Evaluation of the District of Columbia's Demonstration Program,Managed Care System for Special Needs Children: Final Report

Publication Date

May 31, 2000

U.S. Department of Health and Human Services

Evaluation of the District of Columbia's Demonstration Program, "Managed Care System for Disabled and Special Needs Children": Final Report

Abt Associates Inc.

June 2000

This report was prepared under contract #500-96-0003 between the U.S. Department of Health and Human Services (HHS), Health Care Financing Administration (HCFA) and Abt Associates Inc. In addition to HCFA, other support for the study, Evaluation of the District of Columbia's 1115 Waiver for Children with Special Health Care Needs, has been provided by HHS's Office of the Assistant Secretary for Planning and Evaluation, Office of Disability, Aging and Long-Term Care Policy. For additional information about the study, you may visit the DALTCP home page at http://aspe.hhs.gov/daltcp/home.htm or contact the ASPE Project Officer, Gavin Kennedy, at HHS/ASPE/DALTCP, Room 424E, H.H Humphrey Building, 200 Independence Avenue, SW, Washington, DC 20201. His e-mail address is: Gavin.Kennedy@osaspe.dhhs.gov.

EXECUTIVE SUMMARY: KEY FINDINGS AND IMPLICATIONS

I. INTRODUCTION

A. Background of the Demonstration

B. Managed Health Care for Children with Special Needs: Evidence from the Literature

C. Medicaid Managed Care and Children with Special Needs: Current Practice

Overview
Case Studies
Discussion

D. Evaluation Design

E. Organization of the Report

II. DESIGN AND EVOLUTION OF THE DEMONSTRATION

A. Origins of the Demonstration: 1993-1999

B. HSCSN: Startup and Implementation: 1996-1999

C. HSCSN Finances, 1996-1999

Startup Costs
Capitation Rate and Risk Sharing
HSCSN Financial Results

D. CHCF/MAA Oversight of the Demonstration

E. Relationship to the Community and Coordination with Other Service Systems

The DC Public Schools
Residential Treatment Facilities
Foster Care
Public Health Clinics
District Hospitals
Relationships with the Community
Coordination and Collaboration: Conclusion

F. Extensions of the Demonstration and the CASSIP Procurement, 1998-2000

The First Extension, 1988-1999
The Second Extension, 1999-2000

G. Conclusion and Implications of the Implementation Analysis

III. ENROLLMENT AND SELECTION

A. Introduction

B. Enrollment Process

C. Tracking Enrollment: Data Sources

D. Characteristics of Participants

Average Age and Length of Enrollment
Eligibility and Participation Rates
Reasons for Disenrollment
Distance Participants Traveled for Services
Description of PCPs and their HSCSN Caseloads

E. Evidence for Selection Bias in the Demonstration

Selection in Voluntary Demonstrations
Data and Methods
Findings

F. Conclusions and Implications of the Enrollment and Selection Analysis

IV. CARE MANAGEMENT AND PROVIDER FEEDBACK

A. HSCSN Care Management: Goals/Objectives/Structure/Process

B. Evolution of Care Management/Care Managers Roles and Perceptions During the Demonstration

C. Processes of Care

D. Provider Feedback: Changes in Role and Response of Care Management

E. Case Record Review

Introduction
Data and Methods
Findings

F. Families of Demonstration Participants

G. Conclusion and Implications of the Care Management Analysis

V. IMPACTS OF THE DEMONSTRATION

A. Introduction

B. Data and Methods

Caregiver Survey
HSCSN and Medicaid Eligibility and Claims Data

C. Analysis Methods

Impact Analyses: Concurrent Comparisons
Impact Analyses: Change Measurement
Claims Based Analyses: Trends in HSCSN Utilization, Comparisons of Participants and Non-Participants, and the "Costs" of the Demonstration

D. Impact on Access and Satisfaction

Background
Measurement Issues
Findings
Discussion

E. Utilization: Formal and Informal Caregiving

Background
Measurement Issues
Findings
Discussion

F. Preventive and Medical Services Utilization

Background
Measurement Issues
Findings
Discussion

G. Impact on Health Status and Impairments

Background
Measurement Issues
Findings
Discussion

H. Utilization and Expenditure

Background
Measurement Issues
Findings
Discussion

I. Conclusion and Implications of the Impact Analyses

VI. CONCLUSION

NOTES

REFERENCES

APPENDICES

A. Case Study Protocols

B. Enrollment and Disenrollment in the Demonstration

C. Demonstration Participation

D. Reasons for Disenrollment From the Demonstration

E. Providers/Participants by Zip Code

F. Case Record Review

G. Components of Case Records That were Reviewed

H. Caregiver Survey Instrument

I. Acronyms Used in the Report

LIST OF EXHIBITS

Exhibit 2.1: Demonstration Milestones: Planning and Implementation

Exhibit 2.2: Calculation of the Capitation Rate, 1995-2000

Exhibit 2.3: Risk Sharing Corridors Between the District and HSCSN for Different Values of the Medical Claims Ratio

Exhibit 2.4: HSCSN Revenues and Expenses, 1996-1999, and Comparison of 1999 vs. 1998

Exhibit 2.5: Medical Cost Increases Per Member Per Month (PMPM) in 1998 and 1999: Total and Medical Cost Items

Exhibit 2.6: Comparison of the Demonstration Model to the Original CASSIP Concept and the Draft CASSIP RFP

Exhibit 3.1: Number of Demonstration Participants by Quarter: Third Quarter 1996-Second Quarter 1999

Exhibit 3.2: Enrollment by Month: March 1996-June 1999

Exhibit 3.3: Mean Age of Participants and Disenrolled Children and Length of Enrollment: HSCSN FY 3rd Quarter 1

Exhibit 3.4: Demonstration Participation, by Quarter: SSI-Eligible Participants as Percentage of Total SSI Eligible, non-SSI- Eligible Participants as a Percentage of All Participants

Exhibit 3.5: Disenrollment Reasons, by Frequency of Reason Given

Exhibit 3.6: Distance Traveled to Receive Care: Provider Zipcode in Comparison to Participant Residential Location

Exhibit 3.7: Comparison of Provider Organizations by HSCSN Caseload and Zipcode

Exhibit 3.8: Enrollment/Selection Analysis: Steps Taken in Data Preparation and Sample Selection

Exhibit 3.9: Frequency of Diagnoses Recorded for Participants vs. Non-Participants, Before and After Enrollment/Anchor Dates

Exhibit 3.10: Enrollment/Selection Analyses: Independent Variable Means--Total, Participants, Non-Participants

Exhibit 3.11: Enrollment/Selection Analysis: Estimated Effects of Independent Variables on Probability of Enrollment

Exhibit 3.12: Enrollment/Selection Analysis: Estimated Effects of Independent Variables on Probability of Disenrollment

Exhibit 4.1: Explanations for Unclear Treatment Goals in POTs

Exhibit 4.2: Changes in Condition and Evidence of Planned Service Changes

Exhibit 4.3: Noted Service Needs and Service Provision for Specific Health Issues

Exhibit 4.4: Rates of Appropriately Followed EPSDT Guidelines for Vision, Hearing, Dental and Physical Examination

Exhibit 4.5: Acuity of Emergency Room Visits: Emergency, Urgent and Non-Urgent First and Subsequent Visits Per Child

Exhibit 4.6: Delivery of Prescribed Services by Type of Service

Exhibit 4.7: Evidence of Timely Response to Participants' Health Problems

Exhibit 4.8: Frequency of Provision of Special Education to Caregiver/ Child for Specific Health Issues

Exhibit 5.1: Caregiver Survey: Wave One and Wave Two Completions for Participants and Non-Particpants

Exhibit 5.2: Caregiver Survey: Characteristics of Responding Caregivers

Exhibit 5.3: Measures of Perceived Access/Satisfaction: Caregiver Survey

Exhibit 5.4: Perceived Access/Satisfaction: Percent Reporting Provider Attachment, and Usual Provider Handles Needs

Exhibit 5.5: Perceived Access/Satisfaction: Percent With Minimum Indirect Costs of Access to Services

Exhibit 5.6: Perceived Access/Satisfaction: Percent Reporting Unmet Needs for Specific Services

Exhibit 5.7: Perceived Access/Satisfaction: Percent Reporting Good to Excellent Access to Services

Exhibit 5.8: Perceived Access/Satisfaction and Length of Enrollment

Exhibit 5.9: Measures of Formal/Informal Caregiving

Exhibit 5.10: Formal/Informal Caregiving: Percent Using Formal and Informal Care

Exhibit 5.11: Formal/Informal Caregiving: Percent Reporting Mix of Skilled, Unskilled, Informal Care

Exhibit 5.12: Formal/Informal Caregiving: Percent Reporting Specific Measures of Caregiver Burden

Exhibit 5.13: Formal/Informal Care by Length of Enrollment

Exhibit 5.14: Utilization of Preventive and Medical Services: Caregiver Survey

Exhibit 5.15: Preventive and Medical Service Utilization: Percent Reporting Gap of More Than One Year in Receiving Preventive and Mental Health Services

Exhibit 5.16: Prevention and Medical Service Utilization: Percent Using Medical Treatments, Total and At Home Utilization

Exhibit 5.17: Percent Using Preventive and Medical Services Utilization: Multiple Treatments, Total and At Home

Exhibit 5.18: Measures of Health Status/Function: Caregiver Survey

Exhibit 5.19: Health Status/Function: Percent Reporting Problems in Cognitive/Behavioral/Emotional Development

Exhibit 5.20: Health Status/Function: Percent Reporting Problems/ Using Special Equipment Related to Physical/Social/ Overall Functional Capacity

Exhibit 5.21: Health Status/Function: Percent Reporting Problems with Physical/Social/Overall Functional Capacity; Percent with School/Work Limitations

Exhibit 5.22: Health Status/Function: Percent Reporting Good to Excellent Health, Among Disability and Bed Days

Exhibit 5.23: Utilization and Expenditure Analysis Measures

Exhibit 5.24: Service Utilization of Participants and Non-Participants: Selected Categories and Total, Pre- and Post-Anchor: Percent Using Service

Exhibit 5.25: Service Utilization of Participants and Non-Participants: Selected Categories and Total, by Sampling Windows 1 and 4: Percent Using Services

Exhibit 5.26: Utilization by Age, Participant Status: Pre-Anchor and Post-Anchor Periods: Percent Using Service

Exhibit 5.27: Total Expenditure Analysis: Medicaid, HSC and Capitation Expenditure per Month, Participants and Non-Participants

EXECUTIVE SUMMARY: KEY FINDINGS AND IMPLICATIONS

Demonstration Implementation and Management

This Demonstration tests the viability of a program to coordinate and integrate care for SSI-eligible children with special health care needs, financed through capitation payments. Health Services for Children with Special Needs (HSCSN) manages the Demonstration, under a waiver administered by the District of Columbia (DC) Medicaid.
During the Demonstration, HSCSN (1) has undergone a fundamental restructuring, (2) has experienced serious problems coordinating with other DC agencies serving special needs children, and (3) has experienced repeated financial losses.
HSCSN concludes that the capitated model tested in the Demonstration is not viable, in part because the population of special needs children in the District is too small to protect a plan against severe financial losses. Despite a proposal by DC Medicaid to add stop-loss protection, HSCSN has announced that it will not enter the waiver re-competition.

Demonstration Enrollment

HSCSN has enrolled nearly one-half of all eligible children in the District.
Participation in the Demonstration is voluntary. This has meant that families and children who participate are different in several ways from those who did not enroll. In particular, participants tend to be heavier users of services than non-participants.

Care Management

Care manager teams, made up of a team leader, care managers and care manager associates, coordinate with each participating child's primary care provider (PCP).
Problems in communication among major stakeholders (care managers, providers, outside agencies, families) have impeded effective care coordination throughout the Demonstration.
In response to care manager and provider concerns, HSCSN has moved to reduce caseloads for the care manager teams.
Families generally agree that care managers have eased the burden on informal caregivers by assuming many of the logistical functions of coordinating care. Stakeholders share the perception that care management is the most valuable feature of the Demonstration.

Impacts of the Demonstration

The Demonstration may have positive impacts over a range of outcomes, but design and data problems place limits on the evaluator's ability to detect these impacts.
According to most measures, Demonstration participants enjoy equal or better access to services, compared to non-participants. Participants' caregivers are generally more satisfied with their access to services than non-participants.
Despite positive perceptions of the Demonstration expressed by caregivers, there is little evidence that these caregivers benefit from reduced burden, relative to non-participants.
Participant caregivers report somewhat heavier use of many medical services than do non-participant caregivers. However, the most salient differences between groups are in preventive services: participants are more timely in their use of preventive services than non-participants.
There is some evidence that Demonstration participants are more physically impaired than non-participants. However, measures of mental and developmental function show no consistent patterns.
Despite higher reported satisfaction with access to services among participants, participants' caregivers are not more positive about their children's current or future health status.
In general, participants use more of most ambulatory and institutional services than non-participants, and their rates of utilization increase, once in the Demonstration.

Policy Implications

Although the particular model tested in this Demonstration may not be viable, coordination and integration of care, provided by professionals who understand special needs children, is a valuable service.
Policy makers might consider alternative models of financing care management for special needs children -- for example, primary care case management (PCCM) programs that pay a child's principal physician directly to provide coordination and integration.

I. INTRODUCTION

A. Background of the Demonstration

On February 2, 1994, the Hospital for Sick Children (HSC) incorporated Health Services for Children with Special Needs, Inc. (HSCSN). HSCSN proposed to demonstrate the effectiveness of a managed care organization serving only children with special health care needs (CSHCN) who are eligible for Supplemental Security Income (SSI). On March 17, 1994, a formal waiver application was submitted to HCFA for approval. The Demonstration model envisioned that HSCSN would take the fragmented care delivered in the fee-for-service system and -- with the incentives of a fixed capitation rate -- integrate and manage the care actively, across the full spectrum of covered services and in a fashion sensitive to the particular needs of each child.

The Health Care Financing Administration (HCFA) and the Assistant Secretary for Planning and Evaluation (ASPE), in the Office of the Secretary, have contracted with Abt Associates, Inc., to evaluate the District's Demonstration.¹ Results of this work will increase general understanding of what factors go into designing, implementing, and maintaining this type of health care system; which children are willing to enter a health care system such as the one designed by this Demonstration; and what kinds of experiences the organizations, providers, and clients can expect to have when a managed care system like the Demonstration is introduced.

HSCSN planned to have information systems and a system of care management in place to perform the actual integration of care. In some cases, it was expected that HSCSN's management of care would reduce the care beneficiaries received. But selective increases were expected as well. For example, small physical changes in beneficiaries homes (e.g., ramps, telephones, and special forms of transportation) could resolve critical difficulties for some beneficiaries and actually reduce the costs of service. Such non-standard benefits were extremely difficult to fund under Medicaid, but were expected to be readily funded by HSCSN when those benefits made sense. There was in this model an assumption that the Demonstration system could be more efficient, in that beneficiaries could get improved services and costs could be saved at the same time.

Given the general commitment to integration of services, there was also an assumption that the Demonstration model would be effectively spliced into the complex web of public support programs for children with special needs. What this meant in practice was that HSCSN planned to coordinate with other programs and providers at the boundaries of HSCSN's responsibilities -- notably, the schools, the mental health system, and the foster care system -- to provide a measure of integration of those services with the Demonstration services.

The study of the implementation of the Demonstration involves an in-depth case study based on interviews, a longitudinal series of site visits to the District's Medicaid office and HSCSN, and a review of documentary records. This Demonstration is not based on an experimental design, so the evaluation is limited in its ability to detect impacts. The comparison group sampled for impact analyses includes eligibles who decided not to participate. The Demonstration relies on voluntary enrollment. Understanding enrollment patterns will be critical to examining outcomes. The District's Demonstration presents an important opportunity to analyze health plan choice, and the factors important to that choice, within a population with extremely heterogenous and costly health care needs.

This evaluation focuses on four areas:

analyses of the implementation of the Demonstration, from which other states and managed care systems might learn how to better serve similar populations;
analyses of enrollment, which will describe which children enroll in this type of system, as well as disenroll;
analyses of the care management system in HSCSN; and
impact analyses that document the experiences of the District plan, providers, children and their families.

Some researchers have studied the experience of children with special needs in managed care. However, most often this experience reflects policies of main-line managed care organizations, not programs specifically designed for this population. Also, few programs serve the segment of special needs children enrolled in SSI.

Section B of this chapter reviews research on special needs children in managed care. It begins by comparing the disparate available estimates of how many children have special health care needs. It then reviews arguments on the advantages and disadvantages of managed care for these children, drawing on both normative and objective studies. Section C concludes with an overview of state Medicaid policies for children with special needs during a period when states pressed increasing numbers of Medicaid enrollees into managed care. Several case studies, compiled by Hill and his colleagues (1999), are presented to illustrate the range of organizational and financial models developed for Medicaid-eligible special needs children.

Section D describes the design for the evaluation of the DC Demonstration. Threats to the evaluation come from the timing and quality of Demonstration data, and these threats clearly limit possible conclusions about Demonstration impacts across a range of measures of access, satisfaction, impairments and utilization. This section also stresses the importance of combining findings from several analyses, including statistical tests on quantitative data and assessments of qualitative information, to strengthen the overall assessment of Demonstration impacts.

Finally, Section E summarizes the contents of the rest of this report.

B. Managed Health Care for Children with Special Needs: Evidence from the Literature

With the rapid expansion of managed care, CSHCN are being increasingly included in managed care programs. Until now, this population has received services primarily on a fee-for-service basis where care is fragmented and spread among many providers. While there has been limited research into the most effective ways of providing health services for this population, states are beginning to recognize that these children generally need a more comprehensive package of services. While many states have implemented Medicaid managed care programs that include special needs children, few have been evaluated.

Defining "Children with Special Health Care Needs": No single answer

Historically, the lack of agreement on a definition has contributed to a large discrepancy in estimates of the prevalence of children with special health care needs. Over 40 definitions for disability exist at the federal level alone. Identification of children who qualify for SSI provides some consistency across states. However, this represents only a small sub-population of all children with chronic conditions (HRSA, 1999).

Common components of definitions for children's disability have included (1) having particular service needs; (2) carrying a diagnosis of a specific physical or mental condition; and (3) exhibiting specified functional deficits (Westbrook, Silver, and Stein, 1998). However, as the mission of the federal and state Title V programs for children with special health care needs has expanded in recent years, the need for a broader definition for planning and advocacy purposes has become apparent. As a result, the federal Maternal and Child Health Bureau Division of Services for Children with Special Health Care Needs developed the following definition:

Children with special health care needs are those who have or are at risk for a chronic physical, developmental, behavioral, or emotional condition and who also require health and related services of a type or amount beyond that required by children generally (Newacheck et al., 1998).

Using this definition, data from the National Health Interview Survey (NHIS) indicate that approximately 18 percent of all U.S. children younger than 18 years old, or 12.6 million children nationally, suffered from a disabling chronic condition in 1994 (Newacheck et al., 1998). In the same time period, approximately 2 percent of all children suffered from a disabling mental health condition (Halfon and Newacheck, 1999). Other estimates have found that as many as 35 percent of children (depending on what is included as a disabling condition) have special health care needs (American Academy of Pediatrics, 1998). In 1997, SSI provided assistance to nearly 900,000 children under the age of 18, most of whom lived below the poverty line (Social Security Online, 1999). This number is dramatically smaller than other estimates as it represents only the subset of CSHCN who meet SSI criteria. To qualify for SSI, a child must have limited family income and resources and have a physical or mental condition which results in severe functional limitations and is expected to last at least 12 months or to result in death.

While there is disagreement about how to identify children with such needs, it has become evident that the costs of meeting the health care needs for this population are high. One study found that approximately 5 percent of children with special health care needs account for 35 percent of health care costs for all children and adolescents (American Academy of Pediatrics, 1998). This underscores a characteristic of most distributions of health care use and expenditures: that a minority of children with complex medical needs contribute to a large portion of health care costs for children with and without special needs.

Health Services for Special Needs Children: Managed Care Advantages and Disadvantages

Managed care offers several potential advantages for this population. Managed care has the ability to offer improved coordination of care, a major advantage given the varied services that may comprise the care of a special needs child. Additionally, managed care may offer a greater focus on prevention services and health care maintenance (Newacheck et al., 1996). Improved access to primary care, with an increased likelihood of establishing a medical home from which a broader array of services can be provided, is another potential advantage of this kind of health care (American Academy of Pediatrics, 1998).

Researchers have also noted several disadvantages of managed care for children with special health care needs. Access to and choice of specialist services may be reduced in managed care settings if primary care physicians act as gatekeepers. Another potential drawback is the disruption of an existing doctor-patient relationship that occurs if a provider is not part of the established managed care network. If care is sought outside of the managed care plan because it is not included in the package of services offered, or is provided by a provider outside of the network, families will see their out-of-pocket expenses increase (Newacheck et al., 1996). Furthermore, some high-risk children may be excluded from managed care initiatives altogether, due to the cost of treating them (American Academy of Pediatrics, 1998).

Approximately 39 percent of privately insured children with special health care needs were enrolled in health maintenance organizations (HMOs) in 1993 (Newacheck et al. 1996). In a survey to assess how well private HMOs were meeting the needs of the special needs child population, Fox et al. surveyed 59 HMO plans. One of the key findings of this study was that HMOs are more likely to protect families from out-of-pocket costs for their child's health care. HMOs offer financial benefits for families whose special needs children require primarily medical interventions (Fox, Wicks, and Newacheck, 1993a). Services often needed by this population, including ancillary therapies, home health care, and mental health services, were more often covered by HMO plans than other types of insurance. One drawback, the study concluded, was that care coordination was not as comprehensive as it needed to be for this population. Even so, the advantages were found to outweigh the disadvantages (Fox, Wicks, Newacheck, 1993a).

C. Medicaid Managed Care and Children with Special Needs: Current Practice

Building more integrated systems of care for children with special needs has become a priority over the past several years. This is exemplified by the increase in the enrollment of children with special needs into Medicaid managed care. While no single model exists, several program designs have evolved across many states that represent variations on the fully-capitated, partially-capitated, and fee-for-service PCCM approaches. There is national debate over how to best serve CSHCN, as neither government policy makers nor managed care organizations fully understand the best ways to address this population. This section reviews details on some states' programmatic responses to managed care for children with special needs. This is not meant to be an exhaustive review; it merely portrays a cross-section of the various programs that can be found for children with special needs across the United States.

1. Overview

As the largest public insurer for children, Medicaid covers many child recipients eligible through SSI. In 1995, more than one million Medicaid-eligible children were covered under SSI, almost a fourfold increase from 1989. The greatest rates in growth were among children with asthma and mental health conditions other than mental retardation, including attention deficit hyperactivity disorder (ADHD) (Perrin et al., 1999). In total, Medicaid covered approximately 1.5 million children with activity limitations in 1994 (Fox et al., 1997).

So far, there has been limited research on Medicaid managed care serving CSHCN. Consistent with more general findings on managed care, research on the Medicaid Competition Demonstrations of the early 1980s suggested that managed care does not automatically result in improved access to care (Newacheck et al., 1994). The same Demonstrations showed that, after enrolling in managed care, the proportion of children with a visit to a specialist declined an average of 53 percent (Newacheck et al., 1994). In a study of Medicaid Demonstration waivers in four states, Fox and her colleagues found that prior authorization policies were structured such that children were often unable to obtain ancillary therapies, mental health services, durable medical equipment, and pediatric subspecialty services (Fox et al., 1997).

Although states enroll special needs children in Medicaid managed care, few include those who are SSI-eligible. In a survey of state Medicaid agencies addressing their policies for enrolling and serving special needs children, 27 states were found to enroll at least some of this population in HMOs. Of these 27, only eight have policies mandating such enrollment. As targets for special treatment, SSI recipients are completely exempt from HMOs (Fox, Wicks, and Newacheck, 1993b).

Another study that examined Medicaid managed care enrollment and benefit policies in 1995-1996 found that CSHCN are more likely than their healthier counterparts to enroll in Medicaid. Among poor children, 70 percent with an activity limitation had Medicaid coverage compared to 59 percent of poor children without a limitation. Among non-poor children, 15 percent with a limitation compared to 6 percent without a limitation were enrolled in Medicaid (Fox et al., 1997). Non-poor children might have Medicaid as a result of the Balanced Budget Act (BBA) of 1997 which guaranteed continuous Medicaid coverage to children for one year following the determination of eligibility, even if eligibility requirements were not continuously met throughout that time period (Deal and Shiono, 1998). Some states are also more generous with their Medicaid benefits and provide them to children who do not meet eligibility requirements but are still considered low income.

2. Case Studies

Hill and his colleagues performed a qualitative evaluation of alternative Medicaid managed care models in eight states (one of which was the program evaluated in this report, and is not included in this brief review). Information was gathered through structured interviews with various key informants, including in-depth phone interviews with the Medical Directors and Title V officials responsible for the CSHCN component of block grants. Site visits were made to each state and focus groups were held with parents of CSHCN in each state. Medicaid managed care contracts and other descriptive materials were also obtained for each study state. The following includes highlighted results of the evaluation (Hill et al., 1999).

Arizona

Begun in October 1982 under a 1115(a) research and Demonstration waiver, Arizona's Medicaid program, the Arizona Health Care Cost Containment System (AHCCCS) was mandated for all Medicaid-eligible and special needs children. Arizona defined children with special needs as those who are eligible for SSI or Children's Rehabilitative Services (CRS) under the state's Title V program. A separate managed care plan, the Arizona Long Term Care System (ALTCS), provides comprehensive services for adults and children with developmental and physical disabilities at risk of institutionalization.

Eligibility, Identification, and Enrollment: Aside from meeting eligibility requirements mentioned above, there is no other screening for the presence of special needs to steer children to an appropriate health plan that meets their needs. Children eligible for ALTCS undergo financial and medical eligibility review resulting in a detailed assessment of their needs. Results of the assessment are used by the program's case managers to develop a care plan.

Primary Care: AHCCCS' capitated health plans are responsible for providing all primary and acute care services by linking new eligibles with a primary care provider at the time of enrollment.

Specialty Medical Care: Specialty services are "carved out" to the CRS and Regional Behavioral Health Authority. These "carve outs" cause considerable confusion among families and providers. Health plan administrators and policy makers have expressed their concern that this policy fragments care and undermines integration as a result of poor information flow between systems of care.

Links to Mental Health, Early Intervention, Special Education, and Other Support Services: "Carve-out" arrangements are in place for mental health, early intervention (EI), and special education. This was found to be problematic for integrating and coordinating care for children, due to a lack of information exchange between the health and behavioral health systems.

Case Management/Care Coordination: The mainstream managed care program does not have a centralized case management system in place for special needs children. Many of the different systems involved in care provide some limited case management. However, in the ALTCS program, every participant is assigned a case manager who is responsible for planning and coordinating all aspects of that participant's care. Liaisons assist the case managers in coordinating the medical care provided through the health plans, while the community-based care is provided by other components of the ALTCS model.

Financing: Although separate capitation rates for Medicaid eligibles enrolled in SSI are maintained for risk adjustment, separate rates for different case mix categories within SSI are not used.

Family Involvement in System Planning/Implementation/Oversight: Arizona has a long track record of involving parents in planning, implementation, and oversight of special needs programs for their children. Parents have the opportunity to become directly involved with effecting system change at the community level and strengthening family/ professional relationships.

Florida

In 1996, Children's Medical Services (CMS) network launched a PCCM program for Medicaid-eligible children with special health care needs. Care is received from a special network of primary care and specialty providers who are credentialed and included in the state's Title V/CMS system. Specialty providers also receive fee-for-service payment for the care they provide. Intensive case management is provided by local area CMS nurses who work to ensure services are coordinated appropriately.

Eligibility, Identification, and Enrollment: To make families aware of the availability of the special plan, and to assess particular needs of children to help match them with providers equipped to address their needs, CMS nurses conduct outreach in area hospitals and with area physicians to publicize the CMS network. Once a child is enrolled, the nurse case managers work closely with families to fully assess their children's needs and link them with appropriate providers.

Primary Care: Enrollees are linked with a primary care provider who is responsible for all preventive and primary care services, as well as for managing referrals for specialty and ancillary care. Reimbursement is on a fee-for-service basis, with the additional payment of a monthly administrative fee for care management. To address the challenge of finding enough primary care physicians to serve children with special needs, the CMS program places great emphasis on recruiting, credentialing, and enrolling physicians with expertise in serving this population. CMS officers reported that recruitment efforts have been successful.

Specialty Medical Care: Specialty care, organized around a well-established network of pediatric specialists and tertiary care centers supported by case managers, was found to be a strong point of Florida's CMS Network. Care is provided in centralized centers and satellite clinics. Primary care providers are encouraged to grant "blanket" time-limited authorizations to ease the administrative burden usually associated with obtaining referrals for specialty care.

Links to Mental Health, Early Intervention, Special Education, and Other Support Services: The CMS program is the lead agency for EI, where services are well integrated with those provided by the CMS Network of dually-eligible children. At the time of program evaluation, Florida was in the process of revising how it handles mental health services. Traditionally, mental health services were delivered as a "carve-out." However, plans are to include them into the CMS program and network.

Case Management/Care Coordination: The case managers are responsible for developing a care plan, making referrals, coordinating and monitoring the receipt of care, arranging transportation, and ensuring that information flows among various system providers as necessary. The nurse case managers are supported by a group of paid resource parents who assist families in accessing needed resources and offer support to families confronted with the challenges of raising a special needs child.

Financing: Aside from fee-for-service payments for care management and other covered services, psychologists and family therapists may bill Medicaid if they are under contract with Community Mental Health Centers.

Family Involvement in System Planning/Implementation/Oversight: Florida's CMS program involves parents in policy and planning through "parent consultants," two parents who are employed to be "at the table" and take part in policy and programmatic decisions. These parent consultants also participate in the development and review of program materials, provide training to advocacy groups on parent issues, and provide peer support to other parents.

Maryland

In January 1997, Maryland launched HealthChoice, a Medicaid managed care system, under an 1115 waiver. HealthChoice is a modified managed care model that requires almost all Medicaid recipients to enroll in one of eight private managed care organizations that receive risk-adjusted capitation payments in return for providing comprehensive services. Individuals with selected chronic, complex medical conditions who are expensive to treat are eligible to enroll in the state's Rare and Expensive Case Management (REM) program which provides services on a fee-for-service basis, along with intensive case management explicitly included to address care fragmentation.

Eligibility, Identification, and Enrollment: HealthChoice does not include any screening for the presence of special needs. However, a brief Health Risk Assessment form is included in the packets sent to newly eligible families. Self-identified high-risk families receive an Initial Health Visit within 15 days of responding to the Health Risk Assessment to confirm an individual's special needs status. Referral to case management occurs as necessary.

Primary Care: Hill and his colleagues report that the use of managed care has succeeded in providing all special needs child enrollees with a medical home.

Specialty Medical Care: Prior to implementing HealthChoice, Maryland stated that it would only contract with managed care organizations that included an extensive network of pediatric sub-specialists and tertiary facilities. This spawned the development of new relationships between health plans and traditional providers at area hospitals and TitleV/Children's Medical Services program.

Links to Mental Health, Early Intervention, Special Education, and Other Support Services: "Carve-out" arrangements are in place for mental health, EI, and special education. The division of responsibility between physical health and behavioral health is rather vague. HealthChoice plans are responsible for primary mental health and are instructed to refer enrollees to the state mental health agency for more serious conditions. There are no guidelines for what constitutes "primary" mental health, giving the health plans a strong financial incentive to refer all mental health services to the behavioral health system to avoid the costs associated with their provision.

Case Management/Care Coordination: Special Needs Coordinators, typically nurses, serve as the plan's point of contact for enrollees with special needs. These coordinators assess, plan, coordinate, monitor, and arrange the delivery of medically necessary and appropriate health-related services. Finally, HealthChoice managed care organizations are required to provide case management services to enrollees who fall into one of seven special population groups, identified by the state as needing extra support, particularly children with special health care needs.

Financing: Maryland has made significant progress in the development of risk-adjusted capitation rates. Children with chronic conditions who have at least six months of claims in the Medicaid program are assigned to one of nine rate cells, based on the number and types of diagnoses reflected in their claims.

Family Involvement in System Planning/Implementation/Oversight: HealthChoice extensively involved families and family advocates in various advisory groups that steered the development of the plan. The state periodically conducts family focus groups as another method of obtaining input and feedback on how well the system is working for them.

Michigan

In 1998, Michigan implemented a capitated program designed specifically for children enrolled in Children's Special Health Care Services (CSHCS), regardless of their eligibility for Medicaid. This voluntary program is available in six counties and gives families the option to join an HMO (known as Special Health Plan) or the existing fee-for-service system. Medicaid-eligible children receive the full package of Medicaid and EPSDT services, while those solely eligible for CSHCS receive specialty services related to their qualifying diagnosis with the addition of well-child care and immunizations.

Eligibility, Identification, and Enrollment: Since 1994, Medicaid eligibles, including SSI recipients, have been required to enroll in some form of managed care. Children enrolled in CSHCS, the state Title V CSHCN program, were exempt from enrolling in capitated plans. In 1998, a separate capitated program designed specially for children enrolled in CSHCS, independent of their Medicaid eligibility, was implemented.

Primary Care: The Special Health Plan provides all physical health care services and a limited amount of mental health care.

Specialty Medical Care: The Special Health Plan uses the child's Individualized Health Care Plan as standing authorization for a given set of services. It can remain in place for up to one year, to eliminate the burden of having to obtain repeat authorizations. Care coordination under the Special Health Plan is provided by a local coordinator that enrollees choose from among agencies under contract with the Plan. Care Coordination is not included within the state's capitation rate plans. Instead, funds are set aside for this purpose and paid directly to local agencies.

Links to Mental Health, Early Intervention, Special Education, and Other Support Services: Information is not available.

Case Management/Care Coordination: A Family-Centered Care Coordinator works to ensure that plan policies and services are responsive to the needs of families with special needs children. All care is delivered according to an Individualized Health Plan developed jointly by the enrollees' Principal Coordinating Doctor and families.

Financing: The Special Health Plan receives capitation payments based on an enrollee's diagnosis, insurance and Medicaid status, and geographic location. The rates are set for 48 cells. Care coordination is not included within the state's capitation rate. Funds are set aside by the CSHCS program explicitly for this purpose and paid directly to local agencies.

Family Involvement in System Planning/Implementation/Oversight: Since 1988, Michigan's Title V program has sponsored a parent participation program which was designed to make systems of care family-friendly. Parents involved in the program are paid professionals with a CSHCN. This parent program was instrumental in the designing of the Special Health Plans.

Minnesota

Minnesota's Medicaid managed care program, the Prepaid Medical Assistance Program (PMAP), exempts from enrollment anyone who is eligible for SSI, along with several other categories of children with special health care needs. The state, therefore, served as the study's "control" group. These children are served through the traditional fee-for-service Medicaid program, which does not guarantee access to primary care or specialty providers, does not offer case management for special needs children, and does not monitor the amount or quality of care they receive. Other Medicaid-eligible children with chronic conditions or disabilities who do not meet the eligibility requirements for SSI or Tax Equity and Fiscal Responsibility Act (TEFRA) are enrolled in PMAP, although there are no special provisions for their special needs.

Eligibility, Identification, and Enrollment: Of all the state's evaluated, Minnesota was found to have the least capacity to identify CHSCN and assist them with finding an appropriate provider. There is no system in place to identify special needs children. Medicaid eligibles may choose among any available provider who will accept Medicaid. The SSI population does not receive any special assistance in locating providers who are skilled in serving their special needs.

Primary Care: Whereas in the other states evaluated in this study, safeguards were in place to provide special needs children with a primary care medical home, Minnesota does not have any method for achieving this objective for the fee-for-service enrollees. Also, special needs children who are exempt from enrollment into managed care are provided no guarantee of a primary care medical home or any assurance they will receive EPSDT screens.

Specialty Medical Care: Specialty care is not addressed well in PMAP. Because there is no centralized entity to hold accountable (such as a managed care organization (MCO)), state officials have limited ability to exert control over variations in practice among the state's thousands of pediatric providers, and no system to enforce EPSDT guidelines. A further barrier is the requirement of families to obtain prior authorizations for the receipt of physical, occupational, and speech therapy exceeding certain limits.

Links to Mental Health, Early Intervention, Special Education, and Other Support Services: Minnesota's fee-for-service system lacks any formal mechanisms for integrating service delivery across its separate health, mental health, EI, and special education programs.

Case Management/Care Coordination: There is no explicit case management benefit for special needs children or those participating in the Medicaid state plan. The result is that families are left unsupported and on their own to negotiate how to access and coordinate services among the different programs.

Financing: The Medicaid managed care program covers both medically and educationally necessary services. EI should be covered under this. However, various case managers do not always cooperate, leaving the family to assure that plans reimburse for these services.

Family Involvement in System Planning/Implementation/Oversight: Information is not available.

Oregon

The Oregon Health Plan (OHP) is a Medicaid managed care program implemented in 1994 under a section 1115 waiver. Phase I of the program enrolled mothers and children eligible for Medicaid and receiving Aid to Families with Dependent Children (AFDC), providing a limited benefit package based on a "Prioritized List" of covered services and treatment chosen by a multi-disciplinary panel. Phase II of the program was implemented in January 1995, and called for mandatory enrollment of the SSI special needs population.

Eligibility, Identification, and Enrollment: There is no screening for the presence of special needs in OHP. If SSI recipients indicate any problems or disruptions that might result from their enrollment into managed care, social services case workers are instructed to counsel them more closely on the selection of an appropriate plan. There is also an "opt-out" provision for special needs children who wish to maintain an ongoing relationship with a provider who is not part of any of the network's health plans.

Primary Care: The program represents a model for which all Medicaid populations, including children with special needs, are enrolled into mainstream MCOs that must meet all of their acute and ambulatory health care needs.

Specialty Medical Care: According to families participating in the plan, access to specialty care such as habilitative and rehabilitative therapies, durable medical equipment, and supplies and other support services, was not easily obtained.

Links to Mental Health, Early Intervention, Special Education, and Other Support Services: A "clean" carve-out results in health plans being excused of all responsibility for addressing the behavioral health needs of enrollees. The result is that Medicaid recipients can access mental health services without a referral from their primary care provider. A problem with this practice is the lack of information channels between systems. There is boundary confusion that exists over how to handle crisis and emergency care, laboratory tests, and prescribing medications for common childhood behavioral problems. Doctor-patient confidentiality also interferes with the sharing of information among all systems.

Case Management/Care Coordination: An Exceptional Needs Care Coordinator is required to support individuals with particular needs in service coordination. These coordinators serve as advocates. However, they have limited reach and are more reactive than proactive in advocating for enrollees.

