Service Utilization and the Individual, Family, and Neighborhood Characteristics of Children with Disabilities in Illinois

Robert Goerge, Lucy Mackey-Bilaver, Bong Joo Lee, David Koepke and Allison Harris

University of Chicago, Chapin Hall Center for Children

This report was prepared under grant #94ASPE266A between the U.S. Department of Health and Human Services (HHS), Office of Disability, Aging and Long-Term Care Policy (DALTCP) and the University of Chicago. For additional information about this subject, you can visit the DALTCP home page at http://aspe.hhs.gov/_/office_specific/daltcp.cfm or contact the office at HHS/ASPE/DALTCP, Room 424E, H.H. Humphrey Building, 200 Independence Avenue, S.W., Washington, D.C. 20201. The e-mail address is: webmaster.DALTCP@hhs.gov. The Project Officer was Kathleen Bond.

The opinions and views expressed in this report are those of the authors. They do not necessarily reflect the views of the Department of Health and Human Services, the contractor or any other funding organization.

I. INTRODUCTION

The objective of this research project was to describe the demographic characteristics and human service utilization of all children and adolescents who are identified as having a disability in Illinois, as well as to understand the stability of the care that they receive through childhood and adolescence. Given the recent changes in the Supplemental Security Income program for children and adolescents in the 1996 Welfare Reform legislation, the study can serve as a baseline to assess the impact of the changes on service provision to children and adolescents with disabilities.

Our specific aims for this study were: 1) To describe the trends in demographic characteristics of the disabled child population in Illinois; 2) To describe the pattern of movement and stability of care of disabled children across different service settings; 3) To describe the pattern of multi-service use of disabled children; 4) To determine the extent to which type of disability and demographic factors are related to multi-service use; 5) To demonstrate the utility of administrative data for advancing the state of knowledge on children with disabilities; and 6) To determine the policy implications of these findings.

Illinois provides a geographic unit that contains a range of environments, cultures, and socioeconomic statuses. Four and a half percent of our country's children live in Illinois, and their demographic characteristics are representative of the country overall as well as the larger states. For example, the racial/ethnic distribution of Illinois children is very similar to that of the United States as a whole (69% white in the U.S., 68% white in IL, and 14% non English-speaking in both the U.S. and Illinois). Illinois ranks in the middle third of states in percent of children living in poverty and in percent of children living with both parents.

This study is unique because it is based on statewide service data. All children in Illinois who are identified by a set of agencies as having a disability will be tracked through the use of the Integrated Database on Children's Services in Illinois (IDB), which has been shown to be a reliable source of data for policy makers at both the state and the Federal level. The IDB combines individual child and family level data from the late 1970's through the present and is updated at least yearly. The population under study are those children identified as having a disability from July 1, 1989 through June 30, 1994.

We begin the report with a short summary of the policy context during the time period and the data and methods used for the study. We follow this summary with a discussion of our results, both descriptive and multivariate. We have not fully analyzed the data prepared for this study and intend to use it in the future, not only as a baseline to follow up this population, but also to better understand the changes that we saw during the five year time period. We hope that the implications flow easily from these results and end with an implications and future work section.

II. POLICY CONTEXT

The time period analyzed in this study spans several major policy changes that affected publicly provided services to disabled children in Illinois and nationwide. The most significant of these changes was the implementation of the Supreme Court's decision in the Zebley vs. Sullivan case. The Zebley decision, which expanded the list of disabling impairments and included a functional assessment in the disability determination for children, significantly increased the number of children receiving SSI nationwide.

Medicaid funded services in Illinois were also under going changes throughout this time period. In May of 1992, mental health services reimbursed by Medicaid were expanded in Illinois through the implementation of the Medicaid clinic/rehab option. For the first time, individuals could receive an array of Medicaid reimbursed mental health services (i.e. case management, crisis intervention, mental health assessment) from community mental health centers. Managed care was also being phased in to primary health care programs reimbursed by Medicaid. In May of 1993, the Healthy Morns/Healthy Kids Managed Care Program began in the City of Chicago after being started prior to May of 1993 in the downstate regions of Illinois. These changes are reflected in our analyst . Whenever possible, we note the policy change as well as highlight how they affected the patterns of service receipt.

III. DATA--INTEGRATED DATABASE ON CHILDREN'S SERVICES IN ILLINOIS

One of the unfortunate consequences of administering services to children through more than one public agency is that information about children and youth receiving services from the state is scattered among the various agencies. The fact that each agency provides only a specified range of services to children and has limited responsibility toward them inevitably restricts each agency's view of the client and its knowledge of the client's history and participation in the service system as a whole. No public agency or other office of state government has been charged with the responsibility of gathering together all the information about children that is collected across the human services, and none of the agencies has routine access to information gathered and stored by the others.

