Home» Research
Previous and Closed Studies
Recently Completed Projects-2006
Follow-Up Study of Discharged Patients with Severe Mental Illness from Phila State Hospital:
CMHPSR contacts: Aileen Rothbard, Sungeun Lee
Abstract from paper under review
Objective. This analysis is a follow-up of a 1992 study of service use and cost of care of patients discharged from a state hospital. The current study documents the utilization and cost of care in 2002. Method: Study subjects were 150 patients discharged from a state psychiatric hospital to community settings following the closure of the state hospital in 1989. The subjects represented those individuals who were alive at the time of the follow-up and were Medicaid beneficiaries whose cost and utilization of services could be tracked. An integrated database on all mental health and medical services reimbursed by Medicaid and Medicare as well as state-and-county-funded services was used to construct descriptive statistics on service utilization and unit cost measures. Data on mortality, homelessness and criminal arrests was obtained from vital statistics, jail records and shelter admissions. Results: During 2002, 18% of the subjects had a psychiatric hospital stay of 59.6 days annually. Almost all of the subjects received some form of outpatient mental health care other than case management (96%) and 66% were living in publicly funded residential housing. The cost per study subject for a comprehensive package of services including psychiatric, medical, residential and pharmacy was $89,699. Residential care accounted for the largest share of expenditures (53%). On a user basis, it cost approximately $6,054 per month which included residential staff and other housing support costs provided by the programs. Arrest records showed that 6 individuals (4%) were admitted to county jails, and 12 individuals (8%) had been homeless at some point between 1989 and 2005. Thirty seven percent (37%) of the original group of 590 individuals discharged from the state hospital in 1987-89 had died.
Conclusions: This analysis suggests that individuals discharged from state institutions have been integrated into community residential settings and are receiving psychiatric outpatient treatment on a regular basis at a much reduced cost compared to the estimated $169,000 per year cost of state hospitalization in 2002. Cost estimates using the consumer price index for 2002 were $78,773 in 1992 compared to $85,850 for a similar package of services in 2002.
The Impact of Managed Behavior Healthcare on Patterns of Service Utilization and Cost for Alcohol Abusers in a FFS versus MC setting:
CMHPSR contacts: Aileen Rothbard, Eri Kuno
Abstract from published paper This project examined the impact of a mandatory managed care (MC) behavioral health program on access, utilization, and cost of behavioral health care services for high-risk public sector clients on Medicaid diagnosed with alcohol abuse or alcohol dependence. A pre-post study design examines changes in service patterns and costs using two annual cohorts of alcohol treatment users in 1995 (pre MC period) and 1998 (post MC period). The data sets include Medicaid eligibility, administrative claims, and encounter data. Results of the study showed that overnight/24 hour treatment use changed dramatically in the MC site. Use of general or rehabilitation/D&A units in hospital settings showed sharp declines: 40% to 9% for detoxification and 24% to 4% for rehabilitative treatment replaced by non-hospital detoxification (29%) and non-hospital residential rehabilitation (42%) in 1998, resulting in a higher proportion of overnight treatment users in the post MCO period (64% vs. 55% in 1995). The number of episodes of acute hospital detoxification also dropped as did the total annual detoxification days. Similar changes were found for drug related conditions. Per person service costs for behavioral health among alcohol treatment users declined from $7,662 to $5,664 in the MC site, while it increased from $4,871 to $6449 in the FFS site.
Levels of Community Integration and Variability in Integration among Consumers in Supported Independent Housing
Irene Wong, Ph.D., Phyllis Solomon, Ph.D. Status
Abstract of a Paper in Press: In the post-deinstitutionalization era, supported housing has emerged as a housing and service approach considered most conducive to the goal of consumer empowerment and community integration. Although prior research found beneficial effects of support housing, little empirical work has been done on identifying the gaps between the principles and practice of supported housing. Using multiple data sources on 27 supported independent living (SIL) programs for psychiatric consumers in Philadelphia, Pennsylvania, this paper examines the extent to which these programs are implemented in accordance with supported housing. Findings suggest consistency with as well as deviation from the supported housing approach in regard to consumer choice, typical and normalized housing, resource accessibility, consumer control and provision of individualized and flexible support. This study suggests methods for assessing housing programs for psychiatric consumers along these domains. Implications for the development of housing programs that promote consumer empowerment and community integration are discussed.
