULE
33
?PURPOSE
? To determine reliability, esp. for substance
dependent
?METHODS
? Convenience sample (N51) from 2 homeless day
shelters
? Instruments Diagnostic Interview
Schedule/Homeless Supplement CIDI Short Form
(DSM-III-R) ? Test-retest 1-7 days apart (mean,
1.7 SD,1.3) ? Independent interviewers
? Data analysis ?Categorical data kappa
(compared with ?2 tests) ?Continuous data
intra-class correlations (ICCs)
34
SECTIONS OF THE HOMELESS SUPPLEMENT
? RESIDENTIAL HISTORY
? COURSE OF HOMELESSNESS
? SHELTER USE
? TRANSIENCE
35
RESIDENTIAL HISTORY
?RESULTS
Where have you stayed overnight in the past 12
months?
36
COURSE OF HOMELESSNESS
?RESULTS
37
?RESULTS
SHELTER USE
38
?RESULTS
TRANSIENCE
39
?SUMMARY/DISCUSSION
? Poor to excellent reliability ? Substance
dependent subjects less reliable on only 4
variables (more reliable on 1 variable)
Modifications
? Corrected the poor questions ? Designed methods
to anchor homelessness history - e.g.,
calendars, memory aids
40
COMPARING METHODS OF DRUG USE DATA COLLECTION
SCREENER VS. URINE TEST VS. CIDI-SAM
41
PURPOSE
To compare substance use self report with urine
testing in a homeless population sample
42
METHODS
Comparison data ? Screener Are you presently
using/abusing alcohol or drugs? ? Urine testing
(cocaine, heroin, amphetamines, cannabis,
EtOH) ? CIDI/SAM DSM-III-R diagnoses
(current/last 2 wks)
Data analysis ? T-tests, ?2, kappa, Yules Y
43
SCREENER VS. CIDI/SAM
44
RESULTS SELF REPORT (SCREENER) VS. URINE TESTING
45
LIMITATIONS
? Variable time frames (current vs. last 2
weeks) ?thus CIDI/SAM data not compared to
urine test data
? Urine drug testing limited to cocaine, heroin,
amphetamines, cannabis, EtOH
? Most drugs tested detectable in urine only for
2-3 days EtOH poorly detected
? Severity of substance abuse not considered
46
SUMMARY/DISCUSSION
? CIDI/SAM vs. screener - agreement ?good for
drugs (except poor for cannabis) ?fair for EtOH
? Unknown additional substance use may be
undetected by self report and urine test in this
study
47
CONCLUSIONS
Detection of drug use ?test the urine (self
report adds little)
Detection of alcohol use ?not necessary to test
the urine
48
TRACKING
(How do you track HOMELESS people?)
49
SUNCODA approach to tracking - exhaustive
creative ? Future contacts forms - eg,
girlfriends phone, favorite bar ? Photo IDs ?
Reminder letters - to subjects and their
contacts ? Calendars ? Networking and extended
agency contact ? Barbecues ? SUNCODA drop-in
center ? Open house events ? Lotteries ? Monthly
check-in (paid) ? Visits to shelters and day
centers ? Searches of databases (e.g., death
certificates) ? Posters at bus stops, agencies ?
Hanging out - street routes, public areas,
hangouts that were reported at baseline
interviews (plasma/blood donor sites, recycling
centers, areas where clothes/food are given out,
library, museums, zoo, etc.) city festivals

50
(No Transcript)
51
SUNCODA TRACKING STATISTICS Follow-up interviews
completed ?Wave 1 (12 month) - 71 (275 of
386) ?Wave 2 (24 month) - 75 (60 of 80 due so
far) 85 of baseline sample tracked to
date Others 6 deceased 2 in prison long term 3
too ill 4 dropped out of study

52
SUBJECTS FOUND BY TRACKING METHOD
30
25
Wave 1
Wave 2
20
15
Percent of subjects found
10
5

0
unknown
unknown
saw poster
saw poster
letter to subject
letter to subject
located on street
located on street
subject dropped in
subject dropped in
subject called project
subject called project
phone call to subject
phone call to subject
phone call to contact
phone call to contact
located at shelter/day center
located at shelter/day center
53
CHANGE OVER DECADES
54
?PURPOSE
?BACKGROUND CHANGE OVER TIME
Assumption the homeless population is static
over time
Limited evidence for this assumption ? No
longitudinal data available ? Comparison of
studies across time limited by inconsistent
definitions, sampling, focus of study, and
instruments of measure

• To examine rates of psychiatric disorders in the
  homeless population in 1980, 1990, and 2000

