Organizational Integration in the Delivery of Mental
Health Services: Evidence-Based Practice or Holy
Grail
Robert Rosenheck MDProfessor of Psychiatry and Public
Health, Yale Medical School
Senior Associate in Mental Health Services Research, VA New England
MIRECC
Fragmentation of Mental Health Services and Program
Failure The “fragmented service system” is a
universal nemesis in healthcare policy. “Out of the contemporary debate comes one
point on which nearly all parties agree, the need for improved coordination of services. Underlying this consensus is widespread recognition that… programs are fragmented, incomplete and often inefficient. For this reason, they have failed to respond to the…problems of mentally ill persons in the community.”
-- David Rochfort, 1992
Substance Abuse Treatment…
…patients with dual [substance abuse and psychiatric] disorders tend to receive services from one system and not from the other, and they are often excluded from both because of the complicating features of the second disorder (Drake et al., 1998).
…medical care, even screening, is seldom provided as part of substance abuse services and are most often separate and largely uncoordinated (Weisner et al., 2001).
President Bush: New Freedom Commission on
Mental Health “The second obstacle to quality mental
health care [after stigma] is our fragmented mental health service system. Mental health centers and hospitals, homeless shelters, the justice system, and all have contact with individuals suffering from mental disorders…Many Americans fall through the cracks of the current system…And to make sure the cracks are closed, I am honored to announce the Freedom Commission on Mental Health.”
-April 29, 2002
New Freedom Commission on Mental Health: Final
Report“…for too many Americans with mental
illnesses, the mental health services and supports they need remain fragmented, disconnected and often inadequate, frustrating the opportunity for recovery.”
Michael F. Hogan PhD,
Chairman, President’s New Freedom Commission on Mental Health
July 22, 2003
On the other hand… The quest for coordination is “…the 20th
century equivalent of the medieval search of the philosopher’s stone…”
“If we can only find the right formula for coordination, we can reconcile the irreconcilable, harmonize competing and wholly divergent interests, overcome irrationalities…and make hard … choices to which no one will object.”
– Harold Seidman, 1986
Empirical Evidence?
Is there empirical evidence that fragmentation is: Extensive? Harmful to outcomes, morale, or leadership?
Two cross-sectional studies found clients in communities with more centralized MH systems were better served more satisfied (Beiser, ‘85; Milward and Provan, ‘95)
One longitudinal study (ACCESS) found greater interorganizational integration among homeless service providers in 18 communities predicted better housing outcomes after 1 year of case management (but no better clinical outcomes) (Rosenheck et al., 1998)
That’s it (as far as I know)!
Nevertheless…In spite of lack of evidence
of adverse effects of fragmentation there have
been many efforts to correct it by fostering “services integration”.
What is integration?
Dictionary defines integration as “making whole”.
Concept of “bringing together” allows broader range of linkages from making a single whole, to improving communication.
Answer #1. Integration interventions do not
represent …
… novel biomedical treatments; … novel psychotherapies or behavioral
therapies; …interventions based on any model of
psychopathology.
…although they do represent major investments of research and clinical resources.
Answer #2. They do represent…
Organizational interventions Efforts to change in how groups and
workers in an organization: 1) are structured, and 2) interact,
…to improve organizational cooperation or coordination to improve: client access to services client outcomes.
What can “organization” do? Facilitate cooperation and coordination
of action between parties. Prevents “free riding” – let others do the
work – the central problem of social life. “Social capital reduces the cost of
working together: “transaction costs”. BUT cooperation and coordination are
only useful IF there are unrealized synergies or interdependencies.
Civic Culture/Social Capital “Citizens in a civic community are active,
public spirited, equal...helpful, respectful, and trustful towards one another, even when they differ on matters of substance...”
“Social capital refers to features of social organization, such as trust, norms, and networks, that can improve the efficiency of society by facilitating coordinated actions...”-Robert Putnam, Making Democarcy Work, 1993
Social Capital “Like other forms of capital, social
capital is productive, making possible the achievement of certain ends that would not be attainable in its absence... For example a group whose members manifest truistworthiness will be able to accomplish much more than a group lacking trustworthiness.”
