Behavioral Health
Call for Papers
Disparities in Treatment for & Impact of Mental Illness
Chair: Greer Sullivan, M.D., M.S.P.H.
Monday, June 7 • 8:30 a.m.-10:00 a.m.
• Severe Mental Illness and Congestive Heart Failure
Outcomes among Veterans
Jim Banta, M.P.H., B.S., Ronald Andersen, Ph.D., Alexander
Young, M.D., M.S.H.S., William Cunningham, M.D., Gerald
Kominski, Ph.D.
Presented by: Jim Banta, M.P.H., B.S., Ph.D. Candidate,
Health Services, UCLA, 1037 Willis Court, Colton, CA 92324;
Tel: 909.825.2968; E-mail: jbanta1@msn.com
Research Objective: Congestive heart failure is the leading
cause of death among those greater than 65 years of age and
is the single most expensive condition covered by Medicare.
Co-morbid mental illness may lead to negative outcomes
through worsened self-care and interactions with health care
providers. Depression, for instance, is often associated with
increased mortality and health services utilization among
cardiac patients. This study focuses on the effect of severe
mental illness (SMI) on mortality and readmission of heart
failure patients; but also considers outpatient care.
Study Design: Inpatient and outpatient data for several fiscal
years were obtained from the Department of Veterans Affairs’
centralized patient database. The first heart failure admission
of Fiscal Year 2001 was taken to be the index admission for
this study.
Logistic regression was used to test the effect of mental illness
on one-year mortality and readmission. Logistic and linear
regression was also used to examine receipt of outpatient
medical care preceding the index admission. Using a
behavioral model approach, adjustment was done for
personal characteristics, including service-connected disability,
co-morbid medical conditions (as measured by the CHF Index
and Elixhauser method for co-morbid medical conditions),
and contextual measures of the hospital and county of
residence in which care was delivered.
Population Studied: Veterans with a primary inpatient
diagnosis of congestive heart failure during during federal
Fiscal Year 2001 were selected. Patients were identified as
SMI if any one of the ten diagnosis fields in the index
admission or in an inpatient or outpatient service in the twelve
months preceding the index admission included
schizophrenia, PTSD or severe bi-polar disorder.
Principal Findings: Inpatient data was obtained for 15,739
patients in 127 of 130 active VA medical centers or health care
systems. Thirteen percent of patients died within twelve
months of admission and fifty-four percent were readmitted.
Of those living at least one more year, forty-seven percent
were readmitted. Seven percent of the veterans had a severe
mental illness. Twenty percent of those with SMI had cooccuring substance abuse compared to ten percent of those
with no identified mental illness.
Those with SMI were more likely to die than those having no
mental diagnosis (Odds Ratio 1.34). This despite those with
severe mental illness being younger than those with no
identified mental illness (64 versus 70 years of age), and being
less medically severe (CHF index of 2.4 versus 2.5). Severe
mental illness was not significantly associated with
readmission.
Mental illness was not associated with a lower likelihood of
receiving outpatient primary care in the twelve months
preceding the index admission. In fact those with SMI were
more likely to have received any primary care. However,
substance abuse was associated with fewer outpatient
services.
Conclusions: After adjusting for many factors, heart failure
patients with SMI were more likely to die than veterans with
no mental illness.
Implications for Policy, Delivery or Practice: Interventions
to improve medical care in people with SMI or substance
abuse are needed, particularly among those nearing
retirement age. Such efforts could lead to reduced mortality.
Primary Funding Source: SAMHSA
• The Effect of Mental Disorders on Labor Market
Outcomes among Latino Americans
Pinka Chatterji, Ph.D., Margarita Alegria, Ph.D., Mingshan Lu,
Ph.D.
Presented by: Pinka Chatterji, Ph.D., Health Economist,
Cambridge Health Alliance-Harvard Medical School, Center
for Multicultural Mental Health Research, 120 Beacon Street,
4th Floor, Somerville, MA 02143; Tel: 617.503.8449; Fax:
617.503.8430; E-mail: pchatterji@charesearch.org
Research Objective: This paper investigates the effect of
psychiatric disorders on current employment, the number of
weeks worked in the past year, and the number of work
absences in the past month among Latino Americans.
Study Design: We use multiple regression models to estimate
the association between psychiatric disorders and labor
market outcomes. Psychiatric disorders are represented by
four dummy variables indicating whether or not the
respondent met diagnostic criteria for substance
abuse/dependence, major depression, anxiety, or any mental
disorder in the past 12 months. The models include controls
for race/ethnicity, age, education, marital status, foreign born,
and the number of children. Some models include an
indicator of any lifetime psychiatric disorder. Standard errors
are adjusted for clustering on state of residence.
Drawing inference about the effects of mental disorders on
labor market outcomes is complicated due to unobserved
heterogeneity problems. Mental disorders may be affected by
unobserved personal and social characteristics that also affect
labor market outcomes. Moreover, one’s labor market
outcomes can affect his current mental health. To address
these issues, we follow previous research and estimate
instrumental variables models using onset of mental
disorders during childhood as an identifying instrument.
Population Studied: Data come from the National Latino and
Asian American Study (NLAAS). NLAAS is a national
psychiatric epidemiologic study being conducted to measure
psychiatric disorders and mental health service usage in a
representative sample of Asians and Latinos. Unlike most
population-based surveys, NLAAS provides detailed and good
quality data on mental health status, as well as information on
demographics and labor market participation. We limit the
analysis sample to 1,503 Latino respondents between 18 and
65 years old who are not in school and who have complete
survey information. All models are estimated separately by
gender.
Principal Findings: All psychiatric disorders analyzed have a
negative impact on the probability of being currently
employed. This finding persisted in instrumental variables
models, as well as in models that included a control for any
lifetime psychiatric disorder. Among males, having a
psychiatric disorder is associated with a 16 to 37% reduction
in the probability of being employed. Among females, having
a psychiatric disorder is associated with a 14 to 25% reduction
in the probability of being employed. However, among
employed males and females, psychiatric disorders do not
have a statistically significant association with work absences
or the number of weeks worked in the past year.
Conclusions: The findings indicate that psychiatric disorders
have large effects on employment among Latino males and
females. These results are consistent with previous work,
which has focused on majority populations. Psychiatric
disorders appear to have important impact on individuals’
ability to obtain and/or maintain employment.
Implications for Policy, Delivery or Practice: By providing
evidence on the social costs of mental disorders among
Latino Americans, this study has important policy implications
on issues such as mental health insurance coverage policy
and health care resource allocation. The findings imply that
improving access to effective mental health screening and
treatment may have benefits in the labor market.
Primary Funding Source: NIMH, NCMHHD
• Disparities in the Adequacy of Depression Treatment in
the United States
Jeffrey Harman, Ph.D., Mark Edlund, M.D., Ph.D., John
Fortney, Ph.D.
Presented by: Jeffrey Harman, Ph.D., Assistant Professor,
Health Services Administration, University of Florida, P.O. Box
100195, Gainesville, FL 32611-0195; Tel: 352.273.6060; Fax:
352.273.6075; E-mail: jharman@phhp.ufl.edu
Research Objective: There is evidence that disparities in
depression care exist by factors such as age, race/ethnicity,
and type of insurance. The purpose of this study is to assess
whether observed disparities in care are due to differences in
rates of initiating treatment or differences in the quality of
treatment once treatment has been initiated.
Study Design: Data from the 2000 Medical Expenditure Panel
Survey (MEPS) are used to obtain nationally representative
estimates of the rates at which individuals with self-reported
depression utilize antidepressant medications and/or
psychotherapy or mental health counseling. Additionally, for
those individuals who use some depression care, estimates of
the adequacy of care received are obtained and compared by
patient characteristics. Adequate care is defined as filling at
least four prescriptions for an antidepressant medication at a
minimally adequate daily dosage or receiving at least eight
psychotherapy or mental health counseling sessions during
the calendar year. Weighted logistic regression models are
estimated to assess the independent role of patient
characteristics such as age, race/ethnicity, and type of
insurance on rates of initiating depression treatment for
individuals with self-reported depression and on rates of
adequate treatment for those individuals who receive
treatment after controlling for health status, functional status,
education, income, and marital status.
Population Studied: The Household Component of the 2000
MEPS (N=25,096) is used. The MEPS is a nationally
representative survey of households in the United States
conducted annually by the Agency for Healthcare Research
and Quality. All individuals with self-reported depression are
included in the study (N=1,347).
Principal Findings: African-Americans (OR = 0.48) and
Hispanics (OR = 0.47) with self-reported depression were
significantly less likely to fill an antidepressant prescription
than Caucasians, but if at least one antidepressant
prescription was filled, there was no significant difference in
the probability of receiving an adequate trial of antidepressant
medication by race/ethnicity. No significant differences in the
probability of receiving any psychotherapy or mental health
counseling by race/ethnicity was observed, but AfricanAmericans were more likely to receive an adequate course of
psychotherapy than Caucasians (OR = 2.35). Young adults
were less likely to initiate depression treatment (OR = 0.48),
and if initiated, were less likely to receive adequate treatment
(OR = 0.55). No significant differences for elderly persons
were observed. Individuals without insurance were
significantly less likely to receive any treatment than
individuals with private insurance (OR = 0.58), but if
treatment was initiated, no significant difference in the
probability of receiving adequate treatment was observed by
type of insurance. Individuals who received both
antidepressants and psychotherapy/counseling were more
likely to receive adequate treatment than individuals who
received only psychotherapy/counseling or only
antidepressants (OR = 2.92).
Conclusions: Disparities in depression care appear to be
primarily due to differences in rates of initiating depression
treatment, as rates of adequate care do not differ once
treatment is initiated.
Implications for Policy, Delivery or Practice: Interventions
to reduce disparities in depression care should focus on
initiating depression treatment. However, interventions that
improve the adequacy of care once care is initiated are still
needed, as rates of adequate treatment are low for all
individuals with depression.
Primary Funding Source: NIMH
• Racial Differences in Quality of Care for Bipolar Disorder
Amy Kilbourne, Ph.D., M.P.H., Gretchen Haas, Ph.D., Xiaoyan
Han, M.S., Mark Bauer, M.D., Mujeeb Shad, M.D., Harold
Pincus, M.D.
Presented by: Amy Kilbourne, Ph.D., M.P.H., Assistant
Professor of Medicine and Psychiatry, VA Pittsburgh CHERP,
University of Pittsburgh, University Drive C, Pittsburgh, PA
15240; Tel: 412.688.6477; Fax: 412.688.6527; E-mail:
Amy.Kilbourne@med.va.gov
Research Objective: Despite the existence of practice
guidelines, quality of care and subsequent outcomes for
bipolar disorder remain suboptimal. Potential disparities in
quality of care for patients with bipolar disorder have been
relatively unexplored. We therefore assessed racial differences
and other patient factors associated with quality of care for
bipolar disorder.
Study Design: We conducted a retrospective analysis of data
from the VA National Patient Care Database (NPCD) on
patients with either one inpatient or two separate outpatient
ICD-9 diagnoses of bipolar disorder (I, II, or NOS) in FY 2001.
Data on all psychopharmacologic prescriptions were
ascertained from the VA Pharmacy Benefits Management
database, and patient demographic and outpatient use data
were collected from the NPCD in FY 2001. We applied
American Psychiatric Association Guideline-based indicators
to assess quality of care: 1) current prescription for a mood
stabilizer, 2) no antidepressant monotherapy (i.e., prescription
for an antidepressant without concurrent prescription for
mood stabilizer), and 3) >3 outpatient mental health visits
within one year. Multiple logistic regression analysis was used
to determine whether race/ethnicity was independently
associated with quality of care for bipolar disorder, controlling
for patient factors including age, gender, service connection
eligibility, marital status, and number of comorbidities.
Alternative analyses were conducted in which we limited our
sample to patients diagnosed with bipolar I disorder only
(n=2736).
Population Studied: Patients diagnosed with bipolar disorder
receiving care at one of 10 VA facilities located in
Pennsylvania, West Virginia, Ohio, Delaware, and Maryland.
Principal Findings: Out of 252,621 patients receiving care in
VISN 4 in FY 2001, 2958 (1.2%) met criteria for bipolar
disorder diagnosis. The mean age was 52 years, 11% were
women, and 12% African-American. Overall, 80.1% of
patients were currently prescribed a mood stabilizer, 13.4%
were prescribed antidepressant monotherapy, 63.5% received
>3 outpatient mental health visits. About half (53.4%) met all
three criteria. After adjustment for the aforementioned patient
factors, African-Americans were less likely to have a
prescription for a mood stabilizer (OR=.71;p=.02) and more
likely to be on antidepressant monotherapy (OR=1.61;p=.001)
compared to whites. Older (>60 years) patients were also less
likely to have a prescription for a mood stabilizer
(OR=.54;p<.001), more likely to be on antidepressant
monotherapy (OR=1.32;p=.04), less likely to receive >3
outpatient mental health visits (OR=.61;p<.001), and less
likely to meet all three criteria (OR=.54;p<001) after
adjustment. Similar results were obtained when the sample
was limited to bipolar I disorder.
Conclusions: The majority of VA patients with a diagnosis of
bipolar disorder are receiving guideline-concordant care for
bipolar disorder. Nonetheless, potential gaps in quality of
care are evident in continuity of outpatient care; and in
particular, for African-Americans and older patients, even
among a more restricted sample of bipolar I patients.
Implications for Policy, Delivery or Practice: Identifying the
reasons for potential gaps in quality of care, including patient
preferences, diagnostic instability, provider decision-making,
and management factors will inform future quality
improvement interventions to reduce disparities in care for
bipolar disorder. Future efforts include the validation and
application of performance measures to monitor quality
improvement efforts for patients with bipolar disorder across
the U.S.
Primary Funding Source: VA
• Mental Health and Employment Transitions among LowIncome Women
Leigh Ann White, Ph.D. Candidate
Presented by: Leigh Ann White, Ph.D. Candidate, Health
Policy & Management, Johns Hopkins School of Public
Health, 6427 79th Street, Cabin John, MD 20818; Tel:
301.320.9029; Fax: 301.320.1762; E-mail: lwhite@jhsph.edu
Research Objective: This research examines how mental
health affects employment opportunity of women in lowincome families. There is accumulating evidence regarding the
negative relationship between poor mental health status and
employment. However, we know little about how mental
health and its correlates operate in influencing employment
and job stability. This knowledge is an essential component of
developing strategies to address mental health issues in
publicly financed settings.
Study Design: Estimates of the impact of mental health
symptoms on subsequent employment come from probit
regressions of baseline factors on employment status at the
second interview. Subgroup analyses are performed for
women with different work histories at baseline. OLS and tobit
regressions of the proportion of time spent employed between
interviews add information on job holding between interviews.
All analyses focus on the effect of symptoms of prevalent
mental disorders, measured using the 18-item Brief Symptom
Inventory, and interactions between mental health, initial
employment state, and key baseline covariates.
Population Studied: Data are from two waves of interviews
from the survey component of the Welfare, Children, and
Families Three-City Study. The analytic sample consists of
mothers in families with incomes up to 200 percent of the
Federal Poverty Level, in poor neighborhoods of Boston, San
Antonio, and Chicago. The analytic sample is limited to
approximately 1,700 adult women ages 18 to 64 who had
complete data on key variables in both waves of the survey.
Principal Findings: Probit findings suggest a negative effect
of baseline psychological distress on employment entry for the
unemployed. Controlling for covariates, those meeting BSI
case criteria were significantly less likely to be subsequently
employed. Predicted probabilities were .24 vs. .13 among
those with no recent experience, and .58 vs. .51 among the
experienced unemployed. Among the employed, psychological
distress was a negative predictor of subsequent employment;
predicted probabilities were .84 and .71. Findings from twolimit Tobit regressions of the proportion of time in
employment add to the picture. Controlling for covariates, the
distressed group was slightly more likely to be employed at
any time, .74 compared to .71. However, conditional on
employment, they spent less time in jobs. Predictions showed
that the women with symptoms of psychological distress
spent about 43 percent of the period in some job, compared
to 49 percent of the comparison group, a statistically
significant difference. This finding suggests the possibility of
"job churning."
Conclusions: Symptoms of prevalent psychiatric syndromes
may be important barriers to achieving stability in
employment. They may prevent many low-income women,
especially the socially marginalized, from entering
employment altogether. For women with few skills and very
low educational attainment, mental health problems may
derive from and compound stressors that impede
employment entry. Even among the employed, there is
evidence suggesting a relationship between symptoms of
mental disorders and labor force attachment.
Implications for Policy, Delivery or Practice: Future research
using these data will focus on the effects of mental health
symptoms on job entry, job loss, and job-to-job transitions, to
explore the relationship in a more thorough manner. The
research presented here suggests the importance of
continuing to develop better ways to screen and assess for
mental health problems in multiple locations, and to
coordinate across systems such as TANF offices, vocational
rehabilitation centers, and mental health providers. It also
suggests the importance of aiding in employment retention by
addressing mental health issues through job training and on
the job.
Primary Funding Source: NIMH, University of Michigan
National Poverty Center
Call for Papers
Mental Health in Community Context
Chair: Kenneth Wells, M.D., M.P.H.
Monday, June 7 • 4:00 p.m.-5:30 p.m.
• Witness for Wellness: A Community-University
Participatory Research Mental Health Initiative
Loretta Jones, M.A., Ricky Bluthenthal, Ph.D., Marcia Ellison,
Ph.D., Paul Koegel, Ph.D., Katy Minnium, M.P.H., Aziza
Lucas-Wright, M.Ed., Ken Wells, M.D., Ph.D.
Presented by: Ricky Bluthenthal, Ph.D., Social/Behavioral
Scientist, Behavioral Health Sciences, RAND, 1700 Main
Street, Santa Monica, CA 90407; Tel: 310.393.0411 Ext. 6642;
Fax: 310.451.7004; E-mail: rickyb@rand.org
Research Objective: Research indicates that depression
treatment may have even greater positive outcomes among
people of color than whites. Yet, no model has been
developed, to date, to bring mental health Research suggests
that minorities are less likely than whites to receive
appropriate care for depression yet benefit at least as much or
more from interventions that improve depression care. Yet no
model has been developed to increase public engagement in
depression treatment among minority communities. This
presentation involves an intervention and evaluation model,
and preliminary findings from such a community-based effort.