55
3 CROSS-SECTIONAL STUDIES USING SYSTEMATIC,
RANDOMLY SAMPLED HOMELESS ADULTS IN METRO ST.
LOUIS AREA
YEAR 1980 1990 2000
STUDY ECA Homeless Health Survey SUNCODA
DATES April 1981 ?July 1982 April 1989 ?April 1990 Oct 1999 ?Feb 2001
N 1395 ? 828 ? General population 300 ? 600 ? Homeless 98 ? 298 ? Homeless
Sample Residential (69 ? 81 ? homeless) Shelters streets Shelters streets
Homelessness definition - Lifetime - 1. gt1 mo. unplanned travel 2. gt1 mo. no regular abode (from DSM criteria for ASP) - Current - no stable residence, living in public shelter or on streets without regular mailing address x 30 days - Current - no stable residence, living in public shelter or on streets without regular mailing address x 30 days
56
?DATA COLLECTION
1980 1990 2000
ECA Study Homeless Health Survey SUNCODA
? Diagnostic Interview Schedule for DSM-III (lifetime diagnoses) ? Diagnostic Interview Schedule for DSM-III-R (lifetime diagnoses) ? Diagnostic Interview Schedule for DSM-IV ? Composite International Diagnostic Interview/ Substance Abuse Module for DSM-III-R (lifetime diagnoses)
57
RESULTS
?DEMOGRAPHICS
Sample 1980 1990 2000
Study ECA Homeless Health Survey SUNCODA
N 150 900 397
Male 54 67 (predetermined gender ratio) 76
African-American Caucasian Other race 46 (oversampled) 53 1 75 22 3 74 18 8
Mean age (SD) 35 (15) 34 (11) 42 (11)
58
RESULTS
?DIAGNOSES
plt.05 plt.01 plt.001
59
plt.05 plt.01 plt.001
60
(No Transcript)
61
SUMMARY
3 St. Louis epidemiologic studies, 1980, 1990,
2000 with data on psychiatric disorders in
homeless people
?Alcohol abuse/dependence - ?
increasing still more prevalent in ?
(60) ?Drug abuse/dependence - steep increase, ?
? ?Antisocial personality disorder -
flatline ?Major depression - increasing, ?
? ?Bipolar affective disorder - brisk increase in
last decade ?Schizophrenia - increasing slowly in
?, still not common
62
?Implications for services to the
homeless ? Services based on outdated prevalence
rates will be wrong for the populations
needs ? Increasing psychiatric and especially
drug abuse disorders would indicate increasing
need for mental health and drug abuse services
(particularly for women)
?Speculation ? Are increases in drug use
disorders an unintended byproduct of public
policies such as zero tolerance? - flooding
the existing homeless population with new drug
addicted individuals
63
MODELING SERVICE ACCESS
64
SERVICES USED BY CATEGORY
65
PURPOSE

• ? To model self-reported service access in a
  homeless population across multiple service
  sectors, testing three conceptual models of
  service use

66
DATA ANALYSIS

• ? For each service sector, separate logistic
  regression analyses performed with variables
  entered in blocks based on conceptual model
• ? Tests calculated for significant improvements
  in models
• ? A final logistic regression model run for each
  services sector including all variables, then
  tested for improvement over best conceptual model

67
Conceptual Model 2 Services predicted by needs
and prevalence
Conceptual Model 1 Services predicted by needs
Conceptual Model 3 Services predicted by needs,
prevalence, and other services
aimed at attaining stable housing
subsistence while homeless (e.g., soup kitchens,
shelters)
lifetime episodes total years
major depression, bipolar disorder, schizophrenia,
panic disorder, generalized anxiety, ASP
inpatient outpatient
alcohol, cocaine, cannabis, other drug
inpatient outpatient residential
sex race
68
Significant pathways between any variable and
service sector (lifetime)
Note Pathways are for analyses of Model 3
69
Significant pathways between any variable and
service sector (30 days)
Note Pathways are for analyses of Model 3
70
SIGNIFICANT VARIABLES BY SERVICE SECTOR (LIFETIME)
Numbers in parentheses are odds ratios
71
SIGNIFICANT VARIABLES BY SERVICE SECTOR (30 DAYS)
Numbers in parentheses are odds ratios
72
SO WHICH CONCEPTUAL MODEL IS BEST (lifetime)?
In all analyses of Model 3 compared to model with
all available variables, improvement was not
significant for any service sector
73
SO WHICH CONCEPTUAL MODEL IS BEST (30 day)?
In all analyses of Model 3 compared to model
with all available variables, improvement was
not significant for any service sector
74
SUMMARY