Predicting Staying in or Leaving HUD-Funded Permanent Housing
Irene Wong, PhD., Trevor Hadley, PhD., Steve Poulin, PhD., Morris Davis
Abstract from published paper: This is a longitudinal tracking study of PH participants using administrative and interview data. The sample consisted of 953 residents of 27 PH (i.e., supported independent living, or SIL) programs in Philadelphia during 2001-2004. Data on socio-demographics, psychiatric diagnoses, substance abuse treatment history and service use characteristics were extracted from an integrated administrative database comprising of PH stays, public shelter use, Medicaid eligibility and claims, and non-Medicaid publicly-funded behavioral health services. Chi-square, t-test, and stepwise logistical regression analyses were used to conduct the statistical tests. A sub-sample of 96 leavers was recruited and their post-PH careers (including employment status) documented through monthly in-person or phone interviews. The case study method was used to categorize the scenarios of departure from PH and the post-PH careers. Results: Consistent with prior research findings, housing retention rates in PH were 75% in one year, 60% in two years, and 50% in three years. Regression analysis results suggest that socio-demographic and psychiatric diagnostic characteristics are not associated with departure from PH. Instead, behavioral health service use characteristics during participants’ tenure in PH, particularly the use of psychiatric emergency and in-patient services, are significant predictors of leaving. Two types of PH departures were discerned in the data: voluntary departures (61%) involving agreed-upon discharges to more appropriate housing suited to participants’ needs and preferences (including housing with higher levels of independence), and involuntary departures (39%) involving poor resident-staff relationship, relapses and aggravated drug and alcohol problem, and poor housing and neighborhood characteristics of PH. Follow-up comparisons found that the higher level of behavioral health service use among PH leavers (compared to non-leavers) are attributable to those who departed PH involuntarily. As expected, involuntary leavers were more likely than voluntary leavers to experience unfavorable post-PH residential careers including high rates of homeless shelter use and psychiatric hospitalization. The employment rate during the post-PH career was consistently low (around 20%), regardless of the departure status (voluntary vs. involuntary). Implications: Our findings suggest that monitoring behavioral health service use may be an effective mechanism for identifying PH participants at risk for unfavorable discharges. The implementation of an “early warning system” may help to target intensive services for those in greatest need. Interventions are needed to support and facilitate PH participants who are ready to move on to more independent settings. These may include the availability of long-term housing subsidies, encouragement of public-private partnership to develop housing units and provision of specialized services to enhance the employability of PH
participants.
Rates of Opportunistic Infections among persons with HIV, Substance Abuse, and Serious Mental Illness: A Medicaid Cohort
Michael Blank, Aileen Rothbard, David Metzger, So-Young Min
Abstract from a paper under review. This project establishes estimates of the relative risk of receiving treatment for an opportunistic infection in a Medicaid population over an eleven-year period. Subjects were all adult Medicaid recipients in Philadelphia between July, 1986, and June, 1997 (N = 611,725). Medicaid claims records were used both to identify diagnostic groups and to identify denominators of individuals who received treatment for one or more of the target conditions during the study period. Groups were designated based on those who received a diagnosis of severe mental illness, substance abuse, and/or HIV/AIDS. Those with none of the target conditions were used as controls. For most opportunistic infections, persons with all three conditions, (co-morbid SMI, SA, and HIV/AIDS had the highest relative risk, followed by persons who were HIV positive and substance using. Rates of opportunistic infections that were also STD’s were especially high among co-morbid persons. Persons with only SMI or SA only were at greater risk than the general Medicaid population. Health care providers should be aware of the increased risk for adverse health outcomes for persons with co-morbid HIV/AIDS, SMI, and SA. Additionally, preventive efforts should target those who are at increased risk, especially for STD’s. Logical foci for these interventions will be discussed.