The Illinois Integrated Database on Children and Family Services is a state-level, longitudinal, and relational database constructed out of administrative data gathered by public agencies serving children and families. The database tracks all child and family contact with the following programs and services: foster care, child abuse and neglect, special education, mental health, juvenile justice, Medicaid, Food Stamps, and Aid to Families with Dependent Children (AFDC). In the aggregate, these records represent all child contact with Illinois human services over the course of a decade and can be made to yield a detailed and comprehensive picture of children's service provision in the state.

Administrative databases constitute a rich though underutilized source of knowledge about the operation of the children's service system. What is especially important about these databases is that they document the characteristics of entire client populations, that they often preserve a wealth of information about each child, and that in many cases they extend back more than a decade. The utility of the database, and the kinds of analyses it can support, are greatly enhanced by linking together all records pertaining to each individual client, resulting in a complete record of that client's contact with the agency over the years. Each administrative database represents a complete, longitudinal record of the experiences of the entire population of children receiving service over the time span of the data.

A. Data Description

This section describes the datasets that have been integrated for this study, the nature or format of each of those datasets, and the variables that are available for analysis. We also discuss the reliability of the data.

Department of Public Aid, Client Database

Illinois Department of Public Aid maintains a client database that includes Aid to the Aged, Blind, and Disabled (AABD), Medicaid, Aid to Families with Dependent Children (AFDC), and Food Stamps. AABD is the specific program that we focused on for this study. As stated in the Department description, "AABD is specifically for people who are 65 or older, blind, or disabled (blind or disabled children are included). It is available to people who have applied for, or are receiving, a monthly SSI check from the Social Security Administration. It is also available to people who meet the age/disability criteria for SSI, but whose income is from Social Security or a different source." AABD receipt entitles the recipient to Medicaid in addition to supplemental cash assistance. The database contains demographic details (such as client's age, sex, and address), case opening and closing dates, family composition, level of need, and deduction and grant size.

Department of Public Aid, Medical Service Database

The Principal Investigator has requested (in November, 1990) and received access (in March, 1991) to the Medical Services database of the Illinois Department of Public Aid, the state agency that administers the Medicaid program in Illinois. Non-residential services included under Medicaid include, but are not limited to, assessment, treatment planning, medication, crisis intervention, day treatment, individual, family and group psychotherapy, community-based rehabilitation, intensive family-based services, case management developmental rehabilitation, and pre-hospital screening. "Medicaid" services reflect the full range of medical services provided to children by health care providers in Illinois; a "Medicaid" service simply refers to the method of reimbursement.

The quality of the Medicaid data, in general, is thought to be good. Since the records of the database are tied to payments made to providers, all services provided should be included. Also, since there are regular audits of this data for the purposes of data quality and fraud detection, we expect the data to be of good quality. Finally, automated "edit-checking" prevents wrong codes from being entered into the database.

Department of Mental Health and Developmental Disabilities

The inpatient system records the inpatient episodes of children, adolescents, and adults in the state facilities. The strength of the database is that it is longitudinal and that for some cases it may go back as far as five decades. The database contains the admission date, the movement date (if a change in setting occurred), the prior admission date, and the date of discharge. It contains important demographic information, behavioral characteristics, administrative data (e.g. unit, planning area), legal status and social security number.

The information in this database is believed to be reliable and can often be checked from the case files themselves. The database is useful for tracking admissions and readmissions, length of stay, specific diagnoses, and discharge status of children and adolescents that are hospitalized in DMHDD facilities. However, we do recognize that diagnoses are often used to get a child into an inpatient facility and may not be the result of a comprehensive assessment.

Illinois State Board of Education (Special education database)

There are two funding sources under which handicapped children and adolescents are served. The first is Public Law 94-142, which is the Education of the Handicapped Act. The second is P.L. 89- 313, the Education Consolidation and Improvement Act of 1981, which is the source under which children and adolescents are placed in residential care. These two funding sources are tracked separately. The Illinois State Board of Education (ISBOE) maintains a computerized child tracking system of children who receive special education services in the state of Illinois, the Funding and Child Tracking System (F.A.C.T.S.).

The F.A.C.T.S information system contains information on the child or adolescent's demographics, services received, primary, secondary and tertiary handicap, and type of facility in which service is provided. Since 1) the special education population is the largest population of children who are disabled, 2) most disabled children are first identified as such in the schools and 3) P.L. 94-142 puts the responsibility of serving all children with handicapping conditions aged 3-21 with the schools, this database is an important tool for counting children with specific handicaps and for identifying precursors to future problems.