Financing: Although separate capitation rates are set for Medicaid eligibles who are enrolled in SSI, there is no case mix-based rate setting within SSI.

Family Involvement in System Planning/Implementation/Oversight: Information is not available.

Tennessee

Implemented in January 1994, under a Section 1115(a) waiver, Tennessee's Medicaid managed care program, TennCare, is a mainstream, full capitation model that requires all beneficiaries, including SSI-eligible and other special needs children, to enroll in managed care.

Eligibility, Identification, and Enrollment: TennCare does not have an explicit screening process in place to determine the special needs of children and to be able to appropriately steer them to the correct care. Newly-eligible families are simply given information packets and some hands-on assistance in choosing a health plan for their child. Once in an HMO, access to specialty services such as rehabilitative therapies, durable medical equipment, and other support services, is often reported to be limited.

Primary Care: Medical services are provided under contract with nine HMOs for a fixed capitated amount.

Specialty Medical Care: HMOs contract with network providers, including Federally Qualified Health Centers (FQHCs), to offer acute and specialty care. In addition, TennCare provides behavioral health services to Medicaid-eligible children through a service "carve-out" with TennCare Partners who offer a comprehensive package including mental health services and substance abuse treatment.

Links to Mental Health, Early Intervention, Special Education, and Other Support Services: TennCare has limited information exchange between the physical health and behavioral health systems. This is extremely problematic for integrating and coordinating care for children.

Case Management/Care Coordination: There is no centralized case management system for children with special health care needs. Instead, many different systems that comprise a child's total care are involved in providing some form of case management.

Financing: Although separate capitation rates are maintained for Medicaid eligibles who are enrolled in SSI, separate rates are not set within that category. However, Tennessee's educational program is an example of a successful pilot program to assure appropriate Medicaid payment for medically necessary special education services and integration of these services with those provided through managed care plans. The program supports care coordinators in local health departments who gain access to TennCare's information systems, and bill plans according to the information obtained about the medical necessity of services. This model helps to support coordination between special education and medical care.

Family Involvement in System Planning/Implementation/Oversight: Information is not available.

3. Discussion

Few states have implemented comprehensive, specially-designed programs for managing the care of special needs children. Most state programs reported in this review of case studies enroll special needs children in main-line managed care organizations and carve out at least some benefits, particularly behavioral health services. Focused care management is not always provided.

Michigan is an exception, and is most similar to the DC Medicaid experiment in the sense that CSHCS is a separate, capitated, voluntary program for special needs children. Fee-for-service alternatives include Minnesota and Florida's CMS, a PCCM program. Aside from the process assessments reported by Hill and his colleagues, no studies have been completed that evaluate the impacts of these programs on access, health outcomes, utilization and expenditures of children with special needs.

D. Evaluation Design

This evaluation estimates impacts of the Demonstration through concurrent comparisons between participants and non-participants. Because Demonstration enrollment has been voluntary, the evaluator conducted a separate analysis of key differences between participants and non-participants that could provide evidence on biased selection in enrollment. However, because the evaluator lacked access to data on the baseline values of most of the key measures of outcome, it was impossible to correct explicitly for selection effects in analyzing impacts.

In the absence of an experimental Demonstration design, it was necessary to define a comparison group for estimating impacts. Because the DC Demonstration was (and still is) a unique program, and because the District itself is different in many ways from other urban areas in the United States, the option of sampling comparison group members from outside the District was rejected. Children living in DC and eligible for SSI but not enrolled were determined to be the appropriate comparison. This decision assures the relative homogeneity of the total participant/non-participant sample. However, it limits the generality of findings from the evaluation. Also, choosing an internal comparison group in a voluntary Demonstration that enrolls a large fraction of the eligible population virtually guarantees measurable and unmeasurable differences between participants and non-participants due to selection bias.

In order to assess biases created by the enrollment process, participating children, identified through Medicaid and HSCSN enrollment files, were assigned to fixed time periods, beginning with the start of the Demonstration in February 1996, and ending with the last date for which administrative data were available (June 1998). Eligible non-participating children were then sampled within each of these time periods, in numbers equal to the sampled participants. Demographic indicators and measures of acuity/utilization were used to estimate the probability that an eligible child would enroll, as well as the probability that a participant would disenroll. They were also used, together with Medicaid and HSCSN claims data, to analyze some utilization impacts of the Demonstration.

For most of the measures of interest, particularly those related to access, unmet needs, satisfaction, physician, mental and behavioral social impairments, caregiver burden and certain kinds of utilization, data could only be collected through a survey. Therefore, a separate sample of participants and non-participants was drawn and surveyed. Most respondents to the survey were caregivers of the sampled children. Because sampling information was made available only after many months of Demonstration operation had passed, it was impossible to determine the values of health and behavioral measures for these children at some "baseline" period (in particular, before enrollment for participants). Without baseline data, the analysis of impacts could only be conducted through concurrent comparison of key measures between participants and non-participants, a clearly limited test of Demonstration impact. Without knowing, for example, how impaired children were at a point just before enrollment (and at a comparable chronological time for non-participants), it is difficult to determine how much of the difference in impairment after enrollment is due to the Demonstration and how much reflects underlying differences at baseline.²

Because of weaknesses in the evaluation design related to availability and quality of key data and to the non-experimental design of the Demonstration, it was considered crucial to collect qualitative information on the operation of the Demonstration, on perceptions of key stakeholders (families, care managers and primary care providers), as well as documentary evidence of care managed and provider from Demonstration case records. These qualitative data complement and amplify findings from the enrollment and impact analyses.

Throughout this report, the evaluator has attempted to be very conservative in attributing "impacts" to the Demonstration. It would be overstating the case to say that the Demonstration cannot be evaluated, given the data and design problems noted above and in subsequent chapters. However, for some indicators (for example, measures of physical, mental and behavioral impairments), the absence of baseline measures and of supporting information from other analyses (for example, focus groups) limits the analysis to a report of the overall frequency of impairments in the sample and separately among participants and non-participants. For other measures (for example, satisfaction with access to care) the evaluator has available consistent and significant findings from the impact analyses (that participants are generally more satisfied), combined with information from targeted interviews and focus groups, that identify specific positive perceptions of the care management process in the Demonstration. For these measures, the evaluator feels more confident in asserting that the Demonstration had an impact.

E. Organization of the Report

The next chapter summarizes the organizational and financial activities of HSCSN from the planning stage, and reviews in more detail activities and issues that were prominent within the most recent year. Findings in this chapter reflect targeted interviews with HSCSN and DC Medicaid officials on site and by phone, and reviews of program documents. Chapter 3 tracks enrollments in and disenrollments from the Demonstration, and explores changes in the characteristics of enrollees. This chapter also analyzes selection bias by estimating the influence of demographic and other factors on enrollment and disenrollment. Chapter 4 portrays the care management process under HSCSN through the responses of families, care managers and providers in phone interviews and focus groups. Chapter 4 also reports findings from a review of 100 case records that assesses various dimensions of the care management process. Chapter 5 reports estimates of the impacts of the Demonstration, using survey data (to assess impacts on access, satisfaction, utilization of formal and informal home care, caregiver burden, high-tech utilization and impairments) and Medicaid and HSCSN claims data (to assess impacts on additional measures of utilization). Chapter 6 concludes with a brief discussion of lessons learned and implications for policy.

II. DESIGN AND EVOLUTION OF THE DEMONSTRATION

This chapter provides an overview of the history of the Demonstration from its earliest origins through 1999. The early parts of this history have been presented in much greater detail in earlier reports of this evaluation³ and are briefly summarized here. More recent information and events (concerning the events of 1999, when the extended Demonstration was drawing to an end and plans for the recompetition were being finalized) are discussed at greater length. By combining earlier and more recent information, this discussion will provide an integrated history of the findings of the implementation evaluation.

The discussion is divided into seven sections:

Origins of the Demonstration
HSCSN startup and implementation
HSCSN finances
Oversight of the Demonstration by CHCF/MAA
Relationship of the Demonstration to the community and other service systems
Extensions of the Demonstration and the Child and Adolescent SSI Program (CASSIP) procurement
Summary and implications

To help structure this discussion, a timeline of significant events across the entire period from 1993-2000 is presented in Exhibit 2.1. Protocols for the case study on which findings in this chapter are based may be found in Appendix A.

A. Origins of the Demonstration: 1993-1996

This Demonstration grew out of a 1994 proposal from the Hospital for Sick Children in DC (HSC, a primary site of care in the city for children with chronic needs and for poor, unwed mothers) to the Commission on Health Care Finance (CHCF, later renamed the Medical Assistance Administration), lead agency for Medicaid in the District. This was an unusual time in the District and an unusual context for the Demonstration. The Demonstration proposal moved forward:

In a Medicaid program with severe cost problems. The DC Medicaid program was facing severe cost problems in the early 1990s. It was the fastest growing health program in the city facing a major fiscal crisis.
For a target population thought to be poorly served. In the words of one District Medicaid official interviewed by project investigators, the District had "awful" statistics on preventive, chronic, and long-term care. The organization and financing of District human services was thought to be partly responsible. A proposal to integrate services for children with special needs offered help in an area that the District recognized was in need of help.
By agencies that did not have close working relationships before the Demonstration. The key players in putting the Demonstration together -- principally, CHCF, the Hospital for Sick Children, and the Health Care Financing Administration's Office of Research and Demonstrations (HCFA/ORD) -- had not worked closely together prior to this initiative.
Just as major changes in the surrounding context were about to occur. As waiver negotiations began in earnest, DC finances worsened. The federal price for resolving this continuing crisis was a substantial reduction in home rule powers. The increasing powers of the outside Control Board imposed by Congress threatened to change the politics and administration of the District government in large and small ways. Meanwhile, somewhat independent of that upheaval, the Medicaid program itself was about to change: a relatively small, optional, managed care program for mainstream Medicaid enrollees was about to become mandatory for all mainstream enrollees in 1997. While children with special needs were carved out of the mandate, the commitment of the Medicaid program to managed care had spread substantially from the early days of planning for the Demonstration.

EXHIBIT 2.1. Demonstration Milestones: Planning and Implementation (Calendar Year 1993-2000)
Year	Month	Milestone
1993	May	Hospital for Sick Children (HSC) hosts conference where the idea of a managed care system for disabled children is raised.
1993	June	Initial HSC review is favorable.
1994	January	Results of feasibility study reported to HSC board members.
	February	Health Services for Children with Special Needs (HSCSN) is incorporated.
	March	Commission on Health Care Finance (CHCF) submits an 1115 waiver application to the Health Care Financing Administration (HCFA).
	March	HSCSN applies for not-for-profit status (received in May).
	April	District receives one-year Development Grant from HCFA.
	August	[Startup date originally forecast by HSCSN.]
1995	July	CHCF public forum for families of children eligible for proposed Demonstration.
	October	Waiver approved.
	December	Three-year Demonstration formally begins December 1. Mailing of notification package to first group of beneficiaries delayed until mid-December to permit required HCFA review of materials and resolution of outstanding operational issues.
1996	February	First covered services for enrollees.
1997	May	Joint Commission on Accreditation of Healthcare Organizations conducts on-site survey. HSCSN granted 3-year Accreditation.
	October	Davis replaces Tinto as project director; Holt replaces Davies in December.
	November	Startup for new care management information system. End of second year of the Demonstration. 1,901 enrolled members (11/1/97).
1998	March	HSC reorganization takes effect, major change in HSCSN senior management.
	July	MAA request to HCFA for one-year extension of waiver.
	August	Roadmap submitted to HCFA regarding the RFP.
	November	Extension of waiver granted from HCFA. End of original 3-year waiver.
1999	May	Child and Adolescent SSI Program: Concept paper issued for comment.
	August	Draft RFP issued for comment.
	October	Paul Offner resigns.
	December	HSCSN begins services under second extension, with cost reimbursement for medical care.
2000	February	Earliest likely date for release of RFP.
	April	Proposals due (assuming February release of RFP).
	November	Contracts awarded.
	December	Start of CASSIP services.
Sources: Miscellaneous documents and Case Study Interviews, 1997-1999.

Most Demonstrations move forward under at least some circumstances that are unusual or atypical. What is striking in the District is how fundamental some of these circumstances were, going to the basic structure of the city's finances and organization.

The proposal for the Demonstration grew out of a chance meeting organized by HSC. In 1993, HSC hosted a national meeting of administrators of children's hospitals. The meeting's agenda included time for parents of children with special health care needs to present their first-person experiences with the current health care system. The presentations were surprising to all in attendance, as the hospital stays (often for extremely lengthy and complicated procedures) were "glossed over in a second," according to one HSC participant. To those who had to use the system, the real problems of coordinating care occurred once the patient was discharged, involving such matters as transportation, lining up multiple specialist visits in the same day, and seeing different physicians on each visit so that no doctor was familiar with their cases.

After each discussant unexpectedly emphasized this sense of the problem, the idea arose that managed care -- viewed as the active coordination of care, not limitation of care -- could really work with this population and save money in the process. HSC developed the idea, as the hospital had resources to study the idea's feasibility and also had a politically prominent board to push it with the District, the Internal Revenue Service, and others.

HSC called the Commissioner of CHCF (head of the District Medicaid program) and broached the Demonstration idea to him (circa May-June 1993). District Medicaid at this time was looking for ways to capitate streams of expenditure to control costs and expressed interest in pursuing the ideas HSC presented. By early 1994, HSC had completed a detailed, generally positive feasibility study. CHCF agreed to proceed with the concept. In February, HSC incorporated Health Services for Children with Special Needs, Inc. (HSCSN). It began to hire new staff and rented a fairly large space on Connecticut Avenue NW -- expensive space which meant that the HSCSN venture was now a major financial wager.

In March 1994, CHCF submitted a formal waiver application to HCFA. This application described a model based on a number of key decisions:

A form of privatized administration, using the incentives of capitated funding. The waiver application envisioned that -- with the incentives of a fixed capitation rate -- HSCSN would take the fragmented care delivered in the fee-for-service system and integrate and manage the care actively, across the full spectrum of covered services and in a fashion sensitive to the particular needs of each child. Medicaid and other services (e.g., public health, schools, foster care, and mental health) were not viewed as notably deficient in what they formally covered, although Medicaid coverage was not easily available for: (a) a few particularly helpful items (e.g., limited use telephones) for arranging and gaining access to needed care, nor (b) certain other items that had particular leverage on the circumstances or convenience of care (e.g., respite care or home modifications that would allow a child to receive care at home, rather than in an institution). For items like these, a capitated provider was expected to make decisions based on flexible common sense, rather than the more rigid formalities of Medicaid authorization.
Integration with the broader set of agencies and programs serving the special needs population. Given the general commitment to integration of services, there was also an assumption that the Demonstration would be coordinated with other programs and providers at the boundaries of HSCSN's responsibilities -- notably, the schools, the mental health system, and the foster care system -- to provide a new measure of integration of those services with the Demonstration services. Two signs that this hope might be difficult to realize came when: (a) the services provided through the District schools were carved out of the capitation rate, and (b) the District school system failed to cooperate in later coordination initiatives. But the premise of the Demonstration was to improve such linkages, and the proposal to HCFA assumed that such improvements would be pursued.
A somewhat accommodating capitation rate, with risk corridors. The proposal asked for 97.5 percent of the estimated average composite fee-for-service cost, rather than 95 percent or 90 percent, in order "to give this a chance to succeed," according to one of the outside consultants to CHCF. The whole point was not to attempt to garner the largest possible savings right at the start of the Demonstration. HSCSN's major concern was the risk of possibly catastrophic losses. In discussions with the District, HSCSN sought to take the "win big, lose big" element out of the model. This effort led to the creation of risk corridors on both sides of the proposed capitation rate. The actual rate set for the first year of the Demonstration -- a partial year from December 1, 1995, through September 30, 1996 -- was $997.71. This rate was then to be trended for inflation and the "prior year experience" (after much discussion with HCFA, a 4.6 percent inflation rate was selected).
Single-source contract and mandatory enrollment. This provision meant that all special needs beneficiaries were to be automatically assigned to HSCSN. The key reason to use mandatory enrollment was to minimize risks of selection bias. These risks were not trivial: an HSCSN consultant estimated that 5 percent of the beneficiaries would generate roughly two-thirds of the costs, so it was a big issue as to where that 5 percent of the population ended up -- in one health plan or another, in managed care or fee-for-service. There were thus compelling financial reasons to avoid the possible effects of freer enrollment choice by beneficiaries. However, mandatory enrollment in a single-source Demonstration also meant that there would be substantial limits on beneficiaries' choice of providers -- indeed, disaffected beneficiaries could not even change managed care plans, much less avoid the managed care features of the Demonstration, in the model that the District first proposed to HCFA.

In the course of negotiations between HCFA and the District, the mandatory enrollment model later changed to a default model (beneficiaries who failed to exercise a choice for fee-for-service would be assigned by default to HSCSN), then to a voluntary model (enrollees had to choose HSCSN). This evolution to voluntary enrollment transformed the Demonstration model into a marketing model. This change caused concern among HSCSN and its consultants, but the contractor went along with the change.

CHCF and HSCSN apparently expected approval of the waiver to take only a few months. In fact, the negotiation process took 19 months from the date of submission to the date the waiver was approved (October 13, 1995), and there were important changes (notably, in the enrollment model, as described above) after waiver approval. The protracted delay made it difficult for HSCSN to continue to get funding from its parent. Delays also led HSCSN to furlough people, to hold emergency meetings of its board, and to take other measures that were (in HSCSN's view) extraordinary. The intensity of these battles likely would have been sufficient to derail the Demonstration, if HSCSN had not had such extensive commitments to it at this point.

The progress of waiver discussions with HCFA was complicated by the lobbying activities of an emerging group of project opponents known as "Responsible Parents Against DC Medicaid." The group consisted of day treatment centers and other service organizations that provided care to the District's Medicaid population. Opponents focused on six aspects of the proposed waiver: the lack of specific contractual plans given to providers (who were being threatened by managed care for this population for the first time), the alleged misrepresentation of some providers as supporting the waiver, the alleged absence of any demonstrated need for the Demonstration, the use of a sole-source contract ("another DC deal," meaning corruption), the use of mandatory enrollment for so vulnerable a population (thus resulting in parents' loss of "federal guarantees of freedom of choice"), and the lack of community involvement in the design of the Demonstration.

Some of this opposition was doubtless due to the absence of any active effort to involve community or other stakeholder groups in the development of the Demonstration. Some of it was also doubtless due to the self-interest of providers who considered managed care a threat to the arrangements they enjoyed under fee-for-service. But the political opposition may have had an effect on HCFA and within the District. For example, the DC Control Board, which had originally said it would not look at this contract, changed its mind after the waiver was granted, apparently because of the continuing claims of the opposition. Program opponents believe that their efforts played a major role in HCFA's refusal to approve mandatory enrollment, and also led to the increased involvement of the community later in the waiver process. The former conclusion is likely wrong, since HCFA found conflicts between mandatory enrollment and the Americans with Disabilities Act when the District's application was first received. In any event, as an HSCSN manager noted in an interview, the change in the enrollment process that the opposition sought eventually served to provide political cover to the Demonstration -- especially in discussions with the Control Board -- since beneficiaries had to exercise an affirmative choice to join the Demonstration. One leader of the opposition agreed that opposition efforts waned after enrollment became voluntary. In his words, "The rest was about egos."

Once the Demonstration was up and running, there was much less resistance to it from the former opposition. In fact, all of the day treatment centers that were at the center of opposition efforts became providers in HSCSN's network (the last center, PSI, joined in late 1997, the end of the second year of the Demonstration).

In implementing the voluntary system ultimately approved, there were a number of enrollment, marketing, and other operational issues on which HCFA and the District disagreed. After much discussion, compromises were reached. On December 15, 1995, in one of its last acts just before a federal furlough of employees, HCFA's Office of Research and Demonstrations issued a letter approving the startup. At that point, the first group of enrollment packages was mailed to beneficiaries.

B. HSCSN: Startup and Implementation: 1996-1999

HSCSN had hired managers and staff in anticipation of an earlier start date, so the organization was prepared to begin when the go-ahead was received. HSCSN's initial organizational structure was relatively simple. There were six vice-presidents over functional and support units of the organization: strategic planning, program development, provider affairs, finance and administration, information systems, and care management. Each vice-president reported directly to the CEO of HSCSN. In addition, the position of Medical Director reported to the CEO and oversaw clinical issues, as well as quality control and monitoring. Certain units had important dual reporting responsibilities; perhaps most important, care management teams under the Vice-President for Care Management also reported to the Medical Director for oversight of clinical issues in the course of care management business.

To set up HSCSN, five core administrative systems were established:

Internal network (email, etc.),
Payment systems,
Credentialing systems,
Quality management systems,
Care management systems.

According to a senior HSCSN manager, the first four of these systems were relatively easy. The fifth system, which HSCSN considered critical to its role as an integrator, was the most difficult system to implement, as HSCSN chose to be a major systems integrator, rather than to purchase a turn-key system off-the-shelf.

At the time the waiver was awarded, HSCSN found only a few commercial packages available for care management. HSCSN staff were looking for a package that would store planning information and information on contacts between the Care Manager and client. They also wanted a system that could outline schedules from past medical and contact history and track the course of approved care against protocols embedded in the system.

As an interim solution to support care management, systems staff initially chose to rely on an adaptation of Lotus Notes. But Lotus Notes had important limitations. The long-term goal of HSCSN's managers was to have a care management package that would store contact information as well as generate prospective treatment plans for the clients. HSCSN had a hard time finding a developer to meet its needs. Internal efforts to develop such a system were largely unsuccessful before a new Vice-President for Systems arrived in mid-1995. During an industry conference in April 1996, HSCSN was introduced to Managed Care Systems (MCS). MCS had experience building packages for managed care organizations in Visual Basic. HSCSN decided shortly thereafter to work with MCS to develop a care management system tailored to HSCSN's needs. This effort would combine MCS' computer expertise and HSCSN's "domain expertise" to develop a customized package.

This ambitious goal became an expensive financial commitment. HSCSN invested heavily in the development of care management software: by the fall of 1997, the software system had cost $2 million to develop, out of almost $5 million in total startup costs. The system was slated to go live in November 1997, after many delays. As discussed below, the comprehensive system was never actually implemented. The important point here is that this system reflected a particular vision of what HSCSN should be doing -- developing a sophisticated, computer-based case management capability, to make HSCSN a cutting-edge enterprise that could market its capabilities far beyond the District of Columbia. That is how HSCSN implemented the Demonstration and ran its operations the first two years of covered services (1996-1997).

From its founding in 1994 through early 1997, senior management at HSCSN was relatively stable. However, HSCSN appeared to be changing directions in 1997. It lost four senior managers in positions which were vacant by the last quarter of 1997: strategic planning, program development, provider affairs, and care management. In the case of care management, the position had previously turned over less than a year before. Meanwhile, the position of Chief Operating Officer was created at about this time and filled in November 1997 by a former managed care executive.

This strategic move reflected a long-standing commitment by HSCSN senior management. Certainly, the organization was committed to deliver managed care benefits to special needs children as an outgrowth of the historic mission of the HSC. At the same time, however, HSCSN also appeared to view this Demonstration as a way to do product development for an HSC entity that was somewhat behind the times: i.e., to develop and field test a managed care infrastructure to fit the new competitive environment, rather than an improved method for care coordination and cost-effective care as such. Many of HSCSN's activities are more easily explained from this strategic view. For example:

HSCSN made substantial investments in project infrastructure, especially for systems, as noted above. These investments are difficult to explain if the demonstration was an end in itself. HSCSN seemed to be developing and testing technologies of care management, rather than simply following the least-cost path to serving the Demonstration population. If HSCSN's objectives were limited to making this Demonstration work, it might well have invested more in personnel and less in untried systems, especially since the key care management information system would not be available until the last year of the Demonstration.
Early in the Demonstration, HSCSN set up a for-profit subsidiary to market the managed care capabilities HSCSN developed.

A provider would be foolish to invest scarce people and resources in something like this Demonstration without some larger strategic objectives in mind. HSCSN appeared to be looking at this Demonstration as a platform to become something else.

The staffing changes of late 1997 proved to be transitional, as 1998 brought fundamental changes to HSCSN. First, there was a major, formal reorganization of HSC's corporate structure. The old organization maintained HSCSN as a separate entity, with a for-profit subsidiary (SNCS Inc.) for marketing the capabilities that HSCSN developed. Some of this separateness was done deliberately at the outset, to counter the charges of program opponents and the concerns of HCFA that an integrated HSC/HSCSN would have a conflict of interest (e.g., would preferentially refer patients to HSC). But in February 1998, HSCSN was merged into a much simplified structure and the for-profit subsidiary was dissolved. This organizational change was designed to create one board and one integrated organization. Before the reorganization, staff of the hospital and HSCSN in the same functional areas would not talk to each other to mesh their separate operations. This reached to the top: the CEOs of the hospital and HSCSN never really coordinated their efforts, thus making it impossible for the two entities to conceive of being (much less acting like) a system. After the reorganization, greater integration was possible, and HCFA and previous opponents (at this more stable stage of the Demonstration) did not press any self-dealing concerns, after some basic reassurances from HSC.

A few months later, there was a major change in senior managers at HSC and HSCSN. By July 1998, HSCSN's chief executive officer since the very earliest days of the program and the CEO of the hospital were "reassigned," a preliminary to their departures. The Chief Operating Officer of HSCSN -- whose arrival coincided with a change of direction in the fall of 1997, noted earlier -- also departed. The CEO of the HSC Foundation became the chief executive of HSCSN as well as the hospital and related enterprises. A new Vice-President of Finance and Managed Care Operations took over supervisory responsibility for systems and finance, with HSCSN's senior managers in those areas remaining. A new Vice-President for Medical and Patient Services replaced HSC's existing Medical Director and took over direct supervisory authority over HSCSN's Medical Director, who remained.⁴

New senior managers at HSC wanted to change the philosophy of HSC/HSCSN, from a kind of social service philosophy to a philosophy that focused on conserving resources, even down to the care manager level. Their concern was stated to us in terms of survival: the other 50 states were not doing this kind of special needs HMO, so HSC could not simply count on something like the Demonstration to be available for the indefinite future. HSC appears to have viewed the major change of direction in 1998 as a necessity for adapting the HSC system to changes in the health marketplace. Those changes had come relatively slowly to DC, but the new managers at HSC claimed the need to adapt had arrived. Many outside observers saw these changes as a means to get HSC ready to be acquired by a larger entity.

Beneath the dramatic changes in organization and staffing outlined above, the basic operations of HSCSN did not change immediately or dramatically. But there were some changes. First, administrative support services that had been separate began to be integrated. Second, certain expensive non-medical services used by the hospital and HSCSN (notably, transportation) began to be merged. Third, the marketing activities of the former SNCS subsidiary were terminated. This decision reflected a view that HSC's efforts were to be focused on the development of the system's own capabilities, not on providing consulting services to others. Fourth, HSCSN's major systems development venture for care management software was cut back. As noted earlier, HSCSN invested heavily in the development of this care management software. But by 1998, there had been repeated delays, and the system, as implemented, was not nearly the system originally envisioned. This system reflected a particular vision of what HSCSN should be doing: developing a sophisticated, computer-based case management capability, to make HSCSN a cutting-edge enterprise that could export its capabilities. That vision was replaced in 1998 by a more conservative financial approach that reflected less willingness to take speculative risks. HSCSN sought to cut its losses on this investment. HSC's senior managers were blunt; in interviews, they portrayed this development effort as "a bad idea." New managers, less wedded to the system than the original CEO at HSCSN, were quick to pull the plug on it.

Thus, by the end of 1998, as the original three-year waiver drew to a close, HSCSN had finally resolved the ambivalence apparent throughout the Demonstration, opting decisively for a limited operation without any illusions of exporting the capability to other states.

C. HSCSN Finances, 1996-1999

In this section, we review three financial issues raised by the Demonstration: startup costs, capitation rate and risk sharing, and operating results to date.

1. Startup Costs

Actual startup costs were $4,991,484,⁵ a figure many times the original estimates. Some of the overrun was attributable to the protracted development and negotiation period after the waiver application was submitted -- actual startup was almost 16 months later than original projections assumed. But a large part of HSCSN's cost overrun on startup was due to startup activities that were not taken into account in the original estimate. For example, original estimates assumed that all systems would be acquired off-the-shelf, with minimal adaptation expense (e.g., perhaps $0.1 million). As it turned out, development expenses for the care management system alone were approximately $2.0 million, as noted earlier. Meanwhile, from the start, HSCSN has been in premium office space that a stripped-down implementation would not have used. The model as implemented was a substantial commercial venture, not a bare-bones or temporary venture. Startup costs are where the effects of HSCSN's original strategy are most clearly reflected in the financial results.

2. Capitation Rate and Risk Sharing

The capitation rate for this project was based on estimates made from historical claims records. Initially, HSCSN proposed capitation rates in the range of $1,200 to $1,300 per member per month (PMPM).⁶ But after the District carved out certain high-cost items (such as transplants and District public school services), the rate came down below $1,000. As noted earlier, this rate was then to be trended for inflation and the "prior year experience," and a 4.6 percent inflation rate was accepted. Capitation rates for fiscal years over the 1996-1999 period are detailed in Exhibit 2.2 below, along with how they were derived.

EXHIBIT 2.2. Calculation of the Capitation Rate, 1995-2000
Component of Calculation	BasePeriod7/1/94-12/31/94	Year 111/01/95-09/30/96	Year 210/01/96-09/30/97	Year 310/01/97-09/30/98	Extension10/01/98-09/30/99	Extension 210/01/99-09/30/00
Paid Medical Claims	$15,297,036
Average Completion Factor	91.37773%
Total Incurred Claims	$16,740,442
Member Months (Actual)	17,534
Base Period Fee-for-Service PMPM	$954.75	$954.75	$954.75	$954.75	$954.75	$954.75
Trending Factors @ 4.6%	1.00000	1.07179	1.12110	1.17267	1.22661	1.28303
Trended Fee-for-Service PMPM		$1,023.29	$1,070.37	$1,119.61	$1,171.11	$1,224.98
Capitated Rate Adjustment		97.5%	97.5%	95.0%	95.0%	95.0%
PMPM Capitated Rate		$997.71	$1,043.61	$1,063.63	$1,112.55	$1,163.73
Source: Letter to Thomas Chapman, CEO, HSCSN, from Herbert H. Weldon, Jr., Deputy Director MAA, dated December 1, 1999, Table A.Note: Rate periods generally coincide with the DC fiscal year.

The initial rate was $997.71. The rates for subsequent years were as shown in the exhibit. The extension periods represent the two periods of extension after the originally scheduled conclusion of the waiver. As more fully explained in Section II.F, the second extension was a cost-reimbursement arrangement. Current proposals call for a continuation of capitated payment with stop-loss protection, under a new waiver. The capitated payment rate for the second extension was the basis for interim payments, pending year-end reconciliation of medical costs.

This basic capitated reimbursement was adjusted annually according to the provisions of a risk sharing agreement between HSCSN and the District. This agreement set up an 85 percent medical claims ratio as the target for the Demonstration, where medical claims ratio was defined as the ratio of: a) actual claims expense for covered services (computed on a cash basis -- i.e., ignoring such expense items as depreciation and startup costs), to b) the total capitation payments for each year of the Demonstration. The District shares varying levels of risk in corridors around the 85 percent medical claims ratio target, as shown on Exhibit 2.3. Actual or estimated risk sharing payments for each year of the Demonstration are shown near the bottom of Exhibit 2.4. In both of the first two program years, HSCSN made small risk-sharing payments to the District, although the contractor actually had an excess of expenses over revenues of $0.7 million in one of those years. For 1998, the firm lost over $0.2 million on a cash basis, and risk sharing was expected to cut those losses roughly in half. In 1999, HSCSN expected to lose over $1.1 million on a cash basis, and again the estimated risk sharing payment will cut that loss roughly in half.

EXHIBIT 2.3. Risk Sharing Corridors Between the District and HSCSN for Different Values of the Medical Claims Ratio
Actual Claims Expense as a Percent of Capitation Revenues							Percent of Risk Assumed
Lower Corridor Boundaries			Target	Upper Corridor Boundaries			District	HSCSN
						>95%	85%	15%
					95%		50%	50%
				90%			15%	85%
			85%				0%	0%
		80%					15%	85%
	75%						50%	50%
75%							85%	15%
Source: Agreement between DC (CHCF), and HSCSN dated November 30, 1995, Article 7.Note: Medical Claims ratio = Actual claims expenses divided by total capitation payments.

EXHIBIT 2.4. HSCSN Revenues and Expenses, 1996-1999, and Comparison of 1999 vs. 1998
	1996(Audited)	1997(Audited)	1998(Unaudited)	Estimated ^a1999(Unaudited)	Est. 1999Over/(Under)1998
REVENUES (before risk sharing):
Per Capita Revenue	$8,049,994	$21,363,210	$25,154,850	$25,881,541	$726,692
Other Revenues	$57,538	$525,354	$38,967	$9,404	($29,563)
TOTAL REVENUES	$8,107,532	$21,888,564	$25,193,817	$25,890,946	$697,129
OPERATING EXPENSES
Program-Related Expenses
Hospital Costs	$2,374,066	$7,643,337	$9,445,473	$10,262,138	$816,665
Physician Fees	$217,772	$1,056,917	$728,557	$798,048	$69,491
Care Management Costs	$1,632,156	$2,515,946	$2,597,634	$2,588,154	($9,480)
Reinsurance	$291,665	$227,355	$19,102	$117,346	$98,244
Other Medical Costs	$1,854,034	$6,498,641	$9,560,417	$10,561,764	$1,001,347
Total Program Expenses	$6,369,693	$17,942,196	$22,351,183	$24,327,450	$1,976,267
Administrative Expenses	$2,467,947	$3,694,747	$3,091,068	$2,691,347	($399,721)
TOTAL OPERATING EXPENSES	$8,837,640	$21,636,943	$25,442,251	$27,018,797	$1,576,546
EXCESS REVENUE (EXPENSE)
BEFORE RISK SHARING	($730,108)	$251,621	($248,435)	($1,127,851)	($879,417)
Risk Sharing Adjustment	($142,047)	($31,450)	$145,434	$544,000	$398,566
AFTER RISK SHARING	($872,155)	$220,171	($103,001)	($583,851)	($480,851)
Startup Costs	$861,340	$997,380	$3,132,764 ^b	$0	($3,132,764)
AFTER RISK SHARING & STARTUP COSTS	($1,733,495)	($77,209)	($3,235,765)	($583,851)	$2,651,913
Source: HSCSN, 1999.
1999 results are based on a simple annualization of data for January-October 1999. The large charge of startup costs in 1998 was due to a change in governing accounting principles, as reported in the second evaluation report (Abt Associates, 1998).

3. HSCSN Financial Results

Given that HSCSN started the Demonstration with a strategy that later changed substantially, and given other difficulties described later in this chapter (e.g., problems coordinating with other agencies), it is far too simple to say that the financial results of the Demonstration test the practicality of the Demonstration model. In any event, a one-site Demonstration like this one gives us only a few years of financial data -- and fewer still that could be viewed as indications of steady-state or long-term results. But if the financial results must be qualified in all of these ways, it remains true that unfavorable financial results for the contractor are a caution about the assumptions contained in the original approach. As we will discuss in greater detail later in this chapter (see below and Section II.F), that is certainly how the District and HSCSN have interpreted those results.

As shown on Exhibit 2.3 above, HSCSN has had losses:

In all but one of the four years of Demonstration/extension operations, 1996-1999
Even with the exclusion of identified startup costs.

This apparent result tends to refute the more optimistic possibility suggested in the second evaluation report (Abt Associates, 1998b).⁷ Instead of seeming to be a pattern of approximate breakeven results, these results seem to depict a contractor close to the edge of large losses. At no time was HSCSN able to earn an excess sufficient to buffer years of losses; and in 1999, the losses were large, even on a cash basis. Meanwhile, certain outstanding issues with the District and its Medicaid carrier (First Health) threatened to increase historical losses by approximately $1.7 million (see discussion of this issue in Section II.D below).

The primary reason for the deterioration in results in 1999 is clearly identified in Exhibit 2.4 above: an increase in "hospital" and "other" medical costs. In turn, most of the increase in medical costs from 1998 to 1999 is accounted for by cost growth in four areas: inpatient/long-term care, day treatment, home health, and drugs. These increases in cost per member per month are detailed in Exhibit 2.5:

Inpatient/long-term care costs -- the largest single cost in 1998 -- increased 16 percent in 1999. The explanation for these increases was that a few particularly expensive cases had driven up the overall cost.
Day treatment costs -- the second largest cost in 1998 -- increased 20 percent in 1999, as HSCSN sought to use more "wraparound" services to keep beneficiaries out of Residential Treatment Facilities (RTFs) and other institutional forms of care.
Home health costs -- essentially tied with long-term care for the third largest cost in 1998 -- grew 13 percent in 1999, for reasons similar to the increase in day treatment costs.
Drug costs -- the fourth largest cost item in 1998 -- increased 17 percent, largely due to general increases in drug costs.⁸

EXHIBIT 2.5. Medical Cost Increase Per Member Per Month (PMPM) in 1998 and 1999 (ranked from largest to smallest increase): Total and Medical Cost Items
Medical Cost Item	Costs PMPM
	1998	1999(thruAugust)	1999 Over/(Under) 1998
	1998	1999(thruAugust)	Amount	Percent
Inpatient Hospital (acute care)	$203.10	$242.73	$39.64	16%
Day Treatment	136.43	170.00	33.57	20%
Home Health	119.16	137.29	18.13	13%
Drugs	56.53	67.92	11.39	17%
Transportation	29.07	33.42	4.35	13%
Physician	29.88	33.22	3.34	10%
Therapy (PT and OT)	2.36	5.59	3.24	58%
Outpatient Physician	4.14	4.91	0.77	16%
Dental	2.81	3.29	0.47	14%
Supplies	24.81	25.16	0.35	1%
Limited-use Phone	0.19	0.12	(0.07)	(62%)
Emergency	9.47	9.33	(0.14)	(1%)
Clinical Laboratory and Radiology	12.67	11.81	(0.85)	(7%)
Durable Medical Equipment	27.69	26.22	(1.47)	(6%)
Long-Term Care (RTF, SNF, other institutional care)	120.97	112.83	(8.14)	(7%)
TOTAL	$779.26	$883.84	$104.58	12%
Source: Data supplied by HSCSN on 1999 PMPM expense increases through August 1999.