Peer Support for People with Co-Occurring Disorders Reduces Inpatient Bed Days: A Longitudinal, Comparison Group Study
Mark Salzer
Abstract of Published Paper Background: Peer support has been discussed as an important component of care for individuals with co-occurring disorders to promote recovery. Rigorous research on its effectiveness in reducing inpatient bed days for this frequently hospitalized group is limited. This study examines whether a peer support program, The Friends Connection, is associated with reduced inpatient bed days. Methods: Medicaid data from 1991 – 2001 were used to conduct a pre-post study of inpatient bed days for a group of individuals who participated in a 1:1 peer support program between 1993 and 1998 compared to a similar group who did not participate. A series of pre-posttest analyses are conducted to test the hypothesis, including one involving the comparison of data for those in the peer support program versus an equivalent population of individuals who did not participate in the program. This comparison group allows us to account for any natural recovery (i.e., regression to the mean and natural reductions) that might occur following periods of intense hospitalization. Results: Individuals who participated in the peer support program had a significantly larger reduction in inpatient bed days after their involvement in the program versus the similar comparison group. Discussion: These results provide evidence in support of peer support as an important adjunctive treatment for people with co-occurring disorders and as an important component of systems of care in support of the President’s New Freedom Commission report.
Costs of Implementing a Computerized Prescription Systems in pubic mental health agencies to Improve QOC and reduce Disparities
Aileen Rothbard, Eri Kuno, Trevor Hadley
Abstract from paper in press, Psychiatric Services: Four not-for-profit specialty mental health agencies in an urban setting comprised the study population. The number of psychiatrists working at the agency ranged from 5 to 10 FTE. The prescribing system employed in the project was a web-based system that charged monthly fees per prescribing and non-prescribing user. The costs of upgrading the computer system varied from $1,460 to $6,900, depending on the size of the agency and the state of their information infrastructure. The information system staff devoted approximately 116 hours ($3,248) to implement the prescribing system. The human resource costs for training was around $4,500 per agency, including psychiatrists’ time ($1,500 to $2,700). The fees for training and support for system activation was divided by 4 agencies, costing $9,000 each. In addition, we estimated that the pre-implementation decision making process for each agency cost between $2,600 to $3,600. The actual utilization of the system was a gradual process. Full implementation was accomplished in 6 to 9 months following the initial training.. Delays in implementation were due to insufficient clinician buy-in, inadequate workflow assessment and need to upgrade information systems
The total initial costs ranged from $20,572 to $27,549 per agency. Annual ongoing costs were expected to range from $9,877 (5 FTE agency) to $14,677 (10 FTE agency). On a per psychiatrist based expenditure, initial costs per prescribing physician averaged $2,385 ($1,900-3,133) and ongoing cost per physician averaged $1,100 ($1,000-1,162). Compensation of psychiatrists time was a major incentive for public sector agencies to implement the computerized prescription system. The cost analysis shows that the technology expenditure itself is not prohibiting for initial implementation as well as ongoing support. Once the computerized prescription writing becomes routine practice, the system can be expanded to include other clinical information that can be linked to client outcomes. Providing initial financial support to implement a well-designed computerized prescription system is a valuable policy option, along with technical assistance in implementation, considering long-range benefits of the technology for monitoring and improving quality of care for public sector clients.
Mental Health Residential Capacity Planning.
Eri Kuno, Ph.D., Aileen Rothbard, ScD., Naoru Koizumi, Ph.D.
Abstract from published paper: The purpose of this project is to explore the use of queuing network models in developing a decision support tool for mental health residential capacity planning. The major policy objective of this model is to reduce the extent of unnecessary stays in restrictive settings for individuals with serious mental illness. This decision support tool will assist service planners to visualize client flow throughout different parts of the residential system.