B. Record Linkage Methodology

The databases of each agency were linked on the basis of common information on each of the individuals in each of the databases. Determining what data were necessary and reliable provided a challenge. This required extensive review of the existing data and cooperation on the part of the human service agencies to allow access to their administrative data.

Deterministic vs. Probabilistic Matching

The greatest technical challenge involved in creating an integrated database is accurately linking the records of individual clients across agencies. This process is complicated by the fact no single variable can be relied upon to establish the identity of a child from the records of various agencies. Though each child receiving a service is typically given an identification number (ID) unique to a particular program, each agency and department uses its own system of identification numbers. Indeed, a single agency may issue a single client more than one ID, since IDs may be assigned each time a case is opened or a child or family receives services. Other variables that might be used to establish an "all-or-nothing" match are equally problematic: even names and birthdates that "match perfectly" may refer to two different individuals, as a result of incorrectly entered data or other human error.

The most reliable means of matching records proves to be a process called probabilistic record- matching, first developed by researchers in the fields of demography and epidemiology (Newcombe, 1988; Winkler, 1988; Jaro, 1985, 1989; Baldwin, Acheson, & Graham, 1987). Probabilistic record-matching is based on the assumption that no single match between variables common to the source databases will identify a child with complete reliability. Instead, probabilistic record- matching calculates the probability that two records belong to the same child using multiple pieces of identifying information. Such identifying data may include name, birth date, gender, race/ethnicity, and county of residence. When multiple pieces of identifying information from two databases are comparable, the probability of a correct match is increased. A few commercial software programs perform record-matching and can be customized to perform matches between two databases. The software program called Automatch was used for this analysis.

IV. METHODOLOGY

A. Identification of Children with Disabilities

From each of the databases described above, the population of children with disabilities was identified. All children under the age of 18 at the beginning of state fiscal years 1990-1994 were identified as disabled if they were receiving services through special education, the Department of Mental Health and Developmental Disabilities (DMEDD) inpatient treatment, Assistance for the Aged, Blind and Disabled (AABD) disability programs, or Medicaid reimbursed services. Because individual level data on SSI receipt was only available to us for very recent time periods, children receiving AABD were used to approximate the SSI population in Illinois over the five fiscal years.¹

Disability identification was based solely on program participation for children in special education, AABD, and DMH. The Medicaid reimbursed services data, however, contains data on services provided for preventative and well-child care as well as for the more serious inpatient and rehabilitative services. Identifying "disabled children" in this service data was not a straight-forward task. In order to use the least restrictive definition of disability, we selected children who demonstrated a disabling condition in one of three ways. First, we selected all children who had a non-institutional claim containing an ICD-9-CM (International Classification of Diseases Clinical Modification) diagnosis code that indicated a disabling condition according to the Social Security disability evaluation standards.² Second, we selected all children who received a service under a category of service that indicated a special need. Finally, provider types that were associated with disabling conditions or conditions requiring rehabilitation were selected. A detailed description of the ICD-9-CM diagnosis codes used to map to the Social Security Administration's listing of disabling conditions can be found in Appendix A.

Each of the selection methods was designed to identify children eligible for SSI under the most current disability evaluation standards. We recognize that disability determination in children is a complex evaluation that does not depend on diagnosis or program service alone. We expect that our selection criteria err on the side of overestimating the disabled population. Throughout the analysis, the utilization of "Medicaid services" will refer only to Medicaid reimbursed services that indicate a disabling condition as defined by our selection criterion. Likewise, children referred to as participating in the Medicaid program will refer only to children receiving services specified in our selection criterion.

A description of the disabling conditions, categories of service, and provider types that we used in the selection and their prevalence can be found in table 1. Two of the ICD-9-CM diagnoses that are included in the selection were associated with a few restrictions. Asthma, for example, was only used identify a child as disabled when it was accompanied by two or more hospitalizations or three or more emergency room visits. Epilepsy was similarly conditioned on the presence of a hospitalization or three or more emergency room visits within the fiscal year. Inpatient hospitalizations (general, psychiatric, and rehabilitative) also had conditions that were necessary for the disability determination in our selection. Three or more inpatient hospitalizations within a fiscal year or one hospitalization that lasted for 20 days or longer indicated disability in this analysis.

From the ICD-9-CM selection, the significant increase in the number of children that were identified after FY 1992 was due mainly to the increased prevalence of developmental speech or language disorders and other mental disorders. This increased "prevalence" however was primarily due to expanded service availability, particularly for mental health services. Two provider types, mental health service providers and local educational agencies, served significantly higher number of children after FY 1992. Most children were selected by the ICD-9-CM diagnosis code on a particular claim rather than the category of service or provider type. Over time the number of children that met the selection criteria by the ICD-9-CM diagnosis code and category of service also rose significantly (Table 2). This was more likely to happen as services became increasingly available for diagnoses that fell within our selection criterion.