Overall, HSCSN characterizes the unfavorable results for 1999 as a concise reflection of the problems with this Demonstration model. Specifically, HSCSN managers claim that services for this population in this setting are not usefully done under a standard capitation arrangement. The reasons given by HSCSN managers echo reasons given to the evaluator one year ago and include:

Enrollment levels are too low to make capitation cost effective. HSCSN claims that 2,000 members are not a large enough pool to spread the risks of this population. In the words of one HSCSN financial officer, "if you get 10 bad [cases], it can throw off your numbers." This population has outlier cases with enormous costs -- "10 percent of the children account for 90 percent of the costs." That limits the contractor's ability to manage costs within a fixed capitation amount. For the same reason, it was not practical for HSCSN to reinsure against many of the potential losses it faced.⁹
Risk sharing helps to moderate the results under capitation, but is not a substitute for a more direct stop loss arrangement. As noted in Section II.C.2 above, risk sharing has helped to moderate HSCSN's losses in 1998 and 1999. However, it has been surprisingly difficult to administer the terms of the risk sharing agreement, as there have been protracted disputes over the allocation of costs to different categories -- HSCSN claims that the Medical Assistance Administrator (MAA) auditor has not given full credit for key costs such as care management and systems. This dispute was ultimately resolved in late 1999 as part of a more general resolution of many outstanding financial issues, as a prelude to the second extension of the Demonstration¹⁰ (see Section II.F below). The complexities in the risk sharing arrangement, and the need for more direct protection of losses in outlier cases, lead HSCSN to conclude that risk sharing was beneficial as far as it went, but that a more direct stop-loss arrangement was needed.
The "any willing provider" concept is inefficient. As indicated in interviews with project investigators one year ago, HSCSN has argued that its operations would be more efficient -- and care would be of higher quality -- if the current large network were replaced by a smaller network of providers who were more intensively cultivated and incentivized. The current network constitutes a significant administrative cost to maintain, especially in relation to current enrollment levels and the small array of providers whom most enrollees actually use. It is important to note that this issue touches on one of the compromises made at the outset of the Demonstration: i.e., to require a larger network to mollify opposition to what originally was to be a sole source and non-competitive contract with mandatory enrollment.

The key to the success of the HSCSN program is the well-developed care management function to improve beneficiaries' quality of care, not the risk bearing contractor. HSCSN argues that the care management function is indeed helpful to the care of this population -- and to the experience of parents and others who must guide children with special needs through often complex treatment regimes. Moreover, it is essential to develop a unique approach to this population (these children cannot be "dumped" into the conventional system and taken care of in conventional ways). But the risk-bearing of the contractor is not essential to this benefit (it would be possible to give the contractor incentives for efficiency, without asking the contractor to bear the level of risk that this Demonstration required).
Enrollment is expensive and should be separately reimbursed. As noted in the second evaluation report (Abt Associates, 1998b), HSCSN argued that enrollment costs were so high that a separate payment should be made for each enrollment, outside the capitation amount. This argument is based on the contractor's experience that it took, on average, seven or eight attempts before a contact was made with each eligible family. Enrollment costs for this population are thus higher than for the standard Medicaid population and are not fully reimbursed by a capitation rate based on fee-for-service Medicaid costs.
There is a pressing need for better coordination among District agencies. In mid-1998, the CEO of HSCSN argued: "There has been a lack of a clear definition of roles and responsibilities between HSCSN and the various agencies of the District government that deal with children with special needs. Until the City works out these roles and responsibilities, both the City and the managed care provider will waste time and money while the children suffer in the background."¹¹ This last comment reflects the continuing problems of coordination that the MAA experienced (see Section II.E below). The problems have been expensive and exasperating for the contractor, as well as MAA.

HSCSN thus professes considerable ambivalence about the financial aspects of the Demonstration model. While a firm supporter for case management for this population, HSCSN does not like the financial terms under which it has operated. Note that MAA apparently agrees: while the recompetition (for what is termed the Child and Adolescent SSI Program, or CASSIP) was reportedly to be based on capitated payment and risk sharing similar to the current Demonstration, the new payment method will include stop-losses for three types of patients (notably, three types of losses that track much of HSCSN's cost growth in 1999):

Annual inpatient expenditures greater than $100,000 for any child.
Annual pharmaceutical expenditures greater than $25,000 for any child.
Annual home care expenditures greater than $12,000 for any child.

In addition, enrollment in the new Demonstration is to be done by a separate broker, thereby achieving separate reimbursement for the extraordinary enrollment expenses of this population. Finally, contractors in the recompetition must satisfy network requirements that appear to be more accommodating than in the Demonstration: requirements are generally phrased in terms of the contractor's need to establish a specialized network capable of meeting the health care needs of the enrollees, and similar language, rather than in terms of a required openness to any willing provider. More will be said on the CASSIP procurement below, in Section II.F. The important point here is that key tenets of the HSCSN position represent a consensus with the District and other stakeholders, although some parts of the critique (e.g., concerning inadequate coordination) are still unresolved and -- most important from a financial point of view -- the recompetition will award two contracts to serve this population, thus reducing the enrollment further¹² and creating the potential for biased selection of one of the contractors.¹³

In any event, the changes promised for CASSIP apparently are not enough to persuade HSCSN. Senior managers indicate that HSCSN does not plan to bid on the recompetition for the Demonstration. This choice is said to be prompted by the requirement in the new competition that all bidders be qualified HMOs in the District, which HSCSN is not willing to do. (In the Demonstration, HSCSN did not have to be a licensed HMO.)¹⁴ But that choice to go to the expense and trouble of qualifying as an HMO is not independent of the contractor's judgment about and enthusiasm for the model. While the new model goes far in the direction of what HSCSN has argued is needed, it does not appear to go far enough from HSCSN's point of view.

D. CHCF/MAA Oversight of the Demonstration

From the beginning, the Demonstration was overseen by the Commission on Health Care Finance (CHCF), the District agency responsible for the Medicaid program and the Demonstration. In 1998, CHCF was renamed the MAA, still part of the Department of Health. The Demonstration was meanwhile moved from the former Children's Unit of CHCF to the Office of Managed Care (OMC) in MAA. OMC has more experience in monitoring managed care providers and (given the District's move to a mandatory managed care model for TANF-based eligibles) was the locus of major decision-making on managed care and children's health issues.

Accompanying this modest change in organization was a change in staff. The original CHCF project officer was replaced in the Fall of 1997; after only a couple of months, the project officer was changed again, to a new manager with mental health experience. At this point, direct CHCF oversight stabilized, as there was no additional change in project officers through early 2000. Perhaps the most important change otherwise has been the departure in late 1999 of Paul Offner, Deputy Director of MAA and a prime advocate for the Demonstration since its inception (Goldstein, 1999).

As oversight continued under the OMC, oversight shifted somewhat to a greater emphasis on outcome measures and quality assurance processes as the Demonstration moved into its final year. There were three factors behind this change, apart from the change in project directors:

With the organizational shift of the Demonstration to the OMC came the adoption of some of OMC's monitoring approach.
Joint Commission for the Accreditation of Healthcare Organizations (JCAHO) accreditation reviews of HSCSN and the seven TANF HMOS required the collection of more quality and outcome measures.¹⁵
The MAA Medical Director was brought into a more regular oversight role for the Demonstration, and this encouraged a shift of attention to more systematic health indicators. For example, in the past year, the MAA Medical Director was added to HSCSN's quality control committee. In the words of an HSCSN observer, "[MAA has] separated administration from quality so they don't have to go through administration to get to quality issues."¹⁶

Thus, a somewhat different monitoring approach was gradually introduced into the monitoring of HSCSN. From HSCSN's point of view, the changes were not major. It seemed natural that, as the Demonstration entered its final year, HCFA would understandably be asking more questions, so DC project overseers would be getting more involved, asking HSCSN more questions, and wanting more documentation of outcomes.

There were a series of continuing operational issues between MAA, HSCSN, and others that complicated the Demonstration. These warrant a brief summary here. First, there were a series of problems with District systems from the earliest days of the Demonstration. They include: a) problems locating new enrollees, which made enrollment a much more expensive task than originally projected, as discussed above; b) incomplete canvassing of the target population (by the end of 1997, with only a year to go in the Demonstration, almost 30 percent of the total SSI population had not been located); and c) disparities among enrollment records of MAA, First Health Corporation (the Medicaid carrier), and HSCSN, creating a $1.7 million conflict over the total amounts due HSCSN.¹⁷ This latter problem was close to resolution by December 1999.

Second, as the Demonstration proceeded, MAA corrected a series of areas in which it was out of compliance with the terms and conditions of the waiver, including monitoring requirements, reporting, access study, data provision, and payment. By the end of the original three-year waiver in 1998, these disparities were no longer considered serious.

A final area of major concern for MAA (and HSCSN) was in coordinating the Demonstration with other areas of city government and the community. This topic is sufficiently important to warrant separate discussion.

E. Relationship to the Community and Coordination with Other Service Systems

Especially in a Demonstration where integration of services is the fundamental promise of the initiative, it is important to ask how the Demonstration relates to other service systems. Has there been coordination at the important boundaries of the Demonstration, where there is no formal authority or responsibility? The most important potential areas of collaboration include the school system, residential treatment placements, the foster care system, public health clinics, area hospitals, and the community generally.

1. The DC Public Schools

The District public schools have extensive responsibilities for children with special needs (including health responsibilities: by the mid-1990s, the school system actually had became a substantial Medicaid provider). It is easy to imagine some of these overlaps between educational and health activities -- e.g., for children with special needs who require the support or observation of a nurse while they are instructed. In the beginning, CHCF sought to coordinate the Demonstration initiative with the school system, hoping to include school-based health costs in the Demonstration funding. Ultimately, these school-based costs were not part of HSCSN's formal responsibilities. As a result, much of the hoped-for coordination would have to occur by mutual agreement. CHCF sought in 1994-1995 to obtain a memorandum of agreement from the District public schools. According to CHCF and HSCSN, the public schools resisted all overtures from CHCF or HSCSN to collaborate in any way. (They also resisted efforts to talk to school officials familiar with these issues in the first and second years of our investigation.) That resistance continued throughout the Demonstration. According to senior MAA managers, the real lesson of this frustrating experience is that the schools need to be brought into the Demonstration at the outset. The original design of keeping the schools separate and looking to inter-agency agreements to mesh the overlapping interests of the two agencies (MAA and the DC Public Schools) was too optimistic. By 1999, there was still no agreement with the schools, although school personnel were said to be slightly more responsive. In the past, according to MAA, school personnel never came to meetings concerning the Demonstration. Now, they come to the meetings, but still are not forthcoming on substantive cooperation.

2. Residential Treatment Facilities (RTFs)

A second, and particularly troubling, area at the boundary of the Demonstration and other public programs was residential treatment facilities. These facilities provide supervised residential care for patients in temporary need of 24-hour supervision in an institutional setting. District placements in RTFs can be generated from a number of different agencies: the school system, the courts, the Commission on Mental Health Services, and the Commission on Social Services (including child protective services and foster care). There are no RTFs located in the District. All of the providers are out-of-area, some as far away as Texas. The placements have been extremely expensive -- each child in an RTF cost $125,000 per year per child on average, in 1997 -- both because the daily rates were high and because lengths of stay were "tremendous," in HSCSN's words. The key issue for the Demonstration was that a Medicaid-eligible SSI child enrolled in the Demonstration could be placed in a Medicaid-eligible facility for non-Medicaid reasons. In that event, HSCSN was not required to pay for this care. HSCSN insisted that it should not pay for care it could not manage, and such management was impractical when the placements were supervised by other agencies. There was a somewhat different problem with the TANF HMOs: the TANF plans have a fiscal incentive to shift very difficult and expensive cases to RTFs to trigger HMO disenrollment.¹⁸

The District has attempted to improve the coordination among agencies through a Residential Placement Unit Committee (the "RPU Committee"). This committee is Medicaid-wide and put representatives of different agencies around the same table to approve each placement (other than placements by the courts, for which committee recommendations are advisory only). This committee replaces the Commission on Mental Health, which formerly did the residential reviews to determine proper placements.

The RPU Committee is generally described as an improvement. But the kind of prior authorization and review it provides is episodic and is worked out among agencies that have some conflicting interests (e.g., in shifting responsibility for the cases). Having this committee is not the same as having a single entity wholly responsible for these cases and working actively to create community alternatives to expensive institutionalization -- i.e., it is not the kind of proactive, incentivized care management that the Demonstration model was designed to perform. MAA views the special needs HMO as the natural entity to be responsible for the cases. But according to CHCF/MAA, none of the other agencies involved has been willing to entertain that kind of arrangement at any point in the Demonstration.

The Demonstration can claim partial credit for bringing the RTF problem to light: with the Demonstration, a contractor with a charter to integrate services was set loose on a very complicated system of services. In part out of the contractor's self-interest (i.e., its unwillingness to pay if it was not given authority to manage), some very troubling placements were brought to light. Efforts have been made to improve matters, but progress has been slow, in part because of genuine conceptual complexities (notably, in how to define the group subject to RTF placements, for setting capitation rates).¹⁹ In the CASSIP procurement, the current arrangement will be revised only to reflect the consequences of mandatory enrollment: the special needs contractors will be responsible for the RTF placements of all children on SSI (under the Demonstration, the single contractor was responsible only for RTF costs for SSI children who enrolled in HSCSN). In both cases (i.e., Demonstration and CASSIP), however, non-SSI children are not included.

3. Foster Care

A third key area of coordination has been with the foster care program. This program historically has been administered by the Commission on Social Services, although it is now under a receiver (the so-called "LaShawn Receiver"). There are roughly 3,000 foster care children in the District, a population of almost the same size as the children with special needs who are included in the Demonstration. There are points of contact between foster care and the Demonstration (e.g., over Medicaid-funded placements of foster care children in RTFs). In addition, the foster care population is important as a group that could be added to the SSI group now in the Demonstration, to enlarge the pool of beneficiaries. Specifically, adding foster care children to the Demonstration would make it more feasible financially to have multiple contractors (a HCFA requirement for any recompetition for the Demonstration). But efforts to include foster care in the Demonstration and in CASSIP were unsuccessful.

The enduring conceptual difficulty of including foster care has been that most foster care clients are placed in Maryland and Virginia, outside the District and dealing routinely with providers is not easily managed from the District. Moreover, foster care children typically are subject to court-ordered placements and services, which create endless potential conflicts with a capitated manager over who exactly has authority to manage their care. In any event, the schedule delays for the CASSIP procurement placed that procurement beyond the date when the LaShawn Receiver wanted to act. The receiver awarded a contract in 1999 to provide for the management of foster care health services in a fee-for-service environment.²⁰ That arrangement is very different than the arrangements contemplated for SSI children in the Demonstration or in CASSIP. Conceptual papers for the recompetition assumed (as late as May 1999) that foster care children would be part of the new procurement.²¹ The draft RFP for the CASSIP procurement excludes foster care children, although it does apparently state that "another population" could be added to the group in due course. But that is speculation. What is known is that foster care children will not be added in time to affect initial participation decisions of and bidding by contractors.

4. Public Health Clinics

A fourth area of potential collaboration has been with public health clinics. Public health clinics were interested in collaboration initially. The network of clinics was being consolidated, so some clinics were closing. Meanwhile, Title V and other funding for the clinics were declining, and Medicaid had become a more important source of funding. These clinics initially joined the HSCSN network. But two issues arose: credentialing (the clinics did not want to comply with credentialing requirements that, in some form, were necessary for JCAHO accreditation of HSCSN) and plans of treatment (the clinics wanted some exclusions from plan-of-treatment requirements). When these issues could not be resolved, the clinics withdrew from the Demonstration (approximately April 1997). These clinics were important primary care centers for some enrollees, so some members dropped out of the Demonstration after this decision. By 1999, the clinics still did not meet JCAHO requirements. The clinics have not participated as providers in the Demonstration. However, they are to be included as a required part of the network of providers for CASSIP. The CASSIP procurement will not require providers to be JCAHO accredited, thus eliminating the key obstacle.

5. District Hospitals

A fifth area of collaboration has been with area hospitals. There have been frictions between the Demonstration and two area hospitals: DC General and Children's. Relationships with these hospitals improved in the last year of the original waiver, largely as the result of fairly careful and thoughtful engagement of these institutions by MAA and HSCSN over a series of outstanding issues (Abt Associates, 1998b).

6. Relationships with the Community

A final, and particularly important, area of collaboration has been with the community. Here, efforts have been relatively successful. Perhaps as a result of initial program controversies, the District and HSCSN had to make substantial efforts to reassure the community at the outset of the Demonstration. These efforts at damage control launched an active process of community collaboration that has continued. The Demonstration was able to forge linkages with advocacy groups (many of which were also service organizations) by inviting them to join the HSCSN provider network. Representatives from some of these advocacy groups sit on the HSCSN Community Advisory Board, along with caregivers of HSCSN, members, providers, and Care Managers. HSCSN also hired outreach staff from the community to engage the (often hard to reach) beneficiaries and to create a bridge to a community suspicious of managed care. Similarly, the District and HSCSN cultivated indigenous community leaders, attempting to explain the Demonstration and reduce opinion leaders' opposition to it.

After two years of Demonstration operations, it could be reported that the opposition to the Demonstration had gradually abated.²² Advocacy groups had little to say about the HSCSN Demonstration. By 1999, their attention was focused on other managed care programs of the District -- notably, the much larger managed care program based on TANF. There were still some community complaints over Demonstration-related matters -- namely, that some SSI eligibility outcomes were unfair and that the aging out of beneficiaries at age 22 made no sense -- but these matters generally were outside the District's or HSCSN's discretion. The Demonstration appeared by 1999 to have "dropped off the radar screen"²³ of the Control Board, community groups, and provider groups that were involved in the earlier controversies. Most of the interaction of MAA with the Control Board has been over extensions of Medicaid eligibility for near-poor, non-SSI children and their parents, up to 200 percent of the Federal Poverty Level (FPL). Providers no longer actively oppose the program: reportedly, there were only two complaints from providers in 1999, all of the day treatment centers (the center of the opposition) were contract providers by 1999, and special efforts with hospitals in the area have reduced their antagonisms. Thus, the politics of the program and relations with the community achieved a state of relative calm by 1998, which continued in 1999.

7. Coordination and Collaboration: Conclusion

While community collaboration ultimately improved substantially and frictions with hospitals and other providers have eased, most areas of coordination and collaboration have been disappointing. That is not Abt's opinion alone; it reflects what MAA and HSCSN have concluded in interviews with project investigators. Coordination with the schools, a key actor in any program designed to integrate care for special needs children, has not improved. The public health clinics remain outside the program, although they will be included in CASSIP. RTF coordination remains extremely difficult, although the new Demonstration model proposes slightly more complete integration (as SSI children outside the Demonstration come under a mandatory enrollment requirement).

It is difficult to mesh any managed care program with all of the different systems of care and coverage outside its scope. What is notable in this case is how difficult progress has proved to be to date. Although there are some promising possibilities, there is little tangible progress. Integration of services for the children involved is notably difficult, as it can only be done child by child, through ad hoc arrangements between agencies. In all these areas, the issues are serious to the agencies involved and genuinely difficult to resolve conceptually. But these enduring, disjointed connections are doubtless rather puzzling to the parents affected and go to the heart of the coordination task that was one of the principal motivations for the Demonstration. From the current vantage point, integration of the special needs HMO with the broader service system appears to be the work of many years, not a few-year Demonstration.

F. Extensions of the Demonstration and the CASSIP Procurement, 1998-2000

The original waiver for this Demonstration was scheduled to end after three years of operations, on November 30, 1998. The waiver has been extended twice, through November 30, 2000. This section briefly summarizes events leading to the extensions and provides an overview of plans for CASSIP.

1. The First Extension, 1998-1999

When the District originally applied for the waiver for the Demonstration, one point of conflict with HCFA was that services under the Demonstration were to be offered by a single contractor who received the contract as a sole source, non-competitive award. Apart from the appearances of some possible political motivations in the award, this provision created concerns in HCFA that, when vulnerable special needs children faced any service or other difficulties, they would have no alternatives; i.e., no alternative to the single contractor procured by sole source arrangements (moreover, a contractor inexperienced in managed care for this or any population). These concerns and others ultimately led HCFA to insist on a purely voluntary enrollment arrangement. They also led HCFA to insist in the waiver agreement that a sole source contractor would be permitted only for the first three operational years.²⁴ Multiple plan participation was a condition for renewal. (The BBA of 1997 made multiple plans a statutory requirement for Medicaid, in any managed care program with mandatory enrollment.) As a result, following the first two years of program operations (and assuming continuation of something like the Demonstration), the District was required, as a condition of waiver renewal, to implement an RFP to select multiple plans.

The District did not submit a continuation RFP in the fall of 1997 as originally scheduled, nor a phase-down plan thereafter, as required under the terms of the waiver. HSCSN basically agreed that it would continue to work, as long as payments continued.²⁵ The initial delays in processing the recompetition were apparently due to two reasons. First, oversight of the Demonstration changed twice in a few critical months in the Fall of 1997, just as major termination/transition activities were scheduled to occur (the end of the waiver and the application for a new one, the recompetition of the contract, and so on). Second, the District was beginning to have second thoughts about the Demonstration model as originally specified. Rethinking the model would take time and require new staff to be fully trained for the project. As a result, as time began to run out, the District explored the idea of a short (one-year) extension to the existing waiver, to permit the District to move in an orderly way to the recompetition. On July 31, 1998, three months before the original waiver was scheduled to expire, the District submitted a letter to HCFA formally requesting an extension of the existing waiver. In August, the District submitted a "roadmap" for the development of a replacement waiver program for the Demonstration.

HCFA responded in September with a lengthy series of written questions to the District. MAA submitted answers at the end of October, with more complete answers to HCFA's financial questions submitted shortly thereafter.²⁶ On November 24, 1998, HCFA granted a one-year extension of the waiver, effective December 1, 1998, through November 30, 1999, subject to a set of straightforward conditions.²⁷ There was thus no interruption of coverage for Demonstration members, nor, so far as we know, were there any signs of concern from parents, members, or the community that services might be interrupted. It helped that MAA's request for an extension came at a time when the Demonstration otherwise was in state of relative calm, with no internal disputes. Indeed, controversy would have been created if there were an interruption of Demonstration-like coverage.

2. The Second Extension, 1999-2000

Under the terms of this first extension of the waiver, it was expected that the RFP would be issued in the Spring of 1999, with a contractor selected by late summer or early fall. It quickly became apparent that the District could not meet this schedule. There were three apparent reasons:

Conceptual -- If the recompetition had involved only minor changes in the Demonstration model, it would have been a fairly straightforward task to write the RFP, secure agreement within the District government, solicit comments from stakeholders, and commence the procurement process. However, the District had some serious conceptual issues to confront before this process could begin. By late 1998, it became clear that substantial changes were necessary in the Demonstration model, since:
- HSCSN financial results suggested problems for any single contractor operating under the straight capitation plus risk-sharing arrangement.
- Having multiple contractors would make the risk pool even smaller. The most natural way to increase the size of the population being managed was to add foster care children to the program. Bringing foster care into the Demonstration would take time.
- Meanwhile, there were a number of other issues to address in formulating the new Demonstration model, as has been noted throughout earlier sections (see especially the discussion of HSCSN's critique of the Demonstration model in Section II.C.3). For example, enrollment would have to become mandatory for any contractor to be willing to bid on the new project, and an independent enrollment broker would have to be arranged.
Administrative -- There were three primary administrative reasons for the delay. First, according to an MAA senior manager, the procurement process "takes longer than you could ever anticipate."²⁸ In other words, part of the reason for the delay in procurement was not because of conceptual difficulties or (as it turned out) policy disputes within the District government, but rather because -- once MAA knew what it wanted to do -- the ordinary processes of review and clearance took longer than anyone expected. As a result, MAA did not start on the recompetition in time to be able to begin operations under the new arrangements by December 1, 1999. A second administrative reason for the delay was that the model envisioned by MAA would link the special needs procurement to the procurement for the TANF HMOs. The contracts for TANF HMOs ran until March 31, 2000. MAA decided to delay the special needs procurement by four months to coincide with the TANF HMO procurement schedule. A final administrative reason for the delay concerns the Deputy Director of MAA, Paul Offner. The evaluators are not expert in the inner workings of the DC government. But a published report at the time of Offner's resignation (October 1999) suggested that he had become "'radioactive' because of his conflicts with D.C. Council members and officials from the Public Benefit Corp., the quasi-independent city agency that operates D.C. General Hospital" (Goldstein, 1999, p. B4).²⁹ To the extent this characterization was true, the leader of MAA faced difficulties in putting together a consensus on the new procurement, and his departure meant that a final sign-off on the new CASSIP approach would have to come from his replacement.
HSCSN's decision not to bid and the problems of transition -- MAA was fairly certain that it would require bidders in the CASSIP procurement to be licensed HMOs. MAA also knew from HSCSN that the HMO requirement meant that HSCSN would not participate in the bidding. As a result, the transition from the Demonstration to CASSIP would involve 100 percent of the current Demonstration enrollees. That made transition planning more difficult and likely made the recompetition itself potentially more controversial to key stakeholders, especially parents and providers. (According to MAA, when parents were queried about what they would like the new program to be, they fairly consistently said, "make it like HSCSN"). For many or most parents of current enrollees, that is, no change would have been fine. HSCSN's departure would mean a substantial change, in form at least.

For all of these reasons, MAA was slower in advancing the procurement process than it had envisioned when it requested the first extension. A second extension was needed. No waiver was required, as enrollment was to continue to be voluntary. But a new financial arrangement was necessary, as a) HSCSN was unwilling to bear the risk for another year under an extension of existing terms, and b) it was also necessary to clear up a number of payment and other matters that had persisted. The key financial terms of the extension were as follows:³⁰

A new capitation rate was calculated for the period December 1999 through November 2000.
Twenty percent of this rate was to provide final reimbursement for administrative costs, while 80 percent of the rate was to provide interim reimbursement of medical costs.
After the end of the fiscal year, a settlement process would give HSCSN full reimbursement of all medical costs in excess of the interim payment amount.

With this arrangement in place, the continuation of the Demonstration contractor for a fifth year was secured.

In the meantime, MAA had continued to prepare the CASSIP procurement. In May 1999, MAA issued a concept paper that outlined its tentative parameters for the new procurement. This paper was circulated to stakeholders for comment. In August, MAA issued a draft RFP for comment. A revised draft RFP was completed in October. Exhibit 2.6 below summarizes some of the key differences between the Demonstration model, the Concept Paper, and the October draft of the RFP. The most important differences are as follows:

Enrollment -- The expectation for CASSIP has always been that enrollment would be mandatory, a major change from the Demonstration.
Eligible population(s) -- The Concept Paper assumed that foster care children would be added to SSI beneficiaries, and that certain other high-risk populations for which Medicaid was paying might also be added. The RFP conceded that the special needs HMO would be restricted to SSI children, as in the Demonstration, since foster care children and others could not be added. Note that, by the time of the October draft of the CASSIP RFP, the LaShawn Receiver had established a care coordination arrangement for foster care, as discussed in Section II.E.4. Efforts by MAA to resolve conceptual and other difficulties with foster care proved unsuccessful, so that foster care children were excluded from the RFP.
Plan qualification -- HSCSN was not required to be an HMO. CASSIP bidders must be licensed HMOs. MAA was considering requiring accreditation in the Concept Paper, but deleted that requirement from the RFP.
Network requirements -- Where HSCSN had to accept any willing provider, the Concept Paper established requirements of adequacy of the network and named a few specific required providers. The RFP essentially establishes requirements of adequacy, without naming many specific providers.
Financial terms -- Financial terms of the Demonstration contained in the RFP appear to be evolutionary refinements of what was set forth in the Concept Paper. The CASSIP procurement will augment the capitation-plus-risk-sharing of the Demonstration with specific stop-losses for inpatient, drug, and home health costs, as discussed earlier (Section II.C.3). In addition, CASSIP will employ an enrollment broker, thereby offloading one of the unexpectedly high costs of the Demonstration onto a separate contractor. This approach bears out the commitment of the new Deputy Director of MAA, who indicated in an interview that the point of the special needs HMOs is to improve the care of some of Medicaid's most vulnerable beneficiaries, not to create incentives that pressure the care manager to save money. For that to occur, the District needs to bear more of the risk of the costs of care.

The major question now that the RFP terms are settled is how bidders will respond. Will MAA get enough qualified bidders for the CASSIP HMO to award two contracts? Some observers are skeptical, but the best test will be to observe responses to the actual.

G. Conclusion and Implications of the Implementation Analysis

This chapter has summarized the findings of the implementation analysis. It is important in reviewing these findings to recall the original objectives of the Demonstration: to improve the care received by special needs children and the ease of accessing that care for the families of these children. The use of a risk-bearing HMO to these ends was done not to save money per se, but rather as a natural model to incentivize the contractor and (for a District government in serious financial trouble) to make the costs of the Demonstration more predictable.

The implementation analysis, plus other parts of the evaluation (see especially Chapter 4 and Chapter 5 below, which discuss care management and the impacts of the Demonstration), suggest that the Demonstration was an important step in the right direction to finding a better way to manage and coordinate the care these children need. But the results of the Demonstration are in some respects disappointing -- or, perhaps more accurately, in some respects a reminder of how difficult it is: a) to establish a new approach to coordinate care in the midst of a fragmented, often parochial system that poses substantial political and intellectual barriers to coordination; and b) to address the intense health needs of this population within a risk-bearing or insurance model of financing. Each of these merits further discussion.

EXHIBIT 2.6. Comparison of the Demonstration Model to the Original CASSIP Concept and the Draft CASSIP RFP
Issue	Demonstration Model	CASSIP Procurement
		Concept Paper, May 1999	Draft RFP, October 1999, IF DIFFERENT
Enrollment	Voluntary	Mandatory
Eligible populations	SSI	SSI	SSI only
		Foster care
		Possible additional groups: Youth Services Admin., Children at risk of RTF placement, certain institutionalized children, and newborns with special needs
Exclusions	Dual eligibles	Dual eligibles
	Children in institutions	Children in institutions
Number of contractors	One	Minimum of two required; no more than two to be chosen
Plan qualifications	Not required to be HMO	Must be HMO	Must be HMO
		MAA "considering requiring" JCAHO or NCQA accreditation	Accreditation not required
Network requirements	Any willing provider	Network requirements in descriptive terms, except that four providers specifically named:- HSC,- DC Public Schools contracted services,- Special needs clinic of DC General Hospital, and- Public health clinics of the Public Benefit Corporation	Network requirements stated in general terms
Financial terms	Single capitation rate plus risk sharing	Separation capitation rate for SSI, foster children, and other groups if included.	Single capitation rate (for the one group included), plus risk sharing.
		Risk sharing to be developed.	Stop loss for: inpatient costs > $100,000, drug costs > $25,000, and home health costs > $12,000
		Stop-loss for certain extraordinary expenses exceeding $100,000 annually.	MAA reserves right to institute performance bonuses/penalties after first year
		Performance bonuses/penalties being considered.
Schedule- RFP Issued- Proposals due- Contracts signed- Start of services	N/A	Schedule- August 1999- September 1999- January 2000- April 2000	Approximate schedule (per 1/99 interview):- February 2000- April 2000- November 2000- December 2000[Schedule not stated in draft RFP]
Sources: Medical Assistance Administration, "Medicaid Managed Care: Phase II Plan for Children with Special Healthcare Needs," dated May 3, 1999; and Medical Assistance Administration, District of Columbia, Medicaid Managed Care Program: Phase II, DRAFT Joint Request for Proposal for District of Columbia Healthy Families Plans (DCHFP) and Child and Adolescent SSI Plans (CASSIP), dated October 4, 1999.

1. While the Demonstration has made peace with most agencies, providers, and beneficiaries, it has had continuing difficulties in achieving coordination with other agencies and coverage that are critical to this population (notably, schools, foster care, and RTF admissions).

Throughout the Demonstration, the MAA has aspired to achieve a coordination of responsibility, care, and funding with other critical agencies and programs serving this population. The efforts have been persistent but have not rewarded MAA's larger ambitions. For example:

MAA sought to coordinate the SSI population with foster care, in order to create an eligible population large enough to support two contractors and to bring foster care-based Medicaid expenses under better control. As recently as May 1999, MAA expected that the two populations could be joined. But that goal continues to be elusive, in part because the combination is genuinely difficult to figure out, and in part because of predictable difficulties in getting two agencies to combine separate programs and streams of funding.
The schools have resisted collaborating in any active way (indeed the schools were hostile at the beginning). The evaluator formed the superficial impression that relations may be thawing somewhat, and that MAA may build coordination with the schools bit by bit over time. But the grand arrangement never materialized, nor will it in the recompetition. In the CASSIP procurement, contractors are mandated to go forth and coordinate with the schools, but it is not clear that they will be any more successful than MAA has been.

In each of these cases and others, there are "explanations" why coordination did not occur, beginning with the turmoil in DC government throughout the early years, and continuing with such factors as the apparent parochialism of some would-be partners and even minor issues like incompatible schedules between a foster care receiver and MAA.

But stepping back, the real reason is likely that it is extremely difficult to mesh this special needs organization with all the different programs and data systems with which coordination is ultimately desired. Building those relationships takes time -- apparently, more time than (what will be) a five-year Demonstration.³¹ This point is perhaps obvious, but MAA and HSCSN were not able to bring this total concept into being full blown, nor to complete the necessary work in the time span of the Demonstration.

2. Financial results to date tend to support HSCSN's claim that it is difficult to manage risks for a population -- a) this small, and b) with these special healthcare needs -- within a capitated payment and limited risk sharing.

This Demonstration (and the implementation analysis, by itself) cannot really test whether the financing model of the Demonstration could have worked. This Demonstration involves only one case, implemented in very peculiar circumstances, as detailed in the discussion of Demonstration origins above (Section II.A). Meanwhile, the implementation analysis cannot construct alternatives to prove that there was, in fact, a better way to manage care and control costs, except in the most extreme circumstances that make better alternatives obvious. Needless to say, nothing in the implementation analysis points to such extreme circumstances.

Instead, the available evidence offers a first-order caution that quality health care for this population does not readily fit within a more or less conventional risk-bearing arrangement, even when the risk-bearing is partially mitigated by risk-sharing. HSCSN has never had sufficient surpluses to buffer itself against increases in medical care costs, and appears to have lost a lot of money. That result holds even if areas in which HSCSN arguably overspent (such as systems development in the startup) are excluded. The most troubling evidence available is that the worst financial results (on a cash basis) came in the most recent period (1999), when one would expect care management routines to have become more stable and desirable care management approaches to be better understood. Other states that have set up special care management for special needs populations (notably Florida and Michigan) have deliberately avoided the kind of capitated approach used in this Demonstration (they have employed fee-for-service reimbursement with incentives to the care management contractor to achieve savings). The District has made a substantial move in that direction for the CASSIP procurement. Those approaches are consistent with what the implementation evaluation has found for the DC Demonstration.

III. ENROLLMENT AND SELECTION

A. Introduction

Until the District's Demonstration, no other managed care plan had been established with the sole purpose of serving a population of children with special needs. This section describes enrollment patterns and explores the issue of which children in the SSI program in DC are enrolled when a managed care plan is offered as an alternative to the Medicaid fee-for-service system.

Using HSCSN Enrollment and Claims Data and the Medicaid Recipient and Claims Files, the enrollment and selection analyses find that:

HSCSN was successful in recruiting a majority of all SSI eligible children in the District;
Enrollment accelerated in the first year of the Demonstration and then increased at a declining rate;
There were three major reasons for disenrollment: the child moved out of the District, the child aged out of the Demonstration, or the caregiver changed his/her mind about enrollment;
The majority of participants do not reside close to (in the same zipcode as) their provider's place of service;
Participants are younger, more likely to be African American, more likely to be male, and more intensive users of services than non-participants;
Participants who disenroll are older, more likely to be female, and less likely to be African American than those who remain in the program.

B. Enrollment Process

The enrollment process begins when the District's MAA mails out an enrollment packet to the parents of eligible children. Parents then have 45 days to return the selection form to the MAA. If the form is not returned within the allotted time period, HSCSN is allowed to directly market the program to the family.

If HSCSN outreach workers contact these families, they verify eligibility status. Then, if the caregiver is interested in enrolling the child, the outreach worker will arrange for a health assessment to be conducted. Information that is collected from this assessment includes an inventory of information about the child's health status, needs for a variety of medical and social services, and environmental information, such as the child's housing status. After the health assessment has been completed, and HSCSN has verified the child's SSI eligibility, the child is enrolled and begins receiving services through HSCSN's plan. As of November 1, 1997, the District mailed in excess of 4,450 packets.

In 1998, mailings were discontinued because of administrative and staff changes, in addition to problems with coding data and obtaining correct address information. As there were no new mailings, enrollment later in the Demonstration was driven by referrals from the Outreach Department, community service organizations, agencies, hospitals, social workers, family members, and health fairs.

In 1999, a new enrollment coordinator was assigned. HSCSN also met with MAA and Beneva, the District broker for Medicaid, to revise materials and procedures related to the mailing. One major revision so far has been a reduction of the original 45-day window that parents had to respond to mailings to a ten-day waiting period. However, by the end of December 1999, there were still no new mailings, because revisions to the member materials and procedures for the mailings were not yet completed.

C. Tracking Enrollment: Data Sources

Tracking Demonstration enrollment relies on information from HSCSN's Enrollment and Claims Data Files, and from the eligibility and claims records maintained by DC Medicaid' Management Information System (MMIS). First Health is the Medicaid Fiscal Agent for the District.

Limitations in the availability of these data files determine the length of time covered by different analyses:

Total enrollment is tracked from February 1996 through June 1999, the last month for which HSCSN enrollment data were available.
Trends in Demonstration participation rates require information from the Medicaid Recipient File, which was available for the peered February 1996 through December 1998.
The selection analysis requires Medicaid claims, which were available and complete from February 1996 through June 1998.

HSCSN and Medicaid files that were utilized for this analysis were the HSCSN Enrollment and Claims file and the Medicaid Recipient and Claims file.

The HSCSN Enrollment File provides information on the following data elements: basic demographic information such as care manager, PCP, enrollment date, mailings, primary diagnosis, disenrollment date, reason for disenrollment, date of birth, address, phone number, and Medicaid ID number.
The HSCSN Assessment File provides information on primary diagnoses. The Assessment file contains information collected from the health assessment survey that is administered by outreach workers. The Assessment File that Abt Associates receives does not contain any names. Therefore, Medicaid IDs were utilized, which was limiting because the ID information was not always correct.
The Medicaid Recipient Eligibility File helps identify children who were eligible for various Medicaid programs, including HSCSN. This file contains an individual record for each recipient and is updated on a daily basis by the District's Medicaid Agency. In addition to Recipient ID number, other fields provide the following information on each recipient:
- Birth Date;
- Gender;
- Effective Dates of Coverage; and,
- Program Codes indicating the program under which children qualified for Medicaid benefits.