James Coleman, Foundations of Social Theory, 1990
Examples Interorganizational Integration
The “Big Three”: RWJ PCMI, Fort Bragg, ACCESS HUD-VA joint initiative for homeless veterans
(1993-). VA-SSA joint outreach initiative (1992-99).
Intraorganizational Collaborative Care for Depression/Alcoholism
(Katon, Simon, Wells) Integrated Mental Health/Medical Clinics (Druss et
al. 2001) Assertive Community Treatment (Stein & Test) ACT Augmentations (DDX, IPS– Drake et al.---)
ACCESS Demonstration for homeless people with serious
mental illness Targeted for homeless people with mental
illness, a population with multiple service needs: Mental health services Substance abuse services Public support payments Housing subsidies and support Primary medical care Employment assistance
Demonstration based on the assumption that system fragmentation impeded access to these services.
Hypothesized Causal Chain
Funds and
technical
assistance
Implement
Integration
Interventions
Improved
access and
outcomes
More
Integrated
System
ACCESS Demonstration: Study Design
2 similar sites in each of 9 states = 18 sites One site per state randomized to receive
$250,000/year to implement 12 integration strategies, 2nd site to be control.
All 18 sites given $500,000 per year to operate ACT team to serve 100 homeless clients with serious mental illness each year.
Four annual cohorts recruited 100 clients each: followed-up at 3, 12 months (n=7,200).
Twelve ACCESS “Systems Integration” Interventions
Local interagency coordinating body
State interagency coordinating body
Co-location of services Systems integration
coordinator position Cross training Interagency
agreements; memoranda of understanding
Pooled, joint funding Uniform client
applications, eligibility criteria, assessments
Interagency service delivery team
Flexible funding Use of special waivers Consolidation of
programs/agencies
2.19
2.6
0.48
0.88
0
1
2
3
4
5
Site Visit Year 4 (Wave 2, 1996) Site Visit Year 5 (Wave 3, 1998)
Experimental Sites
Comparision Sites
Implementation of Systems Integration Strategies:
Experimental vs. Comparison Sites, Years 4 and 5.
Two kinds of integration
System-wide
Project-centered
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
Wave 1 Wave 2 Wave 3
Project Integration
System Integration
Experimental Sites
Comparison Sites0.43
0.50
0.60
0.59
0.65
0.57
0.290.28 0.29
0.27 0.26
Changes in Integration: Experimental vs. Control Sites
0
1
2
BL 3 Months 12 Months
Cohort
Psy
chia
tric
Sym
pto
m I
nd
ex*
Int 1 Int 2 Int 3 Int 4 Cont 1Cont 2 Cont 3 Cont 4
OUTCOMES AS HYPOTHESIZED BY SYSTEMS INTEGRATION ASSIGNMENT AND BY COHORT:PSYCHIATRIC SYMPTOMS*
0
10
20
30
40
50
60
70
BL 3 Months 12 Months
Cohort
Per
cen
t N
ot
Ho
mel
ess
Int 1 Int 2 Int 3 Int 4 Cont 1Cont 2 Cont 3 Cont 4
OUTCOMES AS HYPOTHESIZED BY SYSTEMS INTEGRATION ASSIGNMENT AND BY COHORT: NOT
HOMELESS FOR 30 DAYS
0
1
2
BL 3 Months 12 Months
Cohort
Psy
chia
tric
Sym
pto
m I
nd
ex*
Int 1 Int 2 Int 3 Int 4 Cont 1Cont 2 Cont 3 Cont 4
OUTCOMES BY SYSTEMS INTEGRATION ASSIGNMENT AND BY COHORT:PSYCHIATRIC SYMPTOMS*
0
10
20
30
40
50
BL 3 Months 12 Months
Cohort
Per
cen
t N
ot
Ho
mel
ess
Int 1 Int 2 Int 3 Int 4 Cont 1Cont 2 Cont 3 Cont 4
OUTCOMES BY SYSTEMS INTEGRATION ASSIGNMENT AND BY COHORT: NOT HOMELESS FOR 30 DAYS
Hypothesized Causal Chain: Summary of
Results
Funds and
technical
assistance
Implement
Integration
Interventions
Improved
access and
outcomes
More
Integrated
System
YES YES NO
0
20
40
60
0.17 0.27 0.37 0.47
System Integration
% N
ot
Ho
mel
ess
Not Homeless Linear (Not Homeless)
One Thing We Learned: System Integration and Housing Outcome*
*Housing data are adjusted for differences in client characteristics
r=.51
Figure 1. Model of The Achievement of Independent Housing at 12 Months
Social Capital
Housing Affordability
System Integration
.36a
.27b
.34c
.36d
.08d
.17d
Public Housing
Agency
Other Services
.41c
BASELINE 3 MONTHS 12 MONTHS
Independent Housing .47d
.59d
.31d .92d
.94d .97d
.35d .93d
Unexplained
variance
Lesson Learned? Integration can make a
difference in client outcomes!