B. Study Time Frame

The state fiscal year was the unit of time selected to analyze the indicators of service provision. The primary reason for choosing the state fiscal year was that the special education program data only reflected a child's activity during a school year (September to June). The beginning and end of the state fiscal year (July 1 to June 30) brackets the entire school year. From each of the databases that were analyzed, we identified children receiving each service for fiscal years 1990 to 1994. The beginning of fiscal year 1990 (July 1, 1989) offered a time point nearly 2 years prior to the implementation of the Zebley decision while the end of fiscal year 1994 (June 30, 1994) provided three full years of post-Zebley follow-up.

The time frame for this study allowed us to analyze five cross sections (FY 1990-1994) and four new entry cohorts (FY 1991-1994). By new entry, we are referring to children who were not participating in the previous year of disability services. We defined entry cohorts both by entry into any one of the four programs and by entry into a particular program. The cross-sections were analyzed in order to understand the prevalence of multi-service use among disabled children in Iffinois. By comparing indicators across time, we described the prevailing patterns of multi-service use. We also described the characteristics of children who use multiple services over the five fiscal years. Through the analysis of the new entry cohorts, we tracked program participation before and after the Zebley decision. In part, our findings reflect the dramatic change in program participation due to Zeble and the expansion of the role of Medicaid in non-traditional health care settings in Hlinois. Given the changes in SSI eligibility due to Zebley and the concern over the growing caseloads as expressed in the recent welfare reform legislation, the impact of multi-service use on the receipt of SSI was of particular interest. Using the risk population available to us within our data, we examined the incidence of AAED receipt while controlling for demographic characteristics and type of disability.

V. RESULTS

A. Children with Disabilities

The number of children who were identified as being disabled from each of the four databases is presented below. The overall number of disabled children identified in this analysis has increased over time from 259,113 children in FY 1990 to 277,689 children in FY 1994. This growth represents approximately a 7% increase. This growth occurred over a time when the AABD and Medicaid program populations were increasing steadily while the special education population, the largest of the four, was decreasing. Under out selection criteria, this population of disabled children represents approximately 8% of all children in Illinois.³

Demographic Profiles

In Illinois, each of the four programs that we examined served very different populations in terms of race/ethnicity, sex, and age (Tables 3-6). Children in the state AABD programs were most often African-American children and were increasingly so over time. Because AABD program participation should mirror SSI participation, we see an increase in the number of children served after Zebley implementation. The post-Zebley time period in Illinois shows increased participation of children of all ages. Special education serves the largest number of children with disabilities in Illinois. Nearly 2/3 of the children in special education are boys. Approximately 70% of the children were white in fiscal year 1990 but were more likely to be of African American or Hispanic origin over time. Children identified with disabling conditions in the Medicaid service data were distinct from the special education and AABD populations. Medicaid program children had very similar percentages of white and African-American children over time. Boys were slightly more likely to have been identified in this population in each of the fiscal years. The growth in this population over time had its biggest affect on the age distribution of children with disabling conditions. In fiscal years 1990 and 1991, our selection criterion identified infants 25% of the time. After FY 1992, the percentage of infant children that we identified decreased. Conversely, the percentage of school age children began to rise. The DME inpatient population is the smallest population examined in this study. Boys were slightly more likely to be among children served in DMH facilities as were white children over time. Children served in DME inpatient facilities were more often children over age 9.

Type of Disability

The administrative data on each of the four programs offered few program generated descriptors of the participating child's type of disability. Moreover, disability categorizations were not uniform over time. In the special education program, an aggregation of the primary exceptional characteristic lead to a general categorization over the five fiscal years. Of the children in special education, approximately one half were identified as having learning disabilities. For example in FY 1994, approximately 49% were categorized as having learning disabilities, 25% as having speech or language disorders, 11% as severely emotionally disturbed, and 11% as mentally retarded. (Table 7). From the Medicaid services data, the ICD-9-CM diagnosis code was available for almost all children selected. Institutional claims records which primarily identify inpatient hospitalizations, could not be categorized by ICD-9-CM diagnosis codes. Table 8 presents the distribution of the ICD-9-CM diagnosis categories. The categorization was based primarily on the section headings contained within the ICD-9-CM, however, some particular diagnoses of interest were included in the distribution. While most of the diagnosis categories fluctuated little over the fiscal years, the percentage of children with mental disorders rose significantlyhrom 20.9% of Medicaid program children in FY 1990 to 40.7% of children in the Medicaid program in FY 1994.