Study Design: “Witness for Wellness” is a communitypartnered participatory research (CPPR) project with active
community involvement in all stages. The project was
initiated by Healthy African American Families, a community
based organization (CBO) in South Los Angeles, UCLA,
Charles R Drew Medical University, and RAND. The stages
are: 1) coalition development; 2) community conferences; 3)
formation of working groups to develop and implement
strategies; 4) participatory evaluation using mixed methods,
such as pre-post surveys of conferences, qualitative studies,
and secondary data sources.
Population Studied: South Los Angeles, CA, a predominantly
African American and Latino community.
Principal Findings: Over 30 CBOs joined and participated in
coalition planning for a year. A local conference attracted 500
community members, 60 of whom later formulated plans for
3 working groups: Talking Wellness (reducing stigma and
increasing community strength); Building Wellness (assuring
assess and services quality); Supporting Wellness (policy and
advocacy to protect vulnerable populations). As an example
of evaluation data, 370 conference participants completed a
survey: 50% worked in a CBO, 20% were providers, 15%
government employees, 10% academics, and 5% grass roots
community members; 50% were African American and 30%
Hispanic; and most (90%) were women. The surveys
suggested that attendees became better educated about signs
and symptoms of depression and the value of and how to
obtain treatments. Forty roundtable discussions were
recorded by trained notetakers and coded by community and
academic partners. Qualitative analysis suggested that
depression is viewed primarily as a community problem. Key
barriers to treatment included lack of knowledge of resources,
social stigma, cultural values (e.g., that spirituality should
suffice, keeping problems in the family, stoicism), endemic
racism and “fear of the dominant culture’s solution to our
problem,” and “normalization” or accommodation to this
illness. Community members reported that they turned for
help most often to non-familial social intimates. The
presentation will provide further detail on the intervention
model and evaluation findings.
Conclusions: Participatory models are feasible for engaging
community members in a research effort to address
depression; but this demonstration has to date been more
successful in engaging African Americans and women.
Additional strategies are needed to reach diversity goals. The
approach yields data that both community and academic
partners are empowered to share. The effectiveness of the
model is the subject of future study.
Implications for Policy, Delivery or Practice: National
clinical research recommendations call for new models of
community participation. This mental health project
demonstrates one such approach that holds potential to
address this research and policy need, through empowering
community members as research partners and empowering
academic investigators as community intervention partners.
Primary Funding Source: NIMH, RAND
• Does Relative Poverty Predict the Need for Mental
Health Care?
Christine Eibner, Ph.D., Roland Sturm, Ph.D., Carole Roan
Gresenz, Ph.D.
Presented by: Christine Eibner, Ph.D., Associate Economist,
Economics and Statistics, RAND Corporation, 1700 Main
Street, P.O. Box 2138, Santa Monica, CA 90407; Tel:
310.393.0411 Ext. 6522; E-mail: eibner@rand.org
Research Objective: We examine whether an individual’s
income status relative to a reference group affects mental
health outcomes. Relative poverty differs from low income
because individuals with high-income reference groups might
be well-off with respect to the general population but still
deprived relative to their peers. Our contributions to the
literature include 1) defining reference groups using
demographic characteristics in addition to geographic area, 2)
looking at an individual’s relative income status rather than
aggregate-level income inequality, and 3) focusing specifically
on mental-health related outcomes. The four mental health
outcomes we examine are depression, anxiety disorder, severe
mental health disorders, and a summary variable indicating
any probable mental health problem.
Study Design: Our primary data source is the national
household survey component of HealthCare for Communities
(HCC), funded by the Robert Wood Johnson Foundation to
track the effects of the changing health care system on
individuals at risk for alcohol, drug abuse, or mental health
disorders. HCC is a complement to the Community Tracking
Survey (CTS) and reinterviews participants of the main study.
To construct relative poverty measures, we used data from the
5% Public Use Micro Data Sample of the 2000 Census. Our
measure of relative poverty is defined using Yitzhaki’s index, a
term that measures the expected income difference between
an individual and others in his or her reference group that are
more affluent. We evaluated the relationship between relative
poverty and mental health using conditional logit models with
reference group fixed effects.
Population Studied: Adult respondents to the 60-site sample
of the HCC study. Although the HCC over-sampled
individuals who were likely to be at risk for mental health
problems, our final results are weighted to be representative
of the population.
Principal Findings: Even after controlling for an individual’s
absolute income status, those with low relative income are at
higher risk for experiencing probable mental health problems.
Our findings hold for all of the mental health outcomes
examined.
Conclusions: Our findings suggest that relative poverty is
associated with an increased likelihood of probable
depression, anxiety, severe mental health problems, and
overall poor mental health. Our simulations suggest that a 25
percent decrease in relative poverty could decrease the
probability of any likely mental health problems by as much as
8.7 percent. The evidence does not suggest that relative
poverty has a markedly stronger impact on one particular
mental health outcome as opposed to the others.
Implications for Policy, Delivery or Practice: Social ordering
and the perceptions of low relative status may contribute to
socioeconomic disparities in mental health. Efforts to
eradicate socioeconomic differentials should take into account
psychological perceptions and self-esteem in addition to
absolute material resources.
Primary Funding Source: NIMH
• Which Dimensions of Continuity of Mental Health Care
Lead to Improved School Outcomes?
Alka Indurkhya, Ph.D., Joy Basu, Ph.D., Jennifer Yu, M.S.,
Nicholas Ialongo, Ph.D.
Presented by: Alka Indurkhya, Ph.D., Senior Research
Scientist, Harvard University, 677 Huntington Avenue, Boston,
MA 02115; Tel: 617.432.0964; E-mail:
aindurkh@hsph.harvard.edu
Research Objective: The school system in the United States
is a significant provider of mental health services with three
out of four children reporting treatment for mental disorders
only from their school. This is the first study to examine the
extent to which different dimensions of continuity of school
based mental health services impact school outcomes such as
grade repetition, suspension or expulsion, and future mental
health service use.
The aims of this research were: (a) to generate a set of
dimensions for continuity of school based mental health care
using parent, teacher, and school mental health professional
reports for children who need/use behavior counseling
services, and (b) to assess the impact of the different
dimensions on future school outcomes.
Study Design: Follow-up from 1993-2000 of a prospective
study cohort of children in school.
The school based mental health services module of the service
assessment for children and adolescents (SACA) was used by
teachers, school mental health professionals, and parents to
report the need/use for behavior counseling services for each
child in the cohort from 1993-2000. Three distinct dimensions
of continuity of care were obtained from the SACA: timeliness,
intensity, and stability.
Information on grade repetition, expulsion and suspension
were obtained from parent, and teacher reports.
Logistic regression analysis was used to determine adjusted
odds ratios and confidence intervals (CI) after adjusting for
race, gender, and socio-economic status of the children for
each of the dimensions of continuity for school based mental
health care.
Population Studied: A cohort of 678 first graders from nine
schools in Eastern Baltimore. 87% of children are African
American, 12% Caucasian, and 1% Native American, Hispanic,
and Asian. The cohort contains 53% males.
Principal Findings: Timeliness:
28.8% of the children were identifed as needing mental health
services in Grade 1. Only one out of three children received
needed services in a timely manner. Findings from logistic
regression indicated that in Grade 6, children who did not
receive services in a timely manner were 4.6 more likely (95%
CI: (3.7, 8.3)) to need mental health services as compared to
those whose needs were met in a timely manner in Grade 1.
These children were also found to be 2.2 times more likely
(95% CI: (1.8, 2.7))to face suspension or expulsion from
school due to disciplinary reasons. The odds for grade
repetition were not statistically significant.
Intensity: 18% of the students received more than 10 hours of
counseling from a mental health professional in Grade 1.
These students were 5.3 times more likely (95% CI: (1.7, 8.4))
to need mental health services in Grade 6 when compared
those that received less than 10 hours of counseling in Grade
1. The odds were not found to be statistically significant for
grade repetition, expulsion or suspension.
Stability: Only 23% of the children whose needs were met in a
timely manner in Grade 1 were found to continue treatment
until Grade 3. These children were twice as likely not to require
more than 3 hours of mental health services by Grade 6 when
compared to the children who received services in Grade 1 but
did not receive needed services through Grade 3. The odds for
grade repetition, expulsion, and suspension were not
statistically significant.
Conclusions: All three dimensions for continuity of school
based mental health care in elementary school are predictive
of mental health service use in Grade 6. Timeliness appears to
be an extremely important dimension in the study of
continuity of school based mental health care that may lead to
improved school outcomes, and reduce the need for future
mental health services. The dimensions studied did not
impact grade repetition.
Implications for Policy, Delivery or Practice: The results
underscore the urgent need for assessing and providing
school based mental health services to children in elementary
school to help improve educational outcomes in middle
school.
Primary Funding Source: NIMH
• Longitudal Relationships among Family Caregiving
Intensity, Caregiver Role Strain and Treatment Outcomes
for Persons with Comorbid Mental Illness and Substance
Abuse
Ingrid Rystedt, M.D., M.S.
Presented by: Ingrid Rystedt, M.D., M.S., Ph.D. Candidate,
Department of Community & Family Medicine, Center for the
Evaluative Clinical Sciences, Dartmouth Medical School, 7251
Strasenburgh Hall, Hanover, NH 03755; Tel: 603.650.1141; Email: ingrid.rystedt@dartmouth.edu
Research Objective: (1) To examine the long-term
relationships between caregiver role strain and the intensity of
caregiving for clients with comorbid severe mental illness and
substance abuse and (2) to evaluate the impact of the
intensity of caregiving and caregiver role strain on the course
of the client's disorder.
Study Design: Secondary data analysis using data from The
New Hampshire Dual Diagnosis Study, a randomized clinical
trial comparing two treatment models. Primary caregivers
who were family members of the clients were included in this
study (n=155). Original study hypotheses: (1.) The higher the
cumulative intensity of informal care provided by the family
caregiver, the greater the current level of perceived role strain.
(2.) The higher the cumulative role strain perceived by the
family caregiver, the lower the current caregiving intensity. (3.)
High caregiving intensity and low role strain are associated
with improvements in client outcomes over time. Analytic
approach: Mixed effects regression analysis was utilized to
model the longitudinal relationships between caregiving
intensity, caregiver role strain and client outcomes.
Population Studied: Client data were collected between 1989
and 1993 in seven mental health centers in New Hampshire.
Clients were followed for three years and were interviewed at
six-month intervals about psychiatric symptoms, substance
abuse, service use, income sources and contacts with friends
or family. At corresponding time points, primary caregivers
were interviewed about caregiving involvement and family
burden. Data on client involvement with the legal system
were collected.
Principal Findings: More intense cumulative caregiving was
strongly associated with higher levels of currently perceived
role strain. Contrary to the initial hypothesis, the current
caregiving intensity was significantly higher when family
members had experienced high role strain over time. All
caregiving variables (cumulatively high caregiving intensity,
cumulatively low role strain and co-occurring cumulatively
high caregiving intensity and low role strain) were significantly
associated with reduced client substance abuse. Cumulatively
low role strain was significantly associated with lower
probability of client arrest.
Conclusions: By illustrating the effects of caregiving on both
family caregivers as well as clients, this study revealed both
the benefits and social costs of informal caregiving. Although
the caregiving intensity was generally sustained, high
caregiving intensity was consistently associated with high
levels of role strain. The previously identified beneficial effects
of informal caregiving on client substance abuse outcomes
were replicated with a longitudinal analysis.
Implications for Policy, Delivery or Practice: The findings
suggest that role strain in itself does not threaten caregiving
involvement. However, the indirect effects of role strain on
caregiver health may affect such involvement in a more longterm perspective. The continued identification of
mechanisms to support family caregivers would contribute to
at least three important goals: (1) improved quality-of-life for
family caregivers, (2) improved client outcomes, and (3) the
development of a more cost-effective treatment system for
comorbid severe mental illness and substance abuse. Future
research should explore how family members are most
effectively enabled to sustain informal care while maintaining
low to moderate levels of role strain.
Primary Funding Source: NIDA, NIMH, AHRQ (Postdoctoral
Fellowships)
• The Effects of Community Risk and Resource Variables
on Mental Health and Substance Abuse Need and Access
to Care
Susan Stockdale, Ph.D., Kenneth Wells, M.D., M.P.H., Lingqi
Tang, Ph.D., Thomas Belin, Ph.D.
Presented by: Susan Stockdale, Ph.D., Senior Statistician, NPI
Health Services Research Center, UCLA, 10920 Wilshire
Boulevard, Suite 300, Los Angeles, CA 90024; Tel:
310.794.3732; Fax: 310.794.3724; E-mail:
sstockdale@mednet.ucla.edu
Research Objective: In spite of increasing concern over the
gap between development of effective treatments for common
behavioral health conditions, such as depressive and anxiety
disorders and alcohol/drug abuse, and their adoption in the
community, development of community-based strategies to
improve care is limited by the lack of availability of
comprehensive national data on how community factors and
their interaction with individual factors affect occurrence of
behavioral health conditions or access to services for these
conditions. This study uses national data to determine how a
range of community risk and resource characteristics shape
individual need for and access to behavioral health care.
Study Design: Two waves of data from Health Care for
Communities, a national household telephone survey fielded
between 1997 and 2001. Dependent variables include any
alcohol, drug, or mental health (ADM) condition or perceived
need for ADM care, and use of specialty mental health and/or
primary care for these conditions. Neighborhood
characteristics were median household income, neighborhood
stressors (violent crime arrest rate and residential crowding
vs. isolation), and resources (churches per 1,000 capita,
health food, alcoholic beverages, personal care facilities,
recreational facilities, gambling establishments, and mental
health care facilities). Individual variables include education,
race/ethnicity, gender, marital status, age, family savings,
social support, and exposure to violence.
Population Studied: Adult Population with telephones in 60
US communities
Principal Findings: We illustrate findings through analyses of
need. Higher neighborhood income was significantly
associated with greater probability of substance abuse.
Residential crowding, and its interaction with individual social
support, were negatively associated with behavioral health
need, such that greater social support particularly affected
(lowered) ADM need in less densely crowded neighborhoods.
Greater availability of social interaction establishments
(churches per capita, personal care facilities, gambling
establishments as a group) was associated with lower ADM
need, but greater availability of stores selling alcoholic
beverages increased substance abuse. Further, individual
exposure to violence particularly affected (increased) ADM
need in high crime areas, while those not exposed to violence
in these areas had lower ADM need. We will also present and
discuss findings related to access to ADM care.
Conclusions: Overall, this study suggests that neighborhood
factors – spanning physical capacities and risk factors, social
resource availability, and physical exposure factors (alcohol
availability) – are uniquely associated with mental health and
substance abuse problems even after controlling for individual
predictors of need. Further, community factors interact with
important individual risk and protective factors, such as social
support and exposure to violence, in shaping ADM need.
Implications for Policy, Delivery or Practice: These findings,
particularly if replicated in other data sets, have potential
implications for targeting of prevention and treatment
interventions to high-risk populations within the context of
their neighborhood characteristics, and for the development of
community resource interventions.
Primary Funding Source: RWJF
Call for Papers
Policy Developments in Mental Health & Substance Abuse
Chair: Nancy Wolff, Ph.D.
Tuesday, June 8 • 11:15 a.m.-12:45 p.m.
• Coverage of Mental Health Benefits and Parity Laws
Myles Maxfield, Ph.D., Lori Achman, M.P.P., Jeffrey Buck,
Ph.D, Judith Teich, M.S.W.
Presented by: Lori Achman, M.P.P., Health Analyst,
Mathematica Policy Research, Inc., 600 Maryland Avenue,
Suite 550, Washington, DC 20024; Tel: 202.264.3464; Fax:
202.863.1763; E-mail: lachman@mathematica-mpr.com
Research Objective: Access to mental health care by
individuals with mental illness depends largely on whether the
individual has health insurance and the generosity of those
benefits. Benefits are influenced by federal and state mental
health parity laws. This paper estimates the number of
individuals with mental health insurance benefits, the
generosity of those benefits, and the number of individuals
subject to state and federal mental health parity laws in 1999.
Study Design: Estimates are based on the March 2000
Current Population Survey (CPS), the 1999 Medical
Expenditure Survey—Insurance Component (MEPS-IC), and
the Mercer Worldwide National Survey of EmployerSponsored Health Plans. Each survey uses a nationally
representative probability sample with 1999 as the reference
period.
Population Studied: The non-institutionalized U.S.
population in 1999. Individuals were categorized according to
their reported primary source of health insurance.
Principal Findings: In 1999, 37.6 percent of the U.S.
population with health insurance had mental health benefits
at full financial parity with their medical/surgical benefits,
meaning equality in dollar limits, cost sharing and utilization
limits. An additional 37.6 percent of the insured population
had mental health benefits, but they were not at parity with
medical benefits. Another 14.2 percent of the population had
mental health benefits, but data was insufficient to estimate
the generosity. Approximately 2.9 percent of insured
individuals had no mental health benefits. No information
was available on the mental health benefits of the remaining
7.7 percent. To ameliorate the disparity between mental
health and medical/surgical benefits, many states, as well as
the federal government, passed laws requiring some level of
parity for mental health benefits. The Mental Health Parity Act
of 1996, which required parity only in dollar limits, applied to
only 42 percent of the U.S. non-institutionalized population in
1999. State mental health parity laws often go beyond the
federal law to also require parity in cost sharing and utilization
limits, but these laws reach far fewer people. In 1999, 13
states had mental health parity laws requiring full financial
parity for the private, employer-sponsored market. These laws
covered only 8 percent of individuals with private, employersponsored health insurance, or 4 percent of the entire U.S.
population.
Conclusions: Access to mental health care for many people
may be limited by a lack of mental health insurance benefits or
by benefits that are not generous. The generosity of mental
health benefits of many health plans is unaffected by federal or
state parity laws because of the exemptions contained in those
laws.
Implications for Policy, Delivery or Practice: Implementing
full financial parity in all federal health programs, reducing the
exemptions from parity laws, and implementing a federal full
financial parity law may be effective tools for increasing access
to mental health care.
Primary Funding Source: SAMHSA
• Cost Effectiveness of Public Sector Substance Abuse
Treatment: Comparison of a Managed Care Approach to a
Traditional Public Sector System
Martha Beattie, Ph.D., Teh Wei Hu, Ph.D., Rui Li, Master of
Medicine, Jason Bond, Ph.D.
Presented by: Martha Beattie, Ph.D., Scientist, Alcohol
Research Group, Public Health Institute, 2000 Hearst Avenue,
Berkeley, CA 94709; Tel: 408.272.6535; Fax: 408.264.2637; Email: mbeattie@arg.org
Research Objective: A primary premise of managed care is
cost containment. Linking cost of care to the outcomes of
that care leads to an examination of quality of care as well.