The ACCESS initiatives failed to harness it to make a difference in client outcomes.
What next?
Look for a theory. What kind of theory?
Something social/organizational. Something that fixes some problem!
Social Capital Industrial capital=machines Human capital=education Social capital=authority, norms, trust
To take advantage of interdependencies By improving coordination/cooperation Lowering transaction costs.
Mental health examples that “worked”
Interorganizational Integration HUD-VA joint initiative for homeless veterans
(1993-). SSA-VA joint outreach program (1992-99)
Intraorganizational Collaborative Care for Depression (Caton,
Simon, Wells) Integrated Mental Health/Medical Clinics
(Druss et al. 2001) Assertive Community Treatment (Stein &
Test)
Federal Interagency Collaboration to Assist
Homeless Veterans In early 1990s Federal Interagency Council on
the Homeless sought to encourage interagency collaborations.
Two VA projects: Social Security Administration (SSA)-VA
outreach to improve access to Social Security Benefits
HUD-VA Supported Housing (HUD-VASH) to pair VA case managers with HUD section-8 housing subsidies.
HUD-VA Supported Housing
HUD set aside Section 8 housing vouchers (50 per site) for homeless veterans.
Section 8 rental subsidy provides fair market rent less 30% of veterans income.
VA funded 2 case managers/site for maximum 25/1 case load.
RCT comparison of three groups at 4 sites: Case management + Voucher (N=182) Case management alone (N=90) Standard homeless support + referrals
(N=188)
Case Management (CM) Services Delivered in First
Three Months CM+Voucher CM Alone Control
Voucher by 3 Mos. 55% 2% 1%Helped Locate Apt. 44% 26% 9%Apts. CM Visited 2.1 0.6 0.0CM met Landlord 71% 45% 0%Helped Furnish Apt. 37% 22% 4%Vet .Terminated 8% 17% 53%
Outcomes in the HUD-VA Supported Housing Program:Percent Stably Housed (Apartment, Room, or House for 60 Days)
0
20
40
60
80
100
Out
reac
h
Baseli
ne
6 mos
.
1 yea
r
18 m
os.
2 yea
rs
3 yea
rs
% h
ouse
d in
pas
t 60
VASH exp. : N=182Case mgt. N=88Std. care: N=187
Outcomes in the HUD-VA Supported Housing Program:Percentage of Days Homeless in Past 60 days
0
10
20
30
40
50
Baseline 6 mos. 1 year 18 mos. 2 years 3 years
% d
ays
hom
eles
s in
pas
t 60
VASH exp. : N=182Case mgt. N=88Std. care: N=187
SSA-VA Outreach SSA Claims Representatives deployed
to VA sites to take claims at the time of outreach contact and facilitate contact with Disability Determination Specialists who evaluate medical evidence.
VA staff facilitate gathering/generation of medical evidence and help veteran follow through on the process.
Observational Study Design
Compare application and award rates at 4 demonstration sites and at 24 control sites.
Merge VA intake data with national SSA files to identify application and award rates.