• ? Types of service access best predicted by type
  of need, prevalence factors (race, gender, and
  comorbidity), and other service use (Model 3)
• ? Psychiatric diagnosis (MH CD) predicted use
  of relevant services
• ? Use of homelessness services not predicted

75
SERVICE IMPLICATIONS

• ? Services should facilitate appropriate
  cross-sector use at the consumer level
• ? Service provision in the homelessness sector
  should be more needs-driven
• ? Service providers should attend to potential
  barriers (e.g., shelters requiring workers to be
  in at certain hours)

76
ALTERNATIVE HYPOTHESIS

• ? Homeless individuals may use housing and
  subsistence services not according to need but
  rather when they are convenient and available

77
LONGITUDINAL SERVICE USE AND DIAGNOSIS
78
PURPOSE

• ? To examine the impact of diagnoses on
  agency-generated longitudinal service use across
  multiple service sectors

79
SERVICE USE

• ? Total service units in 12-month period
  aggregated in monthly increments
• ? Service Units defined naturalistically
• ? Five types of services
• 1) Shelter services ( of nights)
• 2) Inpatient substance abuse ( of nights)
• 3) Outpatient substance abuse ( of direct
  contact hours)
• 4) Inpatient mental health ( of nights)
• 5) Outpatient mental health ( of direct contact
  hours)

80
DATA ANALYSIS

• ? For each of 5 service types, 12-month totals
  compared between those with diagnoses and those
  not meeting criteria
• ? Kruskall-Wallis (non-parametric sum-rank) tests

81
DIAGNOSES (12-MONTH)

• ? Psychiatric diagnoses
• ? Major depression, mania, schizophrenia
• ? Substance abuse/dependence (SA/D) diagnoses
• ? Cocaine, cannabis, opioid, alcohol
• ? Aggregate variables for
• ? Mood/psychotic Dxs
• (depression, mania, schizophrenia)
• ? Any diagnosis

82
SERVICE USE used in past 12 months
83
MEAN SERVICE UNITS PER SECTOR
units of use
84
SERVICE USE AND ANY DRUG ABUSE/DEPENDENCE
DIAGNOSIS
units of use
All relationships significant at p gt .05
85
SERVICE USE AND MOOD/PSYCHOTIC DIAGNOSIS
units of use
All relationships significant at p gt .05
86
SERVICE USE AND SPECIFIC DRUG/ALCOHOL DIAGNOSES

• Cocaine diagnosis (n146)
• ?less shelter use
• ?greater outpatient and inpatient SA
• Cannabis diagnosis (n48)
• ?no differences
• Opioid diagnosis (n12)
• ?less shelter use
• Alcohol Diagnoses (n173)
• ?no significant differences

87
SERVICE USE AND SPECIFIC MOOD/PSYCHOTIC DIAGNOSIS

• Schizophrenia diagnosis (n31)
• ?greater outpatient MH
• Mania (bipolar) diagnosis (n54)
• ?greater outpatient MH
• Major depression diagnosis (n112)
• ?less shelter use
• ?greater outpatient MH

88
QUARTERLY SERVICE USE MOOD/PSYCHOTIC DIAGNOSIS
AND SHELTER SERVICE USE

units of use
plt.05 for quarter
89
QUARTERLY SERVICE USE MOOD/PSYCHOTIC DIAGNOSIS
AND OUTPATIENT SERVICE USE
60
50
40
units of use
30
20


10

0
Q1
Q2
Q3
Q4
SMI Dx/OMH
No SMI Dx/OMH
plt.05 for quarter
90
DISCUSSION

• ? Similar to the modeling data, amounts of
  service use is related to diagnosis, both for
  substance abuse and mental health services.

91
DISCUSSION

• ? Shelter findings somewhat indeterminate

- data unable to separate between individuals
getting housed (decreasing need) and those
expelled or departing (changing accessibility)
92
TOWARDS THE FUTURE.
93
TOWARDS THE FUTURE
CONTINUING RESEACH

• Completion of SUNCODA data will allow us to
  examine the interrelationships among longitudinal
  service use, needs, and outcomes over time.
• Wave I follow-ups on SUNCODA will be completed in
  May 2002, Wave II in May 2003.