This study compares the costs and cost effectiveness of public
sector substance abuse treatment in two California counties
with similar substance abuse treatment system histories, one
of which implemented a managed care approach to treatment
in 1997. The two counties were compared three years after the
system change in MidState – the county that adopted a
management of care perspective – to allow time for the
system to stabilize. It was hypothesized that 1. the cost per
client served would be lower in MidState than in SouthState –
the county whose system remained virtually unchanged; and
that 2. because of its attention to quality of care as well, the
cost effectiveness ratios would show greater effectiveness in
MidState than in SouthState.
Study Design: Baseline and outcome data were collected by
trained research interviewers. At twelve months 82% of
MidState’s 198 subjects and 80% of SouthState’s 256 subjects
were interviewed. Information on service utilization and costs
for clients participating in the study came from the counties’
automated Management Information System -MISdatabases. Analysis is limited to those who could be matched
to treatment utilization data in each county’s MIS – 196 in
MidState and 208 in SouthState.
Population Studied: Randomly selected adult clients entering
public sector substance abuse treatment in two California
counties from June, 2000 to June, 2001.
Principal Findings: During the index episode of treatment,
SouthState averaged significantly more group - 25.7 per user v.
13.9 per user- and individual treatment services - 7.3 v. 3.8 per
user - than MidState. Residential services per client were not
significantly different (48.7 v. 34.8 days per user). Results of
the OLS regression analysis show that MidState costs for the
index treatment were significantly lower even when adjusting
for predictor variables that might tend to explain the
differences in costs between counties. An earlier report found
no significant differences in outcomes between counties. As a
result, MidState has superior cost effectiveness in all seven
Addiction Severity Index functional areas due to its lower cost
per client treated.
Conclusions: Costs of treatment during the index episode
were lower in the county that has adopted a managed care
approach to treatment, yet outcomes did not differ between
that and the comparison county. Because of that lower cost,
MidState can claim greater cost effectiveness for its treatment
dollars. The lower costs were due to lower outpatient
utilization in MidState than in SouthState. Notably, residential
treatment days were not significantly different between the
counties: MidState had explicit policies for residential
programs to average 45 days for their clients, while SouthState
left decision-making on length of stay to their residential
programs.
Implications for Policy, Delivery or Practice: That MidState
utilized fewer outpatient services could be cause for concern,
however this utilization is higher than for typical treatment
programs. It may be that SouthState could reduce their
outpatient utilization with little or no impact on outcomes. In
this time of cost containment and reduction in budgets this
strategy might be applicable.
Primary Funding Source: NIDA
• The Cost Offset of Behavioral Health Treatment in
Medicaid
Jeremy Bray, Ph.D., Todd Grabill, B.A.
Presented by: Jeremy Bray, Ph.D., Senior Research
Economist, Health, Social, and Economics Research, RTI
International, 3040 Cornwallis Road, Hobbs Building,
Research Triangle Park, NC 27709; Tel: 919.541.7003; Fax:
919.541.6683; E-mail: bray@rti.org
Research Objective: For more than 20 years, the potential for
a cost offset of behavioral health treatment has been a key
area of interest to researchers and policy makers. A cost offset
is achieved when utilization of behavioral health services
results in a reduction in expenditures for other medical
services. Cost-offset findings have been used to bolster recent
parity legislation, urging equal insurance coverage for
behavioral health and other medical services. Yet little has
been done to assess the extent of a behavioral health
treatment cost offset in publicly financed health plans. This
study analyzes the claims history of Medicaid recipients with
identified behavioral health disorders in order to estimate a
cost offset of behavioral health treatment.
Study Design: Secondary data analysis is conducted using
administrative claims data from the State Medicaid Research
Files of Michigan, New Jersey, Pennsylvania, and Washington
from 1995 to 1998.
Adults ages 21-64 with an identified mental health and/or
substance abuse - MH/SA - disorder and a comparison
sample of randomly selected non-MH/SA claimants are
followed in each of the four states over the 4-year period.
Their fee-for-service claims are used to generate a monthly
longitudinal database. Multivariate regression models control
for demographic and case-mix differences across individuals.
Time-series analysis is conducted, and mean expenditures are
plotted for pre- and post-treatment months.
Population Studied: This study focuses on the adult
Medicaid populations in Michigan, New Jersey, Pennsylvania,
and Washington.
Principal Findings: Expenditures for non-behavioral health
services rise substantially prior to behavioral health treatment
initiation and peak 1 month prior to treatment. Following
behavioral health treatment initiation, other medical
expenditures decline below pretreatment levels when
measured at 12, 24, and 36 months after treatment initiation.
Total expenditures return to but do not decline below
pretreatment levels, indicating a cost offset but not overall
cost reduction for behavioral health treatment. Other medical
and total expenditures remain above the mean expenditure
levels of the comparison sample.
Conclusions: Several researchers have raised concerns that
the cost-offset hypothesis promotes the justification of
treatment based on cost savings rather than on costeffectiveness or better health, and behavioral health care
therefore is held to a higher standard than general medical
care. In demonstrating a cost offset but not an overall cost
reduction for behavioral health treatment under the Medicaid
program, we echo other researchers’ concerns. Our finding
that long-term total costs do not increase following the
initiation of behavioral health treatment suggests that followup treatment does not impose a serious economic burden.
Implications for Policy, Delivery or Practice: Economic
grounds alone should not determine coverage for behavioral
health services and do not correspond to sound public policy.
Nonetheless, our results suggest that economic concerns
should not hinder the expansion of behavioral health coverage
under the Medicaid program. Policy makers may use these
cost-offset findings to support parity legislation but should be
cautioned that treatment need and cost-effectiveness are more
appropriate parameters for justifying behavioral health care.
Primary Funding Source: SAMHSA/CMHS
• The Association Between Federal Block Grants and
Individual Mental Health and Substance Abuse
Expenditures
Alexander Cowell, Ph.D., Jeremy Bray, Ph.D.
Presented by: Alexander Cowell, Ph.D., Research Economist,
Health, Social, and Economics Research, RTI International,
3040 Cornwallis Road, Research Triangle Park, NC 27709; Tel:
919.541.8754; Fax: 919.541.6683; E-mail: cowell@rti.org
Research Objective: Considerable public resources are spent
on mental health and substance abuse - MH/SA. One key
mechanism by which the federal government addresses
MH/SA needs among the population is the substance abuse
and mental health block grants, which provide annual grants
to states. Despite the block grants being a primary source of
funding to address MH/SA needs, the impact of such public
expenditures and policies on individual behavior has received
little attention in the literature. To help address this research
gap, this project uses data from several sources to assess the
impact of the substance abuse and mental health block grants
on individual mental health and substance abuse - MH/SA expenditures.
Study Design: We conduct secondary data analysis using
administrative claims from the private sector for 1997
matched with data containing a comprehensive list of policy
variables. MarketScan data provide information on plan
characteristics, key demographics, and MH/SA expenditures
for over 380,000 claimants with private, fee-for-service
insurance. Matched to these individual data are several
policy-relevant state-level variables measuring: the amounts of
the federal substance abuse block grant and the mental health
block grant; statutory state fines and jail terms for marijuana
possession; cigarette and beer taxes; and adult and juvenile
crime rates per thousand population by type of crime.
Regression analysis is used to examine the impact of the two
types of block grant on whether individuals seek MH/SA care
and on MH/SA expenditures, conditional on seeking care.
Population Studied: This study focuses on adults covered by
private sector fee-for-service insurance plans.
Principal Findings: Results indicate that individuals are very
responsive to changes in both the mental health and
substance abuse block grants. The results also indicate that
the response to mental health block grants can be very
different from the response to substance abuse block grants.
Conclusions: The federal government relies heavily on two
types of block grant – the mental health block grant and the
substance abuse block grant – to help deliver services to those
in need. This paper represents the first research that
examines how these important funding mechanisms affect the
use of MH/SA care. The findings indicate that both kinds of
block grants affect individual decisions and utilization of
MH/SA care.
Simulations indicate that a 10 percent increase in the mental
health block grant is associated with a 14.1 percent increase in
the likelihood of using MH/SA care, but a total decrease in
MH/SA payments of approximately $22 million. A 10 percent
increase in the substance abuse block grant is associated with
a 19.8 percent decrease in MH/SA care and a total increase on
MH/SA payments of about $4 million.
Implications for Policy, Delivery or Practice: The findings
indicate that the block grant mechanisms operating in the
public domain have spill-over effects on MH/SA utilization in
the private domain.
Primary Funding Source: SAMHSA/CMHS
• Effect of Managed Care on Treatment Costs for a
Medicaid Population with Psychiatric Disabilities
Kristine Jones, Ph.D., Neil Jordan, Ph.D., Huey Jen Chen,
Ph.D., Roger Boothroyd, Ph.D., David Shern, Ph.D.
Presented by: Neil Jordan, Ph.D., Assistant Professor, Mental
Health Law & Policy, University of South Florida, 13301 Bruce
B Downs Boulevard, MHC 2713, Tampa, FL 33612; Tel:
813.974.9243; Fax: 813.974.9327; E-mail:
njordan@fmhi.usf.edu
Research Objective: To examine the effects of differing
financial risk arrangements for health, mental health and
pharmacy services on the costs of these services and on total
social costs, with particular attention to cost containment and
cost shifting for individuals with severe mental illness.
Study Design: Study subjects were interviewed bi-monthly for
15 months following enrollment. Comprehensive service
utilization data were collected, including information about
mental health, substance abuse, physical health, pharmacy,
and social and other human services (e.g., corrections)
provided by both formal and informal providers, and on and
off of the Medicaid budget. Administrative data were used to
validate and supplement the self-report data. Details on
providers allowed us to investigate the differences among
financing conditions in the overall distribution of financial
burden between formal and informal providers, among
sectors (health, pharmacy, social services, etc.) and payors.
Rate information was used to calculate costs, and a two-part
regression model was employed to test for differences among
risk conditions.
Population Studied: Study subjects were adults with severe
mental illness (SMI) who received Medicaid Supplemental
Security Income benefits (n=630). Collectively, 72% of the
subjects were women, they averaged 42 years old, and were
47% white, 32% Black, and 15% Hispanic. Given this was an
observational study and did not involve random assignment,
there were some differences found across the three financing
arrangements with respect to subject demographics.
Subjects were enrolled in either a fee-for-service (FFS), carvein, or mental health carve-out plan. In the FFS arrangement,
all enrollees’ health, mental health, and pharmacy services
were paid on a fee-for-service basis with no financial risk to
providers. In the carve-in arrangement all enrollees’ health,
mental health and pharmacy services were paid using a
capitated per member per month (PMPM) payment, with
providers bearing full risk for these services. In the mental
health carve-out arrangement, all enrollees’ mental health
services were paid using a PMPM capitated rate while their
health and pharmacy services were paid on a fee-for-service,
non–risk basis.
Principal Findings: Total treatment costs were found to
significantly vary depending on risk arrangements. Differential
pharmacy costs were principally responsible for these
differences. No differences in total social costs were found
between conditions, with individuals in the conditions with
financial risk showing a relatively greater use of out of plan
and informal resources than persons in the FFS condition.
Conclusions: Financial risk is significantly related to access to
pharmaceuticals in this sample of Medicaid recipients with
severe mental illness. There was modest evidence of cost
shifting for persons in managed care plans, with greater
financial risk associated with greater off-plan costs.
Implications for Policy, Delivery or Practice: When
evaluating the impact of Medicaid managed care on access
and cost of services for persons with SMI, it is important to
use a societal perspective because overall cost savings within
health budgets may be offset by other social costs.
Primary Funding Source: , Substance Abuse and Mental
Health Services Administration (Grant # 5 UR7 TI11278)
Related Posters
Poster Session B
Tuesday, June 8 • 7:30 a.m.-8:45 a.m.
• Physicians’ Diagnoses Associated with Antidepressant
Use between 1995 and 2001
Ayse Akincigil, Ph.D., Benjamin Druss, M.D., M.P.H., John
Bowblis, M.A., Stephen Crystal, Ph.D.
Presented by: Ayse Akincigil, Ph.D., Research Associate,
Institute for Health, Health Care Policy and Aging Research,
Rutgers, The State University of New Jersey, 30 College
Avenue, New Brunswick, NJ 08901; Tel: 732.932.5348; Fax:
732.932.8592; E-mail: aakinci@rci.rutgers.edu
Research Objective: Describe the trends in clinical diagnoses
underlying the prescription of antidepressant.
Study Design: Cross sectional data from National Ambulatory
Medical Care Survey and National Hospital Ambulatory
Medical Care Survey, 1995 to 2001, a nationally representative
sample of ambulatory care providers in physician offices and
hospital outpatient departments.
Population Studied: Patients aged 18 or older, who were
prescribed antidepressants.
Principal Findings: In 1995, antidepressants were prescribed
at 5.5% of visits, compared to 8.5% in 2001. Of those
receiving antidepressants, 39% were diagnosed with
depression, 5% with anxiety, 12% with some other mental
illness (including bipolar and schizophrenia), and 44% had no
primary or secondary diagnosis of mental illness In 1995, 10
million antidepressant prescriptions were written without any
diagnosis of mental illness, rising to 23 million in 2001. The
most common diagnoses other than mental disorders for
visits at which antidepressants were prescribed were diseases
of the circulatory system (25%), chronic joint pain (16%),
diabetes (9%), chronic back or neck pain (8%), migraine or
headaches (7%), and cancer (6%).
Conclusions: The prescription of antidepressants rose sharply
between 1995 and 2001. Use without a diagnosis of
depression or other mental disorder increased even faster
during the period.
Implications for Policy, Delivery or Practice: The rising rate
of use of antidepressants in patients without mental
diagnoses suggests either rising off-label use of these
treatments for other conditions, or else their use for
unreported (e.g., subsyndromal) mental problems. Under
either scenario, there appears to be a widening use of
antidepressants beyond populations for whom they have been
shown to be effective.
Primary Funding Source: NIMH, AHSR
• Employing Community-Based Research Methods to
Describe the Process of Competitive Employment for
Persons with Serious Mental Illness
Marjorie Baldwin, Ph.D., Rebecca White, M.P.H.
Presented by: Marjorie Baldwin, Ph.D., Professor, School of
Health Administration & Policy, W. P. Carey School of
Business, Arizona State University, Box 4506, Tempe, AZ
85287-4506; Tel: 480.965.7868; Fax: 480.965.6654; E-mail:
Marjorie.Baldwin@asu.edu
Research Objective: This work applies community-based
participatory research methods to develop a comprehensive
model of mental illness and work disability. The research
objective is to identify important determinants of employment
outcomes for persons with serious mental illness that are not
currently measured, or measurable, on national survey data.
Study Design: The study uses qualitative, community-based
methods to assess the experiences of persons with mental
illness in the competitive workplace and identify factors that
facilitate or impede success in competitive jobs. In the first
phase a series of focus groups was conducted with mental
health professionals. In the second phase individual semistructured interviews were conducted with employers and
mental health services consumers. The researchers assessed
how illness characteristics, individual characteristics, family
and community characteristics, productivity-related
characteristics, and work-related characteristics affect
employment outcomes for this population.
Population Studied: This research targeted three stakeholder
groups: (1) mental health services professionals, (2)
supervisors and managers who have hired or employed a
person with mental illness, and (3) persons with serious
mental illness who are working or trying to work in
competitive employment and living in the community.
Principal Findings: The information obtained through
interviews and focus groups paints a rich picture of a range of
employment experiences. Many persons with even the most
serious mental illnesses are working. Persons with mental
illness work in the full spectrum of jobs, ranging from
professional to service to unskilled laborer positions.
Important characteristics related to the individual, his or her
illness, family and community support systems, productivity,
and work environment have been identified as factors that
may influence employment outcomes for this population.
Conclusions: Although there are a number of studies of
persons with mental disorders in transitional or supported
employment settings, this is the first study to provide in-depth
analyses of the experience of persons with mental disorders in
the competitive labor market. Developing and testing a
comprehensive model of mental illness and work disability is a
fundamental step toward advancing mental health services by
recognizing employment as a primary component of the
recovery process.
Implications for Policy, Delivery or Practice: While national
surveys have examined issues related to competitive
employment, mental health issues have been given only
cursory attention, possibly because no one knows the right
questions to ask, or because it is assumed that persons with
serious mental disorders are unable to work. This study helps
to identify the concepts that need to be measured to support a
more comprehensive analysis of persons with mental
disorders in the workplace. The results will inform the
development of a survey instrument that will be used to
obtain sufficient data, with adequate samples of persons with
different types of mental disorders, to support a
comprehensive analysis of the important determinants of
labor market outcomes for persons with serious mental
illness. Potentially, the findings will contribute to improving
employment outcomes for future generations of persons with
mental illness who want to work.
Primary Funding Source: Office of Research and Sponsored
Projects Administration, Arizona State University
• Health Services for Women in Outpatient Substance
Abuse Treatment
Cynthia Campbell, Ph.D., M.P.H., Jeffrey Alexander, Ph.D.
Presented by: Cynthia Campbell, Ph.D., M.P.H., Postdoctoral
Fellow, Psychiatry/Division of Research, UCSF/Kaiser
Permanente, 2000 Broadway, 3rd Floor, Oakland, CA 94612;
Tel: 510.891.3584; Fax: 510.891.3606; E-mail:
cynthia.i.campbell@kp.org
Research Objective: Research suggests that gender-sensitive
services can positively influence outcomes for women in
substance abuse treatment, yet some treatment organizations
fail to provide services that are sensitive to women's needs.
This study evaluates how a sample of outpatient substance
abuse treatment units respond to organizational and
environmental influences by adopting and implementing
health services for women.
Study Design: A national sample of outpatient drug abuse
treatment units were surveyed in 1995 and 2000 about their
organizational structure and treatment practices. The
dependent variables in this paper are health services for
women in substance abuse treatment. The predictors include
organizational and environmental factors that represent
resource dependence and institutional pressures for the
treatment unit. Logistic regression and Heckman selection
models were used to test study hypotheses.
Population Studied: The program directors and clinical
supervisors at each treatment unit, who were interviewed by
telephone in 1995 and 2000.
Principal Findings: Units that depended on specific funding
for women’s programs and that depended on government
funds were more likely to adopt, but not necessarily
implement, women’s services. Methadone units and units
that train more staff to work with female clients were more
likely to adopt as well as implement women’s services. Private
not-for-profit units were more likely to adopt some services,
while for-profit units were less so. However, in general,
neither for-profit or not-for-profit unit significantly
implemented services. There was evidence that the odds of
adopting services were greater in 2000 than 1995 for two
services, but were otherwise stable.