The Medicaid claims files were used to construct measures of "prior utilization" for the selection analysis. Medicaid claims are stored in 13 subfiles for the following types of provider's and/or eligibility categories:

Hospital ( Inpatient)
Long Term Care (LTC)
Outpatient Hospital
Physician
Home Health
Vision Care
Durable Medical Equipment (DME)
Dental Services
Pharmacy
Transportation
Early Periodic Screening Diagnosis and Treatment (EPSDT)
HMO

Limitations of the Data

Among the data problems that the evaluator had to address, most involved determining which records uniquely identify an individual:

determining correct Medicaid IDs and other key identifiers;
resolving conflicts in Medicaid IDs associated with two different names;
identifying duplicate records;
resolving conflicts regarding Demonstration participant status, including allocation of claims by service dates in "improbable" time periods (e.g., HSCSN claims for services many months before or after a child was apparently enrolled in the Demonstration).

The number of Medicaid records received from the District was 6,924. This number, however, does not reflect the actual number of children, as some of them were counted more than once. In addition to the 6,924 records, 55 more records were added from the District's cross-reference file in an effort to accurately link the same children who had different Medicaid IDs. After the data were cleaned, a total of 403 records were eliminated from the Medicaid data files because of conflicting information. The end result was a sample size of 6,576 consisting of 2,511 participants and 4,065 non-participants. Further sampling was done for the selection analyses, as described in Section III.E.

D. Characteristics of Participants

Throughout the Demonstration, enrollment steadily increased from April 1, 1996 to June 30, 1999 (see Exhibit 3.1). Analysis of enrollment and disenrollment by month revealed that retention rates, the percentage of children that continued to participate in the Demonstration from month to month, were steady from February 1996 to June 1999. Among participants, retention rates ranged from 98 percent in July 1996 to 76 percent in June 1999. Net gain and loss, the difference in numbers of children enrolled/disenrolled each month throughout the Demonstration period, was also fairly stable (see Appendix B).

Two programmatic events were expected to have effects on enrollment and disenrollment trends: 1) two mailings from HSCSN to potential participants, and 2) concentrated volumes of First Health eligibility reconciliations. It was hypothesized that HSCSN mailings would be followed by noticeably large increases in enrollment. It was also expected that major First Health reconciliations -- a process through which children without adequate evidence of Demonstration participation in both HSCSN and First Health eligibility files were at risk of being dropped from the program -- would be associated with increased disenrollments.

In January 1997 and August 1997, there were two large mailings by HSCSN. Enrollments increased sharply from March to May. Enrollments also increased steadily after the August mailing, from September through November. There was an increase in enrollment between these months, with the exception of a drop in enrollment in April 1997 (see Exhibit 3.2).

While First Health's reconciliation took place throughout the Demonstration, there were two heavy reconciliations in March 1998 and August 1998. The reconciliation in March 1998 did not have an effect on disenrollments, according to reasons for disenrollment coded by HSCSN. Disenrollments during that month and the following month were a result of some participants moving out of the District or a caregiver changing their mind. The August 1998 reconciliations, however, were associated with a small number of disenrollments. In September 1998, 16 participants were dropped from HSCSN, and First Health reconciliation was listed as the disenrollment reason.

*EXHIBIT 3.1. Number of Demonstration Participants by Quarter: Third Quarter 1996-Second Quarter 1999**
Quarter	Number of Participants
3rd Quarter 19964th Quarter 1996	1,0561,319
1st Quarter 19972nd Quarter 19973rd Quarter 19974th Quarter 1997	1,5611,8261,8782,090
1st Quarter 19982nd Quarter 19983rd Quarter 19984th Quarter 1998	2,1692,2372,3532,425
1st Quarter 19992nd Quarter 1999	2,4782,531
Source: HSCSN Enrollment File, February 1996 to June 1999.Note: *Participants who were enrolled as of the last day of each quarter are considered to be enrolled in that quarter.

After an initial period of accelerated growth, from start-up in February 1996 into the Fall of 1996, the largest subsequent increases were associated with HSCSN mailings. Whether or not additional mailings could have had the same effects, reviving what was clearly a declining growth rate as the Demonstration matured, is unclear. Despite evidence that some children were dropped from the program due to reconciliation of eligibility information, First Health's two largest reconciliations appear to have had no major effects on disenrollment.

EXHIBIT 3.2. Enrollment by Month: March 1996-June 1999: unavailable at the time of HTML conversion--will be added at a later date.

1. Average Age and Length of Enrollment

Average age at enrollment declined over the Demonstration, as children aged out. There was a slight decline in the age of children disenrolling as well. (See Exhibit 3.3.)

EXHIBIT 3.3. Mean Age of Participants and Disenrolled Children and Length of Enrollment (in days)³²: HSCSN FY 3rd Quarter 1996-4th Quarter 1999
	Average Age ofParticipants in theDemonstration	Average Age ofDisenrolled Children	Average EnrollmentDays
1996
1/1/96 - 3/31/96	12.6	16.6	1,027
4/1/96 - 6/30/96	12.0	15.4	1,034.7
1997
7/1/97 - 9/30/97	11.2	14.4	997.5
10/1/97 - 12/30/97	10.9	14.2	969.9
1/1/97 - 3/31/97	10.9	14.3	947.1
4/1/97 - 6/30/97	10.6	14.1	924.6
1998
7/1/98 - 9/30/98	10.5	14.3	935.5
10/1/98 - 12/30/98	10.4	14.0	901.6
1/1/98 - 3/31/98	10.4	14.5	905
4/1/98 - 6/30/98	10.2	14.8	901
1999
7/1/99 - 9/30/99	10.1	15.1	875.2
10/1/99 - 12/30/99	9.9	15.0	856.4
1/1/99 - 3/31/99	9.8	15.6	841.6
4/1/99 - 6/30/99	9.7	14.9	822.1
Source: HSCSN Enrollment File, February 1, 1996 to June 30, 1999.

2. Eligibility and Participation Rates

Throughout the Demonstration, there was a steady increase in the number of eligible participants enrolling. As of June 30, 1998, HSCSN was successful in recruiting about 46 percent of all SSI eligible participants in the District (Exhibit 3.4 and Appendix C). At that time there were 4,753 participants enrolled in the Demonstration, of which 233 cases were enrolled in HSCSN but were not SSI eligible. The explanation for this seems to be that sometimes children enroll and apply for SSI eligibility at the same time. At the time of this analysis it could not be determined how many of the 233 ineligible HSCSN cases had, in fact, been granted SSI eligibility.

EXHIBIT 3.4. Demonstration Participation, by Quarter (Q1, 1996-Q4 1998): SSI-Eligible Participants as Percentage of Total SSI Eligible, non-SSI-Eligible Participants as a Percentage of All Participants
Quarter	SSIEligible inDC	SSI Eligiblein HSCSN	Not SSIEligible inHSCSN	In HSCSN	% of SSIEligible inHSCSN	% in HSCSNnot SSIEligible
Q1, 1996	3,830	151	3	154	4%	2%
Q2, 1996	3,955	597	8	605	15%	1%
Q3, 1996	4,022	1,023	33	1,056	25%	3%
Q4, 1996	4,073	1,265	54	1,319	31%	4%
Q1, 1997	4,110	1,492	69	1,561	36%	4%
Q2, 1997	4,161	1,741	85	1,826	42%	5%
Q3, 1997	4,295	1,786	92	1,878	42%	5%
Q4, 1997	4,390	1,971	119	2,090	45%	6%
Q1, 1998	4,541	2,039	130	2,169	45%	6%
Q2, 1998	4,490	2,066	171	2,237	46%	8%
Q3, 1998	4,590	2,052	201	2,353	45%	9%
Q4, 1998	4,753	2,192	233	2,425	46%	10%
Source: HSCSN Enrollment file, February 1996 to June 1999 and the Medicaid Recipient file through December 1998

3. Reasons for Disenrollment

For the 583 participants who disenrolled, there were three main reasons for disenrollment from the Demonstration (see Exhibit 3.5 and Appendix D). First, some participants moved out of the District. Low-income households, such as those represented in this study, are extremely mobile. Given the geographic characteristics of the District, it is common for these children to move into neighboring states of Maryland and Virginia. (Irvin, 1998). The disenrollment reason "moved out of the District" is given for 203 children disenrolling from the Demonstration. Second, 100 participants reached the age limit for the Demonstration, which is 21. Third, the caregivers of 98 participants changed their mind about keeping the child enrolled in the Demonstration. Finally, 70 children lost eligibility for SSI.

4. Distance Participants Traveled for Services

Most participants must travel to receive services from the five major facilities that see HSCSN cases (see Exhibit 3.6, Exhibit 3.7 and Appendix E). Children's National Medical Center in the 20010 zipcode had 701 HSCSN cases and only 94 participants resided in that same zipcode. Howard University Hospital in the 20060 zipcode had 174 HSCSN cases and no participants who resided in that same zipcode. The same was true for Georgetown University Hospital in zipcode 20007 that had 90 HSCSN cases and Gerald Family Care Associates in zipcode 20017 that had 91 HSCSN cases, both with no participants who resided in either of those zipcodes. DC General Hospital in the 20003 zipcode was the exception. DC General had 92 HSCSN cases with 81 participants who resided in the same zipcode.

EXHIBIT 3.5. Evaluation of DC Managed Care Program for Special Needs Children Disenrollment Reasons, by Frequency of Reason Given: unavailable at the time of HTML conversion--will be added at a later date.

EXHIBIT 3.6. Distance Traveled to Receive Care: Provider Zipcode in Comparison to Participant Residential Location
ProviderZipcode	Enrollees whohave Providersin that zipcode	Averagemiles toProvider	Participantzipcode	Participantsliving in thesezipcodes
20010	908	3.5	20020	446
20003	251	2.75	20019	438
20060	191	3.2	20002	350
20017	172	3.2	20032	296
20032	165	5.3	20011	215
20020	136	4.0	20001	161
20037	92	3.5	20009	117
20002	91	2.75	20010	94
20007	79	7.0	20018	84
20009	24	3.0	20003	81
Source: HSCSN Enrollment File, February 1, 1996 to June 30, 1999, and Distances were calculated from http://bridger.link-usa.com/scripts/miles.exe

5. Description of PCPs and their HSCSN Caseloads

Caseloads for the top ten providers who maintained the majority of HSCSN caseloads ranged from 85 to 54 cases. The major area of concentration for the top ten HSCSN providers was Mental Diseases and Disorders.

Within the classification of Mental Diseases and Disorders the majority of HSCSN providers covered the following types of cases:³³

395 HSCSN cases of Other Learning Difficulty;
287 cases of Mental Retardation Not Otherwise Specified (NOS);
211 cases of Developmental Delays NOS; and
110 cases of Attention Deficit Disorder with Hyperactivity.

The second major area of concentration for HSCSN providers was Diseases and Disorders of the Nervous System. Within that classification the majority of HSCSN providers covered the following cases:

217 cases of cerebral palsy NOS;
63 cases of Speech Disturbance Not Elsewhere Classified (NEC);
54 cases of convulsions;
31 cases of obstructive hydrocephalus; and
17 cases of cerebral palsy NEC.

EXHIBIT 3.7. Comparison of Provider Organizations by HSCSN Caseload and Zipcode
Organization	HSCSNProvider andCaseloads	Participants Livingin Provider'sZipcode
Children's National Medical Center	701	94
Howard University Hospital	174	0
DC General Hospital	92	81
Georgetown University Hospital	90	0
Gerald Family Care Associates, PC	91	0
Capitol Hill Medical Clinic	85	94
George Washington University Medical Faculty Associates	83	0
McKnight & McKnight, MD, PC	75	94
Lee and Javate, MD, PC	68	296
Drs. Crawford & Hudson, MD, PC	62	446
Source: HSCSN Enrollment File, February 1996 to June 1999Notes: *The most predominant zipcodes were selected.

Remaining areas that HSCSN providers had five or more HSCSN cases were: Diseases and Disorders of the Respiratory System; Newborns; Diseases and Disorders of the Blood; Diseases and Disorders of the Ear, Nose and Throat; HIV infection; Diseases and Disorders of the Eye; Diseases and Disorders of the Circulatory system; Injury and Poisoning; and Endocrine/ Metabolism Diseases and Disorders.

E. Evidence for Selection Bias in the Demonstration

1. Selection in Voluntary Demonstrations

This Demonstration was planned as a mandatory program for all SSI-eligible children with special needs in the District. In response to stakeholder pressure, HSCSN and the District Medicaid implemented a voluntary Demonstration. This decision had two immediate effects on the evaluation design. First, it made available an internal comparison group of SSI-eligible children who elected not to participate in the Demonstration. Second, it complicated the evaluator's job of assessing program impacts through comparisons of program participants with non-participants. Participants with certain characteristics that are different from non-participants may self-select into a voluntary program. Also, the program itself may create explicit or implicit incentives that determine who, in the target population, applies to enroll and, among applicants, who gains entry. Whatever the mechanism, processes that encourage systematic selection may bias estimates of program effects.

Can this bias be eliminated? The truthful answer must always be negative. Technically, selection bias "encompasses any differences between the program participants and the comparison group that affect the outcomes of interest." (Orr, 1996) For this Demonstration, outcomes include access, satisfaction, and utilization. Measures that capture differences between groups could include standard demographic indicators like age, gender and race, measures of the type and severity of the condition(s) responsible for a child's disability, as well as certain attitudes, perceptions and behaviors.

Under the best of conditions, the evaluator cannot measure most of these differences. Researchers have developed statistical techniques to adjust impact estimates for inter-group differences. (Heckman and Robb, 1985). However, the conditions for applying these adjustments are stringent indeed. For example, the investigator must find one or more predictors of program enrollment that have no correlation with any of the program's outcomes, a task that rarely rewards the search.

Rather than adjusting estimates to "remove" bias, this analysis concentrates on describing processes in the Demonstration that might have promoted bias and on suggesting the nature of bias through a full understanding of measurable inter-group differences. For example, MCOs provide services for a fixed capitation payment and earn profits by managing service and administrative costs against a monthly cap. Under these rules, MCOs have clear financial incentives to hold down costs, and one plausible way to achieve this objective is to find and admit persons from the "healthier" segment of their target populations. Other incentives may rule, of course. MCOs like HSCSN may be motivated by professional and service goals, knowing that their opportunity in the Demonstration to prove the worth of their model of care will be compromised by credible evidence of "skimming" the most able eligible children. Clearly, baseline measures of illness costs or severity of condition can provide valuable evidence on whether or not selection was, from HSCSN's point of view, favorable (participants were healthier at enrollment than comparable non-participants) or adverse (participants were sicker than non-participants).

Disenrollment is another process that can distort measures of impact. If children who disenroll from HSCSN tend to be sicker (or less sick) than those who stay in the program, impacts based on comparing the "average" participant and non-participant may be biased. This study compares children who disenroll to participants who stay.

In the Second Annual Report on the Demonstration (Abt Associates, 1998), early analyses of HSCSN and Medicaid data suggested that:

Before enrolling, participants incurred more Medicaid and Medicare payments than non-participants;
Much of the difference in payment reflected the fact that non-participants were less likely than participants to use any services; and
Non-participating children were more likely than participants to be in long term care facilities and to use home health services.

This analysis confirms these early findings.

2. Data and Methods

Enrollment and Eligibility Data. Data used in the analysis of selection effects come from enrollment records maintained by HSCSN and from eligibility and claims files from DC's MMIS. HSCSN data provide information on which children enrolled (and disenrolled) and the date of enrollment/disenrollment.

For the selection analysis, investigators tried to assess the enrollment and disenrollment decisions at comparable points in time throughout the Demonstration. To do this, investigators sampled an equal number of participants and non-participants, distributed over that part of the Demonstration for which complete data were available. Three steps were required to prepare the sampling frame:

Investigators defined five fixed-time "windows" that divided the Demonstration period into relatively small parts. The first window began in February 1996, the starting date of the Demonstration, and the fifth window ended in June 1998, the last month for which complete Medicaid claims data were available.
Investigators defined "anchor months" for each participant and non-participant. Anchor months were assigned to divide time for individual children into "before" and "after" periods. For participants, the concept is easy to understand. The anchor month is the month of enrollment. For non-participants, there is no similar construct, and so anchor months were selected randomly for non-participants within the fixed-time windows. Therefore, the "before" periods for non-participants differs from the "after" only in chronology, while the before periods for participants are characterized as "not in HSCSN" and the after as "in HSCSN."
Eligibility for sampling was defined separately for participants and non-participants. Eligible participants had to have enrolled in HSCSN before June30, 1998, had to have been enrolled for at least one month, and had to have been SSI-eligible for at least one month in a six-month period before enrollment.³⁴ Eligible non-participants had to have been SSI-eligible for at least one month in a six-month period before their anchor dates.

EXHIBIT 3.8. Enrollment/Selection Analysis: Steps Taken in Data Preparation and Sample Selection
Steps in Data Preparation		EnrollmentRecords
Total records provided by DC Medicaid		6,979
	Remove records with incomplete linking information, duplicates, internally conflicting information	-403
Total "cleaned" enrollment record data base, consisting of Non-participants: 4,065; Participants: 2,511		6,576
	Remove from participant total persons who enrolled after June 1998 or not SSI-eligible in six months before enrollment	-355
	Participant sampling frame	2,176
Distributed by enrollment data:
	Window 1 (2/96-7/96):	723
	Window 2 (8/96-1/97):	640
	Window 3 (2/97-7/97):	443
	Window 4 (8/97-1/98):	269
	Window 5 (2/98-6/98):	101
Sample 2,176 non-participants from frame of 4,065:
	Window 1 (2/96-7/96):	723
	Window 2 (8/96-1/97):	640
	Window 3 (2/97-7/97):	443
	Window 4 (8/97-1/98):	269
	Window 5 (2/98-6/98):	101
Total enrollment sample		4,352

Having set eligibility criteria and defined sub-periods and anchor months, the evaluator selected all participants with an anchor date in Window One who met other eligibility criteria. An equal number of eligible non-participants was then selected. Once chosen, no sample member could be re-sampled for later periods. Therefore, the numbers sampled for subsequent windows come from a diminishing frame, for both participants and non-participants, as Exhibit 3.8 shows. Samples totaled 4,352 children, with equal numbers (2,176) of participants and non-participants.

Participants differ from non-participants in several respects (Exhibit 3.9 and Exhibit 3.10).

Evidence for the presence, numbers and types of medical conditions provides a basis for contrasting participants and non-participants. For this comparison, a list of ICD-9 diagnoses associated with eligibility for SSI were selected. These diagnoses were grouped into two major categories, "mental disorders" and "physical disorders", and further divided into subcategories as shown in Exhibit 3.9. The evaluator searched Medicaid claims records for all participants and non-participants in the selection sample from January 1995 through the month before each individual's anchor month. Then,claims incurred after enrollment, or post-anchor month for non-participants, were searched in the same way. Qualifying diagnoses were selected whether or not they appeared as principal or secondary diagnoses. Therefore, a child could have multiple qualifying diagnoses.

As Exhibit 3.9 shows, participants differ from non-participants in frequencies across the 14 diagnosis groups, and in the relative frequencies of additional diagnoses coded after the anchor date. In general, participants have higher frequencies than non-participants in all diagnostic categories and both time periods, and participants added more diagnoses after the anchor month of enrollment than did non-participants in a comparable time period. Differences are large and statistically significant for mental disorders, and for most categories of physical disorders. It would be a mistake to invest too much significance in this finding. Having more diagnoses is no sure indication of higher levels of illness or impairment. Participants, as subsequent analyses will show, used more services than non-participants, and so were probably more likely to have generated claims and coded diagnoses in the appropriate codes. Nonetheless, these differences between the groups are substantial and should be taken into account in modeling selection.

EXHIBIT 3.9. Frequency of Diagnoses Recorded for Participants vs. Non-participants, Before and After Enrollment/Anchor Dates
	Pre-enrollment/anchor date		Post-enrollment/anchor date
	Participants(N=2176)(1)	Non-participants(N=2176)(2)	Participants(N=2176)(3)	Non-participants(N=2176)(4)
Mental Disorders
Mental Retardation	297	191***	238	67***
Psychological and Neurological Development Impairment	557	415***	434	139***
Physical Disorders
Infectious Diseases	46	50	20	10*
Neoplasm	48	40	31	17**
Endocrine	50	48	36	25
Nervous System	521	330***	269	101***
Blood and Blood Forming	74	44***	29	9***
Digestive System	167	117***	128	45***
Circulatory System	94	73*	85	36***
Respiratory System	436	296***	203	120***
Musculoskeletal System	184	135***	254	111***
Congenital Anomalies	272	170***	190	67***
Perinatal	78	75	29	14**
Injury and Poisoning	23	11**	14	3***
Source: HSCSN Enrollment File, February 1, 1998-June 30, 1999 and the Medicaid Recipient File through.......Notes: Children may have diagnoses in more than one category. Chi-square measures of participant/non-participant differences pre-enrollment/anchor date (levels of significance shown in Column 2) and post-enrollment/anchor date (levels of significance show in Column 4).Statistical significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .05.

Summary measures of diagnostic history and other measurable characteristics (Exhibit 3.10) were constructed. All differences are significant at five percent and most are significant at one percent. Comparisons between participants and non-participants show:

Participants are more likely to have diagnostic evidence of development/mental health conditions (54.1 percent, compared to 31.8 percent);
Participants are more likely to have been diagnosed with one (27.8 percent, compared to 21.4 percent) or multiple diagnoses (34.7 percent, compared to 23.9 percent) in the six months before the anchor month;
Participants are older than non-participants (41.5 percent aged 5 to 12, compared to 29.2 percent, and 22.3 percent 18 years and over, compared to 34 percent);
Participants are less likely to be female (36.3 percent, compared to 40.9 percent);
Participants are more likely to be African American (93.3 percent, compared to 66.7 percent);
In the six months before the anchor month, participants were more likely to have been SSI-eligible for the entire period (95.4 percent, compared to 82.7 percent);
In the six months before the anchor month, participants were more likely to have used some Medicaid-reimbursed services (83.6 percent, compared to 53.4 percent) and inpatient services (9.9 percent, compared to 7.8 percent), but somewhat less likely to have used long-term care or residential services (1.1 percent, compared to 1.5 percent);
In the six months before the anchor month, participants incurred somewhat higher total Medicaid expenditures per month eligible for Medicaid ($1,151, compared to $1,056) but lower inpatient and long-term care expenditures ($412, compared to $571 and $137, compared to $474 respectively). Of course, because participants were more intensive users of all services, total expenditures for those who used any services were lower for participants ($1,377, compared to $1,994) not shown in this exhibit.

EXHIBIT 3.10. Enrollment/Selection Analyses: Independent Variable Means -- Total, Participants, Non-Participants
Variable	Total(N=4,352)	Participants(N=2,176)	Non-Participants(N=2,176)
Percent Age: 5-12 13-17 18+	35.3%23.028.1	41.5%27.222.3	29.2%*18.834.0**
Percent female	38.6	36.3	40.9***
Percent African-American	80.0	93.3	66.7***
Percent with at least one developmental/mental health diagnosis	42.9	54.1	31.8***
Percent with one specified diagnosis	24.6	27.8	21.4***
Percent with more than one specified diagnosis	29.3	34.7	23.9***
Percent with any prior utilization	68.5	83.6	53.4***
Percent with any prior inpatient utilization	8.8	9.9	7.8***
Percent with any prior long term care/residential utilization	1.3	1.1	1.5**
Total Medicaid expenditure per month (excluding HMO)	$1,104	$1,151	$1,056***
Total Medicaid inpatient expenditure per month	$492	$412	%571***
Total Medicaid long term/residential care expenditure per month	$306	$137	$474***
Percent SSI-eligible for 6 months pre anchor month	89.0	95.4	82.7***
Source: HSCSN/DC Medicaid Enrollment/Claims Data BaseStatistical significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .05

In order to weight the relative importance of different child characteristics in the enrollment and disenrollment decisions, the evaluator estimated logistic regression models of the probability of enrollment and disenrollment. The evaluator estimated the enrollment model for the sample of participants and non-participants. Sample for the disenrollment model included only participants.

3. Findings

Selection and Enrollment. Several characteristics which were clearly different between participants and non-participants also proved to be important determinants of enrollment within a multivariate model. The fit of the enrollment model was not exceptional (chi-square score of 935.7, p=0.0001, implying a pseudo R² of .196). Exhibit 3.11 shows the changes in probability of enrollment associated with specific characteristics. For the most part, these estimates capture differences between participants and non-participants shown in Exhibit 3.10.

EXHIBIT 3.11. Enrollment/Selection Analysis: Estimated Effects of Independent Variables on Probability of Enrollment
Variable	Parameter¹	Pr (chi-square)
AGECAT3	.229	.0606
AGECAT4	.160	.2201
AGECAT5	(.406)	.0014
FEM	(.085)	.2335
RACECAT1	1.576	.0001
DDX_A	.572	.0001
DX_PREC1	(.211)	.0278
DX_PREC2	(.343)	.0010
DMIQX_0	(.098)	.0001
DMIQXI_0	.007	.9598
DMIQXR_0	.094	.8423
MIQX_0	.00003	.2551
MIQXI_0	(.00006)	.0558
MIQXR_0	(.00005)	.0032
DMSSI6_0	1.348	.0001
DSAMP1	(.399)	.0252
DSAMP2	(.269)	.1323
DSAMP3	(.199)	.2772
DSAMP4	(.069)	.7196
Source: HSCSN/DC Medicaid Enrollment/Claims Data Base-2 Log L = 4,971 Pseudo R²=.1961 Negative values in parentheses

Age. Eligible older children are less likely to enroll than younger children. This estimate implies that a 1 percent increase in the percentage of eligible children 18 years and older will lead to a decrease in the enrollment rate of about 10 percentage points.

Gender. When adjusted for other influence, gender has no statistically significant effect on enrollment (though the negative value of the gender coefficients suggests that girls are less likely to enroll than boys).

Race. African American children are substantial majorities of both participant and non-participant samples, but being African American still significantly increases the probability of enrollment (by a little more than 39 percentage points above other racial groups).

Diagnosis. Children with diagnoses identified before the anchor date in the category "mental retardation, psychotic and neurotic disorder, and developmental impairments" are more likely to enroll than children with other qualifying diagnoses or no diagnoses at all. Having adjusted for diagnosis type, the number of qualifying diagnoses (one qualifying diagnosis, more than one) seems to have a negative effect on enrollment.

Prior utilization: probability of use in the pre-anchor month date window: The use of any Medicaid-reimbursed services in the pre-anchor window is strongly associated with enrollment. A 1 percent increase in the percentage of eligible children with any Medicaid utilization would be associated with a 27.4 percentage point increase in the probability of enrollment. Neither inpatient nor long-term care/residential treatment use utilization have any effects independent of total use.

Prior utilization: amount of use in the pre-anchor month window: Total Medicaid expenditure in the pre-anchor window has no effect on the probability of enrollment, but expenditures on inpatient care and on long-term care/residential treatment are associated with small but statistically significant negative effects on enrollment. For example, a $100 increase in average Medicaid inpatient expenditure per month would be associated with a decrease of .006 percentage points in the probability of enrollment.

Selection at Disenrollment. A logistic regression estimated on all participants (N=2,176) that predicted disenrollment fit the data less closely than the enrollment regression (chi-square score = 90.6, p=.0001, a pseudo R² of .042). Estimated parameters, in probability form, are shown in Exhibit 3.12.

Age. Older children are less likely to enroll, and more likely to disenroll, than younger children. A 1 percent increase in the proportion of participants 18 years and older is associated with a nine percentage point increase in probability of disenrollment.

Gender. Although gender has no significant association with enrollment, being female increases the probability of disenrollment, by about 4.4 percentage points for every 1 percent increase in the proportion of female participants.

Race. Race does not significantly affect disenrollment, though the sign of the parameter estimate for race is negative, suggesting a tendency toward less frequent disenrollment among African Americans.

Diagnosis. All diagnosis parameter estimates are statistically significant. Having a diagnosis in the category "mental retardation, psychotic and neurotic disorder, and developmental impairments" reduces the probability of disenrollment by 31 percentage points.³⁵

EXHIBIT 3.12. Enrollment/Selection Analysis: Estimated Effects of Independent Variables on Probability of Disenrollment
Variable	Parameter¹	Pr (chi-square)
AGECAT3	.114	.6538
AGECAT4	.220	.4121
AGECAT5	.655	.0136
FEM	.317	.0092
RACECAT1	(.130)	.5971
DDX_A	(.311)	.0188
DX_PREC1	.310	.0585
DX_PREC2	.621	.0004
DMIQX_0	(.249)	.1566
DMIQXI_0	.402	.0750
DMIQXR_0	.919	.1989
MIQX_0	(.00001)	.8155
MIQXI_0	(.00006)	.3611
MIQXR_0	(.00003)	.4869
DMSSI6_0	.281	.4268
DSAMP1	2.156	.0003
DSAMP2	2.070	.0006
DSAMP3	1.512	.0133
DSAMP4	1.573	.0114
Source: HSCSN/DC Medicaid Enrollment/Claims Data Base-2 Log L = .1851 Pseudo R²=.0421 Negative values in parentheses

Prior utilization, expenditure and eligibility: As might be expected, health services utilization and SSI-eligibility status in the six-month period before enrollment have diminished influence on later events, particularly the probability of disenrollment. Only the probability of pre-anchor inpatient use correlates significantly with disenrollment. A 1 percent increase in the probability of pre-anchor inpatient use is associated with a 5.5 percentage point increase in the probability of disenrollment.

F. Conclusions and Implications of the Enrollment and Selection Analysis

At a given capitation rate, the volume and composition of enrollment contributes substantially to the financial viability, or non-viability, of a health plan. When planning for the Demonstration, HSCSN operated on the assumption that enrollment of the target population would be mandatory, ensuring both the largest and most balanced risk pool for the District. In fact, after negotiations with community groups, mandated enrollment was dropped in favor of a wholly voluntary program. As this and the preceding chapter have shown, HSCSN faced unanticipated enrollment challenges that contributed both to the financial difficulties and to HSCSN management perspectives on requirements for future waivers.

1. HSCSN's proactive enrollment process succeeded in enrolling eligible children early in the Demonstration, but at a substantial cost.

Over a four-year period, HSCSN enrolled nearly half of the eligible children in DC. Because lists of eligible children provided by DC Medicaid were often incomplete and inaccurate, HSCSN experienced serious difficulties (and unforeseen costs) in identifying and contacting potential participants throughout the Demonstration. Initially, HSCSN relied heavily on mailings to families of potentially eligible children -- followed by direct marketing efforts by outreach workers -- that did not respond positively to the mailings within 45 days (later reduced to ten days). Mailings were eliminated in 1998, following management changes at HSCSN.

After an initial period of accelerated growth in enrollments, from start-up in February 1996 into the Fall of 1996, the rate of increase slowed. The largest subsequent increases in enrollment were associated with two large HSCSN mailings, in January 1997 and August 1997. Whether or not additional mailings could have had the same effects as the earlier mailings, reviving what was clearly a declining growth rate, is unclear. What does seem clear is that outreach paid off in enrollment growth, although there was no evidence that HSCSN ever assessed the costs and benefits of intensive outreach. Enrollment costs were simply perceived as excessive.

Disenrollments from the program averaged 8 percent of total enrollments. From start-up in 1996 through mid-1999, 583 children disenrolled from the Demonstration. There is no evidence that disenrollment was associated with serious dissatisfaction. Over half had either moved out of the District or "aged out" of the Demonstration. Seventy participants lost SSI eligibility. For only 98 children, caregivers had "changed their minds" about participation.

2. The Demonstration is voluntary. Families and children who enroll are different from those who do not enroll in several ways. This has affected HSCSN performance and financial viability, and has also complicated the task of assessing impacts of the Demonstration.

As noted, the Demonstration was planned as a mandatory program for all SSI-eligible children with special needs in the District. In response to stakeholder pressure, HSCSN and DC Medicaid implemented a voluntary Demonstration. As reported in Chapter 2, management perceives the relatively small number of participants as contributing to the financial difficulties faced by HSCSN in most years of the Demonstration. This decision also affected the evaluation design. First, it made available an internal comparison group of SSI-eligible children who elected not to participate. Second, by introducing a potential for bias due to self-selection in enrollment, it complicated the evaluator's job of assessing program impacts through comparisons of participants with non-participants.

To assess the extent of selection bias in enrollment and in disenrollment from the program, the evaluator compared several measures derived from Medicaid claims and eligibility data between Demonstration participants and non-participants. These comparisons reveal that, compared to non-participants, participants are younger, more likely to be male, more likely to be African American, more likely to have mental and developmental impairments, more likely to have been SSI-enrolled before the Demonstration and more likely to have multiple conditions diagnosed before enrollment. Conversely, disenrollment is higher among older children, female children, non-African American children, and children with diagnoses other than mental and developmental impairment.

There is no evidence that HSCSN has enrolled selectively from the least-impaired among children with special needs. Participants seem to be heavier users of Medicaid-reimbursed services before enrollment than non-participants, over a similar time period. This suggests that the Demonstration might have experienced adverse selection in enrollment -- Demonstration participants are "sicker" than non-participants. However, the facts do not necessarily support only this interpretation. Although participants use more services, data show that non-participants who use similar services incur higher expenditures. In addition, the higher frequency of diagnoses for participants might correlate with acuity, but might also reflect extensive use: the more contacts a child has with the medical system, the more likely is that child to have a relevant diagnosis recorded. Therefore, instead of (or in addition to) the acuity interpretation of adverse selection, these findings might support a "sophisticated consumer" hypothesis. That is, participants are more likely than non-participants to be long-term SSI program members, and are from families whose relatively extensive contacts with medical and social service providers have given them the knowledge needed to take advantage of the available services. Findings from the survey reported below tend to suggest that differential selection based both on acuity and consumer sophistication may be at work.

If mandatory enrollment (or voluntary enrollment with a substantial outreach effort), combined with more restricted provider choice, are features that are necessary to make a program like HSCSN viable, then the assumption that this model of managed care will work for highly vulnerable populations deserves reexamination. Under the best of circumstances, children with special needs are a population with highly diverse needs. A small number of utilization and expenditure outliers in a risk pool of 2,000 individuals can have devastating effects on program viability.

Options to expand the risk pool (to include presumably healthier foster children or to require enrollment of all SSI-eligible children), to restrict provider choice, and protect against damaging financial losses (through stop-loss protection) have been discussed in negotiations over a new waiver for the Demonstration, as described in more detail in Chapter 2. However, based on the most recent waiver request for proposal, foster children will not be eligible to participate. Mandatory enrollment is not a certain option, though it has been proposed. Restrictions on provider choice, in addition to hindering voluntary enrollment from a population that values regular provider attachments (see Chapter 5), also runs afoul of current disability legislation. Stop-loss provisions offer some financial protection, but managing a modified capitation system increases Medicaid and plan administrative costs. In addition, the perceived need for stop-loss calls into question the assumption that capitation is a viable financial model for children with special health care needs.

IV. CARE MANAGEMENT AND PROVIDER FEEDBACK

A. HSCSN Care Management: Goals/Objectives/Structure/Process

DC's Demonstration project is a voluntary, unique, managed care plan available to any child living in the District who participates in both DC Medicaid and the federally funded SSI program. One of its most distinguishing features is the assignment of a care manager to every participant to help coordinate and negotiate all of the services a participant might need. Care management includes:

24-hour care management services while enrolled;
Development and maintenance of individualized plans of treatment (POTs);
Assistance in arranging all medically necessary services as well as transportation, medical appointments, and all other covered services.

The Demonstration is built on the premise that care management is the most important benefit of the program. With an emphasis on primary care, coordination of services, and supportive services, the following goals were established at the onset of the Demonstration:

Fewer visits to the emergency room and fewer acute episodes will be experienced by children who previously had no usual source of primary care;
Duplication of services and the provision of unnecessary services will be avoided;
The provision of more appropriate services will result from improved knowledge of patient's needs; and
The maximum benefit of medical care will result from the provision of supportive services and the resolution of non-medical issues.

The Demonstration employs a unique staffing model to meet the needs of its participants. Four care management teams are responsible for coordinating the care of all participants in the program. Each team consists of (in order of years of experience) one team leader (TL), three to four care managers (CM) and four to six care manager associates (CMA). Each individual on a team has his/her own caseload. The most complex cases go to the most experienced care managers (team leaders), who consequently, have the fewest cases. Similarly, the least complex cases are assigned to the CMAs who maintain the largest caseloads. Throughout this chapter, except when the distinctions are important, anyone who provides care management is referred to as a care manager.

While each child/family is assigned to a care manager, there is team collaboration through regularly scheduled team meetings to review cases. Occasionally, a participant's needs may change in complexity, in which case it is passed up or down within a team as appropriate. Regardless of the level of the care manager, the care management job is formally defined as to:

Assist in the access, promotion, and maintenance of high quality care through the location, coordination, and evaluation of health services provided to participants;
Monitor delivery of health services to participants for effective and efficient utilization;
Assure that medical services are provided to participants at the appropriate level of care;
Assure that each participant's individualized optimal health level is achieved and maintained;
Facilitate access to all appropriate preventative health services; and
Facilitate health education and community involvement activities that will empower families to become self-reliant.

The remainder of this chapter reviews the operation and evolution of the Demonstration care management system, as perceived by HSCSN management, care managers, providers and families. Section IV.B focuses on organizational changes, reported by program administrators and care managers, that have affected care management. Section IV.C summarizes care manager perceptions of the process of care. Section IV.D describes perceptions articulated by Demonstration providers of HSCSN and their interactions with care managers. Section IV.E represents findings from a review of 102 case records that describe documentary evidence for integration, coordination, and appropriateness of services under HSCSN care management. Section IV.F reviews findings from focus groups conducted with families of participants. Section IV.G presents conclusions.

Findings from these analyses include:

All stakeholders generally share the perception that care management is the most valuable feature of the Demonstration.
Communication, between providers and care managers, between care managers and organizations outside HSCSN, and between care managers and families, remains a key problem that limits the effectiveness of care coordination and integration.
In response to concerns about overburdened care managers, a factor cited in complaints about inadequate communications, HSCSN has made organizational and technological changes. Most notably, staff have been added to facilitate outreach, conduct utilization review, and expand the number of available care managers.
Care managers have eased the burden on families and informal caregivers by assuming many of the logistical functions of coordinating care. Care managers also appear to have been successful in educating families about the appropriate use of emergency rooms and the value of preventive services.