This model application is based on the Philadelphia mental health system. Several administrative data files from the County Office of Mental Health (i.e., residential placement referral records, Medicaid inpatient claims and the State Hospital records from 1997 to 1999) will be integrated to construct queuing model parameters, including 1) hospital admission and residential referral rates, 2) mean length of stay, and 3) transition rates between facility types. The project relies on secondary data that involves no direct contact with patients by any member of the research team. The University of Pennsylvania, Center for Mental Health Policy and Service Research (CMHPSR), in collaboration with staff of the Philadelphia Office of Mental Health (OMH) , will apply the algorithm to create a composite identifier for each person and perform data linkages. The research team will work with this de-personalized data set to derive the rates to be used for the queuing model.
Involvement in the Child Welfare System among Mothers with Serious Mental Illness
Jung Min Park, PhD; David S. Mandell, ScD (Center Investigator); Phyllis Solomon, PhD
Abstract from published paper: Objective: To determine the association between maternal serious mental illness (SMI) and child custody arrangements among a sample of Medicaid-eligible mothers. Methods: Medicaid eligibility and claims data were merged with data from the child welfare system in Philadelphia, PA for fiscal years 1995-2000. The study sample comprised 4,827 female residents of Philadelphia between the ages of 15 and 45 as of 1996, who were first Medicaid-eligible through Aid to Families with Dependent Children (AFDC) between 1995 and 1996, and who had at least one family member less than 18 years of age at the beginning of the study period. Logistic regression was used to determine the association between maternal SMI and child welfare involvement. Results: Fourteen percent of mothers with SMI received child welfare services, compared with 11.7% of those with other psychiatric diagnoses and 4.2% of those without a psychiatric diagnosis. After adjusting for the presence of an inpatient episode, ethnicity and age, mothers diagnosed with SMI were almost 3 times more likely to have any child welfare involvement or to have children who had an out-of-home placement. Conclusions: The results of this and other studies suggest the urgent need for increased planning and coordination between the child welfare and mental health systems, including providing parenting support as part of mental health treatment for mothers.
Excess Sexual Health Risk among Adolescents Receiving Special Education Services
David S. Mandell, ScD (center investigator); Catharine C. Eleey; Julie A. Cederbaum, MSW; Elizabeth Noll, MA; M. Katherine Hutchinson, PhD, RN; Loretta Sweet Jemmott, PhD, RN, FAAN; Michael B. Blank, PhD1,3
Abstract from published paper Objective: to estimate the relative risk of STIs among children identified as having learning disabilities through the special education system
Design: This cross-sectional study used special education data and Medicaid data from Philadelphia, PA, for calendar year 2002. The sample comprised 51,234 Medicaid-eligible children, ages 12 to 17 years, 8,015 of whom were receiving special education services.Main Outcome Measure: Claims associated with diagnoses of STIs were abstracted and logistic regression was used to estimate the odds of STI among children in different special education categories.Results: 3% of boys and 5% of girls were treated for an STI through the Medicaid system in 2002. Among girls, those in the mental retardation category were at greatest risk (6.9%) and those in the emotionally disturbed or “no special education” category at lowest risk (4.9% each). Among males, STIs were most prevalent among those classified as mentally gifted (6.7%) and lowest among those in the mental retardation category (3.0%). In adjusted analyses, boys with specific learning disabilities and girls with mental retardation or who were academically gifted were at excess risk for STIs. Conclusion: The finding that children with learning disabilities are at similar or greater risk for contracting STIs as other children suggests the need to further understand their risk behaviors, and the potential need to develop prevention programs specific to their learning needs.