Conclusions: There is considerable variation in the adoption
and implementation of women’s services. In addition, not all
adopted services were significantly implemented, which could
reflect limited organizational resources and/or conflicting
expectations. This also suggests that referral mechanisms to
these services, and therefore access, may not be adequate.
Government funds and specific funds for women’s programs
are important resources for the provision of these services.
Women’s services appear more available in methadone units,
suggesting that regulation has been influential and that the
recent methadone accreditation system should be evaluated.
Staff training may be one strategy to encourage
implementation of these services. For the most part, the
adoption of services for women did not change between 1995
and 2000.
Implications for Policy, Delivery or Practice: Health services
for women in substance abuse treatment can be important for
improving outcomes, yet we know little about the factors that
lead to the provision of these services. This study found that
continued government funding is critical for adoption. In
addition, efforts to develop and strengthen linkages to other
resources specifically for women’s services should be
undertaken. One option to encourage implementation of
services is to train staff to work with female clients. JCAHO
accreditation could also be useful for improving treatment if
the criteria were broadened to include women’s services.
While overall there were no decreases in the level of services,
it is quite possible that the effects of lower federal funding
have not yet become apparent and this should be monitored
in the future.
Primary Funding Source: NIDA
• Concept Mapping for Mental Health Services Evaluation
and Planning: Results from a Project Involving Providers,
Consumers, and Family Caregivers
Edie Cook, M.S., ABD
Presented by: Edie Cook, M.S., ABD, Evaluation Specialist,
Graduate School of Public Health; Center of Public Health
Practice, University of Pittsburgh, 3109 Forbes Avenue,
Pittsburgh, PA 15260; Tel: 412.383.2400; 412.383.2883; Fax:
412.383.2228; E-mail: cooke@edc.pitt.edu
Research Objective: To enhance the quality of
comprehensive mental health services planning through an
increased involvement of hospital and community providers,
consumers, and family caregivers in the evaluation of mental
health treatment. The Psychiatric Services Planning Project
was implemented as a collaborative effort to improve
transition from hospital to community mental health services.
Study Design: Concept mapping (Trochim, 1989) consists of
a graphical representation of multidimensional scaling and
cluster analyses of group sorting and rating tasks and was
chosen as an innovative tool to facilitate understanding
between groups. Participant groups were actively involved in
all four phases in the project: 1) the development of a
classification of area mental health services, 2) provider,
consumer, and family caregiver interpretation of data
represented by concept maps, 3) recommendations for
improvement, and, 3) a prioritization of focus areas for further
evaluation research.
Population Studied: Participant groups were comprised of
thirty-two psychiatric hospital staff, twenty-seven community
mental health service providers, thirty-three consumers of
mental health services, and eighteen family caregivers of
adults with chronic mental illness.
Principal Findings: Concept Mapping yielded graphical
representations which were helpful during all phases of the
project, but was particularly useful for developing a
classification of area mental health services due to use of
cluster analysis and labeling of clusters in individual group
interpretation sessions. Recommendations from consumers
for improvement reflected a desire for increased educational
opportunities and greater independence, and family caregivers
focused on discharge planning, medication management, and
family rights to information. The prioritization method piloted
in this study generated patient aftercare, family education, and
relapse prevention as the three top focus areas for further
evaluation attention.
Conclusions: Concept mapping can be a useful tool for
facilitating the involvement of multiple stakeholders in a single
study or evaluation project. The data derived through the
concept mapping activities and analyses were used in the
development of a classification of mental health services as
well as in program evaluation and treatment quality
improvement.
Implications for Policy, Delivery or Practice: Health service
planners and policy makers may want to consider concept
mapping for evaluation and planning projects which involve
multiple stakeholders, whether they represent different
disciplines of hospital staff, or community and agency
providers. Projects which call for modeling consumer and
family caregiver perceived treatment needs, or for
conceptualizing aspects involving care or treatment delivery
would benefit from the visual representation of data that
concept mapping provides.
Primary Funding Source: Mayo Clinic Psychiatry and
Psychology Department
• Effect of Medicaid Managed Care on Somatic Health
Care Utilization for Adults with Severe and Persistent
Mental Illness in Maryland
Gail Daumit, M.D., M.H.S., Wenke Hwang, Ph.D., Christopher
Anthony, Anthony Lehman, M.D., Donald Steinwachs, Ph.D.
Presented by: Gail Daumit, M.D., M.H.S., Assistant Professor
of Medicine, Johns Hopkins University School of Medicine,
2024 East Monument Street, Suite 2-500, Baltimore, MD
21205; Tel: 410.614.6460; Fax: 410.614.0588; E-mail:
gdaumit@jhmi.edu
Research Objective: Individuals with severe and persistent
mental illnesses such as schizophrenia and bipolar disorder
have a high burdenof medical comorbid disease, and they
often receive health insurance benefits through Medicaid,
Medicare or both. Beginning in 1997, Maryland capitated
somatic health care services for those Medicaid recipients
without Medicare coverage while maintaining mental health
services in fee-for-service. Reciepients with dual coverage
remained in fee-for-service Medicaid for both somatic and
mental health care. Effects of Medicaid managed care on
medical service utilization for vulnerable, disabled populations
are understudied. This policy change provided a natural
experiment to examine if the transition to managed care had
any adverse impact on utilization and quality of somatic health
care for individuals with severe mental illness in Maryland
Study Design: We performed a cohort study of Maryland
Medicaid recipients with severe and persistent mental illness
before (1996) and after (2000) implementation of Medicaid
managed care for somatic care.Participants without Medicare
dual coverage had managed Medicaid in 2000. Our primary
outcomes were hospitalizations and emergency department
visits for medical diagnoses and hospitalizations for
ambulatory sensitive conditions. We calculated outcomes per
person-time adjusting for age, race and gender of recipients,
and also performed multiple logistic regression using
generalized estimating equations to account for events per
individual in different years.
Population Studied: Two-thousand eight hundred seventynine adult Maryland Medicaid recipients with schizophrenia or
bipolar disorder identified in 1994 and followed through 2000.
Principal Findings: The mean age was 43.3 years; 49% were
women, 42% were African American, and 89% lived in an
urban area. Seventy percent had schizophrenia; 32% had
Medicare and Medicaid dual coverage. The following analyses
are adjusted for age, race, gender and person-years of followup. In 1996, participants with Medicare/Medicaid had 0.87
(95% CI, 0.78-0.95) medical emergency department (ED)
visits/person-year; participants with Medicaid had 0.91
(95%CI, 0.83-1.00) visits/person-year. In 2000, participants
with Medicare/Medicaid had 0.70 (95% CI, 0.63-0.78) medical
ED visits/person-year; participants with Medicaid managed
care had 0.75 (95%CI, 0.68-0.83) visits/person-year. In 1996,
participants with Medicare/Medicaid had 0.28 (95% CI, 0.240.32) medical hospitalizations/person-year; participants with
Medicaid had 0.29 (95%CI, 0.26-0.33) per person-year. In
2000, participants with Medicare/Medicaid had 0.22 (95% CI,
0.18-0.25) medical hospitalizations /person-year; participants
with Medicaid managed care had 0.22 (95%CI, 0.18-0.25) per
person-year. In multivariate analyses adjusted for
sociodemographic and clinical characteristics, the relative
odds of any medical ED visit were 0.82 (95% CI 0.67-0.99) for
participants in Medicaid managed care compared to those not
in Medicaid managed care. The adjusted relative odds of any
medical hospitalization were lower in 2000 (OR 0.74, 95% CI
0.56-0.97) than in 1996. Participants’ ambulatory sensitive
admissions remained constant at 0.05 (95% CI 0.042-0.060)
per person-year from 1996 to 2000 in both Medicaid and
Medicare/Medicaid in 1996 and 2000.
Conclusions: These Medicaid recipients with severe mental
illness had high levels of acute care utilization. We found
lower medical emergency department utilization with
transition to Medicaid managed care. Medical hospitalizations
decreased from 1996 to 2000 for both managed care and feefor-service programs. Rates of ambulatory sensitive
hospitalizations remained high.
Implications for Policy, Delivery or Practice: Medicaid
managed care in Maryland may have contributed to a
reduction in use of the medical emergency department for
recipients with severe and persistent mental illness. While it is
reassuring that admissions for ambulatory sensitive
conditions did not increase, the level of these admissions is
relatively high compared to non-disabled populations. Future
work should examine the effect of Medicaid managed care on
outpatient health utilization for this vulnerable population.
Primary Funding Source: NIMH
• Substance Abuse Treatment among Opioid Pain
Medication Users with Private Insurance
Andrea DeVries, Ph.D., Katherine Harris, Ph.D., Kelli
Dimidjian, M.B.A., Brian Day, Ed.D
Presented by: Andrea DeVries, Ph.D., Head of Health
Services Research, Healthcare Informatics, Research &
Analysis, Highmark Blue Shield, 120 Fifth Avenue, Suite
P7250, Pittsburgh, PA 15222-3099; Tel: 412.544.0794; Fax:
412.544.0700; E-mail: andrea.devries@highmark.com
Research Objective: To (1) determine whether frequent users
of opioid pain medication have an increased likelihood of
seeking substance abuse treatment. (2) identify the
proportion of pain medication users who have a history of
behavioral health problems. (3) identify the extent to which a
history of behavioral health problems increases the likelihood
for substance abuse treatment among pain medication users.
Study Design: Medical and pharmacy claims data are
analyzed for a large insured population over a two-year period.
Members with frequent opioid pain medication use are
identified, where “frequent use” is defined as a >90 day supply
in a 12 month period. Substance abuse treatment rates
among the pain medication users are compared to general
population treatment rates for this same period, adjusted by
age and gender. Among the pain medication users, members
with a recent history of behavioral health problems are
identified, where ‘recent history’ is defined by either mental
health or substance abuse services provided during the
preceeding 12 month period. The rates of substance abuse
treatment among pain medication users with a recent history
of behavioral health problems are calculated separately.
Population Studied: The population is drawn from a large
mid-Atlantic health insurer with more than 3 million members
enrolled in a range of insurance products. We analyzed the
subpopulation of 1.5 million members between the ages of 18
and 65 with both prescription drug and behavioral health
benefits provided through the health plan. There were 22,529
individuals identified as being frequent users of opioid pain
medication.
Principal Findings: Preliminary analyses suggest that
members with frequent use of opioid pain medication had
higher rates of substance abuse treatment. The pain
medication users were 5 times more likely to receive
substance abuse treatment services as compared to the
general health plan population. Pain medication users with a
recent history of behavioral health problems were 15 times
more likely to receive substance abuse treatment services as
compared to the general health plan population. Members
with a recent history of behavioral health problems
represented 17% of the total pain medication users and 47%
of pain users receiving substance abuse treatment services.
Conclusions: Our results are consistent with recent studies
suggesting that official statistics underestimate the problem of
addiction to prescription pain medication. However, a patient
population with vulnerability to addiction may be responsible
for the observed higher treatment rates. Individuals with a
history of substance abuse disorders and mental health
problems comprise a disproportionate share of individuals
who frequently use opioid pain medications.
Implications for Policy, Delivery or Practice: When
prescribing opioid pain medication providers should screen
for risk factors that would increase the likelihood of misuse
and indicate history of untreated or inadequately treated
behavioral health problems. Recognition and treatment of
comorbid mental health problems may reduce the likelihood
of pain medication misuse. Future research is needed on the
relationship between chronic pain, mental health disorders,
and vulnerability to addiction.
• Work Disability and Barriers to Evidence-Based
Psychosocial Treatments
Farifteh Duffy, Ph.D., Joyce West, Ph.D., M.P.P., William
Narrow, M.D., M.P.H., Ilze Ruditis, M.S.W., Lisa Countis,
Darrel Regier, M.D., M.P.H.
Presented by: Farifteh Duffy, Ph.D., Research Scientist, ,
American Psychiatric Institute for Research & Education, 1000
Wilson Boulevard, Suite 1825, Arlington, VA 22209; Tel:
703.907.8620; Fax: 703.907.1087; E-mail: fduffy@psych.org
Research Objective: Disabilities associated with mental
disorders are equally or more debilitating than those
associated with chronic general medical conditions (Von Korff
1994, Wells et al., 1989). This study examines: 1) rates and
correlates of work-disability among adult psychiatric patients;
and 2) rates and correlates of receiving psychosocial services
(i.e., vocational rehabilitation, social-skills training, and casemanagement services) among work-disabled patients.
Study Design: Cross-sectional observational data from the
1999 American Psychiatric Practice Research Network (PRN)
Study of Psychiatric Patients and Treatments (SPPT) were
used. A national sample of 1,596 psychiatric patients 18 years
of age or older (median age=44) were systematically selected
by 615 psychiatrist-members of the American Psychiatric
Practice Research Network (PRN). Data were weighted to
provide nationally representative estimates.
Population Studied: Adult psychiatric patients (n=1596);
approximately 43% of patients were male; 79% white, 10%
African American, and 7% Hispanic.
Principal Findings: 33% of patients were not working due to
disability (i.e., work-disabled). Only 15% of the work-disabled
patients received any vocational rehabilitation (1.1%), social
skills training (4.9%), or case management (12.4%) services
within a 30-day period. Highest rates of work-disability were
observed among patients with: schizophrenia (67%),
substance use (43%), and bipolar (40%) disorders although
substantial levels of disability were also found in patients with
personality (34%), anxiety (26%) and depressive (24%)
disorders. Multiple logistic regression analyses indicated
patients with schizophrenia, co-occurring general medical
conditions, multiple psychosocial problems, public insurance,
older age, and less education were more likely to be workdisabled. Among work-disabled patients, those with
schizophrenia, Axis II comorbidity, and public insurance were
more likely to receive any of the aforementioned psychosocial
services.
Conclusions: One out of every three adult psychiatric patients
treated by psychiatrist is reported to be work-disabled, yet only
one out of every seven work-disabled patients received any
psychosocial services. Socioeconomically disadvantaged
patients and patients with clinically complex conditions were
more likely to be work-disabled, and were more likely to
receive psychosocial services.
Implications for Policy, Delivery or Practice: This study
highlights a number of patient subgroups at increased risk for
work-disability. Despite evidence in support of psychosocial
services for persons with serious mental illness (Surgeon
General’s Report 1999, PORT 2002, Bond et al. 2001), our
findings indicate only a small minority of work-disabled
patients received any vocational rehabilitation, social skills
training, or case management services (15%). Although not
every work-disabled individual would require, seek, or benefit
from psychosocial interventions (depending on duration and
severity of the person’s disability), models of service delivery
which aim to improve the coordination and integration of
services and strive toward the provision of individually-tailored
treatment programs are critical to meet the needs of persons
with work-disability. Current health care financing and services
delivery policies and interventions, however, need to address
the gaps in insurance coverage and enhance the organization
and financing of services that are considered major barriers to
access (Surgeon General’s Report 1999).
Primary Funding Source: , The Center for Mental Health
Services, the American Psychiatric Foundation, the Center for
Substance Abuse Treatment, and the John D. and Catherine T.
MacArthur Foundation.
• An Examination of ED Psychiatric Consultation
Arrangements
Jennifer Field, B.A., B.S.N., M.S.N., Ph.D. Candidate
Presented by: Jennifer Field, B.A., B.S.N., M.S.N., Ph.D.
Candidate, Psychiatric Clinical Nurse Specialist, Health
Administration, Virginia Commonwealth University, 718
Greensboro Avenue, Virginia Beach, VA 23451; Tel:
757.491.7985; E-mail: jlfield@hsc.vcu.edu
Research Objective: The purpose of this study was to
examine the degree to which environmental and economic
factors affect an emergency department's psychiatric
consultation arrangement and how arrangement influences
service effectiveness. It answers two questions:
1) What factors lead hospitals to create an emergency
department psychiatric consultation service, to purchase the
services from an outsider, or to not provide the service?
2) Does variation in the consultation arrangement have a
significant impact on service outcomes
Study Design: The study utilizes a balanced cross-sectional
design. It retrospectively examines hospital ED psychiatric
service arrangements and outcomes. It uses a combination of
primary and secondary data gathered from several sources.
ED encounter data was obtained from two state ambulatory
care and inpatient databases (Maryland and South Carolina).
America Hospital Association and Area Resource File
databases provided information on facility characteristics. A
survey of ED administrators from the participating facilities
was conducted to capture information about facility
psychiatric consultation arrangements. Regression and
multilevel structural equation modeling were the chosen
methods of analysis.
Population Studied: The target sample included all
nonfederal acute care medical-surgical hospitals that have
emergency departments in South Carolina and Maryland
(N=102). Seventy-one facilities responded to the survey
resulting in a 69.6% response rate. Respondents were
categorized by type of PES arrangement (in-house, contract, or
no service) and this characteristic was the focus of subsequent
analysis. Data from 66 randomly selected ED psychiatric
patients was drawn from each participating hospital yielding a
sample N=71*66=4686.
Principal Findings: The hypotheses were derived from a
conceptual model based upon the principles of transaction
cost economics (TCE). The model proposes that
organizations choose the arrangement that incurs the lowest
transaction costs, which are determined by the degree of
environmental and behavioral uncertainty, asset specificity,
and transaction frequency. The results support each TCE
construct as a statistically significant predictor of
arrangement; however, not all indicators demonstrated the
predicted relationships. In addition, the conceptual model
proposed that arrangement influences outcomes (time,
disposition, readmission, and cost). The results
demonstrated arrangement significantly affected outcomes
but only partially supported the proposition that vertically
integrated services are associated with more positive
outcomes.
Conclusions: The results support the application of TCE
theory to explain ED psychiatric service variation and provide
new information about the effects of arrangement on
outcomes.
Implications for Policy, Delivery or Practice: The study
confirms that variation in ED psychiatric consultation
arrangement exists and has an influence on outcomes.
Hospital administrators, accrediting organizations and
insurers should explore this further to confirm the conditions
and type of arrangement that are necessary to produce the
most promising outcomes for the patient and organization.
Policy makers and accreditors may want to impose guidelines
that reduce variation in effort to create more comparable units
for evaluation. Health insurers could use the findings as
performance criteria for network selection. Finally, this study
represents the creative merging of multiple data sources to
develop a comprehensive portrayal of psychiatric service
utilization. It demonstrates the utility of state ambulatory
service data encouraging future use of an HCUP emergency
service database currently under development.