B. Evolution of Care Management/Care Managers' Roles and Perceptions During the Demonstration

In September 1999, for the third time during the Demonstration, the evaluator conducted individual interviews with senior staff and group interviews with care managers at HSCSN. These included interviews with the Acting Medical Director, Chief Operating Officer, Director of Care Management, and each of the four team leaders.³⁶ The purpose of the interviews was to follow up on what had been learned in previous years' interviews and to provide senior staff the opportunity to convey new information and perceptions.

The evaluator conducted three interviews of groups comprised of care managers and care manager associates. This allowed the frontline workers to share their first-hand experiences working with participants. It also afforded an opportunity to learn about the perceived strengths and weaknesses of the Demonstration.

Organizational Change: Director of Care Management

Care coordination is one of the distinguishing features of the Demonstration. Over the three years of the project, the manner in which care coordination has been administered has become more streamlined. In March 1998, a new Director of Care Management was hired. The new Director came to HSCSN with several years of experience in managed care and a strong focus on using data to facilitate care management. Since she assumed this role, HSCSN has instituted a more systematic approach to care management by codifying and improving procedures already in place.

One of the changes made by the new Director was a reduction of some unnecessary positions and the addition of staff to help overburdened care managers. Five individuals from HSC were transferred to assist HSCSN with heavy caseloads. Four of the five eventually became CMs. The fifth individual became a clinical liaison nurse whose primary task was to conduct utilization review of the highest risk/highest cost cases and medical records. As discussed below, a fourth care management team was created to reduce the number of CMs/CMAs per team, and caseloads were redistributed to become more manageable.

The Director of Care Management also implemented new standards to increase accountability. One reported effect has been a more timely completion of POTs and medical records, making care management more efficient. In March 1998, the POT policy became more stringent, requiring that POTs be completed on a monthly basis and updated every six months. In addition to these changes in the POTs, HSCSN established additional committees (explained below) to ensure that other standards are being met. The new Director also instituted weekly review meetings where TLs/CMs/CMAs report a summary of findings on their caseloads.

Finally, all residential caseloads have recently been assigned to one CM. This was done in an attempt to better monitor out-of-state facilities and participants in residential treatment. For the first time, in March 1999, money was allocated for the CM to visit out-of-state facilities. In addition, recognizing that residential care is expensive and reintegration difficult, a residential committee was established in November 1998 to review participants at risk of placement, and discuss participants in placement, as well as those participants who may soon be released.

Changes in Utilization Review Procedures

Recently, in order to reduce workloads of the care management teams, HSCSN has implemented changes in its approach to utilization review (UR). Originally, CMs were assigned the task of UR, using InterQual, a software program (no longer utilized at HSCSN) to determine if participants met criteria for various treatments. CMs would contact hospitals to review medical cases and authorizations for care. However, hospitals did not communicate well with CMs. This lack of communication compromised HSCSN's ability to effectively collect health care utilization data. As a result, a position was created to manage the UR function. Eventually, a TL took over this role and worked to put systems into place to track health care utilization. Prior to this effort, there were no formal policies and procedures to make these data accessible. The TL focused on one variable, length of hospital stay, and discovered much variance between different sites of care.

Over time, HSCSN realized that the UR function was taking the TL away from necessary care management duties. To address this situation, a clinical liaison nurse was hired to assume all UR functions and hospital oversight. The liaison nurse has worked to avoid service delays, anticipate discharges and collect information on in-patient and ER utilizations as well as ensure that CMs are up-to-date on patient needs. HSCSN believes that having a designated liaison on site who reviews medical records and is familiar with what is happening at the hospitals has been beneficial in terms of gathering data and monitoring utilization. Interestingly, some CMs believe that it is a disadvantage to be no longer part of the UR process.

Organizational Efforts to Monitor Care Management and Quality

HSCSN has created several committees in an effort to provide more cost-efficient care and tighten procedures. A Benefits Utilization Committee (BUC), established in November 1998 to review all questionable costs, has worked to develop guidelines for commonly reoccurring authorizations. Members of this committee include the Vice-President, the Chief Operating Officer, Department Directors and Team Leaders.

Another committee, established in the summer of 1998, is the Performance Assessment and Improvement Committee (PAI), now renamed the HSCSN Quality Council. The Council is made up of external physicians and other providers, as well as some internal HSCSN staff. The purpose of the committee has been to examine alternative quality improvement mechanisms (i.e., arranging transportation, timely authorizations, payment of claims, and providers billing correctly). The PAI committee meets every other month and is in the process of being revised to integrate quality improvement within programs and operations of HSCSN.

The Operations Committee, now known as the Performance and Outcomes Improvement Committee (POIC), is a monthly interdepartmental meeting that includes the Chief Operating Officer, Medical Director, Director of Care Management, Co-Directors of Quality Management, department heads from Claims, Outreach, Provider Affairs, and Member Services, Care Management Team Leaders, the Comptroller, Office Manager, the Coordinator for Information Services and the Vice President of Finance, Information and Managed Care Services. The purpose of this committee is to maintain fiscal integrity, improve processes related to the collection and analysis of data findings, continue to improve participant health status and health outcomes, and work to improve the planning and delivery of care coordination services.

Information Technology and Care Management: HSCare

HSCSN acknowledged the need for increased standardized, automated technology to support care management. HSCSN has designed databases to extract information on the access, quality and cost of health care. In June 1998, HSCSN changed their tracking system from Lotus Notes to a new software program, HSCare. This program, designed specifically for HSCSN, has significantly augmented HSCSN's capacity for storing and accessing information. HSCare has allowed for information to be processed more efficiently, and increased access to information and easy on-line documentation.

HSCSN has also improved the tracking system for participants. In an attempt to boost the EPSDT utilization rate for participants, HSCSN developed a tracking system, using claims, that produces a monthly list of participants who will be due for an EPSDT check-up. CMs will be able to see if a visit occurred by examining claims data from encounters. HSCSN is now working on a curriculum to make physicians familiar with this tracking system so that CMs can follow-up and track EPSDT visits accordingly.

The HSCare computer system has enabled better tracking of data overall and increased access to information. Information that was once available only through patient charts can now be accessed through the computer. TLs/CMs/CMAs reported that information has been processed more efficiently and that it has been easier to follow-up on participants using HSCare, compared to Lotus Notes. In an effort to make all systems reproducible and codified, HSCSN information technology and clinical staff have been meeting together to decide what variables to examine.

Despite these improvements, however, the system cannot profile costs and utilization at the patient or provider level. HSCSN has been able to capture much of the mid-level and top utilization data such as aggregate hospital and EPSDT utilization. Lower level data capture, however, such as pharmacy costs, and how much a participant costs per year, continues to be inaccessible. For instance, the data cannot reveal how often a physician referred X patient for a CT-scan, but the data can reveal how many patients overall had CT-scans per year. Additionally, HSCSN recognizes the need for long-term data to examine outcomes and cross-comparisons over time. HSCSN has stated that, if the program continues, there will be discussions about changing the system so that it has the capacity to generate participant-level data.

Policies Regarding Plans of Treatment

As mentioned above, care managers report that POTs are completed in a more timely fashion due to new policies and procedures. However, formal guidelines on POT procedures are needed. For instance, it is still difficult to determine if the PCP reads the POT unless the PCP returns the document with comments or calls to discuss the POT. CMs suggested that computerization would facilitate monitoring the POTs.

Efforts to Reduce Care Manager Work Loads

Before adding a fourth care management team, TLs carried a caseload of approximately 65 cases. Also, at this time, CMs carried approximately 100 cases, while CMAs carried approximately 150 cases.

Despite reduced caseloads, the evaluator found that CMs and CMAs still felt overburdened. Since the fourth team was added, TLs have been carrying caseloads of approximately 6 to 21 cases. CMs have been carrying caseloads of approximately 58-68 cases, and CMAs have been carrying a maximum of 110 cases. CMs reasoned that caseloads need to be further reduced, because their patient population is very reactive and requires many wrap-around services. CMs reported that this population has a hard time planning ahead and is often in a state of crisis.

CMs reported that working at crises management rather than prevention and education detracts from the objectives of care management. CMs noted that accountability has also continued to be a challenge to care management. It has been hard to coordinate so many players without well-defined roles; sometimes, there have been up to ten people trying to coordinate care.

CMs/CMAs have continued to spend much of their time addressing multiple non-medical issues. Paperwork, documentation, and "hidden caseloads", where HSCSN has helped siblings of participants who did not qualify for enrollment, were cited as contributing factors. Due to the volume of non-medical issues, CMs and CMAs have occasionally felt more like a social service agency than an insurance company that provides case management and care coordination. However, care managers at all levels agreed that these social issues must be addressed so that a child can benefit from a stable home environment.

Despite concerns relating to the volume of their caseloads, CMs stated that it is difficult to "download" a case to another CM because of the rapport they have established with the family. However, downloading may occur when:

A participant's physical or mental condition dramatically changes;
A participant's condition is misdiagnosed;
Additional diagnoses are uncovered that were not initially identified;
A CM or CMA leaves; or,
A member requests another CM.

When cases are shifted to another CM, every effort is made to keep the case within the same team.

Policies to Improve Monitoring of Residential Care Management

Investigators discovered significant improvements in care management for participants in residential treatment. First, in September 1998, a masters-level CM at HSCSN took over all the residential treatment cases. Next, in an attempt to establish better oversight of these residential treatment facilities, HSCSN made funds available to the CM to visit out-of-state facilities. These visits allow the CM to have face-to-face interactions with health care providers to collect summary data and to have better opportunities for monitoring care appropriateness. Furthermore, while in the past only partial medical records were reviewed, the CM reviews entire medical records during these on-site visits.

Senior HSCSN staff commented that these improvements have resulted in more focused care. HSCSN has also learned the value of reintegration and the services that need to be in place when a participant comes home. To date, the residential CM has returned home eight participants, none of whom have had to be readmitted.

While there have been some significant improvements, the residential CM has noted three challenges that remain:

Finding the participant an appropriate accredited facility;
Ensuring accountability for participant care with out-of-state providers and agencies; and,
Reducing the volume of paperwork.

Reported Effects of Organizational Policies and Changes on Care Manager Effectiveness

As this section has shown, HSCSN has made changes to its care management system. Having fewer staff on each team and reduced caseloads has allowed TLs/CMs/CMAs to become more effective. CMs/CMAs at all levels have commented that there has been more comraderie and collaboration between team members. TLs reported that since their caseloads have been reduced, they now have time to mentor and work closely with their teams. Review meetings, where cases are presented, have allowed a forum for communication between upper level staff and frontline workers to examine which care management practices work and case outcomes.

TLs have also commented that there is more communication and involvement now with upper level staff. TLs stated that the Director of Care Management has empowered them by involving TLs in the decision-making processes related to care management. The residential CM also commented that since improvements were made to residential care management, HSCSN has learned the value of transition programs. Two positive results that have come from visiting out-of-state facilities are that 1) the residential CM has been able to better develop relationships with contacts outside of HSCSN, and 2) the CM has been able to monitor participants in residential treatment more effectively.

Also, TLs/CMs/CMAs reported that HSCSN's Outreach Department, in handling some of the social issues, has continued to be very helpful and complementary to care management. Despite these efforts, however, CMs/CMAs, have continued to be much more involved in the social services role than originally expected, which they believe has detracted from their responsibilities for coordinating medical care.

C. Processes of Care

Over the past year, neither initial assessments nor decision making processes have seen dramatic changes. However, HSCSN has made organizational changes to promote more direct reporting among the different levels of care managers and their staff. In particular, POTs are monitored on a monthly basis.³⁷

Referral and Authorization Process

The referral process has changed little over the Demonstration period. However, with the newly computerized recording system that replaced the manual system at the end of May 1998, HSCSN has the ability to track referrals and retrieve information in a much simpler, faster manner. This has simplified the time-consuming task of tracking down paper copies of referrals which are not always where they are supposed to be. However, communication with providers about authorized referrals continues to be a problem area for HSCSN. According to HSCSN, the receiving party often "misplaces or loses" the referral before the participant shows up for an appointment, requiring the referral form to be re-faxed. CMs reported that sometimes it can take up to ten faxes before the authorization ends up in the correct hands.

Coordination with Outside Agencies

CMs/CMAs reported continuing frustration with the coordination of services for participants when so many outside agencies are involved. CMs/CMAs commented that they often need to coordinate services with up to ten people, all from different agencies. These can include the participant's social worker, psychiatrist, school, PCP, home care worker and therapist. When coordinating care outside HSCSN, accountability issues are sometimes encountered. There is also more chance of disagreement on treatment plans. Additional frustrations include: duplication of services from lack of communication; failure of home care staff to show up for appointments (CMs believe that reimbursement rates contribute to this problem); and continued shortage of psychiatric technicians for respite care.

Communication with Specialists and PCPs

CMs at all levels stated that communication with PCPs has been satisfactory and that they are in frequent contact with the PCPs. However, CMs also noted that sometimes they know more about participants than the child's PCP because of their established relationship with the caregiver. This detailed knowledge and information that the CM has benefits the PCP who may not be able to obtain the information from participants, and it benefits the CM by keeping the PCP in close contact with the CM.

CMs at all levels reported that there has continued to be a shortage in the HSCSN network of dentists, mental health providers, child psychiatrists, physical therapists, speech therapists and occupational therapists. Care managers noted that policies around dental authorization could be more flexible. HSCSN has recognized that the slow credentialing process for new providers may also contribute to delays in recruiting new dentists. Further, CMs noted that dentists have reported that children with special health needs need twice as much work as normal children, and that sometimes they have been unprepared to treat children who are medically fragile or children with behavioral problems. Dentists have also commented to CMs/CMAs that it has been difficult to work within the restrictions of prior authorizations for specific treatments.³⁸ HSCSN is making attempts to reduce these restrictions and is trying to eliminate prior authorization for scaling and sealants when the service is part of the overall dental protocol.

Educating Participants on Appropriate Emergency Room (ER) Use

HSCSN has worked to educate participants and their families about proper emergency room use. CMs reported that, once caregivers understand the function of the emergency room, they tend to go to their PCP first. One CM commented that some of her asthma and sickle cell cases were used to going to the ER almost every day prior to enrollment at HSCSN. Now that these participants see their PCP regularly, CMs have noticed that some participants will call their PCP first rather than go directly to the ER. CMs noted that using member newsletters as an additional means to educate members about services and other matters has been effective.

HSCSN Relationships with Families

TLs/CMs/CMAs reported that they generally have good communication with families. CMs/CMAs expressed concern that some caregivers are too dependent on the CM, calling several times a day. CMs noted that many of these families tend to be reactive rather than proactive. This may require the care manager to deal with last minute crisis interventions for several participants in a given day. CMs/CMAs stated that they can easily receive up to 40 calls a day. CMs/CMAs further commented that they often end up assisting other members of the family and tending to the "social issues" of the family, which is very time-consuming and detracts from the being able to coordinate medical care.

Handling Important Transitions: Aging Out of the Program

HSCSN recognizes that aging out of the Demonstration can be problematic for participants and their families. CMs found that caregivers sometimes continue to call after participants have aged-out of HSCSN, probably attributable to the fact that there is no care coordination after transition to DC Medicaid. Caregivers often become reliant on care management and accustomed to working with one contact person for all their child's health care needs.

Investigators found that, while procedures are in place to prepare participants for transition back to DC Medicaid, there is variation among CMs/CMAs in regard to when they begin the transition process. CMs/CMAs start the transition process at any time from three months to one year prior to termination. CMs and CMAs stated that their goal has been to ensure that processes are put in place to allow participants to continue to receive care once they have become ineligible for HSCSN, and to try to help participants to become more independent whenever possible. However, CMs at all levels observed that since there is no adult equivalent of HSCSN, there will inevitably be a gap in some services for people with special needs once they become 22.

Care Management and the Costs of Services

Over the span of the Demonstration, HSCSN has gained a heightened awareness of cost issues, as evidenced by changes to incorporate cost-saving strategies into policies and procedures. For instance, HSCSN initially provided home modification (where a participant's residence is renovated to be handicapped accessible), following a needs assessment conducted by the National Rehabilitation Hospital (NRH). Where renovations were recommended, they were typically done regardless of cost. Now, however, HSCSN first considers other options before making expensive home modifications. When appropriate, mobile equipment is installed, as it can be moved from one home to another should a participant change residences.

Additionally, as noted above, HSCSN has established a Benefits Utilization Committee (BUC) to review all questionable cost issues and to identify and develop more clear cut policies for commonly occurring requests. The Medical Director attends all BUC meetings and continues to approve requests for medical necessity, such as durable medical equipment (DME) that exceed $2,500. Expenses, such as home modification, are usually presented to the BUC for authorization. Furthermore, in November 1998, addressing the continuing high costs of residential care, a committee was established to review all participants in residential care.³⁹ These changes have allowed HSCSN to have more effective oversight and monitoring of incurred and potential costs.

CMs have played an increasingly important role in HSCSN's cost-saving strategies. While CMs are more conscious of cost than earlier in the Demonstration, they clearly do not view it as their primary concern when rendering care.

TLs reasoned that, although care management may have been intended to contain costs, savings may be delayed. This is because HSCSN may have had to play "catch-up" for the first two years of the Demonstration, to fill in the gaps of care that many participants were missing. However, once unmet needs have been filled, HSCSN believes the care management policies and procedures will provide cost-efficiency and save money over time. The positive results of care coordination and management have been apparent in the reduction of emergency room visits and hospitalizations noted by CMs.

D. Provider Feedback: Changes in Role and Response to Care Management

For the third year in a row, investigators interviewed a sample of participating providers to obtain direct feedback regarding their experiences with serving children in the Demonstration. Whereas in prior years, interviews were conducted either in person or a combination of in-person and over the phone, the most recent round of interviews was conducted exclusively over the phone. This approach reduced the costs of interviewing and allowed for a more flexible schedule to reach providers over the course of several weeks.

The original group of providers who were first interviewed in 1997 were chosen based on the participants they served. Obtaining a cross-section of providers who served children in the following mix of categories was important:

Average users of services;
Above-average users of medical services (but not ancillary services);
Intense users, including ancillary services; and
Users of mental health services.

The sample included 21 providers with the following representation:⁴⁰

hospitals
clinics
group primary care practices
7 individual PCPs
physician specialty practices
non-physician providers of mental health services
DME/supplies
1 physical therapist
1 occupational therapist
1 dentist

Using the original list of 21 providers who granted Abt Associates interviews during the first round in 1997, providers were re-contacted in the summer of 1999 and asked if they would take part in a third (final) interview. Re-contacting as many providers as possible was important, as it allowed participating providers an opportunity to express not only their current feeling about the Demonstration, but also an opportunity to highlight changes over the course of their involvement. From the original group of 21, 13 providers granted interviews for the third round. Investigators were unable to obtain an interview with a dentist or a physical therapist during this third round. However, all other categories were represented. Overall, the findings are fairly consistent with years past, and similarly encouraging. There was rarely 100 percent consensus on a given topic. Common responses, as well as dramatic contrasts in responses, are highlighted below. Additionally, any changes providers noted from previous years are also highlighted.

The major themes that evolved from this third and final round of provider interviews were:

Cost of treatment is not a factor in prescribing treatment;
Treatment is not dependent on insurance type;
The majority of providers would join HSCSN again if asked today; and,
Providers feel that care managers contribute the distinguishing difference in the quality of care received by HSCSN patients vs. those with conventional Medicaid.

Referral, Authorization, and Reimbursement Processes

The referral process does not pose any particular challenges to medical providers participating in the Demonstration. The volume of paperwork generated for referrals and reimbursement submissions was perceived as manageable and reasonable. The benefits providers receive for doing the necessary paperwork were said to "outweigh the burden." For providers who were interviewed, there did not seem to be any obvious differences in the paperwork requirements between the HSCSN and traditional Medicaid.

The rate at which requested services are approved was generally perceived as being quite reasonable as well. Only one provider felt there was a higher denial rate with HSCSN than with traditional Medicaid. Another provider, a primary care physician, noted that the referral process is cumbersome for most services, excluding those of specialists, but still felt the overall process was tolerable.

Cost of treatment is not a factor that providers consider when addressing patients' needs. A child's needs are the primary consideration. Additionally, most providers feel that the fee schedule is fair and that reimbursement rates are greater than or equal to those of traditional Medicaid. In fact, compared to conventional Medicaid, collecting payment is easier in the Demonstration, and HSCSN covers a wider range of services. However, some payment issues did surface during the interviews. For example, there is no compensation for time spent writing up reports, a significant component of providing mental health services. Providers absorb the costs when this occurs. Another issue that displeased a mental health provider was the reimbursement disparity based on credentials. A more equitable reimbursement rate would be preferred. In general, specialists had more negative comments about reimbursement issues than did primary care physicians.

Patients and Caseloads

Many of the providers interviewed noted that children participating in the Demonstration have many problems. Overall, providers find them to be medically needier than children with conventional Medicaid. Providers typically have no knowledge of whether a child's "total" needs are being addressed, outside the provider's own specialty area.

Some providers noted that their HSCSN patient caseloads have remained fairly constant throughout the duration of their participation in the Demonstration, while others have noticed a steady increase. None of the interviewed providers turn away HSCSN patients who seek their services. HSCSN patients comprise anywhere from one to 60 percent of interviewed providers' total patient population.

Service Availability

The common perception of interviewed providers was that HSCSN makes more services available to children than does conventional Medicaid. The fact that mental health services are included at all within the Demonstration is an important point to highlight. However, providers noted gaps, particularly in arranging smooth transitions out of the program.

Processes of Care

Two providers observed that the POTs were useful. However, one doctor felt that revising them every six months is too frequent because of the large volume of paperwork it entails, making it easy to fall behind.

There was an overall sense of satisfaction with how decisions are made about a child's care. However, providers differed in their views of who should be the appropriate gatekeeper with respect to care decisions. The specialists seemed to be more inclined to think that role belongs to the care managers. What is clear is that there is no agreement between the various types of providers that there is an obvious gatekeeper.

For the most part, providers feel that care managers are very responsive and readily available when needed. Care management was frequently credited with addressing the comprehensive needs of the participant population it serves, unlike conventional Medicaid. One specialist noted that care managers are easier to deal with than pediatricians, from whom they regularly seek service approval in conventional Medicaid. Confusions raised by another provider were the high turnover of care managers and the large caseloads that pose challenges to the formation and maintenance of effective relationships with participants and their families. Interestingly, providers do not typically work in teams to discuss a Demonstration participant. The consultations between providers seem to be on a rare "as needed" basis rather than as regular, scheduled events. Most of the interviewed care providers stated they would join HSCSN again if asked today. They feel it is an important plan for children with special needs that addresses many components of a child's health through care management.

E. Case Record Review

1. Introduction

The case record review was included in the evaluation to address three major objectives. It was specifically intended to 1) determine if the needs of this special population are being adequately met; 2) assess the coordination of health care services by HSCSN; and 3) verify that provided care is accessible, comprehensive, coordinated, and continuous. Non-participants were not included in the sample due to budget constraints, both monetary and temporal. As a result, the findings from the sample of participants have no basis for comparison. Therefore, caution must be used in their interpretation. While this is a potential limitation to the study, it is believed that there is still knowledge to be gained by examining care management and its relationship to the overall care of this special needs population.

It is important to note here that, while the case record review was intended to test the utility and success of a care management component in the continuum of care for special needs children, it may, in fact, be measuring something else altogether. It is possible that the review simply measures the quality and maintenance of records kept by physicians and HSCSN rather than the care itself. This is a factor to take into account when interpreting the findings that follow.

2. Data and Methods

To perform the case record review, a tool was developed and tested to collect data that could be used to identify the ability of HSCSN to promote healthy physical and psychosocial development for the population served by the health plan (see Appendix F). A sample of 102 records was drawn using current enrollment data. Records were selected based on maintaining a sample that was representative of the most common diagnoses served by the health plan. As a result, the majority of participants whose records were included in the case record review had one of the following primary diagnoses: (1) mental retardation; (2) learning disability; (3) respiratory problems; (4) cerebral palsy; 5) HIV/AIDS.⁴¹

For each participant who was selected for the case record review, records from the child's primary care physician and HSCSN care management records were collected. Primary care physicians were asked to submit all pages of the requested child's record from January 1, 1996 through the date of the request. They were asked to include all information on the child's presenting condition and diagnoses, results of routine tests and exams, lab orders and results, referral orders and results, progress notes, information on hospitalizations and visits to the emergency room, and any other information that was collected in the stated time frame. Similarly, HSCSN was asked to send their files for the same participants so that a joint medical/care management file could be created for each child included in the sample. Specifically, HSCSN was asked to forward the initial assessments, all POTs from enrollment to the present (or disenrollment), sections of the child's "desk files" including care management notes, encounter/inpatient notes, referrals/authorizations, correspondences, and specific sections of the communications logs (see Appendix G).

Performing the record review were two Abt employees with clinical nursing backgrounds. It took the nurse reviewers an average of 2.5 hours to complete a review tool for each participant's combined file. However, the range for each completed file was as little as three-quarters of an hour to as many as eight hours, the latter only occurring on two occasions. Once the record review was completed, responses were coded and entered into a database entry program that was designed using Microsoft Access 97 to input the data items collected on the reviewed records. Data field validation was performed during the entry process on the categorical variables. What follows are highlights of the findings that resulted from the analyses and the hypotheses that preceded them.

3. Findings

Findings from reviews of case records suggest that indeed, care management is working, but there is room for improvement in the process (or improved record keeping). The POTs appear to be useful, although goals could be stated more clearly. PCP signatures on POTs could be more frequently obtained, and the prevalence with which they are found in the PCP records could improve. Appropriate services were identified to meet the needs of participants for the majority of the sample, although actual delivery of these services was less clearly documented. Finally, the inappropriate use of emergency rooms appears to decrease as participants continue to stay enrolled in HSCSN. These and other related findings will be detailed below.

The hypotheses and related findings are presented below, grouped by the original objective to which they relate.

OBJECTIVE: To determine if the needs of children with special needs are being adequately met:

Treatment goals will be clearly stated in POTs

To help increase the likelihood that the needs of a patient are addressed, it is useful to establish treatment goals that are clear, reasonable, and available to all persons involved in the care of an individual. While there is no assurance that stated treatment goals will be followed, goals provide a framework for identifying patient needs and tracking the subsequent provision of care.

Treatment goals were stated in POTs for 49 of the participants in the sample. They were not clear for 52 of the participants, and one participant had no treatment goals stated at all. Reasons for lack of clarity recorded by the nurse record reviewers fall into four categories. Twenty-one participants had unclear POT goals because they were not individualized. For example, "EPSDT not up-to-date" does not state a goal but rather the issue that needs to be improved. Another seven participants had unclear POT goals as a result of being vague or not very descriptive. For example, one goal was to "make significant gains in dressing," leaving much room for interpretation of "significant gains." Another nine participants had unclear goals that were the result of an obvious health concern that was not addressed in the POT. Finally, twelve participants had unclear treatment goals for undetermined reasons.

POTs will reflect a change in services prescribed, as participant's needs change

Health is not static, particularly in this study population. Therefore, it is important to recognize that, as health conditions change, so do the need for health services. For 77 children, records recorded changes in condition. Of these 77, evidence of changes in planned services was noted in the POTs of 58 participants. Nineteen participants' POTs did not indicate a change in planned services to correspond to changes in the participant's condition. Finally, no changes were necessary for 24 participants.

EXHIBIT 4.1. Explanations for Unclear Treatment Goals in POTs
	Frequency of Response (N)
Not Individual to Case	21
Goals Vague	7
Obvious Omission	9
No Answer	12
Total	49
Source: Case Record Review File, 1999

EXHIBIT 4.2. Changes in Condition and Evidence of Planned Service Changes
	N
Change in condition noted	77
Evidence of changes in planned services	58
No evidence of change in planned services	19
No noted changes in condition	24
Source: Case Record Review File, 1999

PCP signatures will be on all POTs

Physician signatures on treatment plans are intended to provide an indication that the tending physician is familiar with and agrees with the POT. While a signature is no guarantee of a physician's having read the treatment plan, it provides some indication of the responsibility assumed by the physician. In this study, primary care physicians' signatures were found on all of the POTs in the files of only five participants. In contrast, the POTs of 17 participants in the sample did not have any PCP signatures on file. However, for the remaining 80 participants, PCP signatures were found on some of the POTs in a child's file. On average, 46 percent of all the POTs, or almost half of the POTs in a participant's file, contained PCP signatures.⁴²

POTs will be located in both PCP and HSCSN records

The POTs are important because they provide a blueprint for the care of a participant. Copies of the POTs should be available in both the PCP records and HSCSN records so that these two key players will more likely be on the same page with respect to the treatment of participants. For 79 out of 102 sampled participants, no POTs were found in the PCP records. This was true regardless of how long a participant had been enrolled in the Demonstration. At the other extreme, all POTs for a given participant were located in PCP records for two participants in the sample. The percentage of POTs found in the PCP medical files for the remaining 23 participants were fairly evenly distributed from 17 to 83 percent.

POTs were much more likely to be found in the HSCSN care management records. In fact, for 94 participants in the sample, all POTs were located in the care management records. For one participant, no POTs were found in care management records, in drastic contrast to the PCPs records, where the absence of all POTs was much more frequent. For the remaining participants in the sample, three were missing one POT and one was missing five POTs from the CM records.

POTs will include services to meet the needs of the participant identified through the initial assessment

Based on the initial assessment, the number of participants with one or more of 16 impairments was measured. These 16 impairment categories were taken directly from HSCSN's initial assessment form. The results were compared with the number of participants who received services to address these impairments. This is conveyed in Exhibit 4.3, which also includes the number of participants for whom a health impairment was resolved without further action. The results indicate that the frequency with which needed impairments were not addressed ranged from zero percent to 71 percent.

OBJECTIVE: To assess the coordination of health services by HSCSN, Inc.:

EPSDT guidelines for prevention services will be appropriately followed

With the use of POTs and coordinated care provided by care managers to participants in the Demonstration, EPSDT guidelines were expected to be followed rigorously. In 1996, national averages for all children who qualify for EPSDT ranged from 13 percent for hearing exams to 21 percent for dental exams.⁴³ Indeed, in the case record sample, Demonstration participants had higher EPSDT utilization rates, ranging from a low of 60 percent for dental exams to a high of 82 percent for physical exams. These rates are likely to be even higher than indicated by this analysis, because 4-10 percent of participants' exam schedules could not be determined. Exhibit 4.4 illustrates the EPSDT utilization rates in the Demonstration.

EXHIBIT 4.3. Noted Service Needs and Service Provision for Specific Health Issues
Nature Of Health Issue	# ParticipantsAssessedWithImpairment	ParticipantsWithServicesNoted	Participants WithNo ServicesNeeded (IssueResolved)	ParticipantsWith ServicesNot Addressed
Visual	15	10	5	0 (0%)
Hearing	7	1	1	5 (71%)
Communication	37	25	3	12 (32%)
Cardiovascular	6	3	1	2 (33%)
GID	7	7	0	0 (0%)
Bowel/Blad/Kidney	9	2	1	6 (67%)
Respiratory	31	13	10	8 (26%)
ADLs	26	18	6	2 (8%)
Special Equipment	20	17	0	3 (14%)
Mental/Emot/Behav	45	33	1	11 (24%)
Autism	0	NA	NA	NA
Orthopedic	11	8	0	3 (27%)
Serious Emot Disturbance	15	9	0	6 (40%)
Language	35	21	2	12 (34%)
Traumatic Brain Injury	1	1	0	0 (0%)
Other Health Probs	10	7	1	2 (20%)
Source: Case Record Review File, 1999

EXHIBIT 4.4. Rates of Appropriately Followed EPSDT Guidelines for Vision, Hearing, Dental and Physical Examinations
	Vision	Hearing	Dental	Physical
Timing of exam met EPSDT Guideline	75%	73%	60%	82%
Timing of exam did not meet EPSDT	18%	18%	28%	13%
Unknown	5%	6%	10%	4%
Source: Case Record Review File, 1999

Inappropriate ER use will decrease as participants are enrolled for longer periods of time

One of the goals of the Demonstration was to reduce inappropriate use of the emergency room (ER) as a source of primary care. To assess the achievement of this goal, the number of ER visits per child, as well as the reasons for seeking care through the ER over the course of the Demonstration project, were examined by the record abstractors. Exhibit 4.5 shows the number of visits made by the participants and the acuity of each visit.

Over the course of the Demonstration, as participants had more ER visits, they were less likely to be labeled non-urgent, suggesting more appropriate use of the ER as a participant continued in the Demonstration. There was also a noted reduction in ER visits that were considered emergency in nature, possibly suggesting that the Demonstration was successful in placing an emphasis on prevention services. Participants' enrollment date did not appear to have any effect on how the ER was used for care. The number of participants with one ER visit was 50 out of the sample population of 102.

As Exhibit 4.5 reveals, the overall number of multiple ER visits per child declines at a fairly constant rate after the first visit. While the non-urgent visits declined from the first visit to the last ER visit recorded, it was not at the same steady rate noted in the visits that were labeled "emergency" or "urgent" in nature. Nonetheless, a notable reduction can be seen.

EXHIBIT 4.5. Acuity of Emergency Room Visits: Emergency, Urgent and Non-Urgent First and Subsequent Visits Per Child: unavailable at the time of HTML conversion--will be added at a later date.

OBJECTIVE: To verify that care is accessible, comprehensive, coordinated, and continuous:

Prescribed services will be delivered

In addressing the health issues of children with special needs, identifying what services to provide is only one part of the continuum of care. Ensuring that those services are actually delivered is critical to the well-being of this population. The case record review provides some insight into understanding the frequency with which prescribed services are delivered. Exhibit 4.6 illustrates the findings from this review.

EXHIBIT 4.6. Delivery of Prescribed Services by Type of Service
	Frequency ofPrescription	Delivered	NotDelivered	StatusUnknown
PCP Services	526	253 (48%)	23 (5%)	254 (48%)
Dental Services	516	108 (21%)	15 (3%)	393 (76%)
Specialist Services	375	158 (42%)	16 (4%)	201 (55%)
Medications	186	93 (50%)	----	93 (50%)
Nutrition Services	61	40 (66%)	----	20 (33%)
Lab Services	399	120 (30%)	23 (6%)	256 (64%)
Immunizations	353	107 (30%)	21 (6%)	225 (64%)
Home Care/Nursing	253	61 (24%)	8 (3%)	184 (73%)
Therapy	270	76 (28%)	5 (2%)	189 (70%)
Hospitalization	30	24 (80%)	1 (3%)	5 (17%)
Transportation Services	488	220 (45%)	5 (1%)	263 (54%)
Durable Medical Equipment	53	23 (43%)	----	30 (57%)
Disposable Medical Supplies	71	62 (87%)	----	9 (13%)
Personal/Environmental Adaptive Equipment	59	37 (63%)	1 (2%)	21 (35%)
Emergency Services	20	11 (55%)	----	9 (45%)
Other Outpatient Services	37	14 (38%)	----	23 (62%)
Educational Services	417	154 (37%)	5 (1%)	258 (62%)
Mental Health Services	46	16 (35%)	10 (22%)	20 (43%)
Community Referrals	61	21 (34%)	1 (2%)	39 (64%)
Source: Case Record Review File, 1999

The key to interpreting these findings is to understand that the nurse case reviewers were able to positively identify delivered services only some of the time. They labeled services as "not delivered" when there was actual proof in the record that a prescribed service was not received. There is a large portion of prescribed services that fall into the "unknown delivery" category, when there is insufficient documentation in the record. Therefore, the rates of both delivered and undelivered services are underestimated here. Evidence from the records suggested that services were delivered anywhere from as little as 21 percent of the time for dental services, to as much as 87 percent of the time for disposable medical supplies.

Referred services will be delivered

Evidence from the records suggested that referred services were delivered from 68 percent to 80 percent of the time. There was no evidence that services were delivered for 16 to 32 percent of referrals. Unlike the prescribed services mentioned, there were very few "unknown delivery" for these referred services, making these percentages more stable.

Evidence of timely response to participants' health problems will be noted

In addition to prescribing, referring, and delivering necessary health services, the length of time it takes practitioners to address particular health issues is also an important dimension of care management. Therefore, examining the timeliness of responses⁴⁴ was determined necessary. Having reviewed participants' records, the case reviewers became familiar with the health issues for each participant and how they were addressed. As Exhibit 4.7 shows, for 69 of the sampled population, there was evidence of timely responses for most or many of the participants' health problems. For 14 cases, there was timely action that occurred some of the time or few times. Finally, for 17 participants, there was rarely or never any evidence that problems were addressed in a timely fashion.

EXHIBIT 4.7. Evidence of Timely Response to Participants' Health Problems
Timely Action Rate	Frequency of Typeof Response
Evidence of timely action in most instances (90-100%)	55
Evidence of timely action in many instances (75-90%)	14
Evidence of timely action in some instances (50-75%)	12
Evidence of timely action in few instances (25-50%)	2
Evidence of timely action rare (25%)	9
No evidence	8
Source: Case Record Review File, 1999

Special education will be provided to participants with specific health issues

For particular health conditions, the provision of special education is part of the continuum of care. In the Demonstration case review sample, records were reviewed to determine which children with particular health issues were receiving special education (Exhibit 4.8). The findings indicate that special education was provided to children with specific health problems from seven to 82 percent of the time. The most common health problems that were addressed through special education were orthopedic in nature. Of 11 children with orthopedic disorders, nine received special education. At the other extreme were children with visual impairments. There were 15 in total in the sample population of which only one received special education. Additionally, the one child with traumatic brain injury did not receive any special education.⁴⁵

PCPs will be notified of the results of referrals

To determine the extent to which communication flows between specialists and primary care physicians, a measure of how often PCPs were notified of services provided by specialists was included in the abstraction instrument. The case reviewers found that out of a total of 402 referrals that were made for this sample population, there was documentary evidence that only 196, or 49 percent of the results, were conveyed to PCPs.

EXHIBIT 4.8. Frequency of Provision of Special Education to Caregiver/Child for Specific Health Issues
	Frequency ofCondition	Special EdProvided	PercentProvided
Autism	0	0	NA
Hearing	7	1	14%
Mental Retardation	45	20	44%
Orthopedic Impairments	11	9	82%
Serious Emotional Disturbance	15	10	67%
Specific Learning Disorder	45	24	53%
Speech/Language Impairments	35	20	57%
Traumatic Brain Injury	1	0	0%
Visual	15	1	7%
Other Health Problems	10	4	40%
Source: Case Record Review File, 1999

F. Families of Demonstration Participants

In 1997 and 1998, the evaluator conducted three rounds of focus groups, with each of the three involving five separate group discussion.⁴⁶ Round One included 38, Round Two included 36 and Round Three included 35 parents. Selection of parents was purposive, and based on primary conditions of their children, including mental retardation, respiratory conditions, cerebral palsy, mental disability/developmental disability, and terminal and/or acute illness. Although these focus groups were not repeated in 1999, it is important for completeness to summarize here the findings and conclusions that investigators drew from the earlier discussions. More detailed findings are presented in the Second Annual Report of the Evaluation (Abt Associates, 1998).