Ethnic Disparities in Special Education Placement among Children with Attention Deficit / Hyperactivity Disorder
David S. Mandell, ScD (center investigator); Jasmine K. Davis; Katherine B. Bevans, PhD; James P. Guevara, MD, MPH
Abstract from published paper Objective: to examine disparities in special education placement among children diagnosed with Attention Deficit/Hyperactivity Disorder (ADHD) Methods: We examined special education and Medicaid-reimbursed mental health records for 4,852 children diagnosed with ADHD in Philadelphia, PA during calendar year 2002. Logistic regression was used to estimate the odds of special education placement associated with demographic and clinical characteristics. Results: 38% of the sample received special education. Black children were more likely than others to receive wraparound and less likely to use stimulants. In adjusted analyses, black children were less likely than white children to receive special education (OR = 0.78); among those in special education, black children were more likely to be in the emotional disturbance (ED) category (OR = 1.40). There was a significant interaction between ethnicity and receipt of wraparound; white children with wraparound were more likely to be in special education than white children without wraparound, while black children with or without wraparound did not differ from white children without wraparound in the probability of special education placement. Among those in special education, black children without and without wraparound, and white children with wraparound were more likely than white children without wraparound to be in ED. Conclusions: The results indicate ethnic disparities in special education placement among children with similar clinical profiles, and suggest that mental health and education services substitute for each other differently by ethnicity. Possible reasons include under-treatment of ADHD, differential interpretation of associated behaviors, and differences in parents’ ability to advocate for children’s educational and mental health needs.
The Co-Occurrence of HIV and Serious Mental Illness among Medicaid Recipients Michael Blank, Ph.D., David Mandell, Ph.D., Linda Aiken, RN, Ph.D. and Trevor Hadley, Ph.D.
From Published paper: The objective of this study was to estimate the treated period prevalence of HIV in the Medicaid population and the rate of HIV infection associated with serious mental illness (SMI). Because persons with SMI almost always have their health care reimbursed by Medicaid, studying claims is a useful method of tracking prevalence of illness. This was a cross-sectional study that merged the Medicaid claims data and the welfare recipient files for fiscal years 1994 through 1996. Claims data were merged with welfare recipient files to calculate the treated prevalence of SMI and HIV in the Medicaid population and the odds of receiving an HIV diagnosis given a diagnosis of SMI. The results indicated that the treated period prevalence of HIV among Medicaid recipients without an SMI diagnosis was .4% compared to 2.1% of those with a schizophrenia diagnosis and 2.2% of those with a diagnosis of affective disorder, for a total risk among those with SMI of 4.4%. After controlling for sex, age, race, and time on welfare, the odds of having an HIV diagnosis given a diagnosis of schizophrenia was 1.523 the odds given a diagnosis of affective disorder was 3.87. We concluded that the rate of HIV was significantly elevated among those with SMI, and that risk associated with affective disorder was even high than for schizophrenia.
Prevalence of Serious Mental Illness in Persons with HIV/AIDS in Philadelphia
Michael Blank, Ph.D., Stephen Metraux, Ph.DDavid Mandell, Ph.D., Linda Aiken, R.N, Ph.D. and Trevor Hadley, Ph.D.
From Published paper: This study was intended to improve the estimate from study 1 by crossing that data with the confidential AIDS registry maintained by the Philadelphia Health Department in compliance with CDC active surveillance guidelines. A total of 4,092 people appeared in both data files, many of whom had been previously identified as having had treatment for HIV/AIDS in the claims but many did not. This study substantially improved our estimate of the treated prevalence of SMI and HIV in the Medicaid population and the odds of receiving an HIV diagnosis given a diagnosis of SMI. These results also suggest that other studies relying on secondary data from single sources are likely to vastly underestimate true prevalence. The results indicated that the treated period prevalence of HIV among Medicaid recipients without an SMI diagnosis was .9 % compared with 2.8% of those with a schizophrenia diagnosis and 4.6% of those with a diagnosis of affective disorder, for a total risk among those with SMI of 5.7%. The overall prevalence in the Medicaid population increased from Study 1 from .4% to .9% here, the prevalence in people with schizophrenia increased from 2.1% to 2.8%, and the prevalence for those with affective disorders increased from 2.2% to 4.6%. The total risk for HIV infection among the SMI almost doubled from Study 1 to Study 2 (4.3% - 7.4%). After controlling for sex, age, race, and time on welfare, the odds of having an HIV diagnosis given a diagnosis of schizophrenia was 2.13; the odds given a diagnosis of affective disorder was 4.68. In conclusion, using a more comprehensive sampling technique this study further supported the previous result that the rate of HIV was elevated among those with SMI. Adding the Philadelphia AIDS Registry data had the effect of increasing the prevalence estimates of HIV/AIDS in both the general and the SMI population. We concluded that the rate of diagnosed HIV infection in the SMI population in Philadelphia is between three and five times as high as in the general Medicaid population.