Primary Funding Source: AHRQ
• Mental Illness, Traumatic Brain Injury, and Medicaid
Expenditures
Wenhui Wei, M.A., M.S., Patricia Findley, Dr.PH, M.S.W.,
Usha Sambamoorthi, Ph.D., Stephen Crystal, Ph.D.
Presented by: Patricia Findley, Dr.PH, M.S.W., Asst Research
Professor, Program for Disability Research, Rutgers University,
303 George Street, New Brunswick, NJ 08901; Tel:
732.932.3421; Fax: 732.932.1894; E-mail:
pfindley@rci.rutgers.edu
Research Objective: Use Medicaid administrative data to
analyze variations in rates of mental illness and annual
Medicaid expenditures for individuals with diagnosed
traumatic brain injury (TBI) by their sociodemographic and
health characteristics
Study Design: The study used personal summary information
and paid Medicaid claims data from State Medicaid Research
files for Alabama, Georgia, New Jersey, and Wisconsin. Annual
expenditures for total, inpatient, and non-inpatient services
were derived from personal summary files. Mental illness and
TBI were derived based on International Classification of
Diseases 9th revision, Clinical Modification codes recorded in
paid Medicaid claims. Chi-square statistics were used to test
for subgroup differences in the proportion of individuals with
TBI and mental illness. Kaplan-Meier survival techniques were
used to estimate annual expenditures for different types of
services and t-tests were used to test the statistical
significance of bivariate subgroup differences in annual and
monthly expenditures. Finally, Ordinary Least Squares
regression was used on average monthly expenditures, and
Cox-Proportional Hazards regression was used on annual
expenditures to determine the extent to which differences in
mental illness account for variations in Medicaid expenditures.
The extrapolation of results into current Medicaid policy on
programs with for individuals with TBI was considered.
Population Studied: Continuously enrolled Medicaid
recipients who were aged 21-64 and diagnosed with TBI in
1995.
Principal Findings: Overall, less than 1% (n = 4,266) were
diagnosed with TBI in four states in 1995. Significant
geographic, demographic, racial, and health status differences
were found in the rates of TBI. Among those diagnosed with
TBI, a majority (67%) did not have a diagnosis of mental
illness and 17% were diagnosed with serious mental illness.
Individuals in the age group 40 – 49 years, racial minorities
other than African American and Latinos, diagnosed with
substance abuse, and burdened with chronic comorbidity
were more likely to have a mental disorder. Significant
differences in estimated total, inpatient and non-inpatient
expenditures by mental illness were also noted. In general,
those with serious mental illness had higher expenditures
than those with out any mental illness. Additionally, the
presence of high burden of chronic comorbidity significantly
increased the overall expenses, most significantly for inpatient
expenses.
Conclusions: This study found that expenditures, although
variable across service types, are generally higher for those
with TBI and severe mental illness. This suggests the overall
need for general awareness that differences do exist in the
expenditures for the dual diagnosis, and that this should be
considered in programming and budget planning, particularly
for tightly budgeted programs like State Medicaid programs.
Implications for Policy, Delivery or Practice: Future
Medicaid policy modifications need to take into consideration
the health and mental health care needs of this duallydiagnosed subpopulation, particularly in aspects that address
home and community based programs.
Primary Funding Source: NIMH, National Insitute of Child
Health and Human Development
• Satisfaction of Impaired Health Care Professionals with
Mandatory Treatment and Monitoring
Carol Fletcher, Ph.D., R.N., David Ronis, Ph.D.
Presented by: Carol Fletcher, Ph.D., R.N., Research Specialist,
Center for Practice Management and Outcomes Research,
Veterans Health Administration, Box 130170, Ann Arbor, MI
48113-0170; Tel: 734.769.7100 Ext. 6212; Fax: 734.761.2617; Email: carol.fletcher@med.va.gov
Research Objective: To determine the level of satisfaction of
active participants mandated into the Michigan Health
Professional Recovery Program, HPRP, and the Indiana State
Nurses Assitance Program, ISNAP.
Study Design: Mailed surveys were used to assess overall
program satisfaction, the usefulness of various aspects of
monitoring, several miscellaneous items, e.g., insurance
coverage and whether the person was currently working in a
position requiring licensure, and basic demographics. Mailed
responses, n=263, 43% for the HPRP; n=120, 45% for ISNAP,
were anonymous.
Population Studied: All active participants in the HPRP,
n=620, and ISNAP, n=267, programs in July 2002. The HPRP
covers >300,000 persons in 19 professions including
physicians, nurses, pharmacists and dentists. ISNAP covers
>100,000 nurses.
Principal Findings: Means on the 1 to 5 satisfaction scale
were 3.6, SD=.94, for the HPRP and 4.1, SD=.86, for ISNAP,
indicating overall satisfaction with the programs. Means on
the 1 to 4 monitoring scale were 2.7, SD=.62, for the HPRP
and 2.7, SD=.57, for ISNAP, indicating most aspects of
monitoring were considered useful. 40% of HPRP
respondents and 53% of ISNAP respondents lacked any
insurance coverage for program costs. 39% of HPRP and 35%
of ISNAP respondents stated they had substance abuse or
mental health difficulties prior to their professional career.
Conclusions: Although 20% or less of the HPRP and ISNAP
participants come into the programs by their own choice,
respondents appear to be generally satisfied with these
mandatory programs once they become involved.
Implications for Policy, Delivery or Practice: Maintaining
patient safety by ensuring health professionals continuing
competency to practice is essential. Impaired professionals
can be mandated into treatment, but true success depends
upon their willingness to follow the program. Satisfied
persons are more likely to truly cooperate. This study
demonstrates that it is possible to create satisfaction with
mandated programs.
Primary Funding Source: Michigan Health Professional
Recovery Corporation
• Outpatient Mental Health Service Use: Racial and Ethnic
Variations?
Nell Forge, Ph.D., M.P.H., M. Audrey Burnam, Ph.D., Cathy
Donald Sherbourne, Ph.D.
Presented by: Nell Forge, Ph.D., M.P.H., Associate Behavioral
Scientist, Health, RAND, 1700 Main Street, Santa Monica, CA
90056; Tel: 310.393.0411; Fax: 310.451.7063; E-mail:
Nell_Forge@rand.org
Research Objective: To examine whether there is any racial
and ethnic variation in the use of outpatient mental health
services and to identify pathways to mental health service use
among a national sample of HIV-infected individuals.
Study Design: This study draws upon baseline data from the
HIV Cost and Services Utilization Study, a national
longitudinal survey of HIV-infected adults receiving HIV care.
Logistic regression models estimated the effect of race and
ethnicity on likelihood of mental health service utilization.
Population Studied: The analytic sample is comprised of
individuals who have a need for mental health services.
Principal Findings: Whites were more likely to receive
psychotropic medications, whereas African Americans were
more likely to use mental health support groups. Perception
of mental health need was a robust pathway to mental health
services, whereas financial and knowledge-related barriers
decreased likelihood of use.
Conclusions: Gaps continue to exist in psychotropic
medication even among racial groups who were receiving care
for their HIV status.
Implications for Policy, Delivery or Practice: Increasing
knowledge of mental health need, coupled with providing
financial mechanisms to access care are needed to address
racial variation in mental health service use.
Primary Funding Source: NIMH
• Translation and Adaptation of a Mental Health Outcome
Measure
Susan Eisen, Ph.D., Mariana Gerena, M.A., Dharma Cortes,
Ph.D., Mariana Gerena, M.A., Glorisa Canino, Ph.D., Sergio
Aguilar-Gaxiola, M.D., Ph.D., Cristina Magana, Ph.D.
Presented by: Mariana Gerena, M.A., , Center for Health
Quality, Outcomes and Economic Research (CHQOER), EN
Rogers Memorial Veterans Hospital, 200 Springs Road (152),
Bedford, MA 02464; Tel: 781.687.3257; Fax: 781-687-3106; Email: gerena@bu.edu
Research Objective: To develop a culturally and linguistically
appropriate mental health assessment instrument to assess
outcome of treatment for Spanish-speaking mental health and
substance abuse service recipients. Culturally and
linguistically appropriate outcome measures are especially
needed to determine the presence and direction of health
disparities, and to assess quality and outcomes of treatment
among minority populations. Clinical outcome is widely
recognized as an important indicator of the quality and
effectiveness of mental health care. However, due to a dearth
of mental health outcome assessment instruments that have
been translated and validated, Spanish-speaking mental health
consumers are often excluded from efforts to assess
treatment outcomes.
Study Design: This study is using qualitative and quantitative
methods to adapt, translate, field test and validate the revised
Behavior and Symptom Identification Scale (BASIS-R), a
mental health instrument developed on English speakers in
the United States. The translation and adaptation used
addresses key dimensions of cross-cultural equivalence:
content, semantic, technical, criterion and concept
equivalence. This presentation summarizes results of the
qualitative methods. A professional translator provided a
preliminary Spanish translation of the instrument, which was
back-translated by a second translator, then reviewed by a
multi-national bilingual committee. Focus groups and
individual cognitive interviews were conducted to obtain
feedback about the content and format of the instrument, and
to assess respondents’ understanding of the instructions,
questionnaire items, time frame and response options.
Respondents’ ratings of clarity and importance of each item
were also obtained.
Population Studied: The sample consisted of Spanishspeaking, adult mental health service recipients from three
regions, Boston, Fresno and San Juan, who are of Dominican,
Mexican or Puerto Rican origin. Thirty-six individuals
participated in one of four focus groups, and 45 individuals
participated in cognitive interviews.
Principal Findings: Analysis of focus group data identified a
number of items reported to be confusing or difficult for
participants. These were deleted or re-worded to clarify or
simplify them prior to conducting cognitive interviews.
Analysis of the cognitive interviews indicated that 22 out of 26
items tested were rated as completely clear by at least 80% of
respondents and 24 out of 26 items were rated as very
important by at least 80% of respondents. Respondents’
qualitative explanation of item meaning did not always
correspond with their rating of item clarity. Thus the
qualitative explanations of item meaning were important in
supplementing ratings of clarity. Several “psychosis” items
were identified by Latino participants as ordinary experience
within their culture and religious beliefs, and consequently, are
not necessarily indicative of psychosis in a Spanish-speaking
population (e.g., “hearing voices or seeing things,” and
“having special powers”).
Conclusions: The combined use of qualitative and
quantitative methods is important for determining crosscultural equivalence of mental health assessment instruments.
Implications for Policy, Delivery or Practice: Assessment
instruments developed on Anglo populations may not be
generalizable to minority groups. Use of inappropriate
instruments may lead to erroneous conclusions about mental
health status or outcomes. Consequently, assessment
instruments should be adapted, tested and validated when
used to assess individuals from other countries or cultures.
Primary Funding Source: NIMH
• Automating Clinical Practice Guidelines for Tobacco Use
Cessation – The 5 A’s: Ask, Assess, Advise, Assist and
Arrange
Jack Green, M.B.A., B.S., James Chesney, Ph.D., Patricia
Williams, MPA, Susan Szpunar, Ph.D.
Presented by: Jack Green, M.B.A., B.S., Sr. Systems Analyst,
Office of Public Policy Initiatives, Henry Ford Health System, 1
Ford Place Drive - OPPI 3C, Detroit, MI 48202; Tel:
313.874.3127; Fax: 313.874.3039; E-mail: JGreen1@hfhs.org
Research Objective: To assess the effectiveness of an
Automated Clinical Practice Guideline (ACPG) for Tobacco
Use Cessation (TUC) designed for use within the electronic
medical record. Outcome variables include TUC guideline
adherence, tobacco use quit rates, and staff satisfaction with
the TUC ACPG.
Study Design: TUC guideline adherence was assessed with a
pretest – posttest nonequivalent control group design. An
intervention group received the TUC ACPG and a business
process re-engineering (BPR) whereby patient vital signs and
check-in notes are entered and viewed within the electronic
medical record (CarePlus) instead of on paper. As the
provider views this information, he or she is prompted, when
appropriate, to follow the TUC guideline. The intervention
group is referred to as the TUC sites. Two control groups
served as comparison: one group received the BPR but not the
TUC ACPG (Check-In sites), and the second group received no
intervention (Control sites). Data on guideline adherence,
tobacco use, and patient demographics were collected
through telephone interviews of research subjects and
through ACPG process point data. Provider reaction to the
ACPG was determined through provider surveys and focus
group sessions. Information was collected on the six month
quit rates of post-implementation subjects who reported
tobacco use. Data analyses include Chi-square tests of
association and logistic regression.
Population Studied: Patients aligned to study or control
physicians at Henry Ford Health System in Detroit, Michigan
were asked to consent to an interview. Of the 44,000 patients
who consented, 8,867 patients at baseline and 7,809 patients
post-implementation presented at the study clinics during the
study months and were included for survey. 5,334 baseline
patients and 3,970 post-implementation patients completed
telephone interviews. 700 post-implementation tobacco users
were resurveyed to assess 6-month quit rates.
Principle Findings: Race, age, and education level for
participants varied slightly among study groups. At baseline
all study groups had high adherence rates; however, the TUC
sites had higher guideline adherence rates than either of the
controls. At the TUC sites, post-implementation adherence
rates increased relative to baseline for all five points of the
guideline with the largest increases seen in the ASSESS and
ARRANGE guideline points (17% and 14% respectively).
Logistic regression results indicated that the probability of a
provider ASKing their patient’s tobacco use status was twice
as high in the TUC sites relative to the no-intervention
Control, controlling for time, related co-morbidities, and race.
In addition, being at a TUC site significantly increased the
probability of the ASSESS, ADVISE, and ARRANGE guideline
points being followed over the control site, after accounting
for time period, demographics and co-morbidities. Only the
ASSIST rate was not significantly related to intervention group.
Six-month quit rates showed no differences among the study
groups; however, the TUC sites were more likely to report use
of assistive therapies in their quit attempts. Provider
understanding and acceptance of the Tobacco Use Cessation
guideline varied among the study groups with the TUC sites
demonstrating the highest acceptance of the requirements.
Time constraints were the largest reported barrier to
compliance with the guideline. Providers from TUC sites
positively evaluated the ACPG implementation, with
unanimous support for the automated means to refer patients
for follow-up.
Conclusions: The TUC ACPG was easily assimilated into the
clinic process and promoted the provision of appropriate
healthcare advice at the appropriate time.
Implications for Policy, Delivery or Practice: Increased
provider adherence with the established TUC guideline should
translate into higher quit rates for current users and less
recidivism for recent quitters. Tobacco-related morbidity and
mortality remain major health concerns in the U.S. with large
financial implications, thus further efforts at encouraging
tobacco use cessation are imperative for the health of the U.S.
and the U.S. economy.
• Self-Medication of Mental Health Problems
Katherine Harris, Ph.D., Mark Edlund, M.D., Ph.D.
Presented by: Katherine Harris, Ph.D., Research Fellow,
Office of Applied Studies, Substance Abuse and Mental
Health Services, 5600 Fishers Lane, 16-105, Rockville, MD
20856; Tel: 301.443.0747; E-mail: kharris@samhsa.gov
Research Objective: To evaluate the association between
past 30 day use of alcohol, marijuana, and other illicit drugs
and past year unmet need for and use of mental health care.
Study Design: A subsample of 18,849 respondents from the
2001 National Household Survey on Drug Abuse and the
2002 National Survey on Drug Use and Health. Subjects were
between the ages of 18 and 65 and had least one past year
mental disorder symptom and no past year substance
dependency. Logistic regressions of past 30-day substance use
on past 12-month unmet need for mental health care and past
12-month use of mental health services controlling for clinical
and socio-demographic characteristics. Predicted
probabilities and corresponding standard errors are reported.
Population Studied: Use of illicit drugs other than marijuana
increased with unmet need for mental health care (4.4% vs.
3.2%, p=.046) but was not reduced with mental health care
use. Heavy alcohol use was not associated with increased
unmet need for mental health care, but was higher among
individuals with no mental health care use (4.4% vs. 2.7%,
p<.001). By contrast, marijuana use did not appear
associated with either unmet need or mental health care use.
Principal Findings: Substance use varies with past year
unmet need for mental health care and mental health care use
in ways consistent with the self-medication hypothesis.
Results suggest that timely screening and treatment of mental
health problems may prevent the development of substance
use disorders among those with mental disorders. Use of
illicit drugs other than marijuana increased with unmet need
for mental health care (4.4% vs. 3.2%, p=.046) but was not
reduced with mental health care use. Heavy alcohol use was
not associated with increased unmet need for mental health
care, but was higher among individuals with no mental health
care use (4.4% vs. 2.7%, p<.001). By contrast, marijuana use
did not appear associated with either unmet need or mental
health care use.
Conclusions: Substance use varies with past year unmet need
for mental health care and mental health care use in ways
consistent with the self-medication hypothesis.
Implications for Policy, Delivery or Practice: Results
suggest that timely screening and treatment of mental health
problems may prevent the development of substance use
disorders among those with mental disorders. Further
research should identify subgroups of individuals for whom
timely and appropriate mental health treatment would prevent
the development of substance use disorders.
• Healthcare Utilization, Expenditures, and Financial
Burden of Families with Persons with Severe Mental
Illness: Results from the 2000 Medical Expenditure Panel
Survey
Denys Lau, Ph.D., Thomas Taylor, Ph.D.
Presented by: Denys Lau, Ph.D., Research Fellow, Outcomes
Research, Pfizer / University of Michigan, 315 S. Division, #9,
Ann Arbor, MI 48104; Tel: 734.622.4011; Fax: 734.622.2641; Email: Denys.Lau@Pfizer.com
Research Objective: (1) Describe the socio-demographic and
health characteristics, and the family situations of persons
with severe mental illness (SMI-persons). (2) Examine the
healthcare utilization, expenditure, and financial burden of
families with a SMI-person (SMI-families).
Study Design: Data were from the 2000 Medical Expenditure
Panel Survey (MEPS), which contains self-reported data on the
socio-demographic and health information, as well as
healthcare utilization and expenditures of a nationally
representative sample of the civilian, non-institutionalized U.S.
population. SMI included schizophrenia (ICD-9-CM codes:
295.XX); affective psychoses (296.XX); paranoia disorder
(297.XX); severe neurotic and personality disorders (300.4 and
301.11-301.13); and other functional psychoses (298.XX and
299.XX); all of which were grouped into three clinically
meaningful categories by the Clinical Classification Software
included in MEPS (CCS codes: 69-71). Financial burden for a
family was the percentage of total family income for out-ofpocket health care expenditures. For non-normally distributed
outcome variables, logarithmic transformation was performed
before conducting bivariate analyses, z-test or t-test,
depending on the un-weighted sample size. Only statistical
significant results were reported (p-value<=0.05). Estimates
were adjusted for the complex multistage survey design of
MEPS.