87 7
10
14
1819
23
78
109
1012
1112
0
5
10
15
20
25
Year -4
Year -3
Year -2
Year -1
Year +1
Year +2
Year +3
Year +4
Cohort
Per
cen
t A
pp
lied
Joint Outreach Sites Comparison Sites
SSA-VA Joint Outreach: Rates of Application for Benefits (N=34,431)
Intervention
5861
71
55
75
6057 58
55
69
6167 68 68
6559
0
20
40
60
80
Year -4
Year -3
Year -2
Year -1
Year +1
Year +2
Year +3
Year +4
Year Before/After Program Initiation
Per
cen
t R
ecei
ved
Ben
efit
s
Joint Outreach Sites Comparison Sites
SSA-VA Joint Outreach: Rates of Award Among Applicants (N=3,952)
Intervention
4 45 5
1011 11
13
4
6 6 67
87 7
0
5
10
15
Year -4
Year -3
Year -2
Year -1
Year +1
Year +2
Year +3
Year +4
Year Before/After Program Initiation
Per
cen
t R
ecei
ved
Ben
efit
s
Joint Outreach Sites Comparison Sites
SSA-VA Joint Outreach: Rates of Award Among All Outreach Veterans (N=34,431)
Intervention
Collaborative Care in Treatment of Depression in
Primary Care Most depression is treated in primary care
setting by non-specialists and is of poor quality
Katon, Von Korff, Simon et al., have developed models of collaborative integration of mental health and primary care treatment of depression at Group Health Cooperative of Puget Sound.
Four collaborative models developed by Katon et al. (’95, ’96, ’99, ‘00) plus diffusion model of Wells et al (2000).
Collaborative Model #1: Katon et al. 1995
Research assistants screen all pts, identify depressed and obtain consent
Pt. given booklet and video about depression and structured questions for primary care MD
Primary care MD given ½ day didactic session and monthly case conference and consultation on treatment of depression.
Psychiatrist provide 2-4 direct visits, prescription monitoring, and feedback to primary care MD, esp. about premature discharge
Primary care MD provides ongoing care
Collaborative Model #1: Katon et al. 1995: Results
N=217 Greater adherence to medication regime
for 90 days (76% vs. 50%, p<.01) Greater subjective satisfaction
Overall quality of care (93% vs. 75%, p<.03) With antidepressants (88% vs. 63%, p<.01)
More likely to show 50% improvement in symptoms: (74% vs. 44%, p<.01)
Greater symptom improvement (p<.004)
Collaborative Model #4: Simon et al. 2000
Patients identified by primary care MDs Three conditions:
1. Usual care vs. 2. Computerized feedback to primary care
MD provided with recommendations based on adherence to treatment algorithm vs.
3. Care management (15-20 minute telephone calls at baseline, 8 and 16 weeks).
Results: #3> #1 = #2
Collaborative Models; What doesn’t work)
One-time education from strangers.
Facilitated referral. Computerized feedback to primary
care MD provided with recommendations based on adherence to treatment algorithm (Simon, 2000).
Conceptual Conclusions (1)
Problem of fragmentation is not as well demonstrated as commonly assumed and should be demonstrated before designing integrating interventions.
Conceptual Conclusions (2): Specify unrealized interdependence
Unrealized interdependence allowing access to resources should be specified in designing integrative initiatives.
Specific interdependence must be mapped on to organizational intervention.
Unrealized interdependencies are most likely found between socially distant groups, e.g. between mental health and housing agencies, or police – rather than between mental health and substance abuse agencies.
Conceptual Conclusion (3): Integration efforts can be general or
specific and can be distal (higher level) or proximal (lower level)
Distal Proximal General RWJ/ACCESS ACT
Specific HUD-VA Collab Care SSA-VA (Katon/Wells)
Conceptual Conclusion (4)
More specific and more proximal interventions are more successful at improving clinical outcomes. Interventions are (can be) better targeted at
those who will benefit. Interdependencies are better specified and can be
more explicitly mapped onto organization of services.
But they rely on supportive frame works at higher levels.
Effective interventions rely on feedback more than programming due to high uncertainty.