Patterns of Program Participation

Because each of the four programs provide very different services to disabled children in Illinois, the duration of program participation varies considerably. We would expect AABD (SSI) program participation, for example, to be longer term given that it provides basic human services including income support and health insurance. Medicaid program identification, however, is not likely to be observed year to year unless a particular disabling condition requires constant treatment provided through Medicaid. Because our data did not allow for continuous time estimates of duration, discrete time service receipt histories were analyzed for new entrants to each of the programs by fiscal year. A new entrant to AABD in FY 1992, for example, was defined as a child not observed participating in AABD prior to the fiscal year 1992. Only children who were age 13 or younger at the end of the fiscal year of the first year of service, and thus would not reach age 18 during the observation period, were examined. The survival rates are presented in Table 9.

Both special education and AABD program participation were likely to last longer than Medicaid program or DMH inpatient participation. For children entering the AABD program in FY 1991, 70% were still participating after 4 years. Over time, children entering AABD programs were more likely to continue participation after 1 year. For example, 78% of the AABD children entering in FY 1991 were participating after 3 years while 86% of the children entering in FY 1992 were participating after 3 years. In special education over time, children were slightly less likely to continue participation for a second year. Of children entering special education in FY 1991, 75.7% were participating for a second year while only 71.9% of children entering in FY 1993 were participating. The length of participation in the Medicaid program increased substantially between the FY 1992 and FY 1993 cohorts. 32.2% of the FY 1992 entry cohort was found to be participating for a second year while 44.9% of the FY 1993 entry cohort was participating in the next year.

Service History

Table 10 presents indicators of the past year of service receipt for each program entry cohort. New entry is again defined by the first year of service in a particular program. Overall, AABD entrants are the most "known" to the system. In FY 1991, 47% of children entering AABD had participated in another service last year while just 3% of special education entrants and 14% Medicaid entrants had done so. An examination of the participation rates of AABD entrants over the fiscal years, however, reveals that the degree of participation in other services prior to AABD entry has decreased. By FY 1994, 39.3% of children entering AABD had prior service experience. For special education and Medicaid new entrants, we observe the opposite trend. Children in special education were more likely to have had prior service experience in both AABD and Medicaid. Entrants to the Medicaid program were also more likely to have received other services in the past year. Medicaid entrants who participated in special education in the past year increased from 12.7% of entrants in FY 1991 to 18.3% of entrants in FY 1994. Although these indicators can only approximate the pathways that children take, the trend in recent fiscal years suggests that frequency of transitions between other services to Medicaid and special education is increasing while the frequency of transitions from other services to AABD is decreasing.

B. Patterns of Multi-Service Utilization

Describing the patterns of multi-service utilization of disabled children was a primary goal of this study. At a time when eligibility for SSI will possibly be restricted, it is important to understand the extent to which SSI and other services have been supporting the population of children with disabilities. In addition, indicators of multi-service utilization over time can further illustrate the effects of the Zebley decision.

Cross-Sectional Analysis

A point in time examination of multi-service utilization reveals that the majority of children participate in one of two services: special education or Medicaid services (Table 11). Over time however, multi-service use becomes more prevalent. In FY 1990, 5.3% of the children receiving service were multi-service users, while 15.2% of the children receiving services in FY 1994 were multi-service users. Although a small percentage of the total disabled population, the number of children using some combination of programs increase substantially over time. Only 2,747 children used AABD, special education, and Medicaid services within FY 1990. In FY 1994, this number had increased nearly four times to 10,031 children. The number of children participating in special education and the Medicaid program also increased from 5,501 in FY 1990 to 16,201 children in FY 1994. At the same time that these increases occurred, the nurnber of children participating in AABD or the Medicaid program only increased steadily while the number participating in special education only decreased over time.

By examining the incidence of multi-service utilization by program type, we see that the patterns of multi-service use do vary considerably (Table 12). Children in AABD programs have the highest incidence of multi-service utilization but it is not increasing over time. In FY 1990, 51.3% of the AABD children were in special education while only 42.3% were participating in FY 1994. As a percentage of all children in special education, multi-service use is rare but increasing over the fiscal years. Special education children utilizing Medicaid reimbursed services increased substantially from 3.6% of children in FY 1990 versus 13.3% of children in FY 1994. Similarly, special education children were more likely to participate in AABD. In FY 1990, 2.7% of children were participating in AABD while 9.4% were participating in FY 1994. Children receiving Medicaid services were slightly more likely to participate in either AABD or special education over time. In FY 1990, 13.8% of the Medicaid children were also participating in AABD while 19.8% were participating in both programs in FY 1994.