Population Studied: SMI-persons were estimated to be 1.0%
of the civilian, non-institutionalized U.S. population, or about
2.9 million people. Among SMI-persons, 57.3% or 1.6 million
people were living in a family (household with two or more
persons). Among all U.S. families, 2.1% or 1.5 million families
had a SMI-person, and 39.6% of the SMI-families had a SMIperson as the head of household.
Principal Findings: Compared to the general population,
more of the SMI-persons were between the ages of 18 to 64,
earned less than $10,000, unemployed, on public health
insurance, and reporting poorer self-perceived health, more
functional and role limitations, and more comorbidities.
Compared to other families, more of the SMI-families were
either single adult with children or two adults with no children,
had no private health insurance, and were poor or near poor.
On average, SMI-families collectively had more office-based
visits to healthcare providers (8.2 [SE=0.9] vs. 4.5 [SE=0.1]),
higher overall healthcare expenditures ($2,821 [SE=$329] vs.
$2,294 [SE=$81]), higher pharmaceutical expenditures ($701
[SE=$93] vs. $375 [SE=$13]), and greater financial burden (5.5%
[SE=0.8%]) vs. 3.4% [SE=0.1%]), compared to other families.
Poor or near poor SMI-families on average had four times
greater financial burden as the high-income SMI-families
(8.1% [SE=1.7%] vs. 2.2% [SE=0.3%]). Within SMI-families,
the SMI-person on average had the lion share of the
healthcare utilization and expenditures, with 12.7 [SE=1.7]
office-based visits, $4,856 [SE=$668] overall healthcare
expenditures, and $1,391 [SE=$217] pharmaceutical
expenditures.
Conclusions: Although SMI-families had greater healthcare
utilization and expenditures, and higher financial burden than
other families, the majority of the consumption was incurred
by the SMI-person. Limited by the small un-weighted sample
size, this study could not further investigate how the increased
burden affected the healthcare utilization and expenditures for
family members with no SMI.
Implications for Policy, Delivery or Practice: Adequate
resources may be required for SMI-families, especially the
poor and near poor families, to meet the substantial
healthcare needs of the SMI-person and the social welfare
needs of their family members.
Primary Funding Source: Pfizer, Inc.
• Clinic Variation in Use of Antidepressant Medications for
Primary Care Patients with Depression
Chuan-Fen Liu, M.P.H., Ph.D., Edmund Chaney, Ph.D.,
Duncan Campbell, Ph.D., Yu-Fang Li, Ph.D., Mary McDonell,
M.S., Stephen Fihn, M.D.
Presented by: Chuan-Fen Liu, M.P.H., Ph.D., HSR&D
Investigator, Health Services Research and Development, VA
Puget Sound Health Care System, HSRD (152) 1660 S.
Columbian Way, Seattle, WA 98108; Tel: 206.764.2587; Fax:
206.764.2935; E-mail: chuan-fen.liu@med.va.gov
Research Objective: Depression is under-detected and undertreated in primary care. Yet, antidepressant pharmacotherapy
has been shown to be effective in this setting. There is
apparent variation in antidepressant prescription practices
among settings. Studying the potential sources of this
variation could help guide depression treatment
improvement. This study examined variation in use of
antidepressant medications among VA primary care clinics.
Study Design: This was a retrospective, longitudinal study of
veterans with a positive screen for major depressive
symptoms at entry into the Ambulatory Care Quality
Improvement Project (ACQUIP). Treatment utilization and
pharmacy data were obtained from administrative databases
for each local site. We examined variation in use of
antidepressant medication during one year after the baseline
across the 7 sites. Logistic regression was used to estimate
the probability of any antidepressant use. Zero-inflated
negative binomial regression models were used to estimate
the number fills and the number of days of supply of
antidepressant medications
Population Studied: The study sample included 2,111 patients
in the 7 primary care clinics participating in ACQUIP with
Hopkins Symptom Check List (SCL) scores > =1.75 at baseline
screening, suggestive of significant symptoms of major
depression
Principal Findings: Overall, 52% of patients with significant
symptoms of major depression received at least one fill of
antidepressant medication during the year (range 35- 58%
across clinics). Across clinics, patients with depression
averaged 4.5 fills of antidepressants with a mean supply of 160
days. The average number of fills and the number of days of
supply per patient ranged from 2.8 to 5.7 and 101 to 168 days
across clinics, respectively. After controlling for patient
characteristics, case-mix, and baseline SCL score, depression
history, prior use of antidepressant, and outpatient care
utilization, treatment location (i.e., Primary Care Clinic) was
significantly related to the likelihood of receiving at least one
antidepressant fill, the number of fills, and to the number of
days of supply.
Conclusions: In this sample, only about half of primary care
patients with significant symptoms of major depression were
treated with antidepressant medications. After adjusting for
patient characteristics, there was significant variation in
antidepressant use among primary care clinics, likely reflecting
different practice patterns among the clinics.
Implications for Policy, Delivery or Practice: Further
research is needed to improve practice patterns among
primary care clinics in delivering evidence-based depression
treatment.
Primary Funding Source: VA
• Financial Incentives and Gaming in Alcohol Treatment
Mingshan Lu, Ph.D., Ching-to Albert Ma, Ph.D.
Presented by: Mingshan Lu, Ph.D., Associate Professor,
Department of Economics, University of Calgary, 2500
University Drive, N.W., Calgary, T2N1N4; Tel: 403.220.5488;
Fax: 403-282-5262; E-mail: lu@ucalgary.ca
Research Objective: Whereas a simple fee-for-service (FFS)
allows much flexibility and minimal interference on the
physician-patient relationship, managed care seeks to impose
controls to achieve objectives deemed necessary by health
plans or payers. Against the cost and quality controls in
payment regimes other than FFS, clinicians may find ways to
game the system. Our paper provides direct testing of the
impacts of financial incentives on clinician gaming.
Study Design: We present two data sets on client information
in alcohol treatment episodes. Each of the two data sets
originated from the same clinician. One set consisted of
reports required by the Maine Addiction Treatment System for
administrative and funding assessments; the other was
abstract from actual medical records. Performance Based
Contracting was a policy implemented by the state of Maine
around the middle of the period for our data sets. MATS data
were used by PBC to assess program performance, which then
determined future funding.
There are discrepancies in the two data sets. We hypothesize
that the information inconsistencies are due to clinicians
misreporting information to MATS to game the system. The
implementation of PBC resulted in a financial incentive for
clinicians to misreport information to MATS. Good
performance recorded in MATS can be due to over-reporting
of a client’s alcohol use at admission, or under-reporting at
discharge. We identify over-reporting and under-reporting by
comparing MATS data against medical record abstracts. We
identify gaming incentives by the time dummy of the
implementation of PBC.
We are able to directly test gaming because of our unique data
sets. The comparison between the MATS and medical abstract
data is a straightforward way to identify gaming. The
identification of financial incentives on gaming was due to the
implementation of PBC. Our results call for attention on
provider reactions against incentive mechanisms: information
manipulation should not be ruled out when such actions have
financial consequences.
Population Studied: About 1,000 alcohol abuse treatment
episodes in the state of Maine covering the period from
October 1990 to June 1995.
Principal Findings: We find strong empirical evidence for our
hypotheses. PBC was found to have a significant and positive
effect on clinicians over-reporting client alcohol use at
admission and under-reporting at discharge. Because
performance evaluation is from assessing alcohol use
reduction and abstinence, these misreporting help boost
treatment performance where clients actually have not
improved.
Conclusions: We find that the introduction of PBC has
increased gaming. The pattern of gaming is consistent with
clinicians exaggerating treatment performance to obtain
higher financial rewards.
Implications for Policy, Delivery or Practice: Our study
suggests two policy implications. First, and perhaps more
directly, auditing should be used more often when regulatory
authorities must rely on information supplied by providers for
financial and funding decisions. Appropriate auditing may
deter gaming, and gives more reliability to the veracity of
reports. Second, and perhaps less directly, establishing a gold
standard should be considered whenever it is feasible. In our
case, the analysis was possible precisely because we were able
to compare the administrative reports against an appropriate
standard, namely the medical record abstracts. Having an
independent and reliable data source for testing the reliability
of administrative data may seem obvious but appears to have
received less emphasis. Data collection methods should
consider obtaining the same information in more than one
way.
Primary Funding Source: NIAAA, Alberta Heritage
Foundation for Medical Research
• Clinical Supervision Issues in the Implementation of
Randomized Clinical Trials
Sarah Turcotte Manser, M.A., Joseph Guydish, Ph.D., Barbara
Tajima, MEd, Martha Jessup, R.N., Ph.D.
Presented by: Sarah Turcotte Manser, M.A., Senior
Administrative Analyst, Institute of Health Policy Studies,
University of California, San Francisco, 3333 California Street,
Suite 265, Box 0936, San Francisco, CA 94143-0936; Tel:
415.476.3378; Fax: (415)476-0705; E-mail:
sarahtu@itsa.ucsf.edu
Research Objective: This qualitative study investigated the
dissemination of evidence-based interventions in substance
abuse treatment settings. We studied the extent to which
evidence-based substance abuse treatments were adopted by
programs participating in multi-site clinical trials. This study
explored issues and challenges that emerged in the
implementation of research initiatives within communitybased clinic sites and the factors that influenced adoption of
the new treatment intervention. This study investigated: 1) the
process and extent to which drug abuse treatment
interventions are implemented and adopted by programs
participating in multi-site Randomized Clinical Trials (RCTs),
and 2) the barriers to intervention adoption and factors that
increase or decrease the likelihood of adoption. This paper
focuses on the issues and themes that emerged in the process
and nature of the clinical supervision that occurred during the
RCT.
Study Design: We studied these issues within the research
initiative of the Center for Substance Abuse Treatment’s
(CSAT) Methamphetamine Treatment Project (MTP) which
tested a manualized intervention for methamphetamine users
at seven clinic sites. Forty-two individual, semi-structured
interviews were completed at seven levels of clinical trial
participation. Interviews were audio-taped and conducted inperson or, if necessary, by phone. Respondents were asked
about their experiences with the clinical trial and issues related
to project implementation and adoption of innovative
interventions.
Population Studied: Respondents consisted of 42 staff
participants in the CSAT MTP clinical trial. These respondents
were identified within seven organizational levels of the
research initiative: study funders, steering committee
members, study coordinating center staff, site principal
investigators, site evaluators, clinic directors, and clinicians.
Principal Findings: The supervisory model used in this RCT
was a decentralized model where weekly conference calls with
the study coordinating center and project clinical staff were
used for monitoring of protocol adherence and case
consultation. Critical challenges were identified in applying
the distance supervision model within an existing supervisory
model at the participating clinics. Issues and tensions arose
between the project clinical component and the RCT clinical
supervision component with the use of the distance
supervision model.
Conclusions: Themes related to the process and nature of the
model used for clinical supervision in the RCT were: 1)
supervisory relationships that were not well developed, 2) loss
of autonomy, 3) lack of flexibility, 4) conflict with individual
needs of clients and needs of client population, and 5) issues
of intervention fidelity.
Implications for Policy, Delivery or Practice: Identification
and discussion of these themes will inform specific
recommendations to the planners and implementers of RCTS
and enhance the opportunities for the adoption of evidencebased interventions in substance abuse treatment.
Primary Funding Source: NIDA
• Contracting For Coordination of Behavioral Health
Services in Privatized Child Welfare and Medicaid
Managed Care
D. Richard Mauery, M.P.H., Julie Collins, M.S.W., LCSW, Jan
McCarthy, M.S.W., Charlotte McCullough, MEd, Sheila Pires,
MPA
Presented by: D. Richard Mauery, M.P.H., Senior Research
Scientist, Department of Health Policy, GWU School of Public
Health and Health Services, 2021 K Street, N.W., Suite 800,
Washington, DC 20006; Tel: 202.530.2376; Fax: (202) 2960025; E-mail: rmauery@gwu.edu
Research Objective: This study examined coordination
between privatized child welfare initiatives and Medicaid
managed care systems for the delivery of behavioral health
care services for children and families in the child welfare
system. Specific objectives: 1) to assess how states’
expectations as embodied in their contract documents are
actually happening during program implementation; and 2) to
identify promising approaches for delivery of coordinated
behavioral health care services and contracting that can be
shared with other stakeholders. The goal was to highlight the
real world experiences and lessons learned that others may
draw upon when designing and implementing similar
contracts for systems of care.
Study Design: Two linked qualitative research methods: 1)
content analysis of Medicaid managed care and privatized
child welfare contract documents that specify requirements for
care coordination and interagency collaboration, and 2) site
visits in four states to interview key stakeholders about how
these paper requirements were playing out in practice.
Population Studied: Staff of state child welfare, behavioral
health, and Medicaid agencies; managed care organizations;
family organizations; and foster and birth parents of children
in child welfare systems in El Paso County, Colorado, Franklin
County, Ohio, Massachusetts, and Missouri.
Principal Findings: Despite varying degrees in scope and
specificity of language regarding collaboration and care
coordination, the child welfare contracts all encompassed
service delivery expectations that acknowledge the importance
of having access to an array of behavioral health services for
children and their families (including supportive wraparound
services) to maximize the likelihood of achieving successful
permanent placements within shorter periods of time. All
contracts covered standard behavioral health treatments.
Specialized services, such as domestic violence treatment and
sexual abuse/offender treatment, were far less likely to be
covered in the contracts, particularly the Medicaid contracts.
The success of interagency collaborations depends largely on
key leadership having the authority, and taking the
responsibility, for development and implementation of
programs and systems that stress a coordinated approach to
holistic care that integrates health, behavioral health, and
permanency issues. Multiple categorical funding and
reimbursement streams can create treatment silos that can
hamper care coordination when children are perceived as
belonging to one funding stream or another. Even well-written
contracts cannot overcome shortages of providers with child
and adolescent treatment expertise that result from low
reimbursement rates and/or under-funding and under-staffing
of public and private child welfare and behavioral health
systems.
Conclusions: Agencies and contractors are encountering
difficulties fulfilling their contractual care coordination
obligations, largely due to external factors such as the
inadequate supply of specialized treatment providers and
insufficient case rates and Medicaid reimbursements.
Interagency collaborations serve to bring together decisionmakers who can apprise their colleagues of the intended and
unintended effects of their actions on their programs.
Implications for Policy, Delivery or Practice: Contract
specifications, while critical, are only an essential first step to
ensure effective care coordination for children with behavioral
health needs and their families. Ongoing attention must be
paid to implementation issues and to the factors that impede
adherence to contract specifications and potentially threaten
the achievement of positive outcomes for children and their
families.
Primary Funding Source: Center for Health Care Strategies,
Inc.
• Healthcare Cost Waste in the Treatment of Depression
Stephen Melek, B.A. Math, FSA
Presented by: Stephen Melek, B.A. Math, FSA, Consulting
Actuary, Health, Milliman USA, 1099 18th Street, Suite 3100,
Denver, CO 80202; Tel: 303.672.9093; Fax: (303) 299-9018; Email: steve.melek@milliman.com
Research Objective: To identify and quantify potential
healthcare cost savings associated with inappropriate and
ineffective use of antidepressants, primarily among those
prescribed by PCPs.
Study Design: Individual prescription drug claim records were
examined for services provided between 7/1/02 and 5/31/03
(based on claim service dates). Claimants that received
prescriptions for any of 10 different popular anti-depressants
during this period, regardless of diagnosis, were identified and
chosen as the study group. We analyzed the treatment
patterns for these antidepressant-taking patients by age and
gender, by prescribing doctor, and by filling pharmacy in an
attempt to identify areas of major concern for high levels of
inappropriate or ineffective treatment – areas where
behavioral healthcare quality is of major concern. Using data
from the managed behavioral carve-out vendor, we separated
patients that had also received treatment from a behavioral
health specialist from those that did not, in order to identify
differences in Rx treatment patterns for these two patient
groups. We also analyzed other medical claims incurred by the
antidepressant-taking patients during the experience period,
including coded primary and secondary diagnoses, and types
of services provided for these patients. This effort was to help
quantify the potential amount of additional medical and
diagnostic services provided to these same patients that may
also be inappropriate or ineffective treatment costs.
Population Studied: State of Louisiana employees and
dependents.
Principal Findings: The following potential annual savings
were identified:
$2.3 million in direct costs of antidepressant drugs,
$10.2 million in direct costs of substituting generic for brand
antidepressants,
$0.3 million in direct costs of low dosage use.
Nearly 15% of the State of LA adult population (ages 19+)
filled an antidepressant prescription (annualized basis). That
is comparable to recent AMA findings that 6.6% of adults in
the U.S. suffer from major depression annually, and only
about one-third of them receive medications.
Approximately 67% of all patients that began treatment with
antidepressants discontinued prior to the minimum six month
duration stated in clinical guidelines; 25% of all patients that
were prescribed an antidepressant never refilled their
prescription.
We also identified additional healthcare costs incurred by
patients discontinuing antidepressant treatment too early that
were associated with behavioral and psychosocial diagnoses
in the claims data. Depending on the definition of behavioral
and psychosocial disorders, for these patients:
Between $9.9 million and $55.2 million of additional nonbehavioral healthcare costs are incurred annually;
Between $2.3 million and $12.9 million of additional nonantidepressant prescription drug costs are incurred annually.
Conclusions: Depressed patients too often get the wrong
treatment by the wrong providers. Disruptive innovation to the
current healthcare delivery system is needed to solve the
multi-faceted problems with the way depression is treated
today.
Implications for Policy, Delivery or Practice: There is a
substantial opportunity for quality improvement and increase
in healthcare cost effectiveness through more effective
identification, education and treatment of patients suffering
from depressive disorders. There is also a substantial savings
opportunity through more effective and appropriate use of
antidepressants for the patients that truly may benefit from
such treatment, and the elimination of inappropriate use of
such drugs.
Primary Funding Source: State of LA OGB
• A Trial of Incentives to Improve Depression Therapy
Appointment Adherence among Low-Income AfricanAmericans
Edward Post, M.D., Ph.D., Jeff Harman, Ph.D., Mario Cruz,
M.D.