These sessions produced the following findings:

Early in the Demonstration, during the first focus groups, parents expressed their frustration with the burden of coordinating their child's care before enrollment, and expressed pleasure with the improvements and reduced burden achieved immediately after enrollment.
Parents in the first focus groups were impressed with the caring attitude of CMs and with aspects of the process that resulted in a more equitable balance between medical and non-medical services.
Later in the Demonstration, some parents expressed a variety of frustrations with the process, including
- difficulties contacting CMs, limited (mostly phone) contact, and in a few cases, no contact;
- inadequate CM follow-up on referrals;
- CM contact largely confined to crises or specific problems; and
- problems obtaining services/equipment due to poor communication with CMs.
Parents began reporting problems with CMs in the second round of focus groups, and the frequency of reported problems, as well as problem types, did not change between the second and third rounds.
Reports of satisfaction with individual CMs varied widely, with some parents expressing great enthusiasm and others deep disappointment with the same CM.

Although initial enthusiasm for the Demonstration dissipated somewhat as parents experienced day-to-day difficulties in using the care management system, most continued to recognize that their children were better off than they were before they enrolled. In particular, CMs were credited with reducing time parents had to invest in the logistics of coordinating their children's care.

G. Conclusion and Implications of the Care Management Analysis

Despite many problems, all major stakeholders value the care management system of the Demonstration.

1. The Demonstration implemented a unique care management staffing model to meet the needs of its participants.

Care management goals include gaining greater understanding of individual participants need, in order to (1) reduce the frequency of inappropriate use and duplication of services, and (2) to maximize the benefits of medical care by providing supportive services and resolving non-medical issues. To accomplish these goals, CMs assist in (1) locating, coordinating and evaluating health services, (2) monitoring the delivery of services to participants, (3) assuring that services are appropriate to participants' needs, (4) assuring the attainment and maintenance of every child's optimal level of health, (5) facilitating access to preventive services, and (6) facilitating health education and family involvement in community activities that will promote family self-reliance.

HSCSN organized four care management teams that are responsible for coordinating the care of all participants in the program, in collaboration with each child's PCP. Each team includes one TL, three to four CMs, and four to six CMAs. Fewer, and more complex, cases go to the team leader, while care manager associates carry larger numbers of less complex cases. Teams meet regularly. If the complexity of a participant's case changes, responsibility can be passed up or down the team hierarchy, as appropriate.

2. Inadequate communications between providers and CMs, between CMs and organizations outside HSCSN, and between CMs and families, have limited the effectiveness of care coordination and integration throughout the Demonstration.

In general, providers and CMs express satisfaction with their respective roles and with levels of communication. However, some providers observe that high turnover of CMs and heavy caseloads may keep CMs from forming effective relations with participants and families. Some providers express frustration with delays in the credentialing process and in obtaining prior authorizations.

Incomplete documentation is a source and symptom of poor communication, among members of the CM team and between CMs and providers. Reviews show that less than one-half of case records include clear treatment goals in the POT. PCP signatures are on all POTs for only five of 100 participants. In about three-quarters of records maintained by PCPs, no POTs are included. There is documentary evidence that the results of less than one-half of all referrals are conveyed to PCPs.

As amply documented in Chapter 2, the Demonstration has never successfully coordinated with other DC agencies that have medical or social service responsibilities for SSI-eligible children with special health needs. CMs express frustration with barriers they face in trying to coordinate with agencies outside HSCSN, many of which, like the school system, have never accepted the idea that care needs to be coordinated.

CMs see themselves as having generally good communication with families, but express several specific concerns. These include 1) excessively dependent caregivers, who monopolize a CM's time; 2) caregiver tendencies to be reactive rather than proactive, forcing CMs to spend excessive amounts of time in crisis interventions; 3) lack of consistent procedures for effecting a smooth transition when participants age out of the program; and 4) the extra burden of handling "hidden caseloads," social needs of other family members which, if not addressed, may impede the achievement of POT goals for the child.

Family perceptions of care management have changed somewhat as the Demonstration matures. Early in the Demonstration, having been frustrated with the barriers they faced operating, largely unsupported, in a fee-for-service environment, parents and other caregivers were grateful for reductions in burden experienced after enrollment and the "caring attitude" of HSCSN CMs. Over time, families express more dissatisfaction with CM communications, including limited contact, inadequate care manager follow-up on referrals, CM contact limited to crisis interventions, and specific problems of obtaining services or equipment.

In spite of these problems, families generally agree that CMs have eased their burdens by assuming many of the logistical functions of coordinating care. Although initial enthusiasm for the Demonstration has dissipated somewhat as parents experience day-to-day difficulties with the care management system, most continue to recognize that their children are better off in the Demonstration than they were before.

3. In response to concerns about overburdened CMs, a factor cited in complaints about inadequate communications, HSCSN has made organizational and technological changes.

Most notably, in 1998, (1) staff were added to facilitate outreach and conduct utilization review under a new Director of Care Management, (2) a new software system for tracking care provision and referrals was installed, (3) a fourth care management team was added, and (4) a CM was recruited to focus exclusively on residential care.

Despite organizational problems at the highest levels of HSCSN, the care management system in the Demonstration is much valued by providers, families and CMs. As these stakeholders note, however, there is still room for improvement through further increases in staff to reduce caseloads, added technological tools to support effective care management, and improved documentation of treatment needs and services. Had the Demonstration not been as troubled at the top, and had management and funds not been diverted to develop what ended as a failed care management software tool (described in greater detail in Chapter 2), progress on these improvements might have been more substantial.

The HSCSN care management system is a work in progress, but one with an uncertain future. When implemented, the Demonstration had no well-tested template available for designing an efficient and effective care management program. Despite organizational turmoil, management has learned on the job, recognizing and beginning to address barriers that hinder CMs and providers. However, since HSCSN appears reluctant to enter the competition for a contract under a new waiver, it is uncertain if the improvements initiated during the last years will be extended and strengthened.

Clearly, a longer and more stable experiment, in several sites, would provide a more reliable test of the value of this model. Nonetheless, most of the problems in communication and coordination encountered in this Demonstration are not unique to DC. HSCSN's experience offers useful lessons for other programs that serve children with special health care needs.

V. IMPACTS OF THE DEMONSTRATION

A. Introduction

Under a Medicaid waiver, HSCSN organized a program, financed by capitated Medicaid payments, to improve coordination and integration of care by linking eligible children with special health care needs and their families to care managers and primary care providers. Stakeholder feedback on the care management process has generally been positive, as reported in an earlier chapter. Therefore, in HSCSN, compared to the fee-for-service Medicaid environment in the District, children are expected to gain greater access to services that meet their diverse health care needs, and higher levels of satisfaction with access. Improved access should lead to improved health and function.

Expectations for the net cost and utilization impacts of this experiment are less clear, and may be quite different when viewed from the social, program and organizational perspectives. The social costs of caring for children with special health care needs include both direct (mostly Medicaid-subsidized) and indirect costs. Although Medicaid-eligible children and their families have access to a wide range of formal services free of charge, both in HSCSN and in the fee-for-service environment, families and other informal caregivers of these children may also face substantial indirect cost burdens in lost income, lost career opportunities, negative physical and emotional consequences of caregiving duties, and other effects. The Demonstration should be capable of reducing caregiver burden by lifting some of the burden of coordinating across disparate providers and services, and potentially by improving access to supportive services in the home. This evaluation provides some information on caregiver burden.

Medicaid program costs of the Demonstration would seem to be relatively predictable, determined by the monthly capitation payment rate and numbers of children enrolled in the Demonstration.⁴⁷ Payments to HSCSN may or may not be higher than Medicaid would have paid, had the participants not enrolled. If payment rates accurately reflect acuity and demand for services in the target population, in particular those in the population most likely to enroll in a voluntary Demonstration, then an objective test, controlled for differences between participants and eligible non-participants, should show the Demonstration to be budget-neutral for Medicaid. At best, rates should be set to encourage slight saving to the program, based on managed care efficiencies realized by HSCSN. However, since information on needs and utilization patterns of children with special health care needs is scarce, there is no a priori reason to believe that rates can be set accurately with all the desired incentives. Data problems make it impossible to estimate accurately what DC Medicaid would have paid for participants had they been in a fee-for-service setting. However, it is possible to compute the total amount Medicaid spent for children in the Demonstration and for children eligible but not enrolled in the Demonstration.

For HSCSN, the "true" organizational costs of providing services vary for each child, and depend on the different needs children bring when they enroll, actual service utilization rates in the Demonstration, the prices HSCSN must pay to secure these services, and the availability and strength of informal caregiving supports at home. If children enroll with unmet needs and pent-up demands for services, one might expect to see greater utilization, at least initially, than these children would have experienced in a fee-for-service environment. HSCSN has incentives to manage and coordinate services to fulfill children's real needs in the most appropriate and least costly setting. Comparing HSCSN to other Medicaid alternatives in the District, this might mean reductions in obviously inappropriate utilization (e.g., non-urgent emergency room visits) for HSCSN participants, but an increase in access to and use of specialty services. The net effect could be an increase in utilization (and perhaps costs) or a reduction. Evidence on utilization effects of the Demonstration is presented in this chapter.

Analyses in this chapter attempt to identify plausible impacts of the Demonstration on a wide range of satisfaction, health and utilization measures. Section V.B describes the survey and claims data used in the impact analyses. Section V.C describes several analytic methods used to exploit the particular strengths (and minimize the weaknesses) of the available data. Section V.D, Section V.E, Section V.F, Section V.G and Section V.H present evidence for Demonstration impacts on access to services, satisfaction of children and families, health status and function, family and formal caregiving, and utilization. Section V.I presents evidence for utilization effects and describes components of HSCSN and Medicaid expenditures, using Medicaid and HSCSN claims and eligibility/enrollment data.

B. Data and Methods

To generate data for the impact analyses, Demonstration evaluators conducted a survey of families with eligible children, including both Demonstration participants and non-participants. A copy of the survey instrument is in Appendix H. In addition, evaluators acquired basic eligibility data for children who were residents of the District and eligible for Medicaid and/or Supplemental Security Income (SSI), as well as Medicaid and HSCSN claims for these children. These administrative records provide more detailed information on formal utilization of health care services and of payments made for services by Medicaid and HSCSN.

1. Caregiver Survey

Over a 17-month period (June 1998 through October 1999), the evaluator conducted two survey efforts simultaneously:

"Wave One" and, approximately six months later, for a subset of Wave One respondents, "Wave Two" surveys of a Comparison Sample of participants and non-participants, designed to be the major source of data for impact analyses;
Surveys designed to capture data on New Participants in the Demonstration were conducted shortly after enrollment. As discussed below, this effort did not achieve its original objectives of identifying and interviewing children soon after enrollment. As a result, completed surveys of the "new" participants in the survey were merged with Waves One and Two of the Comparison Sample.

The Survey Instrument

The evaluator designed a survey instrument to be administered either to the sampled child's principal caregiver or to the child.⁴⁸ This instrument included questions that addressed: 1) access to care (provider attachment, the indirect or time-related costs of access, unmet needs) and satisfaction with access, 2) utilization (of formal skilled and unskilled care in the home, as well as informal care by family and friends, principal caregiver burden, timing of preventive visits, the child's use of mental health providers and of a range of "high tech" medical and other services such as tube feeding), 3) medical/behavioral/cognitive and social outcomes, and 4) respondent/child demographic and socioeconomic characteristics. Sources for most of the questions were the 1987 National Medical Expenditure Survey (NMES), the 1993 National Health Interview Survey, and an instrument developed by the Oregon Health Plan.

Comparison Sample

The comparison sample was implemented in two waves (cross sections), with the follow-up wave administered to a sub-sample of respondents to the first "baseline" wave. Wave One was not a true baseline, because the evaluator had no way of controlling the timing of this survey in order to capture respondent data just before the key decision point for this Demonstration (to enroll or not enroll in HSCSN). Respondents in the Demonstration may have been in the program for several months when contacted, and there was no way to know when, if ever, responding non-participants had considered and rejected enrollment. In fact, by the time the Design Report had been drafted, well before the survey began (November 1996), 1,731 children had already enrolled in the Demonstration. However, with appropriate adjustment for the timing of the survey relative to the initial month of the Demonstration and, for participants, relative to the month of enrollment, the survey provides some evidence of the relationship between exposure to the Demonstration and key measures for a subset of the survey population.

In addition, the two waves of the survey were expected to provide evidence for changes in certain variables over time within and between the participant and non-participant groups. By themselves, and without reference to other impact analyses, the change estimates do not support inferences about Demonstration impacts. Wave Two sample sizes are small, and both uncontrolled and controlled estimates of change effects tend to be small, statistically insignificant, and uncertain in direction. They are used sparingly in these analyses.

Using data from an HSCSN marketing database called the "Prospect File" (derived from DC Medicaid's records of SSI-eligible children and used to identify non-participants), HSCSN's Enrollment File, and later, an eligibility file maintained by DC Medicaid, the evaluator constructed a frame for drawing the initial comparison sample that included 1,549 participants and 1,042 non-participants in Wave One, and 597 participants and 202 non-participants in Wave Two, as shown in Exhibit 5.1.

EXHIBIT 5.1. Caregiver Survey: Wave One and Wave Two Completions¹ for Participants and Non-Participants
	Wave One		Wave Two
	Participants	Non-Participants	Participants	Non-Participants
Sample	1,549	1,042	597	202
Complete	740	391	323	80
Phone	735	332	323	80
Field	5	59	---	---
Source: Abt Associates' Survey Research Group1. Completions may differ from observation counts in impact analyses, because some surveys were removed due to missing data.

Survey staff administered the instrument by phone and, when phone contact could not be established, in the field. Over both waves, 1,470 interviews were successfully completed by phone, while 64 had to be completed by field interviewers.

Interviewers found that completing interviews with participants required considerably less effort than interviewing non-participants. In general, contact information for participants from HSCSN's Enrollment File was more complete and accurate than information on non-participants from the Prospect File. (In this respect, the evaluator's problems mirrored HSCSN's own problems in identifying and contacting beneficiaries.) Several problems listed below, though identified in the design phase of this project, proved larger than anticipated. To solve them, the evaluator first worked with those resources that were initially made available (mostly HSCSN data). Later, the evaluator established a helpful working relationship with MMIS and other staff at DC Medicaid that improved the accuracy of eligibility and contact information for eligible non-participants.

To obtain telephone numbers for non-participants (telephone numbers were not included in the SSI eligibility files), the evaluator used a matching service and a request for a telephone number included in the initial letter to the comparison sample. In May 1998, just prior to implementing the survey, returns from the mailing and matching provided phone numbers for only one-quarter of the initial group of non-participants scheduled for contact. Efforts continued to improve information on correct phone numbers and addresses. After several months, DC Medicaid agreed to release contact information on SSI children that proved to be more complete than data available from the Prospect File. However, many households in the District are without phones. Assuming that phone ownership is less frequent in low-income households, these figures should be much higher in the target population of this Demonstration. Therefore, the evaluator requested and received additional funds from HCFA to increase the percentage of sampled households contacted in the field.
Identities of non-participants' parents and caregivers were difficult to verify from Medicaid and SSI records, since surnames of children, parents and other caregivers are often different. This increased the challenge both of securing phone numbers and of identifying current addresses, factors that complicated both telephone and field survey efforts.
Eligibility changes increased the time and effort required to complete interviews. Shortly after initiating the survey effort, after a large proportion of households selected for field contact proved to be ineligible for the survey, the evaluator asked DC Medicaid to verify the eligibility of the initial sample. Seven months after survey implementation, DC Medicaid determined that nearly one-third of non-participants sampled for the first round of surveys was ineligible. Thereafter, the evaluator worked closely with DC Medicaid to verify eligibility on a regular basis.

Finally, unsettled conditions within HSCSN during 1998 led to a break-down in delivery of needed Demonstration eligibility/enrollment data to the evaluator. This delay, and the time required for DC Medicaid to process the evaluator's requests for verification of non-participant eligibility, meant that the supply of potential respondents was exhausted by the Fall of 1998. Phone and field interview efforts were suspended in November 1998 and not resumed until January 1999. Interviewing continued through the second week in October 1999. In order to maximize the size of Wave One, given the limited time frame available for the survey, it was decided during the Summer of 1999 that Wave Two efforts would be suspended and surveyor resources redirected to contacting and interviewing Wave One respondents.

Characteristics of the Comparison Sample

Participants differ from non-participants in the survey in several measures. Selected characteristics of participants and non-participants in the Wave One sample are shown in Exhibit 5.2.

Participants are likely to be younger than non-participants. Slightly more than 18 percent of all participant children are under five years of age, twice the fraction of very young non-participants (9.9 percent). Additionally, 8.8 percent of participants are 18 years and older, compared to 14.8 percent of non-participants.
Female children are less frequent in the participant sample (33.6 percent) than in the non-participant sample (39.6 percent).
A larger fraction of the participant sample is African-American (95.8 percent, compared to 92.3 percent of non-participants).
On a simple measure of socioeconomic status (SES, measured as having four or more of eight indirect measures of income, wealth, and public program participation), participants appear to be lower in SES than non-participants (19.2 percent, compared to 13.9 percent).

New participant sample

Because it was believed that the comparison sample would underrepresent newly eligible SSI/HSCSN children, the evaluator designed a process to identify and interview new participants within a month of enrollment. Tracking key measures from these surveys over time should show how the characteristics of new participants changed during the Demonstration. The evaluator's original design specified a two-wave model, similar to the comparison sample design, with most interviewed by phone and a small percentage contacted in the field. Initially, monthly enrollment of 53 children was expected, all of whom would be contacted. Assuming a 75 percent completion rate, 40 children a month would be interviewed in Wave One. Allowing for attrition in the sample and the same completion rate, 24 children would be contacted for Wave Two.

In fact, due to the delayed delivery of HSCSN eligibility data cited above and problems encountered in making contact with potential respondents, monthly interviews of new participants averaged 18, for a total of 302 Wave One completed surveys. Because of these delays, analysis of time of interview relative to time of enrollment revealed that only a very small percentage of respondents were contacted within three months of enrollment. Therefore, these observations were not analyzed separately, but were included with other participant surveys.

EXHIBIT 5.2. Caregiver Survey: Characteristics of Responding Caregivers (Wave 1)¹
Measure	Participants(N=740)	Non-Participants(N=323)
Age
4	18.0%	9.9%***
5-12	48.2	47.4*
13-17	24.9	27.9**
18+	8.8	14.8***
Gender
Male	66.4	60.4***
Female	33.6	39.6***
Race
African American	95.8	92.3**
Other	4.2	7.7
Four or more¹ measures of low SES	19.2	13.9***
Source: Caregiver Survey (Wave One)Statistical significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .05.1. Measures: Household 1) does not own car/truck, 2) does not own dwelling, 3) has moved more than once in past six months, 4) member gets food stamps, 5) member gets WIC assistance, 6) member in TANF, 7) member other than child in SSI, 8) member in SSDI.

2. HSCSN and Medicaid Eligibility and Claims Data

Claims and eligibility data contributed to analyses of selection effects in enrollment and to expenditure and utilization analyses. The evaluator used the same sample of eligible children in both the selection and cost/utilization analyses. Data and methods for the selection analyses, as well as the basic characteristics of participants and non-participants, are described earlier, in Section III.E. To summarize, the evaluator sampled equal numbers of participants and non-participants within each of five windows of time, beginning with February 1996, the starting month of the Demonstration, and ending in June 1998, the last month for which complete Medicaid claims were available. "Anchor months" were defined for each sample member. For participants, the anchor month was defined as the month of enrollment. For non-participants, the anchor was randomly assigned to a month within the window of time in which the non-participant was sampled. Medicaid expenditures in the six-month "utilization period" prior to these anchor months defined a measure of acuity used in the selection analysis. For the expenditure and utilization analyses, similar measures were defined for periods after the anchor months.

To build the "post-anchor" portions of the expenditure and utilization analysis files, the evaluator merged HSCSN and Medicaid claims with enrollment and eligibility data. For each child on the file, five post-anchor utilization periods were defined, beginning with February 1996 (the beginning of the Demonstration) through June 1998 (the last month for which Medicaid claims were available). The first four periods are six months in length, and the last period has, at most, five months of data. Technically, only a child with February 1996 as an anchor month can have utilization and expenditure data in all five utilization periods.

C. Analysis Methods

The evaluator selected analysis methods designed to accomplish two broad objectives:

Estimate the impact of the Demonstration: For the most part, evaluators used simple chi-square tests, supplemented by multivariate techniques to adjust estimates of impact for measurable differences between participants and non-participants, and to test for the relationship of length of enrollment to selected dependent variables. Two kinds of comparisons were made:
- cross-sectional comparisons of participants and non-participants, using data gathered in Wave One surveys and, for some utilization analyses, data created from claims and eligibility/enrollment data;
- change analyses, comparing changes in the differences between participants and non-participants from Wave One to Wave Two for selected variables.
Describe patterns of expenditure and utilization for participants and non-participants, using data created from claims and eligibility/enrollment data.

1. Impact Analyses: Concurrent Comparisons

The survey captured multiple measures within each of several major "domains," including access/satisfaction, health status and impairments, caregiver burden and utilization. To provide manageable comparisons between participants and non-participants, differences were computed and tested for statistical significance using the chi-square statistic. However, because participants and non-participants differ in several measurable respects (see the selection analyses reported in Section III.E), simple contrasts of dependent variable measures between participants and non-participants will not produce unbiased estimates of Demonstration impacts. Therefore, for participant/non-participant contrasts that were statistically significant, the evaluators re-estimated the contrasts adjusting for the influence of several independent variables.

Logistic regression was used to adjust predictions of categorical dependent variables (two-valued variables that characterize an "either/or" situation -- e.g., satisfied = 1/not satisfied = 0). The general form of the logistic regression is:

(5.1) L = B₁ + B₂*P + B₃*X

where the dependent variable L is the log of the odds that the categorical dependent variable will take on a value of one, B₁ is the intercept, B₂ is the coefficient of a participation measure P (which equals 1 for a Demonstration participant, and 0 for a non-participant) and B₃ represents coefficients of X (several independent variables that adjust the impact estimate). B₂, which captures the effect of P on L, adjusted for X, represents an estimate of the Demonstration's impact.

In addition to regressions estimated on the entire Wave One sample, the evaluator produced separate estimates for two groups of participants, pooled in each case with all non-participants: 1) participants enrolled for six or fewer months at the time of the interview, and 2) participants enrolled more that six months. These estimates provide a relatively simple test of the proposition that shorter-stay participants behave differently from longer-stay participants. Although it would be inappropriate to infer that differences by length of enrollment show impacts (by relating outcome measures to how long a child has been exposed to the Demonstration program), such findings may be used to support or raise questions about impact based on other analyses.

2. Impact Analyses: Change Measurement

To assess changes in participant/non-participant contrasts over time, the evaluator used a "double difference" model, that is:

(5.2) (P(2) - P(1)) - (N(2) - N(1))

where the change in a measure from 1(Wave One) to 2(Wave Two) is contrasted between P(participant) and N(non-participant).

Logistic regression was used to model this change:

(5.3) L = B₁ + B₂*P + B₃*X + B₄*F + B₅*(F*P)

where variables L, P and X and coefficients B₁(1), B₂(2) and B₃(3) are as defined above in equation 5.1, and F = ( = 1 if Wave Two measure, 0 if Wave One).

The interaction of F and P, captured in the estimate of B₅, represents a measure of the direction and statistical significance of a change in L.

In reporting findings from the Caregiver Survey below, responses to questions used to construct measures are often described as if the child made them (e.g., participants were more satisfied than non-participants). The reader will remember that most respondents were parents or other caregivers. Therefore, responses are generally the perceptions of caregivers and should be interpreted as such. Where questions specifically address caregivers' own issues (for example, job-related problems), this is made clear in definitions of measures and in the discussion.

3. Claims Based Analyses: Trends In HSCSN Utilization, Comparisons of Participants and Non-Participants, and the "Costs" of the Demonstration

Limitations of the claims data, discussed below in Section V.G.2, make comparisons of expenditures between participants and non-participants problematic, but allow for cautious analyses of claims-based measures of utilization. Simple contrasts between participants and non-participants were employed to describe utilization patterns. Ordinary least squares regression was used to explore factors that influence HSCSN expenditures on participants and Medicaid expenditures on non-participants.

D. Impact on Access and Satisfaction

1. Background

Though the term "managed care" creates an expectation that care coordination will happen and will lead to improved access to appropriate services, research suggests that better access has not always been forthcoming. Szilagyi (1998) reviews studies of Medicaid managed care and finds no consistent evidence of improvements. In fact, there has been some evidence that chronically-ill children encounter reduced access to specialty services in Medicaid managed care programs.

Nonetheless, it is still reasonable to expect that HSCSN will have significant positive impacts on measures of access and satisfaction. Previous research studies the experience of main-line HMOs contracting with state Medicaid agencies. In a managed care entity designed explicitly to address the special needs of children eligible for SSI, participants should:

be more likely than non-participants to have a usual place and a usual health care professional for medical and preventive services;
be more likely to have, as their usual health care professional, a specialist;
face lower "indirect cost" barriers in total time needed to access services;
report fewer unmet needs; and
report greater satisfaction with access to services and products.

2. Measurement Issues

Measures of access, defined in Exhibit 5.3, have been created from answers to questions on the Caregiver Survey and closely follow the order and grouping of these questions. Survey data provide measures of access to and satisfaction with services in four domains:

provider attachment which documents "usual" arrangements for medical and preventive services with places of service and health care professionals;
time required to access services, in making appointments, traveling to the place of service, and in waiting once at the place of service;
unmet needs across a wide range of services and products; and
satisfaction with access to a comparable range of services and products.

Two observations are in order:

Researchers often use ambulatory visits to measure access to care. These data are not available from the survey. Claims-based measures discussed below provide some support for the conclusions that participants use more physician services than non-participants. However, these data show that participants use more of almost all services, once enrolled, so that the special value of physician utilization as a measure of first entry into the system is simply part of a pattern of higher use, and not very instructive.
The Caregiver Survey asked respondents to distinguish between providers of medical services and providers of preventive services. Fewer than 1 percent of respondents identified separate "prevention providers," too few to yield useful information. Therefore, analyses of provider attachment reported in this section refer only to medical service providers.

EXHIBIT 5.3. Measures of Perceived Access/Satisfaction: Caregiver Survey
Measure	Description
Usual Professional/Place of Care
Usual place	Percent of children with usual place for medical services
Usual/office	Percent of children whose usual place is a physician's office
Usual ER	Percent of children whose usual place is emergency room
Usual doc	Percent of children with usual physician for medical services
Usual GP	Percent of children whose usual physician is a general practitioner
Usual spec	Percent of children whose usual physician is a specialist
Usual doc/off	Percent of children with a usual physician (specialist) in the child's usual place (physician's office)
Handles Needs
Ref need	Percent for whom usual physician or place makes needed referrals
Info needs	Percent for whom usual physician or place provides needed information
Prevent. needs	Percent for whom physician or place handles child's prevention needs
Most needs	Percent for whom usual physician or place handles most of the child's needs
All needs	Percent for whom usual physician or place handles all the child's needs
Indirect Costs of Access to Services
Min appt. time	Percent of children that wait less than four days or walk in for appointment at usual place/professional
Min travel time	Percent of children that travel 15 minutes or less to usual place/professional
Min wait time	Percent of children that wait 15 minutes or less in waiting room at usual place/professional
Min access time	Percent of children with minimum appointment time and minimum travel time and minimum wait time
Unmet Needs (referring to the past six months)
Spec need	Child has unmet needs for specialist services
Therapy need	Child has unmet needs for physical or other therapies
Immun need	Child has unmet needs for immunizations
Dental need	Child has unmet needs for dental services
Hospital need	Child has unmet needs for inpatient hospital services
ER need	Child has unmet needs for emergency room services
Mental health need	Child has unmet needs for mental health services
Home care need	Child has unmet needs for home health care services
Glasses need	Child has unmet needs for glasses
Drug need	Child has unmet needs for prescription drugs
Supply need	Child has unmet needs for medical supplies
DME need	Child has unmet needs for durable medical equipment
Mult need	Child has more than one unmet need
Satisfaction
Routine satisfaction	Percent of children who recently sought routine physician care reporting good to excellent access to routine care
ER satisfaction	Percent of children who recently sought emergency room services reporting good to excellent access to emergency room services
Specialist satisfaction	Percent of children who recently sought care from physician specialists reporting good to excellent access to specialists
Mental health satisfaction	Percent of children who recently sought mental health services reporting good to excellent access to mental health services
DME satisfaction	Percent of children who recently tried to acquire durable medical equipment reporting good to excellent access to DME
Home care satisfaction	Percent of children who recently sought home care reporting good to excellent access to home care
Social services satisfaction	Percent of children who recently sought social services reporting good to excellent access to social services
Overall satisfaction	Percent of children reporting good to excellent overall access to services sought over the past six months
Complaint	Percent of children filing a complaint in the past six months
Respect	Percent of children satisfied with the respect with which they are treated at their usual place/professional

3. Findings

Provider attachment. Most children (96.6 percent) have a usual place for medical care, and most (85.2 percent) have a usual physician, as shown in Exhibit 5.4.⁴⁹ For 66.8 percent of children with a usual physician, that physician is a general practitioner. Much more rarely (29.8 percent of children), the usual physician is a specialist. Caregivers of a minority of children report having a usual physician based in a "usual" office setting (33 percent); only 7.5 percent have a specialist as their usual office-based physician.

Participants are more likely than non-participants to have a usual physician (87.8 percent, compared to 79.3 percent). Though differences are not statistically significant, participants also appear more likely to have a usual place for services, have that usual place be a physician's office, have a usual physician based in a usual office setting, and have a specialist as this physician. They are less likely to report an emergency room as their usual place of care (by 1.2 percentage points), though few children overall report this arrangement (2.7 percent), and the difference is statistically insignificant.

Most caregivers agree that their child's usual place or professional handles most specific needs (percentages range from 94.3 to 97.3 percent). A slightly smaller percentage (86 percent) believe that all needs are addressed in the usual place or by the usual doctor. In general, differences between participants and non-participants in these assessments are small and statistically insignificant. However, there is one exception. Participants are more likely than non-participants to report that their place/physician addresses all their needs for preventive services (98 percent, compared to 95.9 percent).⁵⁰

EXHIBIT 5.4. Perceived Access/Satisfaction: Percent Reporting Provider Attachment, and Usual Provider Handles Needs
Measure	Total(1)	Participants(2)	Non-Participants(3)	Difference¹(2) - (3)
Usual place	96.6%	97.2%	95.4%	1.8 pct. pts.
Usual/office	36.9	38.0	34.4	3.6
Usual ER	2.7	2.4	3.6	(1.2)
Usual doc	85.2	87.8	79.3	8.5***
Usual GP	66.8	67.2	65.6	1.6
Usual spec	29.8	29.7	30.1	(0.4)
Usual doc/off	33.0	34.1	30.7	3.4
Usual spec/off	7.5	8.2	5.9	2.3
Ref needs	94.3	94.0	94.9	(0.9)
Info needs	95.7	95.5	96.2	(0.7)
Prevent needs	97.3	98.0	95.9	2.1*
Most needs	94.4	94.0	95.2	(1.2)
All needs	86.0	86.1	85.7	0.4
Source: Caregiver Survey (Wave One)Statistical Significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .05.1. Negative values in parentheses

Time as a cost of access to services. Although no children experience substantial time barriers to arranging appointments (four days or more, or inability to walk in for an appointment), about one-third enjoy minimum time costs of travel to the place of service (15 minutes or less) and waiting time at the place of service (15 minutes or less), as shown in Exhibit 5.5. However, fewer experience minimum time barriers in all three categories together (12.9 percent). Though differences are not significant, fewer participants seem to experience minimum travel times, and participants are less likely to face minimum time barriers in arranging, traveling to and waiting for appointments.

EXHIBIT 5.5. Perceived Access/Satisfaction: Percent With Minimum Indirect Costs of Access to Services
Measure	Total(1)	Participants(2)	Non-Participants(3)	Difference¹(2) - (3)
Min appt time	100.0%	100.0%	100.0%	0 pct. pts.
Min travel time	33.7	33.3	34.6	(1.3)
Min wait time	35.2	35.2	35.2	0
Min access time	12.9	12.4	13.9	(1.5)
Source: Caregiver Survey (Wave One)Statistical Significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .051. Negative values in parentheses

Unmet needs. Of all sampled children, the frequency of specific unmet needs ranges from 1.1 percent for hospital care to 13.4 percent for dental services, and 8.9 percent report more than one unmet need (Exhibit 5.6). In eight of twelve specific services and products, participants are less likely to report unmet needs than non-participants. In only one instance (therapies) is this difference (4.9 percentage points) statistically significant. However, participants are significantly less likely to report multiple unmet needs (7.8 percent, compared to 11.5 percent).

Satisfaction. Consistent with previous research, reports of good to excellent satisfaction with access to services and products are high among survey respondents, generally averaging between 80 and 90 percent -- the frequency of overall satisfaction is 89.8 percent (Exhibit 5.7). Interestingly, satisfaction with access to home care and social services is considerably lower than for other categories, at 68.6 and 64.1 percent respectively.

Participants are, for the most part, more satisfied than non-participants. For seven specific categories of services and products, participants are more satisfied in five, and differences from non-participants are statistically significant for four of these categories. Participants are also more satisfied overall, with 91.6 percent expressing good to excellent overall satisfaction, compared to 85.8 percent of non-participants.

Two additional indicators are shown, one of which supports findings about satisfaction discussed above, and one of which does not. On the one hand, most caregivers (98.7 percent) report that they are satisfied with the respect with which they are treated by their child's usual physician or at their usual place for receiving services. Participants are more likely to be satisfied than non-participants (99.2 percent, compared to 97.5 percent). On the other hand, few report filing a formal complaint about their program or provider in the past six months (4.1 percent), and, though the difference is not statistically significant, participants seem more likely to file than non-participants (4.2 percent, compared to 4.0 percent).⁵¹

EXHIBIT 5.6. Perceived Access/Satisfaction: Percent Reporting Unmet Needs for Specific Services
Measure	Total(1)	Participants(2)	Non-Participants(3)	Difference¹(2) - (3)
Spec need	9.4%	9.5%	9.3%	0.2 pct. pts.
Therapy need	11.5	10.0	14.9	(4.9)**
Immun need	8.7	9.0	8.1	0.9
Dental need	13.4	12.3	15.8	(3.5)
Hospital need	1.1	1.2	0.9	0.3
ER need	2.9	2.6	3.7	(1.1)
Mental health need	8.3	8.1	8.7	(0.6)
Home care need	7.7	7.2	9.0	(1.8)
Glasses need	7.1	6.5	8.4	(1.9)
Drug need	5.2	4.7	6.2	(1.5)
Supply need	6.2	6.4	5.9	0.5
DME need	5.2	4.7	6.2	(1.5)
Mult need	8.9	7.8	11.5	(3.7)*
Source: Caregiver Survey (Wave One)Statistical Significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .051. Negative values in parentheses

Access, satisfaction and length of enrollment: multivariate analyses. On several measures, participants enrolled over six months are more likely to cite favorable access and satisfaction, relative to non-participants, than participants enrolled six or fewer months. Exhibit 5.8 shows this to be true for measures of unmet need. All participants are less likely to express unmet therapy needs and multiple unmet needs. Participants enrolled for six articulate unmet needs for therapy and non-participants are equally likely to express unmet therapy needs, or to report multiple unmet needs. Long-stay participants are much less likely than participants to express these needs. On measures of satisfaction, these data show that, for most service-specific measures and for overall satisfaction, longer-stay participants are relatively more satisfied than short-stay participants. Provider attachment provides a modest exception to this pattern. Relative to non-participants, long-stay participants are somewhat less likely than short-stay participants to have a usual physician.

EXHIBIT 5.7. Perceived Access/Satisfaction: Percent Reporting Good to Excellent Access to Services
Measure	Total(1)	Participants(2)	Non-Participants(3)	Difference¹(2) - (3)
Routine satisfaction	89.2%	89.4%	88.8%	0.6 pct. pts.
ER satisfaction	88.8	89.4	87.4	2.0
Specialist satisfaction	85.5	87.3	81.6	5.7**
Mental health satisfaction	83.5	85.1	79.8	5.3**
DME satisfaction	80.1	83.8	71.0	12.8***
Home care satisfaction	68.6	71.2	62.6	8.6**
Social services satisfaction	64.1	65.1	61.6	3.5
Overall satisfaction	89.8	91.6	85.8	5.8***
Complaint	4.1	4.2	4.0	0.2
Respect	98.7	99.2	97.5	1.7**
Source: Caregiver Survey (Wave One)Statistical Significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .051. Negative values in parentheses

4. Discussion

Though many of the contrasts reported above do not meet accepted standards for statistical significance, the overall pattern, confirmed occasionally by truly significant differences, suggests that participants are more attached to providers and places, do have fewer unmet needs, and are more satisfied with their access to services than non-participants. There is also some evidence that longer exposure to the Demonstration, captured in comparisons of short- and long-stay participants, is associated with reduced frequency of unmet needs and increased satisfaction.

These findings, particularly findings of higher participant satisfaction, are not surprising -- family focus groups conducted in 1997 and 1998 and reported in Chapter 4 confirm that, even for caregivers who complain about specific problems with HSCSN's care management system, the Demonstration is an improvement over what they had before.

EXHIBIT 5.8. Perceived Access/Satisfaction and Length of Enrollment
Measure	One to Six Months Enrolled			Over Six Months Enrolled
Measure	Participants	Non-Participants	Difference	Participants	Non-Participants	Difference
Access
Usual doc	88.0%	79.1%	8.9 pct. pts.*	87.7%	79.6%	8.1 pct. pts.***
Prevent needs	99.6	99.6	0	99.0	97.9	1.1*
Their need	12.5	13.2	(0.7)	8.0	13.9	(5.9)***
Mult needs	9.4	10.1	(0.7)	6.9	10.4	(3.5)*
Satisfaction with access
Specialist sat	88.0	82.0	6.0	87.6	81.7	5.9***
MH sat	83.4	80.1	3.3	85.5	79.9	5.6*
DME sat	77.9	71.6	6.3	85.0	71.1	13.9***
Home care sat	72.6	63.9	8.7	70.8	62.8	8.0**
Overall sat	88.9	86.0	2.9	92.3	86.3	6.0***
Source: Caregiver Survey (Wave One)Statistical Significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .051. Negative values in parentheses

E. Utilization: Formal and Informal Caregiving

1. Background

Most children with special needs live at home. For these children, caregivers who may be parents, other relatives or friends provide emotional and financial support, administer medications and treatments, monitor activities to protect the child from harm, and provide personal care and other supportive services. The strain or burden of constant attention can affect a caregiver's health, social life and job prospects.