Entry Cohorts

Although multi-service use increased overall and especially for children in special education and Medicaid, the impact of children new to disability programs (especially because of Zebley implementation) could not be assessed in the cross sectional analysis. New entry cohorts, defined by the first year of participation in any of the four services, were analyzed for FY 1991-1994. Although the overall number of disabled children identified from the four programs has grown, the overall number of new entrants to one of the four programs has decreased from 89,193 children entering in FY 1991 to 59,014 children entering in FY 1994 (Table 13). This trend is due primarily to the decrease in special education enrollment. None of the multi-service categories increased substantially over the fiscal years. Only a small percentage of the new entrants participated in more than one service during the fiscal year of entry. For example, 2.7% of the FY 1991 entry cohort participated in multiple services during FY 1991. In FY 1994, only 5.4% of the new entrants participated in multiple services. This was true despite the radically changing composition of the entry cohorts. In FY 1991, single service special education children made up 70.5% of the cohort. In FY 1994, only 23.8% of the entry cohort were single service special education children while 61.2% were single service children identified in the Medicaid reimbursed services data.

The trends seen by program type in the cross-sectional analysis are observed again for the new entrants. Over the new entry cohorts, children who began their service receipt in AABD became increasingly less likely to participate in multiple services within their first year of service (Table 14). New entrants participating in special education were slightly more likely to participate in multiple services, especially Medicaid. Children in the Medicaid program for the first time were slightly more likely to be multi-service users. Although the new entrants look similar to the cross-sections in terms of the trend in multi-service use, they contribute little to the overall incidence of multi-service use. In FY 1991, multi-service users that were new entrants made up approximately 1% of the 16,213 multi-service users observed in FY 1991. In FY 1994, new entrants using multiple services continued to represent I% of disabled children using multiple services in Illinois. It appears that although multi-service utilization has increased since FY 1990, the new entrants contributed little to the trend.

The time frame for this study allows us to follow the FY 1991 cohort of new entrants up to three years. This longitudinal analysis can illustrate the program participation of single service children in the future. As presented in table 15, each of the single service groups from each cohort was identified by multiple service use in the future. The patterns overall and by service type are not clear. Overall, an additional 8.1% of the FY 1991 cohort and 10.6% of the FY 1992 cohort was found to be multi-service users after 2 years. The percentage of multi-service users identified after one and two years appears to rise until except for a sharp decrease for the FY 1993 cohort. For children only participating in AABD in FY 1991, 16.9% became multi-service users after 1 year while only 10.8% of the children in the FY 1993 cohort became multi-service users. After 2 years, however, the percentage of children becorriing multi-service users rose slightly from 23.2% of the FY 1991 cohort to 26.2% of the FY 1992 cohort. Special education and Medicaid program single users were even less likely to become multi-service users in the future.

C. Characteristics of Multi-Service Users

In order to approximate which disabled children participated in more than one service, the descriptive variables available to us were examined by single service and multi-service categories. In table 16, the distribution of the primary characteristic is presented for all special education children, as well as special education children participating in more than one service. Only multi- service use involving AABD and the Medicaid program is presented because it represents the majority of the multi-service utilization. In each of the fiscal years, the majority of children in both special education and AABD were those with mental retardation. Over time, however, children with learning disabilities and severe emotional disturbance were more likely to participate in AABD an special education simultaneously. By FY 1994, 35.1% of the children in both special education and AABD were mentally retarded, 30.9% were learning disabled, and 15% were severely emotionally disturbed. In terms of the utilization of the Medicaid program and special education, the patterns of service utilization for different categories of the children varied over time. Mentally retarded and SED children became less likely to use Medicaid reimbursed services while participating in special education. In FY 1990, 25.4% and 25.3% of the children participating in both services were respectively mentally retarded and SED while in FY 1994 only 19.5% and 17.4% of children in both services were respectively mentally retarded and SED. Learning disabled children were more likely to represent children participating in special education and receiving Medicaid reimbursed services over time. The increase was especially pronounced between FY 1992 when 15.9% of children in both services were learning disabled and FY 1993 when 24.4% of children were indicated as learning disabled in special education.

Children participating in the Medicaid program were similarly examined by single and multiple service categories. In FY 1990, the children who received Medicaid reimbursed services and special education looked very different from children who received Medicaid reimbursed services and AABD (Table 17a-17e). The majority of children participating in both the Medicaid program and special education were receiving service under special categories of service (14.6%), for multiple diagnoses (14.2%), and for other mental disorders (13.9%). Children receiving Medicaid reimbursed services and AABD were most likely to be treating infantile cerebral palsy (16.5%) and multiple diagnoses (15.0%) within fiscal year 1990. In terms of the type of Medicaid recipient with the highest degree of multi-service utilization, infantile cerebral palsy was highest with 80.9% of children in FY 1990 participating in multiple services. Children with multiple diagnoses and developmental speech or language disorders also had substantial multi-service use.