Presented by: Edward Post, M.D., Ph.D., Assistant Professor
of Medicine and Psychiatry, Medicine and Psychiatry,
University of Pittsburgh, 230 McKee Place, Suite 600,
Pittsburgh, PA 15213; Tel: 412.692.4846; Fax: (412) 692-4838;
E-mail: postep@msx.upmc.edu
Research Objective: To evaluate the effect of nominal
payments to offset incidental expenses (e.g., transportation)
of attending regularly scheduled therapy appointments for
low-income African-American patients with depressive
disorders.
Study Design: The study uses an interrupted time-series
design: appointment adherence is tracked without monetary
incentive during an initial twelve-week period, followed by a
second twelve-week period of $10 payments at regularly
scheduled therapy appointments (usually weekly, but at the
discretion of the therapist), and a third twelve-week period of
appointment adherence assessment without monetary
payments. Baseline and exit surveys are conducted to collect
demographics, information on attitudes and barriers to care,
and to assess depression severity with the PHQ-9. Fifty
subjects have been enrolled and are being monitored during a
36-week period of follow-up that begins at the time of study
enrollment for each individual.
Population Studied: Patients with coded billing diagnoses
(either prevalent or incident) of major depression or
depression NOS, who seek care at a community mental health
center that serves a predominantly African-American
neighborhood in Pittsburgh, Pennsylvania.
Principal Findings: The subjects’ mean age is 47 years
(standard deviation=9; range=28 to 69); 94% are AfricanAmerican; 86% are female; and 88% are currently prescribed
psychoactive medication. The mean PHQ-9 score obtained
during the baseline survey was 13.8, indicating on average a
significant level of ongoing depression symptomatology. 72%
of subjects travel to appointments via bus; the median oneway fare was $1.50. 78% report that their therapy sessions
make them feel better, and 52% report that they would attend
a therapy session to receive $10 even if the session was not
helpful to them. Among follow-up accrued to date, data
indicate that in the aggregate, subjects kept 66.8% of
appointments in the baseline twelve-week period. Monetary
payments during the second twelve-week period are
associated with a halving of the nonattendance rate, as
aggregate appointment adherence rose to 83.6%. Analysis of
individual trends shows that relative to the first follow-up
period, 65.4% of persons improved their attendance rate;
15.4% remained the same; and 19.2% had lower attendance
rates in the second period. Among all of these subjects, the
median absolute increase in appointment adherence is 21.4%.
This improvement in individual appointment adherence is
statistically significant (p< 0.0004 using a paired t-test).
Additional data will be presented on adherence in the third
period of follow-up.
Conclusions: Nominal incentive payments can improve
appointment adherence for therapy sessions among
depressed, low-income African-American patients. The
subjects are evenly split over whether they would attend a
therapy session based on the incentive in the absence of a
personal belief in its utility. Data on appointment adherence
following the incentive period are currently being accrued.
Implications for Policy, Delivery or Practice: Facilitating
adherence to proven treatments for depressive disorders is
central to improving outcomes for low-income AfricanAmericans with depression. Additional studies are needed to
replicate, evaluate and implement both patient-level and
provider-level incentives to improve quality of care.
Primary Funding Source: Staunton Farm Foundation
• Building Organizational Capacity in Substance Abuse
Treatment Agencies
Andrea Burling, Ph.D., Karen Shore, Ph.D., Mary Anne Lahey,
Ph.D., Christine Timko, Ph.D., Thomas Burling, Ph.D., Edward
Carlson, M.A., MFT
Presented by: Karen Shore, Ph.D., Senior Research Scientist, ,
American Institutes for Research, 1791 Arastradero Road, Palo
Alto, CA 94304; Tel: 650.843.8121; Fax: 650.858.0458; E-mail:
kshore@air.org
Research Objective: To build organizational capacity of
substance abuse (SA) treatment agencies, through
comprehensive organizational assessments and identification
of specific capacity building projects.
Study Design: The Building Effective Substance Abuse
Treatment (BEST) Initiative is a series of projects funded by
the Charles and Helen Schwab Foundation. To date, there
have been two phases to the initiative. The first involved 12
SA agencies in San Mateo County, CA and the second
involved 6 agencies in other San Francisco Bay Area counties.
The Foundation’s program officer identified agencies for
potential inclusion. Each BEST agency participated in an
assessment and planning (AP) process led by a
multidisciplinary consulting team. The AP process for each
agency involved a comprehensive, multimethod
organizational assessment that included 1) an initial planning
meeting with the agency executive director; 2) document
collection and review; 3) a site visit involving multiple
individual and group interviews; and 4) a provider selfassessment. Each agency was assessed in eight
organizational system areas: purpose and goal attainment,
clinical care and quality improvement, governance and
leadership, staff development and human resources, structure
and communication, business operations, infrastructure, and
community context and connectedness. Each agency received
an AP report summarizing their organizational strengths,
problems or challenges, areas for improvement, and
recommendations for appropriate organizational capacity
building projects. Subsequent to the AP process, each agency
submitted a proposal to the Foundation for funding of a
specific capacity building project.
Population Studied: 18 non-profit SA treatment agencies
were studied. 20 were invited to participate in the first two
phases of BEST and 1 declined in each phase.
Principal Findings: The AP process revealed a variety of
strengths in the participating SA agencies, as well as a variety
of needs. Some of the most common agency strengths
included a shared sense of mission within each agency;
passionate, committed staff; and solid relationships with their
communities. Some of the most common problems
concerned purpose and goal attainment and governance and
leadership systems (e.g., little or no strategic planning,
inactive/ineffective Boards of Directors, overly centralized
decision making); data/quality management systems (e.g.,
multiple data systems that could not communicate with one
another, poor or non-existent systems for tracking client
process and outcome measures of quality); human resource
systems (e.g., low staffing levels, low salaries, difficulties in
recruiting and retaining staff ); and clinical care systems (e.g.,
insufficient agency and/or community resources to address
client needs such as mental health).
Conclusions: Funding to address organizational and
infrastructure needs are rarely available to non-profit SA
treatment agencies. While they have passionate, dedicated,
and hardworking staff providing services to many who are in
great need, these agencies face significant challenges and
problems in multiple organizational system areas.
Implications for Policy, Delivery or Practice: Non-profit SA
treatment agencies can enhance their organizational
functioning and capacity by engaging in organizational
assessments via self-assessments and/or with the assistance
of consultants. Increased funding should be made available
to assist these agencies with these assessments, as well as the
design and implementation of capacity building projects.
Primary Funding Source: Charles and Helen Schwab
Foundation
• Do Pleasing Pictures Have a Healing Effect on Seriously
Ill Patients?
Kazue Takayanagi, M.D., Ph.D., Tatsuo Kumazaki, Director of
Hospital
Presented by: Kazue Takayanagi, M.D., Ph.D., Associate
Professor, Health Services Administration, Nippon Medical
School, 1-1-5, Bunkyo-ku, Sendagi, Tokyo, 113-8602; Tel:
81.3.3822.2131; Fax: 81-3-3822-8144; E-mail: taka-y@nms.ac.jp
Research Objective: We investigated the healing potential of
viewing pleasing pictures in patients with terminal illness.
Study Design: A focus group, consisting of one art curator,
two hospital decorators, two nurses, and one doctor, chose
120 pictures out of 9000 in the art catalogues of the hospital.
Six factors - brightness of color, sharpness of line, content,
abstractness, landscape quality and action - were analyzed for
each picture. On this basis, 30 pictures were selected: 10 as
pleasing, 10 as neutral and 10 as not pleasing.
Patients themselves assessed the 30 pictures as pleasing,
neutral and not pleasing. They then chose one pleasing
picture to observe for 15 minutes. A Profile of Mood test
(POMS) was administered2), and pulse, blood pressure,
chromogranin A in the saliva, and NK cell activity in the blood
were measured, both before and after observation of the
selected picture.
Population Studied: Thirty-five private patients with terminal
illness (average age 60.9 years ± 13.5), in the university
hospital of Nippon Medical School, participated in the study.
Informed consent was obtained from all the participants.
Principal Findings: The pictures characteristically chosen by
the patients as pleasing were representational landscape or
nature paintings with bright colors, gentle lines and a positive
content. Abstract or action pictures with dark colors, harsh
lines and a negative content were not selected. The patients
particularly favored landscape pictures with a water motif.
When analyzed using the t-test, Profile of Mode Score
(POMS) showed significant decreases in tension -anxiety,
depression (p<0.01), anger (p<0.001), fatigue (p<0.005), and
confusion (p<0.05), and borderline significance for increased
vigor (p=0.052). Pulse and blood pressure fell significantly
(p<0.05), as did concentration of salivary chromogranin A (p<
0.05). Twenty-five patients permitted blood samples to be
taken. These revealed no significant difference in NK activity
before and after observing the picture, due to extreme
variability of response (before 100%; after 113.2±77.3%, range
21.3 to 420.0%). An increase in NK cell activity occurred after
looking at the picture in 10 patients who had chosen a picture
with a natural and gentle landscape, while serum NK cell
activity decreased in 14 patients who had chosen a picture
which had emotional message. Examples are Botticelli’s “Birth
of Venus ” and F.E.Church’s “Sunset”.
NK activity is not always increased by the most pleasing
picture, although POMS and Chromogranin A improved.
Conclusions: Pictures of landscapes with bright colors and
gentle lines were most often selected as pleasing by patients
in this study. Looking at pleasing pictures altered
psychological state, and may reduce sympathetic tone, in
patients with terminal illnesses. Observation of the picture did
not necessarily enhance immunity, especially if impressive
pictures were chosen . Pictures for the rooms of seriously ill
patients should be chosen to enhance their physiological and
psychological well-being.
Implications for Policy, Delivery or Practice: Pictures for the
rooms of seriously ill patients should be chosen as a
landscape or nature paintings with bright colors, gentle lines
and a positive content, but not with emotional message.
Primary Funding Source: University
• Adopting New Approaches to Addiction Treatment: What
Matters to Clients and Counselors?
Cindy Thomas, Ph.D., Dennis McCarty, Ph.D., Marilyn Daley,
Ph.D., Traci Reickmann, Ph.D., Brett Fuller, Ph.D.
Presented by: Cindy Thomas, Ph.D., Senior Scientist,
Schneider Institute for Health Policy, Brandeis University, 415
South Street, MS 035, Waltham, MA 02454; Tel: 781.736.3921;
Fax: 781-736-3905; E-mail: cthomas@brandeis.edu
Research Objective: The federal government has devoted
considerable resources to developing new medications to treat
addictions. Little information is available, however, on how
members of treatment communities view new
pharmacological approaches, and intend to adopt them as
part of treatment plans. Our objective is to examine attitudes
and beliefs of clients in treatment and providers toward new
treatment medications, and identify factors associated with
their intention to incorporate them into treatment, in order to
enhance efforts to promote adoption.
Study Design: A survey was administered to small groups of
counselors and clients in addictions treatment (programs
selected based on size, willingness to participate, and
location), to determine predictors of intention to recommend
and/or use four current or potential pharmacological
approaches to treatment: methadone, buprenorphine,
clonodine, and ibogaine, an alternative substance that is
classified as a nutritional supplement. Data were analyzed
using descriptive statistics to identify differences between
modalities and clients versus counselors, and multivariate
regression analyses to develop models to predict the intention
to use each of the above pharmacotherapies in the future.
Population Studied: We surveyed 377 addictions treatment
counselors, and 1083 clients in treatment for opiate addiction
in Massachusetts and Oregon. Treatment modalities
represented were: methadone programs, other outpatient
programs, and residential programs.
Principal Findings: Preliminary analysis indicates that
counselors were more receptive than were clients toward three
of the four medications, while clients were more receptive to
use of an alternative nutritional supplement. Half of clients at
this time said they were unlikely to take buprenorphine; a third
of counselors said they were unlikely to recommend it
currently. Attitudes toward various medications in addiction
treatment vary considerably by the type of treatment setting:
outpatient programs appeared more receptive than
methadone clinics or residential programs to adopting new
pharmacological treatments such as buprenorphine. After
controlling for treatment setting, attitudes and subjective
norms had more influence on clients’ and counselors’ intent
to use these new medications, than did demographic
characteristics, or beliefs about the particular medication’s
attributes. Peer support is crucial for both counselors and
clients when designing a treatment plan: when clients make a
decision about taking a new pharmaceutical, the counselors
are the individuals whose advice they trust most, followed by
friends and family members. Counselors are most influenced
by patients’ individual needs and the influence of coworkers,
rather than by clinical supervisors.
Conclusions: Providers and clients are interested in new and
emerging treatments that will improve effectiveness of
treatment. However, they are strongly influenced by peers and
social norms in their intentions to adopt these therapies. In
encouraging counselors to consider incorporating new
medications into treatment, an approach that includes the
input of peers who have had successful experience with such
medications will be important, rather than a top-down-only
approach. Additionally, educational programs and
interventions have to be tailored to specific modalities of
therapy, and supported by provider organizations.
Implications for Policy, Delivery or Practice: The results of
this study are of considerable value when designing
interventions to promote use of new pharmacological and
other treatments in addiction. In the past, some medications
such as naltrexone were not widely adopted by providers and
clients, in part due to features of the medication itself, but in
part due to the (often “top-down”) approaches taken to
encourage adoption. This study is particularly important
considering the recent FDA approval of buprenorphine in
office practice, and current federal efforts to promote its use.
We identify those influences that are greatest in leading to the
intention to use buprenorphine, both from the provider and
the patient perspective.
Primary Funding Source: NIDA
• Social Support in the Pain-Depression Relationship
Shana Traina, M.A., Grace Park, M.P.H., Harold E. Paulus,
M.D., Weng Kee Wong, Ph.D., Catherine MacLean, M.D.,
Ph.D.
Presented by: Shana Traina, M.A., Health Services
Researcher, Medicine/Rheumatology, UCLA, 1000 Veteran
Avenue Room 32-59, Los Angeles, CA 90095; Tel:
310.825.9071; Fax: 310.206.1227; E-mail:
straina@mednet.ucla.edu
Research Objective: To describe the relationship of pain,
depression, and social support in patients with rheumatoid
arthritis, or RA.
Study Design: Ongoing observational cohort study of patients
with early RA. At baseline and approximately every six months,
patients complete mail surveys that include a depressed
mood index, the Center for Epidemiologic Studies Depression
Scale or CES-D; a physical function scale, the Health
Assessment Questionnaire or HAQ; a pain component using
a visual analog scale or VAS; and a social support index, the
Lubben Social Network Scale. Demographics such as age,
gender, ethnicity, and level of education were also collected.
Baseline data were analyzed to describe the relationship of
pain, depression, and social support in early RA patients. Data
from 322 study patients were analyzed using Student’s t-tests,
chi-square tests, and linear regression models. After baseline
data were subset above and below the median on age, disease
activity score, disease duration, income, and HAQ, t-tests
were used to compare the means of Lubben, CES-D, and pain.
Population Studied: Subjects were recruited through a
network of community-based rheumatologists in the Western
United States. Eligibility criteria for enrollment were fulfillment
of the 1987 revised ACR criteria for RA, symptom onset within
one year and no prior therapy with disease-modifying antirheumatic drugs, or DMARDs. Subjects have been recruited
on a rolling basis since 1993 and prospectively followed for an
average of 3.5 years with a range of 2 to 10 years. The average
subject age is 50.3 years, 77 percent are female, and 76 percent
are Caucasian, with an average education level of 13 years.
Principal Findings: Depression scores were worse among
patients under age 50 than among older patients, 17.3 vs. 13.5
CES-D units, p = 0.01, respectively. Pain scores were worse
among patients with disease duration less than six months
than among patients with a longer disease duration, 55.0 vs.
63.7 mm, p = 0.02, respectively. Depression scores were
better, 11.0 vs. 19.8, and pain scores were better, 44.8 vs. 71.2
mm, among patients with better physical function scores, that
is HAQ below 1.25. Social support and depressed mood were
inversely correlated. Higher depression scores are associated
with more pain, worse physical function, and lower social
support.
Conclusions: Social support may mediate the paindepression relationship in early RA patients. Those who
exhibited more depressed mood were also likely to have less
social support and have more pain. Younger patients were
more depressed than older patients, and people with better
function were less depressed and reported less pain than
people with worse function.
Implications for Policy, Delivery or Practice: Providing
social support to RA patients may help to improve their pain,
function, and depression.
• Community-Based Participatory Research: A Systematic
Review of Research Quality, Community Involvement and
Intended Outcomes
Meera Viswanathan, Ph.D., Alice Ammerman, Dr.P.H.,
Eugenia Eng, Dr.P.H., Gerald Gartlehner, M.D., Kathy Lohr,
Ph.D.
Presented by: Meera Viswanathan, Ph.D., Research Analyst,
Health, Social and Economics Research, RTI International,
3040 Cornwallis Road, RTP, NC 27513; Tel: 919.316.3930; Fax:
919.541.7384; E-mail: viswanathan@rti.org
Research Objective: AHRQ commissioned the RTI-UNC
Evidence-Based Practice Center to conduct a systematic
evidence review on community-based participatory research
(CBPR) approaches to improved health. This paper focuses
on two questions covered by the systematic review:
1) How has CBPR been implemented to date with regard to
the quality of research methodology and community
involvement?
2) What is the evidence that CBPR efforts have resulted in the
intended outcomes?
Study Design: We employed well-established methods of
evidence review. We generated a list of search terms for our
systematic search strategy. We then searched standard
electronic databases for all years: MEDLINE®, Cochrane
Collaboration resources, Psycinfo, and Sociofile. We
conducted hand-searches of the reference lists of relevant
articles and sought additional citations from a panel of
experts. We abstracted information from individual or groups
of articles pertaining to each study into standardized forms,
and transferred the information onto evidence tables. We also
created quality rating forms to assess each study’s research
methods and adherence to CBPR principles of community
collaboration.
Population Studied: We included peer-reviewed reports on
CBPR studies that: were published in English language;
conducted in the United States and Canada; and included at
least one community collaborator. We included in our review
a total of 123 articles that make up 60 studies.
Principal Findings: We classify studies according to whether
or not they intended to implement and evaluate interventions,
and by study design. We summarize research design,
methods and community involvement in research for all 60
studies. Further, for the smaller subset of studies that had
completed and fully evaluated interventions, we assess the
quality of research and community participation and the
nature of health outcomes.
Conclusions: We classified studies according to whether or
not they intended to implement and evaluate interventions,
and by study design. We summarized research design,
methods and community involvement in research for all 60
studies. Further, for the smaller subset of studies that had
completed and fully evaluated interventions, we assessed the
quality of research and community participation and the
nature of health outcomes.