Conceptual Conclusions (5)
Integration need not involve structural change if processes can be changed.
Resource enhancement seems to be necessary in most cases – pure integration is rarely effective.
Applications to Criminal Justice System Mental Health System
Integration McGuire J et al Health status, service use, and costs among
veterans receiving outreach services in jails or community settings. Psychiatric services 2003;54(2):201-207.
McGuire J and Rosenheck R. Criminal history as a prognostic indicator in the treatment of homeless people with severe mental illness. Psychiatric Services 2004;55(1):42-48.
Tejani N, Rosenheck, Tsai J, Kasprow W, McGuire JF. (under review). Incarceration histories of homeless veterans and progression through a supported housing process
Stephanie Hartwell et al Predictors of Accessing Substance Abuse Services Among Individuals With Mental Disorders Released From Correctional Custody. Journal of Dual Diagnosis 2013; 9: 11-22
From Prison to Work: Proposal for National Prisoner Re-Entry Program. Bruce Western: 2008
Service Use Among Homeless Veterans Contacted in LA Jail or in the Community: Baseline (McGuire and Rosenheck, 2003)
LA Jail (N=1,676) LA Street (N=6,560)
Drug Abuse 62% 39%Alcohol Abuse48% 42%Psychiatric Dx50% 22%
Mood dx 35% 23%Personal dx. 21% 13%Schizophrenia 11% 6%PTSD 5% 6%
Medical problems 33% 37%
Service Use Among Homeless Veterans Contacted in LA Jail or in the Community: Service Use in Next Year (McGuire and Rosenheck, 2003)
LA Jail (N=1,676) LA Street (N=6,560)
Any VA Svces 38% 84%MH OP 30% 73%MH IP 3% 6%Residential Tx 4% 11%Med Surg OP 29% 60%Med Surg IP 3% 6%Avge. VA Cost $5.503 $7,821
Outcomes of Homeless People with MH Disorders (ACCESS: McGuire and
Rosenheck 2004
No Incar HX (1,195) < 6 Mos (N=2,007) >6 mos (N=1852)
Male 42% 59% 83%Fam Inst. 4.4 5.2 6.2Conduct sx 1.7 2.4 3.6Diagnoses
Psych 80% 84% 87% Drug 25% 37% 51%
Alc 26% 44% 57%Schz 36% 35% 39%Maj Dep 49% 48%48%Dual Dx 21% 36% 44%
Outcomes of Homeless People with MH Disorders:Quarterly Mean Services Used
@ 12 months (ACCESS: McGuire and Rosenheck 2004
No Incar HX (1,195) < 6 Mos (N=2,007) >6 mos (N=1852)
Empl .15 .15 .12Housing3.39 2.21 2.47Med Surg OP 4.1 4.2 5.4Psych OP 8.5 7.6 7.1SA OP 5.2 7.3 8.0Health Costs $891 $832 $790Criminal Justice $52 $99 $298
Outcomes of Homeless People with MH Disorders:Quarterly Mean Outcome Scores@ 12 months (ACCESS: McGuire and Rosenheck 2004
No Incar HX (1,195) < 6 Mos (N=2,007) >6 mos (N=1852)
Days Homless Reference +.53 +1.2ns
Days employed Reference .20 -.27ns
ASI Psych score Reference .06* .07*ASI Alch score Reference .007 .008
nsASI Drug score Reference .007 .002
nsDays in Jail Reference .576 4.7***
Baseline Characteristics of Homeless Veterans Admitted to HUD-VASH Housing Assistance Program (Years 2008 - 2009)
Never Incarcerated (No Incarceration History N=5023)
Incarceration ≤ 1 yr (Short
Incarceration History N=6324)
Incarcerated > 1 yr (Long Incarceration
History N=3210)
% difference between
incarcerated <1 yr and
never incarcerated
% difference between