94
TOWARDS THE FUTURE
CONTINUING RESEACH

• Cost-per-service data are being generated for all
  service providers, allowing extensive examination
  of costs over time
• Data on subset of SUNCODA families has been
  collected (M. Polgar, PI) and are being analyzed

95
TOWARDS THE FUTURE
GRANT APPLICATIONS

• St. Louis City Needs Assessment
• Short-Term Assertive Community Treatment
  (ST-ACT)
• Towards a Spatial Explanation of Homeless Service
  Use
• SUNCODA Continuation

96
ST. LOUIS CITY NEEDS ASSESSMENT

• Purpose To conduct a service needs assessment
  for St. Louis DHHS
• Relation to previous research SUNCODA Data,
  Decades Paper Results
• Data include
• SUNCODA prevalence data
• Focus groups of stakeholders (providers,
  participants, community members)
• Current Status Ongoing grant from St. Louis DHHS

97
SHORT-TERM ASSERTIVE COMMUNITY TREATMENT

• Purpose To implement a stage-based model of
  ACT, maximizing treatment efficiency for a
  comorbid MI/DA homeless population
• Relation to previous research Builds on St.
  Patrick study
• Community partners Behavioral Health (BJC), St.
  Patrick Center
• Status Under development

98
ST-ACT MODEL
Pre-engagement/ Engagement Stage
Strategic Moment Stage
Outcome achieved
Consolidation Stage
Drop-In Services
Treatment Readiness group
Community Resources Appropriate Aftercare
Brief-ACT
99
TOWARDS A SPATIAL EXPLANATION OF HOMELESS SERVICE
USE

• Purpose To test whether homeless service use
  over time can be predicted by convenience of
  service access
• Relation to previous research Alternative
  explanation to modeling results
• Data include SUNCODA monthly service data, GIS
  system mapping all St. Louis agencies
• Current Status Under review NIDA (February 2002)

100
SUNCODA II

• Purpose
• ? To follow the SUNCODA sample over 5 more years
• ? To recruit additional homeless individuals and
  a comparison sample of sociodemographically
  equivalent housed individuals
• ? To study not only impact of drug abuse on
  homelessness but also impact of homelessness on
  drug abuse
• Current Status Under review NIDA (February 2002)

101
TOWARDS THE FUTURE
MANUSCRIPTS UNDER REVIEW

• Pollio DE, North CS, Foster DA, et al. A
  comparison of agency-based and self-report
  methods of measuring services across an urban
  environment by a drug abusing homeless
  population.
• North CS, Eyrich KM, Pollio DE, Cottler LB,
  Spitznagel E. The homeless supplement to the DIS
  Reliability and validity results.
• North CS, Eyrich KM, Pollio DE. Are rates of
  psychiatric disorders changing over time in the
  homeless population?
• Pollio DE, North CS, Eyrich KM. Modeling service
  use in a homeless population.
• Cowell AJ, Pollio DE, North CS, et al. Deriving
  service costs for a psychosocial rehabilitation
  clubhouse Case study and methodological
  considerations.

102
CHALLENGES TO PROVIDERS

• ?CREATING SERVICES RESPONSIVE TO UNIQUE
  INDIVIDUAL NEEDS
• ?DESIGNING AND IDENTIFYING SYSTEMS OF CARE
  APPROPRIATE TO CHANGING POPULATION
• ?COORDINATING CARE ACROSS MULTIPLE TREATMENT
  SECTORS

103
PUBLIC POLICY CHALLENGES

• ?IDENTIFYING UNINTENDED BY-PRODUCTS OF POVERTY
  POLICY

will lifetime welfare receipt limits be next?
?CREATING COST-EFFICIENT SYSTEMS OF
CARE ?EXAMINING REALISTIC SOLUTIONS
ending the game of musical chairs
104
(No Transcript)
105
Investigators
Enola Proctor, PhD Michael Polgar, PhD Karin
Eyrich, MSW, MPE
Carol S. North, MD, MPE (PI) Ed Spitznagel,
PhD Linda Cottler, PhD
106
COMMUNITY PARTNERS
ADAPT of Missouri Archway Community Treatment
Center Barnes-Jewish Hospital BJC Behavioral
Health St. Louis Veterans Administration Grace
Hill Neighborhood Services Black Alcohol Drug
Service Information Center Christian Service
Center Haven of Grace Hope House Hopewell MH
Center Metropolitan St. Louis Psychiatric
Center Our Ladys Inn Peter and Paul Community
Services Places for People Queen of
Peace Lutheran Medical Center St. Alexius
Hospital YWCA Phyllis Wheatley Program
Missouri Department of Mental Health St. Louis
DHHS Community Alternatives Salvation Army Family
Haven Salvation Army Harbor Light
Center Salvation Army CSTAR St. Louis Psychiatric
and Rehabilitation Center St. Patrick Center St.
Phillipine Emergency Shelter Sunshine
Ministries St. Louis University Hospital
(Tenet) BREM Catholic Social Ministries Drug and
Alcohol Rehabilitation Good Samaritan Center for
the Homeless Harris House Independence Center New
Beginnings CSTAR United Methodist Ministries
Shalom
107
Any Questions
?