Implications for Policy, Delivery or Practice: CBPR is a
collaborative approach to research that combines methods of
inquiry with community capacity-building strategies to bridge
the gap between knowledge produced through research and
what is practiced in communities to improve health. Although
many researchers and practitioners offer definitions and
principles for this approach, no clear consensus has emerged
to move the field forward. Moreover, during a time when
interest is growing rapidly for academic institutions, health
agencies, and communities to form research partnerships, few
guidelines exist as to how CBPR proposals should be
evaluated and what resources are required to promote
successful collaborative research efforts. This paper will help
to establish the nature of the current literature. It will also
assist academics, community participants and funders by
identifying gaps in implementing this approach.
Primary Funding Source: AHRQ, National Cancer Institute
• Continuity of Anti-Psychotic Therapy under Capitation
Financing in Colorado
Neal Wallace, Ph.D., Joan Bloom, Ph.D., Teh-Wei Hu, Ph.D.,
Anne Libbey, Ph.D.
Presented by: Neal Wallace, Ph.D., Assistant Professor, Public
Administration, Portland State University, P.O. Box 751,
Portland, OR 97207-0751; Tel: 503.725.8248; Fax: 503-725-8250;
E-mail: nwallace@pdx.edu
Research Objective: This study investigates the impact of
Colorado’s Medicaid mental health capitation payment
program on the continuity of anti-psychotic prescriptions
among persons diagnosed with Schizophrenia. Loss of
medication continuity has been found in studies of other
public mental health capitation programs such as Tenncare
(Ray et al, 2003) but creates somewhat of a quandary.
Medication therapy, typically paid for outside of the capitation
program, is protective of costly psychosocial interventions,
and thus its continued, if not increased, use would be
consistent with the financial incentives of these programs. In
1995, Colorado implemented a capitation payment program
covering psychosocial treatments in part of the state while
leaving some areas under existing Fee-For-Service (FFS)
reimbursement. Prior studies by Bloom et al (1998,2001) and
Cuffel et al (2001) found significant reductions psychosocial
treatments after capitation without decrements in a wide
range of measured clinical and social outcomes.
Study Design: This study uses a difference in difference
approach to evaluate the impact of capitation on antipsychotic prescription patterns compared to FFS. Antipsychotic continuity is measured as the probability of any use
and the number of months with filled prescriptions in a
period. Probability of any psychosocial treatment and cost of
treatment given use are also measured along with change
scores for the Schizophrenia Subscale of the Brief Psychiatric
Rating Scale (BPRS). Explanatory variables include indicators
for two capitation organizational models, one pre- and two
post-capitation periods, and their interactions along with age,
gender, ethnicity and prior treatment cost profile as subject
controls.
Population Studied: The study sample consists of 235 of
Bloom’s original study subjects with a diagnosis of
Schizophrenia but not enrolled in a medical HMO. Pharmacy
claims were not consistently available for these subjects.
Subjects are nearly equally distributed across the three
models, similar in age and gender, but with some differences
in ethnicity and prior treatment costs. During the precapitation period, over 95% of the subjects received some
psychosocial treatment.
Principal Findings: Probabilities of anti-psychotic use
increased in both post-capitation periods for both
organizational models relative to FFS. These effects were
moderate to large but only one was statistically significant.
Duration or intensity of anti-psychotic use, measured in
months with filled prescriptions appeared unchanged. Effects
were small and of mixed sign. In comparison and consistent
with prior findings, subjects in both capitated areas had
significantly decreased probabilities of psychosocial treatment
use with deceases in treatment cost only in the MBHO areas.
Difference scores for the Schizophrenia Subscale showed
consistent, moderate positive effects but these did not reach
statistical significance.
Conclusions: Within the Colorado capitation program,
continuity of anti-psychotic therapy was not impaired, despite
significant decreases in the continuity and intensity of
psychosocial treatments. Positive symptoms generally
amenable to anti-psychotic therapy where also preserved.
Capitation financing does not inherently risk loss of
medication continuity by consumers.
Implications for Policy, Delivery or Practice: Medication use
appears to be a better process indicator of program success
than psychosocial treatment patterns and is more consistent
with outcomes. Psychosocial treatment can be selectively
reduced without interfering with medication therapy. Further
research on the conditions that allow this is warranted.
Primary Funding Source: NIMH
• Are Anxiety Disorders Underdiagnosed and Undertreated
in Routine Psychiatric Practice?
Joshua Wilk, Ph.D., Joyce West, Ph.D., M.P.P., William
Narrow, M.D., M.P.H., Donald Rae, M.A., Lisa Countis, Darrel
Regier, M.D., M.P.H.
Presented by: Joshua Wilk, Ph.D., Research Scientist, Practice
Research Network, American Psychiatric Institute for Research
and Education, 1000 Wilson Boulevard, Suite 1825, Arlington,
VA 22209; Tel: 703.907.8618; Fax: 703-907-1087; E-mail:
jwilk@psych.org
Research Objective: 1) To assess rates of detection and
diagnosis of anxiety disorders in routine psychiatric practice;
2) To examine patterns of treatment of patients with anxiety
symptoms or diagnoses.
Study Design: 615 psychiatrists from a full range of clinical
settings participated in the American Psychiatric Practice
Research Network’s 1999 Study of Psychiatric Patients and
Treatments (SPPT) generating nationally representative
clinical data on 1,843 patients. Comparisons were also made
with household respondents in the National Comorbidity
Survey (NCS) who reported using specialty mental health
services.
Population Studied: 615 psychiatrists from a full range of
clinical settings and household respondents in the National
Comorbidity Survey (NCS) who reported using specialty
mental health services
Principal Findings: The prevalence of anxiety disorders
reported by psychiatrists in routine practice was much lower
than the prevalence found in self-reported specialty mental
health service users ascertained through a structured
household interview. Phobias were 22 times more likely to be
diagnosed in the NCS sample compared to the SPPT sample.
PTSD was diagnosed at 5 times the rate in the NCS sample
compared to the SPPT sample, GAD at over 3 times the rate,
and panic disorder at more than twice the rate. When SPPT
patients with a DSM-IV anxiety disorder were combined with
patients who had no reported anxiety disorder but with
moderate or severe anxiety symptoms, the percentage in the
SPPT sample increased from 23.8% to 54.7%, closer to the
level of anxiety diagnoses in the NCS sample (68.1%). Over
10% of SPPT patients with panic disorder were receiving
neither psychotherapy nor anti-anxiety medication (anti-anxiety
medications were categorized broadly to include SSRIs and
venlafaxine as well as conventional anxiolytics such as the
benzodiazepines), while 8% of patients with PTSD were
receiving neither anti-anxiety medication nor psychotherapy.
14% of patients with no anxiety diagnosis, but moderate to
severe anxiety symptoms were not receiving psychotherapy or
anti-anxiety medication. While over 75% of patients with
generalized anxiety disorder or panic disorder were receiving
anti-anxiety medications, less than two-thirds of patients with
a phobia diagnosis and slightly over two-thirds of patients
with PTSD were receiving these medications. The most
commonly prescribed anti-anxiety medications included
Alprazolam (16.7% of patients with anxiety disorders),
Clonazepam (16.4%), and Paroxetine (15.9%).
Conclusions: These findings suggest that anxiety disorders
are underdiagnosed and sometimes undertreated in routine
psychiatric practice. Further research is needed to examine
associated factors or clinical rationale for the findings such as
clinical significance, comorbidity, and prevailing standards of
diagnostic assessment. Longitudinal research that includes
an in depth assessment of treatment patterns, sequencing,
and outcomes is needed to assess the extent to which not
diagnosing and treating anxiety disorders and symptoms
results in poorer outcomes.
Implications for Policy, Delivery or Practice: Not diagnosing
and treating anxiety disorders and symptoms could result in
poorer outcomes and suggests the need for interventions,
including increased screenings for anxiety disorders.
Primary Funding Source: CSAT, American Psychiatric
Foundation, the John D. and Catherine T. MacArthur
Foundation, the Center for Mental Health Services
• Can Pharmacy Data Identify Severely Mentally Ill Patients
Needing Help with Adherence?
Marcia Valenstein, M.D., M.S., Emily Woltmann, M.S.W.,
Janet Kavanagh, M.S., John Zeber, M.H.A., Patricia Schraner,
M.S.W.
Presented by: Emily Woltmann, M.S.W., Health Science
Research Specialist, Severe Mental Illness Training Research
and Evaluation Center (SMITREC), Ann Arbor Veterans Affairs
Healthcare System, 2215 Fuller Court, Ann Arbor, MI 48105;
Tel: 734.769.7100 Ext. 6259; E-mail: ewoltman@umich.edu
Research Objective: Patients with schizophrenia and poor
antipsychotic adherence are at increased risk for relapse and
re-hospitalization. In organized healthcare systems, pharmacy
data might be profitably used to identify patients who would
benefit from adherence interventions.
However, it is not clear if pharmacy data will be an accurate
and efficient method for identifying at-risk patients. We
examined whether patients who were flagged as poorly
adherent by pharmacy data also appeared to be poorly
adherent upon chart review.
Study Design: Pharmacy data from two VAs were used to
identify patients with schizophrenia/schizoaffective disorder
who filled < 80% of the oral antipsychotics needed for
continuous 12-month coverage (N=537). We reviewed medical
charts to determine potential explanations for patients’ failure
to fill antipsychotics, such as transferring care, starting longterm depots, patient death, or clinician discontinuation of
medications.
Population Studied: Patients were included in the study if
they had a current diagnosis of schizophrenia or
schizoaffective disorder and had at least two outpatient
mental health visits in the previous 12 months the VA
facilities.
Principal Findings: Chart data indicated that approximately
33.5 % of all veterans flagged as poorly adherent (n=185), may
have been “false positives”. These patients’ notes indicated
they had formally transferred their psychiatric care, moved or
lived out of state, had had their medications discontinued for
clinical reasons, had been receiving depot antipsychotics for
most of the year or were deceased.
Of the 66.5% patients who were “true positives” for poor
adherence, only16% were already receiving interventions to
improve their adherence, such as supervised housing
placement, community outreach, treatment with depot
medications in response to recent poor adherence, or
prescription of clozapine.
Conclusions: Chart data indicated that approximately 33% of
patients with an administrative diagnosis of schizophrenia
who are flagged as poorly adherent by pharmacy data may
have been adherent. The remaining 67% appear to be true
positives. Few true positive patients were receiving strong
interventions to improve adherence.
Implications for Policy, Delivery or Practice: Pharmacy data
may be a useful first screen for poor adherence but must be
followed by chart review before outreach efforts are made.
Primary Funding Source: VA,
• Atypical Antipsychotics and Schizophrenia with
Comorbid Substance Abuse
John Zeber, M.H.A., Laurel Copeland, Ph.D., Frederic Blow,
Ph.D., David Smelson, Psy.D.
Presented by: John Zeber, M.H.A., Research Associate,
HSR&D, Department of Veterans Affairs, 928 Rose Drive, Ann
Arbor, MI 48103; Tel: 734.769.7100 Ext. 6260; Fax: (734) 7612617; E-mail: jzeber@umich.edu
Research Objective: The advent of second generation, or
atypical, antipsychotic medications has offered a wider and
potentially more beneficial variety of treatment options for
patients with schizophrenia. Recently these newer drugs have
also been expanded for use in other psychiatric populations,
with early promising results. Veterans with schizophrenia and
a comorbid drug dependency are more expensive and
challenging to treat than patients with only one psychiatric
condition. These individuals tend to be less stable, especially
in terms of medication compliance and inpatient utilization,
compared to similar patients without a substance abuse
diagnoses. However, atypical antipsychotics (AP) may help
stabilize dopamine levels, diminish cravings, and be more
efficacious than conventional drugs. We explore the effect of
atypical APs among patients with dual diagnosis.
Study Design: The VA’s National Psychosis Registry provided
data on psychiatric and substance abuse diagnosis
(schizophrenia, plus either cocaine or alcohol dependence)
and the prescription of APs in FY99. Initial analyses examined
the number of medications prescribed and compliance rates
among 56,816 patients with and without a substance abuse
diagnosis. Then, among just the veterans with a dual
diagnosis (N=8,199), we examined differential effects of
atypical or conventional APs on service utilization,
compliance, and relapse (re-admission within 90 days).
Population Studied: A national population of all veterans
with schizophrenia alone, and individuals who also received a
diagnosis for comorbid substance abuse.
Principal Findings: Of the entire population of veterans with
schizophrenia, 14 percent of patients were also diagnosed
with comorbid substance abuse dependence. Compared to
patients with schizophrenia alone, the dually diagnosed
veterans were prescribed more APs (patients on two or more
drugs: 37 percent vs. 25 percent), were nearly twice as
medication non-compliant, and used significantly more
healthcare services. They also more commonly received
atypical APs (70 percent vs. 55 precent). Within just the more
limited sample on dually diagnosed patients, the use of
atypical antipsychotics was not associated with decreased
inpatient use, nor did these patients experience better shortterm relapse rates. However, medication noncompliance rates
were significantly lower (multivariate OR=.82, CI95 .74-.92).
Conclusions: Compared with conventional APs, atypical
antipsychotic agents appear to improve compliance for these
difficult to treat and high-utilization patients, an important
outcome given the generally poor adherence rates among this
vulnerable population.
Implications for Policy, Delivery or Practice: Atypical APs
may prove to be particularly effective in the treatment of dual
diagnosis patients to improve compliance, relieve substance
abuse cravings, and potentially reduce psychiatric stays and
relapses over a longer period. Further research should
examine atypical APs and symptomology, focusing on the
burden of illness from both schizophrenia and addiction.
Primary Funding Source: VA
• Efforts to Recruit Patients with Serious Mental Illness
into a Randomized Research Study
John Zeber, M.H.A., Marcia Valenstein, M.D., M.S., Emily
Woltmann, M.S.W., Janet Kavanagh, M.S., Patricia Schraner,
M.S.W.
Presented by: John Zeber, M.H.A., Research Associate,
HSR&D, Department of Veterans Affairs, 928 Rose Drive, Ann
Arbor, MI 48103; Tel: 734.769.7100; Fax: (734) 761-2217; Email: jzeber@umich.edu
Research Objective: Despite an NIH mandate for increased
participation of minorities and women in health stuides, these
groups are often under-represented. This scenario may result
from problems in recruitment and enrollment. These
difficulties are particularly salient when research involves
patients with serious mental illness (SMI), representing a
challenge in the ability to contact these individuals and
subsequent recruitment success. Patient characteristics that
can influence recruitment efforts include gender, age,
comorbid substance abuse, illness severity and social support.
Differential recruitment and drop-out rates can introduce
significant research bias, less reliability, and low generalibility
regarding observed results. Within the VA, a veteran’s “period
of military service” and level of service connectedness is an
important predictor for successful recruitment. The objective
of this study is to examine reasons for differential contact and
recruitment success during a randomized trial to improve
medication compliance. Recruitment efforts included
telephoning patients or next of kin, and meeting patients at
their clinic appointments.
Study Design: Research staff attempted to recruit all patients
identified as having a serious mental illness and poor
antipsychotic adherence at three VA medical centers (two
urban, one suburban site). Patients were initially contacted by
letter, followed by a telephone call or being approached during
their next clinical appointment. We used bivariate and
multivariate logistic regression to analyze differences in staff’s
“ability to contact” patients and refusal rates by location,
diagnosis, patient demographics, medication compliance, and
level of VA service-connection.
Population Studied: Veterans with serious mental illness
(schizophrenia or bipolar disorder) and poor antipsychotic
adherence, meeting eligibility criteria for a multi-site RCT to
improve medication compliance (N=241).
Principal Findings: Study personnel were able to contact 76
percent of the identified patients. In multivariate analyses,
researchers were significantly more likely to successfully
contact married patients (OR: 2.55), those living in the
suburban site (OR=1.71), and patients with a service-related
disability (OR: 1.80). However, there were no differences in
contact rates by race, diagnosis, or substance abuse
comorbidity. Of the 183 patients contacted, 71 (39 percent)
agreed to participate in the study. Patients with a bipolar
rather than a schizophrenia diagnosis, lived in a suburban
location, had lower medication adherence rates, and who had
a service-related condition were more likely to agree (ORs:
3.07, 3.97, 2.17, and 2.32, respectively). There were no
differences by race, marital status, substance abuse, or gender
in the likelihood of participation.
Conclusions: In this study, minority patients appeared as
likely to participate as non-minority patients, with adjustment
for urban location, service connection, and type of psychiatric
diagnosis. However, other patient characteristics did influence
the ability to contact these individuals and their likelihood of
agreeing to participate in the research study.
Implications for Policy, Delivery or Practice: Challenges
may arise during contact attempts and subsequent
recruitment efforts, with differences by patient characteristics.
Once contacted, under-represented groups appear just as
willing to participate in a health services study. However,
recruitment in urban areas may require special outreach
efforts to locate patients. Vulnerable SMI veterans may offer
unique difficulties during the course of randomized projects.
Researchers should devote attention to patient differences in
ensuring appropriate representation and accurate
interpretation of results.
Primary Funding Source: VA
Invited Papers
Prescription Drugs & Behavioral Health
Chair: Haiden Huskamp, Ph.D.
Sunday, June 6 • 5:00 p.m.-6:30 p.m.
•
Panelists: Thomas Croghan, RAND; Patricia Deverka,
Medco Health Solutions, Inc.; Benjamin Druss, Emory
University; Stephen Soumerai, Harvard Medical School
and Harvard Pilgrim Health Care; Julie Zito, University of
Maryland, Baltimore (no abstracts provided)
Invited Papers
Challenges in Using Evidence-Based Practices in
Substance Abuse Treatment
Chair: Mady Chalk, Ph.D.
Monday, June 7 • 2:00 p.m.-3:30 p.m.
•
Panelists: Kevin Hennessy, Substance Abuse and Mental
Health Services Administration; Todd Molfenter,
University of Wisconsin, Madison (no abstracts provided)
Invited Papers
Screening & Intervention for Alcohol & Drug Use
Problems in Trauma Patients: If They Improve Public
Health & Save Money, Then Why Don’t We Do Them?
Chair: Harold Perl, Ph.D.
Monday, June 7 • 4:00 p.m.-5:30 p.m.
•
Panelists: Lawrence Gentilello, UT Southwestern Medical
School; Dan Hungerford, Centers for Disease Control and
Prevention; Dennis Kelso, San Diego Health and Human
Services Agency; Michael Sise, Scripps Mercy Hospital
(no abstracts provided)