The Caregiver Survey collected information on both formal and informal caregiving and caregiver burden. HSCSN's care managers should be capable of recognizing and mitigating caregiver burden, in part by connecting the caregiver with sources of formal support services. Of course, families differ in their willingness to accept "outsiders" as care providers, so that even if help is offered, it may be refused.

Nonetheless, it is hypothesized that: (1) participants in the Demonstration will receive more formal skilled and unskilled services at home than non-participants; (2) participants will have less need for informal caregivers to help with the provision of medical and social services and (3) caregiver burden will be less among participants.

EXHIBIT 5.9. Measures of Formal/Informal Caregiving
Measure	Description
Formal Care
Skilled care home	Percent of children using any skilled home care in past six months
Unskilled care home	Percent of children receiving unskilled home care from more than one person in the past six months
Informal Care
Informal care total	Percent of children regularly receiving informal care from more than one person
Informal care -- personal	Percent of children receiving personal care from one or more informal caregivers
Informal care -- medication	Percent of children receiving help with medications from one or more informal caregivers
Informal care -- treatment	Percent of children receiving help with medical treatments from one or more informal caregivers
Informal care -- equipment	Percent of children receiving help with special equipment from one or more informal caregivers
Informal care -- monitoring	Percent of children receiving help through monitoring child's activities from one or more informal caregivers
Informal care -- tutor	Percent of children receiving help with schoolwork from one or more informal caregivers
Caremix -- none	Percent of children using no formal or informal care
Caremix -- skilled	Percent of children using skilled, but no unskilled or informal care
Caremix -- unskilled	Percent of children using unskilled, but no skilled or informal care
Caremix -- skilled and unskilled	Percent of children using skilled and unskilled formal care, but no informal care
Caremix -- informal and unskilled	Percent of children using informal and unskilled care, but no skilled care
Caremix -- informal and skilled	Percent of children using informal and skilled care but no unskilled care
Caremix -- informal	Percent of children using informal but no formal care
Caremix -- all	Percent of children using formal and informal care
Caregiver Burden
Never worked	Percent of principal caregivers who never worked since becoming a principal caregiver
Job not taken	Percent of principal caregivers who passed up a job opportunity to take care of child
Quit job	Percent of principal caregivers who quit job to take care of child
Change job	Percent of principal caregivers who changed jobs to take care of child
Change hours	Percent of principal caregivers who changed hours to take care of child
No promotion	Percent of principal caregivers who turned down a promotion to take care of child
Fewer hours	Percent of principal caregivers who worked fewer hours to take care of child

2. Measurement Issues

In principle, home formal and informal caregiving and burden should be quantifiable -- for example, in numbers of formal care visits per month or week, or numbers of hours spent by informal caregivers providing informal care evaluated at some "opportunity cost" of lost earnings. Although the survey gathered information on visits and hours, there were substantial amounts of missing data in these fields. As a result, the limited data collected were not considered valid or representative. Therefore, all measures of caregiving and burden used in these analyses are dichotomous (the respondent reports the use of some skilled care or some unskilled care) rather than continuous measures of visits and hours (Exhibit 5.9). Also, in an effort to reduce respondent burden of the survey, no attempt was made to collect information on caregivers' current or previous occupations, for purposes of imputing estimates of lost earnings. Burden is captured in various measures of how frequently certain problems appear, not in the "indirect costs" of informal caregiving. "Formal care" was defined for survey respondents as care provided by persons other than the principal caregiver, friend or relative. Formal care could be provided by paid or volunteer staff.

3. Findings

Formal and informal care utilization. Very few children use any formal care at home (Exhibit 5.10). Only 5.9 percent report using any formal skilled care in the past six months, and 1.8 percent use unskilled care. Although these data also suggest that participants use more formal skilled and unskilled home services, as hypothesized, the differences are not statistically significant.

About one-half of the sample receives informal care at home from more than one person. Participants may be more likely to use multiple caregivers, but the difference is small and not significant. Across the range of functions that informal caregivers fulfill, participants are somewhat more likely to receive personal care and assistance with medications, while non-participants are more likely to receive help with medical treatments, special equipment, monitoring for safety (a significant difference) and schoolwork.⁵²

The mix of formal and informal care services, described in Exhibit 5.11, shows a high overall dependence on informal caregivers. One-fifth of all children report using no formal or informal care. This frequency is nearly the same for participants and non-participants. The largest category relies only on informal care (69.1 percent), somewhat less for participants than non-participants (68.5 percent, compared to 70.3 percent). Participants are most likely to use combinations of informal and skilled care, and informal and unskilled formal care. None of the participant/non-participant differences are statistically significant.

EXHIBIT 5.10. Formal/Informal Caregiving: Percent Using Formal and Informal Care
Measure	Total(1)	Participants(2)	Non-Participants(3)	Difference¹(2) - (3)
Formal Care
Skilled care home	5.9%	6.1%	5.6%	0.5 pct. pts.
Unskilled care home	1.8	1.9	1.6	0.3
Informal Care
Informal care, total	50.4	50.8	49.5	1.3
Informal care, personal	55.0	55.3	54.5	0.8
Informal care, medication	49.2	49.3	48.9	0.4
Informal care, treatment	25.9	25.5	26.6	(1.1)
Informal care, equipment	19.1	18.7	20.1	(1.4)
Informal care, monitoring	44.1	42.0	49.2	(7.2)**
Informal care, tutor	54.0	52.7	57.0	(4.3)
Source: Caregiver Survey (Wave One)Statistical Significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .051. Negative values in parentheses

EXHIBIT 5.11. Formal/Informal Caregiving: Percent Reporting Mix of Skilled, Unskilled, Informal Care
Measure	Total(1)	Participants(2)	Non-Participants(3)	Difference¹(2) - (3)
Caremix, none	21.1%	21.0%	22.0%	(1.0) pct. pts.
Caremix, skilled	0.9	0.5	1.6	(1.1)
Caremix, unskilled	1.2	1.4	0.9	0.5
Caremix, skilled/unskilled	0.3	0.1	0.6	(0.5)
Caremix, informal/unskilled	2.6	3.1	1.6	1.5
Caremix, informal/skilled	4.1	4.5	3.4	1.1
Caremix, informal	69.1	68.5	70.3	(1.8)
Caremix, all	0.7	1.0	0	1.0
Source: Caregiver Survey (Wave One)Statistical Significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .051. Negative values in parentheses

Caregiver burden. The Caregiver Survey collected information on caregivers' problems related to their jobs and to the care they are providing to the child. Exhibit 5.12 shows that roughly one-third of all caregivers have not had a job from the time the child's health condition was identified. A larger proportion of participants' caregivers have been without a job than non-participants' caregivers (37.5 percent, compared to 28.3 percent). Of those still in the labor force, many have experienced problems related to their child's needs for assistance: 30.3 percent passed up job opportunities, 25.3 percent quit their jobs, 14.0 percent changed jobs, 31.3 percent altered their working hours, 19.6 percent passed up chances for advancement or promotion, and 31.5 percent worked fewer hours. Among those still in the work force, participants were less likely to have experienced any of these specific job-related problems (with statistically significant differences for passing up promotions and working fewer hours), and were less likely to have mentioned two or more job-related problems.⁵³

Caregivers responded to several questions about possible problems encountered in caring for their children. In the whole sample, 62.1 percent reported one or more problems, and there was no significant difference between participant and non-participant caregivers on the measure of multiple problems, or on the individual problems which included:

problems lifting or moving the child (17.7 percent);
the child receives special medical care at home and only the caregiver or a medical professional can give this care (23.4 percent);
taking care of the child limits social life and free time (46.8 percent);
taking care of the child has a negative impact on the caregiver's health (7.8 percent); and
the caregiver needs a break but cannot get it because the caregiver cannot find anyone to help out (37.8 percent).

EXHIBIT 5.12. Formal/Informal Caregiving: Percent Reporting Specific Measures of Caregiver Burden
Measure	Total(1)	Participants(2)	Non-Participants(3)	Difference¹(2) - (3)
Work-Related
Never worked	34.7%	37.5%	28.3%	9.2 pct. pts.***
Job not taken	30.3	29.4	32.5	(3.1)
Quit job	25.3	25.2	25.5	(0.3)
Change job	14.0	13.1	16.1	(3.0)
Change hours	31.3	30.0	34.1	(4.1)
No promotion	19.6	17.6	24.2	(6.6)**
Fewer hours	31.5	29.9	35.4	(5.5)*
Job problems	37.7	36.9	39.6	(2.7)
Child's Care-Related
Care problems	62.1	62.3	61.6	0.7
Source: Caregiver Survey (Wave One)Statistical Significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .051. Negative values in parentheses

Formal and informal care and length of enrollment: multivariate analyses. Based on re-estimates of statistically significant measures separating short and long-stay participants, long-stay participants are less likely to use informal care for monitoring than short-stay participants, and less likely to have experienced problems with promotion and hours (Exhibit 5.13).

EXHIBIT 5.13. Evaluation of DC Managed Care Program for Special Needs Children Formal/Informal Care by Length of Enrollment
Measure	One to Six Months Enrolled			Over Six Months Enrolled
Measure	Participants	Non-Participants	Difference	Participants	Non-Participants	Difference
Inf Care Monitoring	47.1%	49.9%	(2.8) pct. pts.	36.8%	50.5%	(13.7) pct. pts.***
Job-related Problems
Never worked	41.3	27.9	13.4**	33.7	28.9	4.8
Fewer hours	31.7	34.8	(3.1)	30.2	35.1	(4.9)
No promotion	22.7	23.0	(0.3)	16.4	22.5	6.1**
Source: Caregiver Survey (Wave One)Statistical Significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .051. Negative values in parentheses

4. Discussion

While there is some evidence that caregivers for children participating in the Demonstration experience fewer job-related problems than caregivers for non-participants, there is no difference in the frequencies of problems more directly associated with caring for the child. As hypothesized, participants use more formal home care than non-participants, but few in the sample use formal care, and differences are small and not statistically significant. It seems clear that caregivers of both participants and non-participants are heavily invested in their roles and, for whatever reason, are reluctant to request help from formal service providers.

F. Preventive and Medical Services Utilization

1. Background

The Caregiver Survey gathered focused information on utilization not readily available from HSCSN or Medicaid claims. This information includes measures in several domains:

timeliness of preventive services utilization. Participants should report more timely routine physical/immunization visits, preventive dental visits and routine eye examinations.
regular use of mental health services. Participants should not differ from non-participants in the use of mental health services. Survey data suggest no difference between participants and non-participants in the frequency of mental health problems (Section 5.7.3), but claims data suggest a higher frequency of mental health related diagnoses among participants (Section 3.4).
use of certain "high-tech" medical treatments and therapies. Participants appear to be frequent users of all services before enrollment, and by some measures appear to be somewhat more physically impaired than non-participants, so participants should be more likely to use high-tech services than non-participants.

These data, information from the Case Record Reviews (Chapter IV), and information from utilization estimates based on Medicaid and HSCSN claims, form a relatively comprehensive picture of utilization under the Demonstration.

2. Measurement Issues

As was true for formal and informal care use reported in Section V.E, measures of preventive and medical services utilization do not capture volumes (days, visits), as shown in Exhibit 5.14. Most measures represent frequency (did/does the child use any of the named service). In the case of preventive services, time (in months) since the child's last visit is converted to a measure of "low" utilization (percent of children for whom the last visit was over a year before the interview).

EXHIBIT 5.14. Utilization of Preventive and Medical Services: Caregiver Survey
Measure	Description
Job problems	Percent of principal caregivers mentioning two or more job-related problems
Care problems	Percent of principal caregivers mentioning one or more problems related to: 1) lifting/moving child, 2) needed special treatments, 3) health of the caregiver, 4) limits on social life, free time, 5) no respite
Preventive Services
Immunization (Year +)	Percent of children for whom last routine physical/immunization was over a year ago
Dental (Year +)	Percent of children for whom last dental exam was over a year ago
Eye (Year +)	Percent of children for whom last routine eye exam was over a year ago
Mental Health Services
Psych	Percent of children now visiting a psychiatrist or psychologist regularly
Counseling	Percent of children now visiting counselor or social worker regularly
Complex Medical and Other Services
IV	Percent of children using parenteral or IV medications
IV (home)	Percent of children using parenteral or IV medications at home
Suction	Percent of children using suctioning
Suction (home)	Percent of children using suctioning at home
Oxygen	Percent of children using oxygen
Oxygen (home)	Percent of children using oxygen at home
Inhale	Percent of children receiving inhalation therapy
Inhale (home)	Percent of children receiving inhalation therapy at home
Tube	Percent of children being tube-fed
Tube (home)	Percent of children being tube-fed at home
Bowel	Percent of children receiving bowel or bladder rehabilitation
Bowel (home)	Percent of children receiving bowel or bladder rehabilitation at home
PT	Percent of children receiving physical therapy
PT (home)	Percent of children receiving physical therapy at home
Treat (1)	Percent of children receiving one or more treatments/therapies
Treat (1-home)	Percent of children receiving one or more treatments/therapies at home
Treat (2)	Percent of children receiving two or more treatments/therapies
Treat (2-home)	Percent of children receiving two or more treatments/therapies at home

3. Findings

Preventive and mental health services. A minority of all respondents report a gap of over one year in receiving three kinds of preventive services (Exhibit 5.15). For 3.9 percent, over one year had passed since their last routine physical examination or immunization. Comparable frequencies for routine dental visits and eye examinations are 6.1 percent and 10.8 percent respectively. Participants are more timely than non-participants in all three categories, and significantly more timely for physical examinations or immunizations (2.6 percent reporting more than a year's gap, compared to 6.8 percent of non-participants) and dental care (3.7 percent, compared to 11.8 percent).54

Psychiatric and psychological services are regularly used by 22.1 percent of the sampled children, and 25.0 percent regularly visit a counselor or social worker. As hypothesized, there are no significant differences between participants and non-participants.

EXHIBIT 5.15. Preventive and Medical Service Utilization: Percent Reporting Gap of More Than One Year in Receiving Preventive and Mental Health Services
Measure	Total(1)	Participants(2)	Non-Participants(3)	Difference¹(2) - (3)
Preventive Services
Immunization (Year+)	3.9%	2.6%	6.8%	(4.2) pct. pts.***
Dental (Year+)	6.1	3.7	11.8	(8.1)***
Eye (Year+)	10.8	10.5	11.5	(1.0)
Mental Health Services
Psych	22.1	21.8	22.9	(1.1)
Counseling	25.0	25.0	25.0	0
Source: Caregiver Survey (Wave One)Statistical Significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .051. Negative values in parentheses

Medical treatments and therapies. With the exception of oxygen, participants use more medical treatments and therapies than non-participants, in total and in home settings (Exhibit 5.16). Except for physical therapy (used by 27.7 percent of all children) and inhalation therapy (used by 11.9 percent), fewer than three percent of children use any one of the other named treatments.⁵⁵ The largest and most significant differences were in tube feeding (used by 3.4 percent of participants and 1.2 percent of non-participants) and physical therapy (used by 30.0 percent of participants and 22.6 percent of non-participants).⁵⁶

EXHIBIT 5.16. Preventive and Medical Service Utilization: Percent Using Medical Treatments, Total and At Home Utilization
Measure	Total(1)	Participants(2)	Non-Participants(3)	Difference¹(2) - (3)
IV	3.2%	3.4%	2.8%	0.6 pct. pts.
IV (home)	2.1	2.4	1.2	1.2
Suction	2.0	2.2	1.6	0.6
Suction (home)	1.9	2.0	1.6	0.4
Oxygen	2.9	2.7	3.4	(0.7)
Oxygen (home)	1.9	2.0	1.6	0.4
Inhale	11.9	12.3	10.8	1.5
Inhale (home)	10.5	11.6	8.1	3.5*
Tube	2.7	3.4	1.2	2.2*
Tube (home)	2.4	2.8	1.2	1.6
Bowel	1.5	1.8	0.9	0.9
Bowel (home)	1.3	1.5	0.9	0.6
PT	27.7	30.0	22.6	7.4**
PT (home)	5.6	6.1	4.6	1.5
Source: Caregiver Survey (Wave One)Statistical Significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .051. Negative values in parentheses

Also notable is the finding that participants are more likely than non-participants to use treatments at home, in all categories of service. In some instances (such as intravenous medications, oxygen, and inhalation therapy), the percentage point difference is larger than it is for use of these treatments in all settings.

Multiple treatments are more frequent among participants than non-participants (Exhibit 5.17). Of all children surveyed, 36.5 percent use one or more treatments, and 18.1 percent use two or more. Comparable estimates for multiple home treatments are 10.3 percent and 4.8 percent respectively. In all four categories, participants exceed non-participants, and in two of four (one or more treatments, and two or more treatments in a home setting), the difference is statistically significant.

EXHIBIT 5.17. Percent Using Preventive and Medical Services Utilization: Multiple Treatments, Total and At Home
Measure	Total(1)	Participants(2)	Non-Participants(3)	Difference¹(2) - (3)
Treat (1)	36.5%	38.4%	32.2%	62. pct. pts.*
Treat (1-home)	10.3	11.2	8.1	3.1
Treat (2)	18.1	19.2	15.5	3.7
Treat (2-home)	4.8	5.5	3.1	2.4*
Source: Caregiver Survey (Wave One)Statistical Significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .051. Negative values in parentheses

Preventive and medical treatments/therapies and length of enrollment: multivariate analyses. No significant relationships between length of enrollment and use of preventive and medical treatments/therapies were found.

4. Discussion

Survey data generally support findings reported earlier in Chapter III (that participants are more intensive users of services before enrollment than non-participants). Later in this chapter, claims-based measures will confirm that participants remain intensive service users after enrollment as well. This difference is most striking in measures of the timeliness of routine physical/immunization and utilization of dental services, a result supported by other findings in this report that suggest the Demonstration has had a real impact on preventive services use.

G. Impact on Health Status and Impairments

1. Background

If the Demonstration improves perceived access to both preventive and therapeutic services, as an earlier section suggests, then one might expect to find improvements in health status and stabilized or reduced levels of impairment among participants, relative to non-participants. The research literature offers no evidence on the relationship between managed care participation and health care outcomes for children with special health needs. For several reasons, any improvements, if they occur, will probably be unobservable.

First, the data available to the evaluator from the Caregiver Survey cannot capture changes in health status and impairments from before exposure to the Demonstration to after exposure. Follow-up data from Wave Two of the Survey provide limited information on changes in status over a few months but cannot be used in isolation to measure impacts of the Demonstration.

Second, participants and non-participants are seriously impaired individuals, most of whom cannot expect major improvements in functional capacity or health status. Relative "improvement" may mean that one group remains stable, while another declines in function, or that the improved group declines at a slower rate than others. Precise measures of modest relative improvements in a highly-impaired population require condition-specific measures based on clinical and social assessments provided by trained professionals. For this evaluation, only general measures of health status and capacity, provided by caregivers in the Caregiver Survey, are available.

Third, the interpretation of concurrent data for participants and non-participants is complicated by selection issues which are discussed above in Chapter III. Selection analyses in this chapter show that participants, before enrolling in the Demonstration, were more frequent users of Medicaid-reimbursed services than non-participants, after adjusting for months of eligibility. Three possible interpretations of this finding were presented: 1) before enrolling, participants were more severely impaired than non-participants; 2) participants, through their families and caregivers, better understood how to maximize use of the system of fee-for-service care delivery before enrolling; and 3) participants were more impaired and more sophisticated users than non-participants. Without a better understanding of how utilization rates and impairment levels are associated in this population, it is difficult to predict how the apparent increases in utilization for participants after enrollment would have led to improved outcomes.

With these uncertainties in mind, and based on findings reported in the analysis of selection in Chapter III, it is hypothesized that Demonstration participants will be more impaired than non-participants. However, satisfaction findings reported earlier suggest that reported expectations for the child's future health status should be more optimistic for participants than for non-participants.

2. Measurement Issues

Measures of impairment levels used in this analysis, defined in Exhibit 5.18, are based on information reported by caregivers on the survey, and may be different, perhaps systematically different, from assessments conducted by trained professionals. For example, caregivers may tend to understate the level of their child's impairments. However, there is no particular reason to believe that caregivers of participants would be systematically more likely either to over- or under-report impairments.

EXHIBIT 5.18. Measures of Health Status/Function: Caregiver Survey
Measure	Description
Cognitive/Behavioral (All Children)
Cognitive	Percent of children with problems or delays in learning or mental development from half to all the time
Emotional	Percent of children with problems managing their emotions from half to all the time
Cognitive/Emotional	Percent of children with cognitive and/or emotional problems
Children Five Years and Older
Behavioral	Percent of children with difficulty getting along with others their age from half to all the time
Violent	Percent of children displaying frequent violent behavior from half to all the time
Cognitive/Emotional	Percent of children with one or more cognitive/behavioral problems
Activities of Daily Living (All Children)
Mob	Percent of children with problems that limit ability to crawl, walk or run
Mob equipment	Percent of children using mobility-assisting equipment
Breath	Percent of children experiencing difficulty breathing
Breath equipment	Percent of children using equipment to help with breathing
Eat	Percent of children experiencing difficulties chewing, swallowing or digesting
Eat equipment	Percent of children of children using equipment to help with eating
Blind	Percent of children who are blind
Vision	Percent of children not blind but vision-impaired
Deaf	Percent of children who are deaf
Hearing	Percent of children not deaf but hearing-impaired
Physical problems	Percent of children with more than two physical impairments
Physical/cognitive problems	Percent of children with more than two physical problems and one or more cognitive/emotional problems
Activities of Daily Living (Children Five Years and Over)
Strenuous act	Percent of children with difficulties participating in strenuous activities, compared to other children
Toileting	Percent of children with toileting limitations
Communication problems	Percent of children with difficulties communicating with persons outside the family from half to all the time
Physical problems (5+)	Percent of children with five (of eight) or more physical impairments
Physical/cognitive problems (5+)	Percent of children with five (of eight) or more physical impairments and one (or four) or more cognitive/emotional problems
Health Status
Good health now	Percent of children reported in good to excellent health at time of interview
Good health (6)	Percent of children reported in about the same to excellent health six months before the interview
Good health (6+)	Percent of children expected to be in about same to excellent health in six months
Disability days	Percent of children missing any days from play, school or work in typical month, due to condition(s)
Bed days	Percent of children with any days spent (half day or more) in bed due to condition(s)
Limitations on School (Children Age Five and Over)
School	Percent of children attending school
Special school	Percent of children attending special school/classes related to their condition(s)
Special need	Percent of children of caregivers expressing need for special schooling
Limited school	Percent of children with limited school attendance due to condition(s)
Limitations on Work (Children Age 14 and Over)
Work	Percent of children with job
Limited work	Percent of children with condition(s) that limit amount or type of work

3. Findings

Cognitive and emotional development and behavioral problems. Over one-third of all children in the Caregiver Survey sample have reported frequent problems or delays in learning or mental development (35.6 percent), while slightly more than one-quarter have frequent problems managing their emotions (26.7 percent), as shown in Exhibit 5.19. Forty-four percent have one or both of these problems. Although participants seem less likely to have cognitive problems (lower than non-participants by 3.2 percentage points, an insignificant difference), they are more likely to have emotional problems (higher by 3.6 percentage points), and about equal to non-participants in the frequency of one or both problems.

Of all sampled children age five and older, slightly under one-quarter have frequent difficulties getting along with other children their age, while a small minority (2 percent) display frequent violent behavior. Though participants are little different from non-participants in the frequency of behavioral problems, the frequency of violent behavior is significantly lower among participants (by 2.7 percentage points). About one-half of sampled children age five and older have one or more of four problems (getting along, violent behavior, cognitive problems and emotional problems). Though this frequency is lower among participants than non-participants (by 1.9 percentage points), the difference is not significant.

EXHIBIT 5.19. Health Status/Function: Percent Reporting Problems in Cognitive/Behavioral/Emotional Development
Measure	Total(1)	Participants(2)	Non-Participants(3)	Difference¹(2) - (3)
All Children
Cognitive	35.6%	34.6%	37.8%	(3.2) pct. pts.
Emotional	26.7	27.8	24.2	3.6
Cognitive/Emotional	44.2	44.2	44.3	(0.1)
Children Five Years +
Behavioral	23.6	23.4	24.1	(0.7)
Violent	2.0	1.1	3.8	(2.7)***
Cognitive/Emotional 5+	50.9	50.3	52.2	(1.9)
Source: Caregiver Survey (Wave One)Statistical Significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .051. Negative values in parentheses

Physical and overall functional capacity. The Caregiver Survey requested information on a wide range of impairments. Frequencies of physical impairments in all children range from blindness (3.8 percent) to breathing (31.8 percent), as Exhibit 5.20 shows.

Just under 5 percent are "severely impaired physically" (more than two physical impairments), and 3.5 percent are "severely impaired physically and mentally" (more than two physical impairments and one or both cognitive and emotional problems). For most of the measures, impairments are more frequent among participants, though not significantly higher. (There is a higher percentage of children who are blind among participants, but among those not blind, more non-participants appear to have impaired vision.) Participants tend also to be more likely than non-participants to use equipment to assist with mobility, breathing and eating. However, the differences are not statistically significant, with the exception of equipment to assist with eating.

Additional measures for older children tend to confirm impressions of greater impairment among participants (Exhibit 5.21). For children five years of age and older, slightly more than one-third are limited in strenuous activities most of the time, more so for participants (39.0 percent) than non-participants (35.1 percent). Only 7.2 percent of children have toileting problems, but these problems are significantly more frequent among participants (9.2 percent) than non-participants (3.1 percent). In contrast, participants are apparently less likely to have problems communicating with persons outside the family. Overall, participants have a higher frequency of multiple physical and physical/cognitive/emotional problems than non-participants, though the differences are not significant.

Most children attend school (89.9 percent), attend special schools or classes (72.5 percent), and are limited in school attendance due to their disabling conditions (90.0 percent). Few caregivers (5.5 percent) express an unmet need for special schooling. Though differences are not significant, participants are less likely than non-participants to attend school. However, of those who do attend, participants are significantly less likely to be limited in school attendance (88.8 percent, compared to 92.4 percent).

EXHIBIT 5.20. Health Status/Function: Percent Reporting Problems/Using Special Equipment Related to Physical/Social/Overall Functional Capacity (All Children)
Measure	Total(1)	Participants(2)	Non-Participants(3)	Difference¹(2) - (3)
Mob	25.1%	26.5%	22.0%	4.5 pct. pts.
Mob equipment	12.0	13.0	9.9	3.1
Breath	31.8	32.0	32.0	0
Breath equipment	16.1	16.8	14.6	2.2
Eat	10.0	10.8	7.7	3.1
Eat equipment	2.5	3.0	1.2	1.8*
Blind	3.8	4.5	2.2	2.3*
Vision	9.1	8.5	10.4	(1.9)
Deaf	6.2	6.1	6.5	(0.4)
Hearing	9.7	10.4	8.3	2.1
Physical problems	4.6	4.7	4.3	0.4
Physical/cognitive problems	3.5	3.9	2.5	1.4
Source: Caregiver Survey (Wave One)Statistical Significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .051. Negative values in parentheses

Roughly one-fifth of sampled children age 14 and older have jobs. Of those, 41.1 percent are limited in the amount and types of work they can perform. As with schooling, fewer participants of working age have jobs, but of those who do, the percent with work limitations is lower for participants (33.3 percent, compared to 52.0 percent).

EXHIBIT 5.21. Health Status/Function: Percent Reporting Problems with Physical/Social/Overall Functional Capacity (Older Children); Percent with School/Work Limitations
Measure	Total(1)	Participants(2)	Non-Participants(3)	Difference¹(2) - (3)
Children Five Years +
Strenuous activity	37.7%	39.0%	35.1%	3.9 pct. pts.
Toileting	7.2	9.2	3.1	6.1***
Communication problems	13.2	12.5	14.8	(2.3)
Physical problems	5.1	5.9	3.4	2.5
Physical/cognitive problems	4.0	4.6	2.8	1.8
School	89.9	88.8	92.1	(3.3)
Special school	72.5	72.0	73.5	(1.5)
Special need	5.5	6.1	4.5	1.6
Limited school	90.0	88.8	92.4	(3.6)*
Children 14 Years+
Work	19.1	17.6	21.9	(4.3)
Limited work	41.0	33.3	52.0	(18.7)
Source: Caregiver Survey (Wave One)Statistical Significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .051. Negative values in parentheses

Health status, and disability and bed days. Most survey respondents (78.9 percent) report that their children are in good to excellent health at the time of the interview (Exhibit 5.22). Compared to six months ago, a higher percentage (95.8 percent) see no change or an improvement in the child's health, and 94.1 percent expect no change or an improvement in the child's health six months in the future. Participants appear to be somewhat more likely to foresee stability or improvement in the future (94.5 percent, compared to 93.2 percent), even though they seem to be less positive about current health. None of the participant/non-participant differences are statistically significant.

EXHIBIT 5.22. Health Status/Function: Percent Reporting Good to Excellent Health Among Disability and Bed Days
Measure	Total(1)	Participants(2)	Non-Participants(3)	Difference¹(2) - (3)
Good health now	78.9%	78.8%	79.3%	(0.5) pct. pts.
Good health (6)	95.8	96.0	95.4	0.6
Good health (6+)	94.1	94.5	93.2	1.3
Disability days	40.9	41.5	39.6	1.9
Bed days	24.1	23.3	25.7	(2.4)
Source: Caregiver Survey (Wave One)Statistical Significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .05.1. Negative values in parentheses

In a typical month, 40.9 percent of sample members miss one or more days (disability days) from play, school or work due to their conditions. A smaller fraction (24.1 percent) spend at least half a day a month in bed (bed days). Differences between participants and non-participants are small and not statistically significant. Participants seem more likely to have disability days, but less likely to have bed days.

Health status and function and length of enrollment: multivariate analyses. Multivariate analyses that re-estimated statistically significant measures of health status and function showed no relationship of these measures to length of enrollment.

4. Discussion

As hypothesized, particularly in measures of physical function, participants appear to be more impaired than non-participants. However, other measures of function and health provide an unclear message. Some measures suggest lower non-physical impairment among participants, and others do not. Participants seem less likely to attend school or hold jobs, but those who do either are less limited in their activities at school or work than non-participants. Participants and non-participants are equally likely to report disability and bed days.

Predictions that participant caregivers would be more optimistic about the future health of their child are not borne out. And although participants are more likely than non-participants to expect no change or an improvement in the child's health over the next six months, the difference is small and not significant.

H. Utilization and Expenditure

1. Background

Self-reported survey information portrays children in the Demonstration as heavy users of a range of preventive and therapeutic services, relative to non-participants, in all settings and in the home. Participants use slightly more formal care to assist informal caregivers in providing services at home, but the difference between participants and non-participants in this regard is small. Evidence from case record reviews and survey reports suggest that participants have a much higher propensity to use preventive services than non-participants. This section of the report uses claims data to assess whether or not participants use more of a broader range of institutional and ambulatory services than non-participants.

It had been expected that the analysis would also use Medicaid and HSCSN claims to compare expenditure levels between participants and non-participants, to approximate the relative program costs of the Demonstration. As discussed in the next section, problems with claims data make such a comparison impossible to interpret. Instead, the section concludes with a discussion of the program components of Medicaid expenditures for the Demonstration and for members of the comparison group.

2. Measurement Issues

Expenditure and utilization measures were created from claims by aggregating detailed utilization categories taken from HSCSN and Medicaid classification systems. The following selected major categories are used in analyses later in this chapter:

acute inpatient hospital services
early prevention/screening/diagnosis/treatment (EPSDT) services
pharmacy
dental services
durable medical equipment (DME)
outpatient hospital
emergency room (ER)
physician office
vision care
home health
total utilization (includes additional categories)

Measures used in the analyses are described in Exhibit 5.23.

As often happens, reconciliation of different claims and eligibility/enrollment systems produces anomalies that require judgment to correct. For several Demonstration participants, there were HSCSN claims with service dates both before and after the months during which they were actually enrolled. After inspecting the data, it was decided that any HSCSN claims with service dates within six months of a participant's enrollment period would be added to the enrollment month (for claims incurred before enrollment) or the disenrollment month (for claims incurred after disenrollment). For a relatively small number of participants, HSCSN claims beyond these points were dropped from the file.

EXHIBIT 5.23. Utilization and Expenditure Analysis Measures
Measure	Description
Utilization	Utilization greater than 0, based on either Medicaid or HSCSN claims. Measured pre-anchor month and post-anchor (period 1) for categories: Inpatient acute EPSDT Pharmacy Dental DME Outpatient hospital Emergency room Physician office Vision Home health Total (includes additional categories not separately measured)
Total Medicaid expenditure/month	Total Medicaid expenditure per person per month eligible for Medicaid, excluding HMO payments adjusted for medical care price inflation (October 1997 = 100), pre-anchor and post-anchor
Total HSC expenditure/month	Total HSCSN expenditures per person per month eligible for Medicaid and HSC adjusted for enrolled medical care price inflation (October 1997 = 100), post-anchor
Total expenditure/month	Total Medicaid and HSCSN expenditures per person per month eligible, excluding HMO payments adjusted for medical care price inflation (October 1997 = 100), pre-anchor and post-anchor
HSC capitation payments/month	Total capitation payments per HSC participant per month eligible and HSC enrolled
Medicaid capitation payments/month	Total Medicaid HMO capitation payments per non-participant per month eligible for Medicaid and HMO enrolled

For 68 percent of participants, in addition to monthly capitation payments, DC Medicaid apparently paid regular (non-HSCSN) Medicaid claims during the months these participants were enrolled. Early in the Demonstration, HSCSN worked with Medicaid to resolve a perceived problem with physician claims incurred by 420 children. This joint effort was named (and appears in HSCSN claims data as) the Physician Claims Project. Some success was achieved in identifying claims that Medicaid should have paid (for children who further analysis showed had not yet been enrolled), and those that Medicaid should not have paid. However, the evaluator was unable to determine whether or not HSCSN and DC Medicaid had reconciled their claims records, so that the dollar value of Physician Claims Project claims appearing in HSCSN and Medicaid records would reflect the reconciliation. It is quite possible that Medicaid and HSCSN still have duplicate expenditure records in this special category on their files. In addition, there may be other instances of duplicate billing and payment in other categories that HSCSN did not investigate. It was decided that all regular Medicaid claims incurred by participants during enrollment would be retained on the file, to capture total Medicaid expenditures on participants, whether correctly billed and paid or not.

These disturbing findings raise questions about the validity of any estimates of participant "total expenditure" created from merged HSCSN and Medicaid claims. For this reason, no attempt will be made to compare total spending between participants and non-participants. However, one can be somewhat more confident in measures of whether or not particular services were used. Even if a record of payment for a child's service appears as a claim both on Medicaid and HSCSN data files, it is reasonable to infer that the child used that service, whether Medicaid or HSCSN paid for it. So service utilization rates are compared between participants and non-participants.

A final note concerns the use of multiple periods after the anchor months to compare utilization rates for participants and non-participants. Both samples show some attrition from the Demonstration and from SSI-eligibility after the first six months (after enrollment, for participants, and after the assigned anchor month, for non-participants), and this attrition may have effects on utilization and other measures. In order to simplify the presentation, comparisons are based on the pre-enrollment period and first post-enrollment six-month period for participants, compared to the pre-anchor period for non-participants.

3. Findings

Patterns of utilization: participants and non-participants. As discussed earlier, and as confirmed in Exhibit 5.24, Demonstration participants appear to have been heavier users of Medicaid-reimbursed services than non-participants before enrollment in the Demonstration. Contrasts of participants with non-participants in the "pre-anchor" period, and between participants after enrollment with non-participants in the pre-anchor period, are all statistically significant at a p-level of .05 or lower.⁵⁷ For each category, the percent of participants using the service exceeds the percent for non-participants, ranging from a small difference of 1.5 percentage points in EPSDT use (2.4 percent of participants and 0.9 percent of non-participants) to 12 percentage points for physician office use (36.6 percent of participants, compared to 24.6 percent of non-participants). Overall, participants exceed non-participants by 29.5 percentage points in total or "any" utilization (87.6 percent, compared to 58.1 percent).

EXHIBIT 5.24. Service Utilization of Participants and Non-Participants: Selected Categories and Total, Pre- and Post-Anchor: Percent Using Service
Measure	Participants(percent)(1)	Non-Participants(percent)¹(2)	Difference(1) - (2)
Inpatient acute Pre-anchor Post-anchor (1)	9.9%11.5	7.8%7.8	21. pct. pts.3.7*
EPSDT Pre-anchor Post-anchor (1)	2.420.0	0.90.9	1.519.1***
Pharmacy Pre-anchor Post-anchor (1)	34.746.2	32.432.4	2.313.8**
Dental Pre-anchor Post-anchor (1)	15.224.3	8.38.3	6.9*16.0*
DME Pre-anchor Post-anchor (1)	12.917.1	7.87.8	5.19.3*
Outpatient hospital Pre-anchor Post-anchor (1)	37.933.9	24.524.5	13.4*9.4*
ER Pre-anchor Post-anchor (1)	11.915.2	7.77.7	4.27.5*
Physician office Pre-anchor Post-anchor (1)	36.655.5	24.624.6	12.0*30.9*
Vision Pre-anchor Post-anchor (1)	6.310.2	2.82.8	3.57.4**
Home health Pre-anchor Post-anchor (1)	4.913.4	2.92.9	2.010.5***
Total Pre-anchor Post-anchor (1)	87.699.3	58.158.1	26.5*41.2*
Source: HSCSN/DC Medicaid Enrollment/Claims Data BaseStatistical Significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .05.1. Same period rate used for non-participants.

Once enrolled, participant utilization increases in all categories in Exhibit 5.24 except outpatient hospital. Nearly all participants (99.3 percent) use some Medicaid or HSCSN-reimbursed services during the first six months in the program. Large relative increases are observed for EPSDT (17.6 percentage points, an increase of over 600 percent) and home health (8.5 percentage points, an increase of 173 percent).