These patterns begin to change in FY 1992. In FY 1992, children participating in the Medicaid program and special education are more likely to have been identified by multiple diagnoses, other mental disorders, and developmental speech and language disorders. By FY 1994, the majority of children participating in the Medicaid program and either special education or AABD were being treated for mental disorders. By 1994, the percentage of children with both mental disorders and multiple diagnoses who used special education or AABD had increased during the period. This increase, however, is due in part to the changing composition of the Medicaid population after the Medicaid community mental health services program began. Without controlling for other factors, it is not clear whether the increase is due to changes in multi-service use or to changes in the composition of the population. This hypothesis will be tested using multivariate methods with regard to AABD participation specifically.

D. Factors Affecting the Incidence of Multi-Service Utilization

The descriptive analysis revealed a pattern of increasing multi-service utilization over the time period. The analysis by program type indicates that multi-service utilization is increasing for the special education and Medicaid service populations, but is the decreasing for the AABD service population. In order to further confirm the trends observed in the descriptive analysis, logistic regression was used to model the probability that a child would use multiple services during a fiscal year while controlling for demographic and disability characteristics.

Table 18 presents the odds ratios from each of the three models. Each program type was modeled separately because of the variation in the incidence of multi-service use. DMH multi-service utilization was not examined in the multivariate analysis because it was so rare. The odds ratios describe the effect on the probability of being a multi-service user during a fiscal year relative to a baseline group while holding all other factors constant For example in the model for AABD program participants, females are 20% less likely than males to be multi-service users during a fiscal year. Minority children in AABD were less likely to be multi-service users than were white children. The odds ratios of the age categories indicate that children between the ages of 8 and 13 are slightly more likely than children ages 6 to 7 to be multi-service users. It appears that the very young and teenage children were less likely to be using multiple services during a fiscal year. The direction of the indicators of fiscal year indicate a discontinuous upward trend in the likelihood that children in AABD used multiple services during the fiscal year. In FY 1991 and FY 1992, children were nearly two times more likely to be multi-service users than children in AABD in FY 1990. In FY 1993, however, children were slightly less likely to be multi-service users than children in FY 1990. This drop coincides with the dramatic increase in the AABD population after Zebley and the decreasing likelihood of new entrants to have had prior service experience in any of the other three disability programs. Although this decreasing likelihood of multi-service use persisted in the descriptive trend, after controlling for all other factors the direction of the FY 1994 indicator showed that children were more likely to have participated in multiple services than in FY 1990.

Model 2 presents the odds ratios for children participating in special education during each of the fiscal years. Unlike AABD, females in special education were slightly more likely than males to be multi-service users. Minority children were 2 to 3 times more likely to be multi-service users. With the exception of infants, the likelihood of being a multi-service users decreases with age. Children over the age of 7 are approximately 20% less likely to be multi-service users than children ages 6 and 7. The fiscal year indicators point to a slightly increasing likelihood of multi-service use among the special education population between FY 1992 and 1993. In FY 1994 however, children were 2.25 times more likely to be multi-service users than children participating in FY 1990 holding all other characteristics constant. In addition to the program and demographic characteristics already mentioned, the primary characteristic for each child per fiscal year was included in the model. The odds ratios show that mentally retarded children, physically disabled children and children with other disabilities were approximately 2 times more likely to be multi-service users than were severely emotionally disturbed children. Children with learning disabilities and speech or language disorders were nearly 70% less likely to be multi-service users during a fiscal year than were severely emotionally disturbed children.

In Model 3, the likelihood of being a multi-service user during a fiscal year is modeled for children participating in the Medicaid program. The effects of demographic and program indicators for the Medicaid program population tend to look similar to the effects in the Model 1. Females, for example, are 25% less likely than males to be multi-service users in a fiscal year. Hispanic and African-American children are slightly more likely than white children to be multi-service users during a fiscal year. In the Medicaid program population, children over the age of 7 are more likely to be multi-service users than are children ages 6 to 7. Like the special education population, children participating in the Medicaid program after FY 1990 are more likely to be multi-service users during the fiscal year than are children who participated in FY 1990. The description of the selection into the Medicaid program population was also included in the model in order to control for any effect of the type of disabling condition. Children selected due to mental diagnoses or a combination of mental and non-mental diagnoses were more likely to be multi-service users during a fiscal year than were children selected because of one or more non-mental diagnoses.