incarcerated >1 yr and
never incarcerated
Variable mean ± SD or N (%) mean ± SD or N (%) mean ± SD or N (%)
CLINICAL STATUS Psychiatric Diagnoses Alcohol Abuse/Dependency 1594 (32.1%) 3632 (57.9%) 2002 (62.9%) 81% 96% Drug Abuse/Dependency 1259 (25.3%) 3054 (48.8%) 2201 (69%) 93% 173% Any P.T.S.D. 1461 (29.4%) 1815 (29%) 903 (28.5%) -1% -3% P.T.S.D. From Combat 749 (15.1%) 786 (12.6%) 298 (9.4%) -16% -38% P.T.S.D. From Non-Combat Trauma 826 (16.7%) 1173 (18.8%) 679 (21.5%) 13% 29% Schizophrenia Or Other Psychosis 384 (7.7%) 638 (10.2%) 397 (12.5%) 32% 62% Bipolar Disorder 555 (11.2%) 899 (14.5%) 495 (15.6%) 29% 39% Depressive Disorder 2424 (48.9%) 3262 (52.3%) 1550 (49%) 7% 0% Other Psychiatric Disorder 496 (10.1%) 635 (10.3%) 261 (8.4%) 2% -17% Serious Psyc Prob 2932 (58.8%) 3845 (61.1%) 1848 (58.1%) 4% -1% Psychiatric Or Substance Problem 3671 (73.4%) 5432 (86.1%) 2915 (91%) 17% 24% Dual Diagnosis 1199 (24%) 2755 (43.7%) 1520 (47.4%) 82% 98% Vet Inpt Of Res Tx When Admitted 828 (16.5%) 1494 (23.6%) 819 (25.5%) 43% 55%
Progression of Homeless Veterans through the Housing Attainment
Process (Years 2008 - 2009)
Never Incarcerated (No
Incarceration History N=5023)
Incarceration ≤ 1 yr
(Short Incarceration
History N=6324)
Incarcerated > 1 yr
(Long Incarceration
History N=3210)
% difference between
incarcerated <1 yr and
never incarcerated
% difference between
incarcerated >1 yr and
never incarcerated
Variable mean ± SD or N (%)
mean ± SD or N (%)
mean ± SD or N (%)
PROGRESSION THROUGH HOUSING PROCESS (IN DAYS) Admission To P.H.A. Visit 35.00 ± 62.58 28.54 ± 50.76 29.92 ± 55.04 -18% -15% Admission To Voucher 54.29 ± 101.42 44.77 ± 88.79 48.15 ± 101.59 -18% -11% Admission To First Look 66.53 ± 77.31 59.92 ± 70.04 60.26 ± 68.03 -10% -9% Admission To Lease 104.92 ± 86.93 94.75 ± 72.60 98.17 ± 76.06 -10% -6% Admission To Move In 109.50 ± 90.44 97.74 ± 73.74 101.58 ± 78.64 -11% -7% P.H.A. To Voucher 20.98 ± 83.39 17.13 ± 72.97 18.66 ± 82.89 -18% -11% Voucher To First Look 35.46 ± 59.55 33.12 ± 56.23 34.74 ± 55.37 -7% -2% First Look To Lease 55.17 ± 624.66 40.64 ± 45.02 57.22 ± 679.69 -26% 4% Lease To Move In 4.86 ± 14.41 3.77 ± 12.48 17.35 ± 590.80 -23% 257% Successfully Housed 2868 (57.1%) 3745 (59.2%) 1873 (58.3%) 4% 2% Housing Choice Scale 0.48 ± 0.31 0.44 ± 0.30 0.42 ± 0.30 -8% -13%
Hartwell et al 2013. Predictors of Accessing Substance Abuse Services Among Individuals With Mental Disorders Released From Correctional Custody. Journal of Dual Diagnosis Patients discharged from criminal
justice system who used mental health system.
69% had diagnosis of substance abuse disorders
Within 24 months 61% used substance abuse services.
Correlates: female, white, released under correctional supervision
From Prison to Work: Proposal for National Prisoner Re-Entry Program. Bruce Western: 2008
Big three: Housing, Employment, Drug Abuse Treatment
Personal Mentoring Less harsh response to violations in
order to support reintegration Fewer benefits restrictions Data from demonstrations shows
modest reduction of recidivism:10-15%