Early enrollees may differ in prior utilization from later enrollees, if selection is occurring. Because sampling was done sequentially, beginning with participants and eligible non-participants early in the Demonstration, participants in the first "window" will be the earliest to enroll in the program, while participants sampled later will be more likely to have enrolled later. Exhibit 5.25 shows that later-sampled participants (Window 4) are more likely to use most services than participants sampled early in the Demonstration (Window 1). Exceptions are DME, ER, vision care, and total utilization. After enrollment, Window 4 participants use more inpatient, EPSDT, pharmacy, dental, ER, physician office and vision care than Window 1 participants. Total utilization is, in fact, virtually the same for participants sampled in both Windows. Interestingly, non-participants sampled in Window 4 are uniformly less likely to use any services than Window 1 non-participants. These data suggest that selection has a time dimension. Earlier enrollees are different from later enrollees, and the average non-participant changes over time as well, as participants enroll and leave the non-participant population. However, the evidence is mixed. Even though a majority of individual measures suggest that those least likely to use services enrolled first, some measures, particularly total utilization, suggest the opposite.

Participation in the Demonstration has different effects on the "age profiles" of service utilization. Exhibit 5.26 shows utilization rates for selected utilization categories and for "any" (or total) utilization, in four age groups. Utilization rates for participants in six month periods before and immediately after enrollment are compared to rates for non-participants over a comparable six-month period. Although each utilization category is different, there are several generalizations that can be made:

EXHIBIT 5.25. Service Utilization of Participants and Non-Participants: Selected Categories and Total, by Sampling Windows 1 and 4: Percent Using Service
UtilizationCategory	Window 1			Window 4
UtilizationCategory	Participants(1)	Non-Participants(2)	Difference(1)-(2)	Participants(1)	Non-Participants(2)	Difference(1)-(2)
Inpatient acute Pre-anchor Post-anchor (1)	8.3%10.5	8.9%8.9	(0.6) pct. pts.1.6	9.3%13.8	5.2%5.2	4.1 pct. pts.*8.6*
EPSDT Pre-anchor Post-anchor (1)	1.918.4	1.71.7	0.216.7**	3.027.5	1.51.5	1.526.0***
Pharmacy Pre-anchor Post-anchor (1)	49.455.0	33.233.2	16.2*21.8*	53.967.7	30.130.1	23.8*37.6*
Dental Pre-anchor Post-anchor (1)	13.626.1	9.09.0	4.617.1*	22.728.6	6.76.7	16.0*21.9*
DME Pre-anchor Post-anchor (1)	14.719.8	8.78.7	6.0*11.1*	11.216.7	4.14.1	7.1*12.6*
OP - Hospital Pre-anchor Post-anchor (1)	39.138.4	25.925.9	13.2*12.5*	42.837.5	19.719.7	23.1*17.8*
OP - ER Pre-anchor Post-anchor (1)	13.114.2	9.09.0	4.15.2	12.619.7	5.25.2	7.4*14.5*
Phys office Pre-anchor Post-anchor (1)	36.760.0	25.625.6	11.1*34.4*	39.861.7	19.019.0	20.8*42.7*
Vision Pre-anchor Post-anchor (1)	7.99.1	3.63.6	4.35.5*	5.69.7	2.22.2	3.47.5***
Home health Pre-anchor Post-anchor (1)	4.018.3	2.62.6	1.415.7***	4.810.4	1.11.1	3.79.3*
Total Pre-anchor Post-anchor (1)	89.899.7	58.158.1	31.7*58.1*	85.999.6	53.553.5	32.4*46.1*
Source: HSCSN/DC Medicaid Enrollment/Claims Data BaseStatistical Significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .051. Negative values in parentheses

In almost every age group and utilization category, participants use more services before enrollment than non-participants.
Age profiles often differ in shape between participants and non-participants, even before participants' enrollment in the Demonstration.
- For two services, (EPSDT and home health) there is a negative linear relationship between age and use. Younger children use more services than older children, a characteristic pattern for both participants and non-participants, albeit at different absolute levels.
- For most services, the age profile more nearly resembles an "inverted-U", with relatively low utilization by children both under five years of age and adolescents 18 and over. Examples of this pattern for non-participants are pharmacy, dental, DME, outpatient hospital, physician's office and total utilization. For participants, this pattern characterizes dental and total utilization. However, for pharmacy, DME, outpatient hospital and physician's office, the youngest participants utilize services at rates that equal or slightly exceed use by older children, a characteristic of this population that tends not to occur among non-participants.
- For a few services, the oldest children (18+) tend to have higher utilization rates than the youngest children (0-4). For participants before enrollment and non-participants, this is true for dental services, DME, ER and vision services. Interestingly, for these categories, enrollment in the Demonstration raises the utilization rates of the youngest participants to equality or above rates for the oldest children.
After enrollment, the absolute (percentage point) difference between participants and non-participants increases in almost all age and utilization categories.
Enrollment appears to flatten age profiles for some utilization categories. This is most obvious for total utilization. Nearly 100 percent of participant children in each age category use some services during the six-month period after enrollment.

EXHIBIT 5.26. Utilization by Age, Participant Status: Pre-Anchor Periods: Percent Using Service
UtilizationCategory	Participants
UtilizationCategory	Pre-Enrollment(1)	Post-Enrollment(2)	Non-Participants(3)	Difference(1)-(3)	Difference(2)-(3)
Inpatient acute 0-4 5-12 13-17 18+	21.4%8.96.810.9	20.9%11.66.813.4	9.5%11.33.96.1	11.9 pct.pts.*(2.4)2.91.8	11.4 pct.pts.*0.32.97.3*
EPSDT 0-4 5-12 13-17 18+	6.63.11.50.6	33.224.616.210.9	1.31.90.50.1	5.3**1.21.00.5	20.2*22.715.710.8*
Pharmacy 0-4 5-12 13-17 18+	56.155.438.246.1	66.360.047.554.7	17.739.533.233.7	38.4*15.95.012.4***	48.6*20.514.321.0*
Dental 0-4 5-12 13-17 18+	3.121.113.211.5	28.128.819.919.5	0.513.212.26.1	2.67.9**1.05.4	27.6*15.67.713.4*
DME 0-4 5-12 13-17 18+	17.918.87.86.2	30.123.98.310.1	4.612.66.16.4	34.6*11.24.110.6***	28.5**4.60.412.1***
OP - Hospital 0-4 5-12 13-17 18+	49.544.128.233.3	43.437.524.534.8	14.932.924.122.7	34.6*11.24.110.6***	28.5**4.60.412.1***
OP - ER 0-4 5-12 13-17 18+	9.211.410.115.8	26.015.011.515.6	4.47.54.111.5	4.8*3.96.0**4.3	21.6*7.57.4*4.1
Phys office 0-4 5-12 13-17 18+	54.640.128.732.5	62.861.849.548.4	19.028.821.225.8	35.6*11.37.56.7*	43.8*33.028.322.6*
Source: HSCSN/DC Medicaid Enrollment/Claims Data BaseStatistical Significance (prob value): *Less than .01, Less than .05 and greater than/equal to .01, *Less than .10 and greater than/equal to .051. Negative values in parentheses

Despite the variations across categories in age profiles and pre/post enrollment utilization, the key finding remains that, in almost every instance, utilization rates increase for participants, both above their pre-enrollment levels and relative to non-participants.

Medicaid and HSCSN expenditures. DC Medicaid spent considerably more on the Demonstration than capitation payments, in the form of claims paid for services directly to providers, as Exhibit 5.27 shows.⁵⁸ It has already been noted that participants incurred higher Medicaid expenditures before enrollment than non-participants, measured in a comparable time period. Before the anchor month, Medicaid was paying an average of $1,151 per person per eligible month in fee-for-service claims for children who eventually enrolled in the Demonstration, compared to $1,056 for non-participants. Payments for non-Demonstration managed care averaged a negligible $6 per person per month, reflecting the fact that few SSI-eligible children not enrolled in HSCSN are enrolled in other managed care organizations.

After enrollment, during participants' first six months of enrollment in the Demonstration, DC Medicaid paid HSCSN an average $840 per participant per month in capitation, but also paid providers an average $562 per participant per month for services billed directly to MAA. This capitation payment is lower than the capitation rate ($997.71, trended forward each year at 4.6 percent), because DC Medicaid did not pay for every HSCSN participant-month. Medicaid spending per month for participants totaled $1,402, compared to $1,062 during a comparable period for non-participants. At the same time, according to claims records maintained by HSCSN, the Demonstration paid providers an average of $686 per participant per month. How frequently the program was double-billed by providers is unknown.

EXHIBIT 5.27. Total Expenditure Analysis: Medicaid, HSC and Capitation Expenditure per Month, Participants and Non-Participants
Measure	Participants		Non-Participants
Measure	Pre-Anchor	Post-Anchor (1)	Non-Participants
Medicaid claims payments expenditure per month	$1,151	$562	$1,056
HSC claims payments per month	---	$686	---
Total claims expenditure per month	$1,151	$1,248	$1,056
Medicaid capitation payments/month	$6		$6
HSC capitation payments/month	---	$840	---
Total expenditure Medicaid HSC	$1,157---	$1,402$686	$1,062---
Source: HSCSN/DC Medicaid Enrollment/Claims Data Base

This exercise shows that program expenditures for the average Demonstration participant were much higher than expenditures for non-participants, but it does not suggest how much of this expenditure differential was driven by greater demand or need among participants and how much simply reflects failure to eliminate inappropriate billing by providers. As noted earlier, because of the failure fully to coordinate eligibility information between the Demonstration and DC Medicaid, and the resulting confusion among providers as to who was or was not a participant, duplicate billing and expenditures was a significant problem, though the size of this problem is unknown. Therefore, it is impossible to analyze "total spending" in any meaningful way.

4. Discussion

Claims-based data, supported by data from the Caregiver Survey, suggest that participants in the Demonstration used more of almost all services after enrollment than in the period before enrollment. Whether or not this increase represents better access to appropriate services within HSCSN, more but not necessarily appropriate utilization, data problems or some combination, is impossible to say with certainty. Some trends based on claims -- for example, the large increase in EPSDT utilization -- are supported by findings from the Caregiver Survey and from the Case Record Review. Nonetheless, the uncertain validity of both HSCSN and Medicaid claims records suggests extreme caution in interpreting these estimates.

Although the question of whether or not the Demonstration was "cost-effective" cannot be answered with available data, it is clear that government expenditure on the Demonstration far exceeded the capitation payments specified in the waiver agreement. Moving beyond this straightforward assertion is difficult. If one could sort out the claims that are not duplicates and that are justified and appropriate to the care objectives of the child, one could gain insight into the real resource costs of care under the Demonstration. Unfortunately, there is probably no way to accomplish this task without extraordinary expenditure of time and resources and, in the end, no amount of effort is likely to be able to resolve all uncertainties.

I. Conclusion and Implications of the Impact Analyses

Based on analyses of survey and claims data, the Demonstration appears to have had a positive impact on caregiver satisfaction. However, with the exception of measures that show participants to be more likely to use preventive services than non-participants, findings on other potential impacts (health status, functional outcomes, costs, utilization) are inconclusive.

Although many estimates show differences that are apparently in the "right direction," (e.g., lower relative burden for participants' caregivers), most do not achieve accepted levels of statistical significance. Problems with significance are partly a function of the relatively small size of the Caregiver Survey sample (740 participants and 323 non-participants). In addition, however, fundamental weaknesses in the research design may place limits on the evaluator's ability to detect effects. Key problems include 1) voluntary participation in the Demonstration leads to potential selection bias which can never be wholly eliminated from impact estimates; and 2) lack of true baseline measures of outcomes in the survey, which makes it impossible to say definitively that any differences between participants and non-participants under the Demonstration did not exist earlier, before the Demonstration. In addition, concerns about possible double billing raise questions about the reliability of DC Medicaid and HSCSN claims, which, in turn, limit the value of claims-based cost and utilization measures.

With these limitations in mind, the evaluator compared participants to non-participants over a wide range of outcome domains.

1. According to most measures, Demonstration participants enjoy equal or better access to services, compared to non-participants. Participants also generally express greater satisfaction with access to services.

Participants are more likely than non-participants to have usual places and professionals that provide medical and preventive services, and less likely to report unmet needs. Some, but not all, of these measures of access are statistically significant. Findings for satisfaction with access are more persuasive. Of seven specific categories of services and products, participants are more satisfied than non-participants with access to five. Participants are also more satisfied overall, with 91.6 percent expressing good to excellent satisfaction, compared to 85.8 percent for non-participants. Generally, the inference that satisfaction is associated with participation in the Demonstration is strengthened by the fact that for most service-specific and overall measures, participants enrolled in the Demonstration for more than six months are more satisfied than participants with shorter stays.

These findings are not surprising. Family focus groups show that, despite some specific concerns, families of participating children generally view the Demonstration as a major improvement over what they had before.

2. Despite positive perceptions of the Demonstration expressed by caregivers, there is little consistent evidence that the Demonstration is seen to reduce caregiver burden.

While there is some evidence that caregivers for children participating in the Demonstration experience fewer job-related problems than non-participants' caregivers, there is no statistically significant difference between groups in the frequencies of other problems. Also, participants use more formal home care than non-participants, but the differences are small and not statistically significant -- few (5.9 percent of all families) use any formal home care at all. It seems clear that most caregivers of both participating and non-participating children are heavily invested in their roles and, for whatever reason, are reluctant to request help from formal service providers.

3. Participant caregivers report heavier use of medical services than non-participant caregivers. However, the most salient difference is in preventive services: participants are more timely in their use of preventive services than non-participants.

Claims data used in Chapter III to assess the extent of selection bias show participating children to be more likely than non-participants to use Medicaid-reimbursed services before enrollment than non-participants over a comparable time period. Data from the survey support the finding that participants are "high users." The difference is most striking in measures of timeliness of preventive services utilization -- fewer participants are not timely in routine/immunization visits, dental visits, and eye examinations. These findings are further strengthened by information abstracted from case records and reported in Chapter IV. Records showed that participants are much above the national averages in timeliness of EPSDT exams, with percentages ranging from 60 percent timely for dental exams to 73 percent for hearing exams (compared to 1996 national EPSDT averages of 21 percent and 13 percent respectively).

4. In general, Demonstration participants appear to be more physically impaired than non-participants, but other measures of mental, emotional and developmental impairment provide no clear pattern. Despite higher reported satisfaction with access, participants' caregivers are not more positive about the child's current or future health status.

Lacking baseline measures, there is no satisfactory way to estimate the extent to which the Demonstration does or does not improve participants' health status and functional levels. Claims data, and survey reports of somewhat heavier use of medical services, suggest that participants might be more impaired than non-participants. This finding should be confirmed in the survey data. However, 1) there is no consistent pattern of differences between groups in measures of cognitive, emotional development and behavioral problems; and 2) over several measures of physical function, impairments are somewhat more frequently reported for participants than non-participants, though the differences are not statistically significant.

Most children are reported to be in good to excellent health. Caregivers of participants seem somewhat more likely to foresee stability or improvement in their child's health in the future (99.4 percent, compared to 93.2 percent for non-participants), but the difference is not statistically significant.

If the Demonstration has any impact at all on function or health status, it might be observable as more favorable reported outcomes for longer-stay (over six months) participants, compared to shorter-stay participants. However, no statistically significant findings were detected.

5. Participants use more of most ambulatory and institutional services than non-participants, both before and after the Demonstration. Although it is possible to compute how much the Medicaid program spent on the Demonstration, it is impossible to compare expenditures between participants and non-participants, due to claims data problems.

Before enrollment, participants are heavier users of almost all institutional and ambulatory services than non-participants. Overall, 87.6 percent of participants use "any" service, compared to 58.1 percent of non-participants. Once enrolled, participant use increases in all categories except outpatient utilization, and nearly all participants (99.3 percent) use some Medicaid or HSCSN-reimbursed service. Age is associated with utilization patterns. For example, younger children are more likely to use EPSDT services, and older children are more likely to use emergency room services. However, regardless of age, participants use more services than non-participants.

DC Medicaid spent more per child per month for participants after enrollment than before, an average of $1,402 after, compared to $1,151 before. (Non-participants incurred $1,056 in Medicaid-reimbursed services, a reflection of relatively lower utilization rates.) However, both DC Medicaid and HSCSH reimbursed providers for similar services during the Demonstration. Clearly, total expenditure data may include services that are double-billed, making it impossible to use actual claims data to analyze expenditure impacts of the Demonstration.

Claims data supported by information from the Caregiver Survey suggest that participants in the Demonstration use more of almost all services after enrollment than before enrollment. Whether or not this increase represents better access to appropriate services within HSCSN, more but not necessarily appropriate utilization, data problems, or some combination of the above is impossible to say with certainty. Some trends derived from claims -- for example, the large increase in EPSDT utilization -- are supported by findings from the Caregiver Survey and from the Case Record Review. Nonetheless, the uncertainty surrounding payments recorded by HSCSN and Medicaid for Demonstration services suggests the wisdom of using extreme caution in interpreting these estimates.

Although the question of whether or not the Demonstration was "cost-effective" cannot be answered with available data, given weaknesses in the Demonstration design, it is clear that government expenditure on the Demonstration far exceeded the capitation payments specified in the initial waiver agreement. Moving beyond this straightforward assertion is difficult. If one could sort out the claims that are not duplicates and that are justified and appropriate, given the care objections of the child's POT, one could gain insight into the real resource costs of care under the Demonstration. Unfortunately, there is probably no way to accomplish this task without extraordinary expenditure of time and resources, and, in the end, no amount of effort is likely to resolve all uncertainties.

VI. CONCLUSION

For nearly four years, under a Medicaid waiver, HSCSN has provided managed care services to a majority of the SSI-eligible children in DC. When first implemented, this Demonstration was unique. Though other, similar programs now operate elsewhere in the country, findings from this experiment are still the first evidence available on whether or not a specialty managed care model for children with special needs will succeed. As one might expect, the evidence is mixed and inconclusive.

Enrolled children are different from eligible children not enrolled. It is no surprise that, through enrollment and disenrollment, there has been self-selection into and out of this voluntary Demonstration. However, the nature of this selection was unexpected. Participating children appear to be "sicker" than non-participants. Though definitive measures of baseline health of participants and non participants are unavailable, measures of utilization before enrollment for participants, compared to similar time periods for non-participants, show participants to be higher users in total and in almost all institutional and ambulatory service categories. Though few measures of impairment from the Caregiver Survey show statistically significant differences between participants and non-participants, caregivers tend to report higher levels of physical impairments among participants. Although it is always risky to conclude that higher use means greater health needs, and although caregiver reports are somewhat inconclusive, it does seem clear that selection effects in the Demonstration have been neutral to adverse, not favorable.

HSCSN does appear to "add value" for many caregivers. Focus group participants and survey respondents were generally positive about the Demonstration. Caregivers of Demonstration participants are more satisfied with their ability to gain access to needed services, and have achieved some stability in relating to a usual physician to handle most of their child's needs. There is some evidence that caregivers of participating children are somewhat more likely than non-participant caregivers to mitigate the burden of their responsibilities, by engaging formal skilled and unskilled assistance in the home. Because no true baseline measures are available, it is impossible to be confident that these differences represent impacts of the Demonstration. Nonetheless, one can conclude that a positive impact on caregiver satisfaction was highly probable, based on the size, statistical significance, and consistency among multiple measures.

Given problems with claims data, there is no way to measure the relative costs incurred to serve children in the Demonstration. Enrolling children in the Demonstration is probably more costly for Medicaid than keeping them in fee-for-service. Claims-based data show that the average participant uses more of most services after as well as before enrollment. Nonetheless, problems with claims records make it impossible to estimate Medicaid expenditures in order to compare participants to non-participants.

HSCSN achievements at the caregiver/child level were gained despite turmoil and apparent inefficiency at the top of the organization. Over the course of the Demonstration, HSCSN suffered from misguided investment decisions (to develop original care management software) and management instability (including nearly total turnover in top management). In three of the four years of operation, HSCSN apparently suffered sizable financial losses. Demonstration officials believe that the Demonstration model is not sustainable, because the participant base is too small to spread risks in a way that makes capitation feasible. HSCSN does not now plan to compete for a contract under the new waiver.

Although the managed care model tested in this Demonstration may not be viable, care management can be a valuable service for children with special needs. In the first year of the Demonstration, DC Medicaid paid HSCSN $997.71 per enrolled child per month, based on estimates from historical claims, with certain high-cost items (e.g. transplants, public school services) carved out. DC Medicaid increased this rate annually by 4.6 percent to adjust for inflation. In addition, to protect HSCSN against unforeseen financial losses, the waiver agreement includes a risk sharing arrangement, keyed to the medical claims ratio (actual claims expenses, as a percent of total capitation payments), with a target ratio of 85 percent. If claims expenses increase relative to capitation payments above the target, DC Medicaid assumes an increasing share of potential losses through additional payments to HSCSN. If claims expenses fall and the actual ratio drops below the target, DC Medicaid shares the savings through payments from HSCSN. In fact, although HSCSN did make risk-sharing payments to the District in the first two years, actual expenses exceeded revenues in all years, and the District made risk-sharing payments to HSCSN of nearly $700,000 in the last two years.

This experience forced the District to re-examine the financial model for the Demonstration re-competition (called the Child and Adolescent SSI Program, or CASSIP). Although based on capitated payment and risk sharing similar to the current Demonstration model, the new payment system will include stop-loss protection to cap annual claims expenditures for high-cost patients in three categories of service: 1) inpatient expenditures greater than $100,000 for any child, 2) pharmaceutical expenditures greater than $25,000 for any child, and 3) home care expenditures greater than $12,000 for any child.

Even if a full capitation system cannot reasonably be expected to succeed with a population of small size and complex needs, some method of encouraging more comprehensive care management for SSI-eligible children with special needs should be considered. Current plans in DC to open competition for these contracts to main-line MCOs may not achieve the desired objectives. While HSCSN may have avoided the expected pattern of favorable selection, main-line MCOs that contract to serve DC Medicaid-eligible children may not be able to avoid the temptation to enroll children with relatively low expected resource needs. If special needs children become part of the larger population of covered lives, paid under a common capitation rate, limited research suggests that barriers to access may actually increase. If special needs care is "carved out" and paid on a fee-for-service or partial capitation basis, then it is not clear why, except for administrative convenience, services under the carve-out should be provided through arrangements with MCOs.

Children in this Demonstration appear to have benefited from care management by professionals who understand their needs. The fact that this system operates under capitation may or may not affect provider, family and plan behavior, but the assumed efficiency-enhancing effects of capitation are not at all obvious. Further analysis of special managed care initiatives in other sites, and of programs that main-line MCOs have implemented to address the needs of these children, could provide better information on the relative importance of financial and other incentives on behavior and clinical outcomes.

In the meantime, policy makers might consider alternative models of financing care management. Programs that pay a child's principal physician directly to provide coordination and integration have been used by state Medicaid programs in rural areas not served by MCOs. There is currently no information on how effective these programs have proven to be. Alternatively, Medicaid might contract with other health care professionals, for example, groups organized by nurses and social workers with specialized knowledge of children with special needs, to provide care management services directly. Neither of these proposals avoids thorny questions of how to price and pay for care management services. There should still be incentives, perhaps based on measures of performance, for physicians or care management professionals, to achieve appropriate objectives of cost-effective care coordination and integration.

NOTES

HCFA Contract Number 500-96-0003 T.O. #3.
In addition to cross-sectional surveys of participants and non-participants, the evaluator attempted to follow up a sub-sample of both groups with the same instrument, roughly six months after the initial survey. Logistical issues, discussed in more detail in Chapter 5, reduced the numbers of follow-up surveys completed. Therefore, the impact analyses use limited information on change measures, largely to corroborate findings from cross-sectional analyses. Also, the evaluator planned to survey new participants soon after enrollment into the program, in order to track changes in participant characteristics and, hopefully, attain better information on key indicators at baseline. This effort failed, due to inadequate contact information, as described below in Chapter 5.
See Abt Associates, 1998a.
Note that there were a number of other staff changes at the hospital at this time, including the hospital's head of marketing; a top financial official; the chief professional fund-raiser; the directors of social work, occupational therapy, and physical therapy; and others.
This figure is based on later information than, and differs slightly (by $4,587) from, the figure reported in Abt's first evaluation report (Abt Associates, 1998).
Details of the financial terms in this paragraph are from Agreement between the District of Columbia (CHCF), and Health Services for Children with Special Needs, Inc., dated November 30, 1995, Article 7. The inflation terms are detailed in HCFA, "Health Care Financing Administration Special Terms and Conditions," project no. 11-W-00021/3, dated October 13, 1995.
The second evaluation report noted that -- in 1997 and 1998 -- HSCSN more or less had broken even on a cash basis, following large losses in the first year. One possible interpretation was plausible: a) large losses were incurred in the first year (1996), when a new way of working had to be learned, and costly difficulties (e.g., in locating beneficiaries) were discovered and had to be resolved; but b) something like break-even results were obtained thereafter and might continue to be obtained. That was the "optimistic" interpretation offered in the second evaluation report.
Reasons for cost increases from interviews with HSCSN, January 1999.
A senior manager at HSCSN noted one small consequence of the small population served in the District: the number of enrollees was too small to make a decision based on differential outcomes. In the opinion of this manager, it would help if similar programs across the country could pool there data to permit more robust outcomes analysis, to identify best practices.
This resolution is summarized in Letter to Thomas Chapman, CEO, HSCSN, from Herbert H. Weldon, Jr., Deputy Director MAA, dated December 1, 1999; and Commission on Health Care Financing, DC Department of Human Services, Administrative Ruling, with respect to Agreement with Health Services for Children With Special Needs, Inc., dated December 1, 1999.
Letter from Thomas Chapman, CEO, HSC Foundation, to Paul Offner, Commissioner, Commission on Health Care Finance, dated May 27, 1998, re: five issues key to the future success of the program.
The principal compensation for this reduction in the size of the enrollment pool will be the possibility of some economies of scale for bidders that (unlike HSCSN) also bid to become TANF HMOs (i.e., HMOs in the conventional Medicaid program).
Pardoxically, as was possible in the Demonstration, a contractor that excels at managing difficult cases might attract the most difficult cases -- and thus suffer selection bias except as relieved by the risk-sharing and (in CASSIP) stop-loss provisions.
The City Council exempted the Demonstration contractor from the HMO requirement. Apparently, in a Demonstration with voluntary enrollment and other protections, the Council and CHCF were comfortable with waiving the requirement. For the recompetition, the District has decided that the contract requirements are to be "normalized": if a contractor is taking risks under the terms of the HMO statute, it should be a licensed HMO.
HSCSN, "DC Medical Managed Care for Children with Special Needs: Annual Report, 1997 (Final)," report submitted to the District of Columbia Commission on Health Care Finance, undated.
Interview at HSCSN, 1999.
This latter conflict is complex. Basically, First Health paid some claims for HSCSN enrollees that they should not have paid (because the children were Demonstration enrollees, not Medicaid fee-for-service beneficiaries). But HSCSN takes issue with the payments because some of the payments should not have been made -- HSCSN should have been in a position to control these payments, if it was to be liable for them. District systems did not make that kind of control possible.
To give some idea of the complexities of different coverage involved here, consider the following: Foster care covers a full spectrum of medical needs, including outpatient substance abuse treatment. Medicaid does not cover outpatient substance abuse treatment. For a Medicaid patient with only a substance abuse diagnosis in a TANF HMO, there is no way even to submit the case to the Residential Planning Unit Committee that reviews RTF placements. HSCSN, whose coverage derives from Medicaid, does cover outpatient substance abuse treatment, as a cost-effective means to avoid more expensive treatments.
The essence of this problem is that, in order to set a capitation rate and manage care, it is necessary to specify a group in advance. But outside the SSI population, it is almost impossible to specify in advance who is at risk of RTF placement, i.e., who is in the population to be managed and whose costs should be put into a rate.
The care coordination program was to be operated by a partnership between the Public Benefits Corporation and other entities. The partnership developed a network of fee-for-service providers in the District, Maryland, and Virginia and was to provide care planning, care coordination, and network management services for foster children, as well as special reporting on utilization to the foster care receiver.
Medical Assistance Administration, District of Columbia, "Medicaid Managed Care: Phase II Plan for Children with Special Healthcare Needs," dated May 3, 1999.
Perhaps the clearest sign that the dust had settled was when the last holdout among the day treatment centers, PSI, signed on as a provider to HSCSN in late 1997.
The words of one senior official in the MAA.
HCFA, "Health Care Financing Administration Special Terms and Conditions, Managed Care System for Disabled and Special Needs Children," project no. 11-W-00021/3, dated October 13, 1995, Section IV.C.
Interview at HSCSN, 1998.
Medical Assistance Administration, "Answers to HCFA's One-Year Assessment Questions," October 30, 1998.
Letter from Smith to Offner, dated November 24, 1998, re: granting one-year extension. The principal conditions of the extension were that: (1) relevant provisions of BBA were to be incorporated into the extension, once HCFA issued the regulations; and (2) by the end of the extension, final year two and preliminary year three data on budget neutrality were to be submitted, managed care plans were to be selected through a competitive process, and assurances were to be provided that Medicaid enrollment reported by HSCSN matched the member months on which capitation was paid.
Interview, December 1999.
The Public Benefit Corporation also oversees the public health clinics in the city.
Terms of the extension are from Letter to Thomas Chapman, CEO, HSCSN, from Herbert H. Weldon, Jr., Deputy Director, MAA, dated December 1, 1999, re: contract amendments.
One caution is appropriate in labeling this Demonstration as a five-year effort: parties to the Demonstration could not rely on its being more than a three-year venture (the two additional years were sequential expendients), and thus could hardly make plans or investments on a five-year basis.
Numbers include new and existing participants for that quarter.
This information was obtained from the HSCSN Enrollment File and the HSCSN Assessment File. These diagnoses were taken at the time of enrollment, and are not necessarily the same as diagnoses captured on claims (see Exhibit 3.9).
Some children may not have been SSI-eligible until they enrolled in HSCSN. These will not be represented in this sample.
In isolation, this is a meaningless number, because the sample mean disenrollment rate is .16.
The Acting Medical Director has been in his current position since March 1999, but has a long history of involvement in the Demonstration as Vice President of Medical Affairs and Patient Care Services at HSC Health Care System, the parent organization overseeing HSCSN.
See Section IV.B for more detail.
Endodontics, periodontics, orthodontics, oral surgery, root canals, sealants, prosthodontics, and scaling are dental procedures that require prior authorization.
This is discussed in the preceding section.
These categories are not mutually exclusive, thereby accounting for why the sum total is greater than the number of providers interviewed.
HIV/AIDS was much less prevalent than all other diagnoses. However, it was included in the sample because of the great need for services associated with this diagnosis.
This finding is based on 58 of the 102 participants in the sample but is believed to be an accurate representation of the entire sample. For these 58, records were hand-counted to measure what percentage of POTs contained PCP signatures.
Adapted from Olson K, Perkins J, and Pate T. Executive Summary. Children's Health under Medicaid, A National Review of Early Periodic Screening Diagnosis and Treatment. National Health Law Program, August 1998.
This was determined by the nurse case reviewers who used clinical judgement to determine if a particular health problem was addressed in an appropriate timely manner.
For the purpose of this analysis, details of the disorder were not taken into account to understand which children received special education. Therefore, it is unknown if the brain-injured children was too injured to have received special education.
Round one also included one session with parents who had declined enrollment in the Demonstration.
Of course, this assumes that DC Medicaid paid capitation payments for children only while they were enrolled in HSCSN. As shown later in this report, this assumption was incorrect.
The child was determined to be the eligible respondent if the answer to the question, "Who in your household provides most of the care for (CHILD) and knows the most about (his/her) health care needs?" was the child. In fact, one 2 percent of all respondents were sampled children.
Adjusted for child age, gender, race, relationship to the respondent, and household socioeconomic status, the inter-group difference in the percent with a usual physician declined somewhat (to 7.7 percentage points) but remained highly significant.
Adjusted for child age, gender, race, relationship to the respondent and household socioeconomic status, this difference declined to one percentage point, but remained significant.
All measures of satisfaction for which inter-group differences were statistically significant at 0.1 or lower were re-estimated, adjusting for child age, gender, race, relationship to the respondent and household socioeconomic status. Estimated differences either remained the same or declined slightly, but in no case did the direction of impact change or level of statistical significance decline.
The difference in monitoring activities, when subjected to adjustment for child age, gender, race, relationship to the respondent and household socioeconomic status, decreased to 2.4 percentage points and was no longer statistically significant.
When adjusted for child age, gender, race, relationship to the respondent and household socioeconomic status, differences between participants and non-participants in hours and promotion measures shrank, to 4.7 and 5.3 percentage points respectively, and were no longer statistically significant. The difference in frequency of caregivers who never worked also declined, to 7.1 percentage points but remained statistically significant at a five-percent level.
When re-estimated using multivariate regression that adjusts for child age, gender, race, relationship of the respondent and household socioeconomic status, the absolute differences for immunizations and dental services decline (to 2.6 and 6.6 percentage points respectively), but both remain statistically significant at the five-percent level.
Three services listed in the survey were dropped from the analysis because they were used by four or fewer children in the total survey sample. These were dialysis, chemotherapy and ostomy care.
Frequencies for inhalation therapy at home, tube feeding and physical therapy were re-estimated adjusting for child age, gender, race, relationship to the respondent and household socioeconomic status. Absolute differences declined, to 1.3, 1.4 and 4.9 percentage points respectively. Of the three adjusted differences, only the difference for tube feeding was statistically significant, at a ten-percent level of significance.
Non-participant utilization should not change from before to after the anchor month, because selection of the anchor month is determined by sampling rules, not (as in the case of participants for whom the anchor month is the Demonstration enrollment month) by a major decision/activity that might affect utilization. Therefore, participant utilization measures (pre- and post-anchor) are always contrasted to non-participant pre-anchor measures.
Non-participant expenditures per month did not change post-anchor. Therefore, for simplicity, participant pre- and post-anchor expenditures are compared to non-participant pre-anchor data in Exhibit 5.27.

REFERENCES

Abt Associates Inc. Evaluation of the District of Columbia's Demonstration Program, Managed care system for disabled and special needs children. Second Annual Report. December 18, 1998.

American Academy of Pediatrics. Managed care and children with special health care needs: a subject review. Pediatrics. 1998;102(3):657-660.

Deal LW, Shiono PH. Medicaid managed care and children: an overview. The Future of Children. 1998:8(2):93-104.

Division of Services for Children with Special Needs Maternal and Child Health Bureau. Children with special health care needs in managed care organizations: definitions and identification, family participation, capitation and risk adjustment, quality of care. Summaries of Expert Work Group Meetings. 1996.

Fox HB, Wicks LB, Newacheck PW. Health maintenance organizations and children with special health needs: a suitable match? American Journal of Diseases of Children. 1993a:147:546-552.

Fox HB, Wicks LB, Newacheck PW. State Medicaid health maintenance organization policies and special-needs children. Health Care Financing Review. 1993b:15(1):25-37.

Fox HB, McManus MA, Almeida RA, Lesser C. Medicaid managed care policies affecting children with disabilities: 1995 and 1996. Health Care Financing Review. 1997:18(4):23-36.

Goldstein A. D.C. Medicaid director resigns after 4 years. Washington Post. October 9, 1999: B4.

Halfon N, Newacheck PW. Prevalence and impact of parent-reported disabling mental health conditions among U.S. children. Journal of the American Academy of Child and Adolescent Psychiatry. 1999:38(5):600-609.

Health Resources and Services Administration. Achieving service integration for children with special health care needs: an assessment of alternative Medicaid managed care models. Volume 1: Synthesis of results. July 1999.

Heckman J, Robb R. Alternative methods for evaluating the impact of intervention: an overview. Journal of Econometrics. 1985:30:239-67.

Hill I, Schwalberg R, Zimmerman B, Tison W. Achieving service integration for children with special health care needs: an assessment of alternative Medicaid managed care models. Volume 1: synthesis of study results. Prepared for the Division of Children with Special Health Care Needs Maternal and Child Health Bureau. 1999.

Newacheck PW, Hughes DC, Stoddard JJ, Halfon N. Children with chronic illness and Medicaid managed care. Pediatrics. 1994:93(3):497-500.

Newacheck PW et al. Monitoring and evaluating managed care for children with chronic illnesses and disabilities. Pediatrics. 1996:98(5):952-958.

Newacheck PW et al. An epidemiologic profile of children with special health care needs. Pediatrics. 1998:102(1):117-123.

Olson K, Perkins J, Pate T. Executive Summary. Children's Health under Medicaid, A National Review of Early Periodic Screening Diagnosis and Treatment. National Health Law Program, August 1998.

Orr L. Why experiment? The rationale and history of social experiments. DHHS working paper. July 1996.

Perrin JM et al. Changing patterns of conditions among children receiving supplemental security income disability benefits. Archives of Pediatrics and Adolescent Medicine. 1999:153:80-84.

Social Security Administration. Management of the supplemental security income program, today and the future. On-line report. 1998 at: http:/www.ssa.gov/reports/ssi/ssitoc.htm

Westbrook LE, Silver EJ, Stein REK. Implications for estimates of disability in children: a comparison of definitional components. Pediatrics. 1998:101(6):1025-1030.

Topics

Long-Term Care Financing | Access to Services and Benefits & Services Integration

Populations

People with Disabilities | Children

Location- & Geography-Based Data

State Data

Program

Medicaid

Evaluation of the District of Columbia's Demonstration Program,Managed Care System for Special Needs Children: Final Report

Abt Associates Inc.

June 2000

TABLE OF CONTENTS

Demonstration Implementation and Management

Demonstration Enrollment

Care Management

Impacts of the Demonstration

Policy Implications

1. Overview

2. Case Studies

3. Discussion

1. Startup Costs

1. The DC Public Schools

2. Residential Treatment Facilities (RTFs)

3. Foster Care

5. District Hospitals

6. Relationships with the Community

7. Coordination and Collaboration: Conclusion

1. The First Extension, 1998-1999

2. The Second Extension, 1999-2000

1. Average Age and Length of Enrollment

2. Eligibility and Participation Rates

3. Reasons for Disenrollment

4. Distance Participants Traveled for Services

5. Description of PCPs and their HSCSN Caseloads

1. Selection in Voluntary Demonstrations

2. Data and Methods

3. Findings

1. Introduction

2. Data and Methods

3. Findings

1. Caregiver Survey

2. HSCSN and Medicaid Eligibility and Claims Data

1. Impact Analyses: Concurrent Comparisons

2. Impact Analyses: Change Measurement

3. Claims Based Analyses: Trends In HSCSN Utilization, Comparisons of Participants and Non-Participants, and the "Costs" of the Demonstration

1. Background

2. Measurement Issues

3. Findings

4. Discussion

1. Background

2. Measurement Issues

3. Findings

4. Discussion

1. Background

2. Measurement Issues

3. Findings

4. Discussion

1. Background

4. Discussion

1. Background

2. Measurement Issues

3. Findings

4. Discussion

Connect with Us