E. Predicting AABD Service Utilization

Understanding the factors associated with AABD service utilization is of particular interest in light of the policy changes that are likely to emanate from welfare reform. As we have seen, multi-service use has increased over time among the disabled children identified in this study. In order to assess the risk of AABD program participation over time in Illinois, we have focused on the population of children identified by our Medicaid service selection criterion. We would expect all of these children to be at risk for AABD. We recognize that there are certainly children at risk who would not have prior Medicaid program participation. Again, logistic regression was used to predict the probability of AABD participation within 3 years of participation in the Medicaid program. All children receiving Medicaid reimbursed services in FY 1990 through 1992 were included in the model.

Table 19 presents the results of this analysis. Females were nearly 20% less likely than males to participated in AABD within three years. Espanic and African American children were approximately 1.3 times more likely than white children to participate in AABD. Children over the age of 7 were more likely to become AABD participants within 3 years than were children ages 6 and 7. Children under age 6 were less likely to become AABD participants. The method of the Medicaid selection was also included in the model. Interestingly, children included in the Medicaid program population because of a combination of mental and non-mental diagnoses during a fiscal year were nearly three an a half times more likely to become AABD participants than children included because of one or more non-mental disorder diagnoses. This strong effect indicates that children with the more complex conditions were most likely to receive AABD. The likelihood of participating in AABD appeared to increase over fiscal years 1990 to 1992. Children participating in the Medicaid program in fiscal year 1991 were over two and a half times more likely to receive AABD within three years than were children in Medicaid in FY 1990. Children in Medicaid in FY 1992 were over three times as likely.

VI. IMPLICATIONS AND FUTURE WORK

This study shows very strikingly that the population of children served by public programs for disabilities in Illinois has changed dramatically over a five year period. The expansion due to Zebley to include more children with emotional disorders has opened a new door into services provided through the Medicaid program. The Medicaid program has been expanded simultaneously to address the needs of children with severe emotional disturbances by providing services in settings other than traditional health care settings. It also seems, however, that the new SSI population is slow to use additional services. In other words, children in the SSI population do not quickly become participants in multiple programs. The special education population is also decreasing as there is a move to bring children with handicapping conditions into the regular classroom; the poor fiscal condition of schools also provides little incentive to identify more children as requiring special education. This is our interpretation of the diverging trends in program eligibility and utilization between the SSI and Medicaid programs and the special education program.

We also found that multiple program participation reflected intuitions about how that might occur. In general, children with less severe conditions (language disorders or learning disabilities) were less likely to participate in multiple programs that children with more severe or chronic disabilities (e.g. severe emotional disturbance). This, however, may be due to the fact that we focused on public programs, which are primarily used by the poorer segment of the population; this means that the children with less severe conditions in special education would be less likely to appear in the Medicaid or SSI population.

Given the changes that have been made to SSI eligibility criteria in 1996 and the potential future changes in Medicaid and welfare programs in general, we hope that this study provides a baseline for future monitoring of participation in programs for disabled children. While we have not included every program for children with disabilities, we have included the major public programs which would reflect the trends in the service provision to the population. The addition of other programs into our analysis could provide important information on those programs. In particular, the addition of the new early intervention programs would be an important addition. Better tracking of these children from birth is also key to more complete and better monitoring.

ENDNOTES

1. Throughout this report, AABD receipt is used as a proxy variable for SSI receipt. The table below presents the published numbers of children receiving SSI in the state of Illinois in December of each year. The measure in December represents the midpoint of each fiscal year. The AABD figures map nicely to the midpoints even during the period of rapid growth beginning in 1992.

   	Number of Children in Illinois Receiving:
   	December 1989	December 1990	December 1991	December 1992	December 1993
   Supplemental Security Income (SSI)	12,951	15,387	21,178	31,874	41,225
   	July 1, 1989   to June 30, 1990	July 1, 1990   to June 30, 1991	July 1, 1991   to June 30, 1992	July 1, 1992   to June 30, 1993	July 1, 1993   to June 30, 1994
   Aid to the Aged, Blind, and Disabled (AABD)	12,184	14,839	19,986	33,555	43,941
   SOURCE: Social Security Bulletin, Annual Statistical Supplement, 1989-1993

2. An error was found in the diagnosis field of our medical claims database for services received January 1993-June 1995. The error involved only institutional claims made during the time period, thus we did not expect our diagnosis code selection to be substantially affected. In order to measure the error introduced, we compared the number of children that we identified as disabled using all claims versus all non-institutional claims in fiscal year 1990. We found that using all types of claims in the diagnosis code selection gave us 42,147 Medicaid program participants versus 34,541 using only non-institutional claim records. This represents 82% of the total Medicaid population available to us if the diagnosis code of all claim records could have been used.

3. The population incidence of 8% is based on 3,338,418 children living in Illinois according to the 1990 U.S. Census.

TABLES