Revised Estimates of the Cost of Care for Persons with HIV/AIDS in 1996
Aileen Rothbard, ScD, Michael Blank, Ph.D., Kay Miller, PhD.
Abstract from draft paper Objective: This paper improves on previous cost estimates of treating persons with SMI and HIV/AIDS and examines the extent of disparity in HIV/AIDS drug treatment in the co-morbid SMI/HIV population. Methods: Subjects are adult Medicaid recipients from Philadelphia who received a diagnosis of severe mental illness and/or HIV/AIDS between 1985 and 1996, and were Medicaid recipients in 1996, the year the costs estimates were constructed. A comparison of utilization and cost of services was done for those with and without mental illness using Medicaid claims records from State Medicaid Research Files. This study improves on the cost estimates of a prior study by including more comprehensive use of ambulatory care, laboratory, pharmacy and long term nursing home care in addition to inpatient, and outpatient services. Results: Persons with co-morbid SMI and HIV/AIDS had the highest annual medical and behavioral health treatment expenditures at $20,038 per person followed by persons with HIV/AIDS only at $14,713, while the SMI only group was $9038 per person. The control group had expenditures of $6613 per person. Cost of medication in the co-morbid group ($3461) was found to be lower than the combined costs of medication in the HIV and SMI groups ($5176), however, there were fewer persons using antipsychotic drugs in the co-morbid group due to a lower prevalence of schizophrenia. No disparity was seen in the medication treatment and cost for HIV/AIDS in the co-morbid population. Also interesting to note was that, though they did not meet criteria for SMI, many individuals in the AIDS only group had psychiatric and/or drug dependence problems as seen by their use of antidepressant(58%) and anti anxiety (35%) medication as well as inpatient and outpatient psychiatric services.
Conclusions: Results are consistent with our previous study showing the co-morbid group to be the most costly population requiring significant care. Also, their expenditure pattern indicates that they are receiving a similar type and amount of medication as those HIV patients without serious mental illness. The AIDs only group, though not co-morbid for serious mental illness, had substantial psychiatric costs. Costs for the two HIV/AIDs population groups average $16,223 per person which is similar to the costs found in the Hopkins study of Medicaid AIDS patients and a four state Medicaid study by NERI looking at AIDS/MH/SA individuals. A follow-up study is needed to see the extent of substitution of drugs for ambulatory and inpatient services for HIV/AIDS related illness and if the SMI are using the same level of antiretroviral drug combinations as other HIV/AIDS patients.
Impact of Housing Policies: Assessing the Residential Segregation of People with Psychiatric Disabilities through a Geographic Information Systems Study
Métraux, S, Caplan J, KlugmanD, Hadley TR (in press). "Assessing Residential Segregation Among Medicaid Recipients With Psychiatric Disability in Philadelphia
Abstract from published paper: This study assessed the extent of residential segregation among 15,246 people diagnosed with psychiatric disabilities and receiving Medicaid (MA) in Philadelphia, and an identically sized group of MA recipients serving as matched controls. Results indicate that overall levels of residential segregation among this group were modest at their most extreme; were not markedly different from a control group of Medicaid recipients without any record of treatment for severe mental illness; and were substantially reduced after taking poverty into account. There were, however, localized areas in Philadelphia that showed distinct concentrations of persons with psychiatric disability, suggesting there may be a subgroup that is more at-risk for living in areas with elevated concentrations of persons with serious psychiatric disability.