Managed Care in Michigan’s Child Welfare System William Meezan Bowen McBeath University of Michigan Paper Presented at CWLA’s Tools the Work and National Research Conference Miami Florida November, 2003
Agencies Involved • Catholic Social Service of Wayne County • Judson Center • Homes for Black Children • Lutheran Child and Family Service • Orchards Children’s Services • Spectrum Human Services • Evergreen Children’s Services • Lutheran Social Services of Michigan • St. Francis Children’s Services
Sources of Funding • The Aspen Institute/Michigan Nonprofit Research Fund • Michigan Family Independence Agency (FIA) • W.K. Kellogg Foundation, Global Program for Youth • The Skillman Foundation • Horace H. Rackham Graduate School, University of Michigan • Office of the Vice Provost for Research, University of Michigan • Office of the Associate Dean for Research, School of Social Work, University of Michigan
New State Initiatives • To control caseloads and costs • Many states have requested federal Title IV-E waivers in order to set up innovative programs • A few states have detached investigative from service delivery functions within their child welfare systems, and assigned the former to law enforcement agencies (e.g., Florida) • Other states have embraced privatization, and have contracted out all or specific portions of their public child welfare systems to private nonprofit agencies • Some states have adopted new service delivery models such as performance contracting and managed care
Spread of Managed Care in Child Welfare • As of 1998, 29 states were operating one or more managed care initiatives for a total of 47 managed care child welfare programs across the US • These initiatives vary considerably in their • Scope: State initiatives covered anywhere from 100 to 35,000 children • Auspices: State initiatives range from being entirely privatized to being fully publicly-provided • Risk structure: Performance contracting, capitation, case rate setting, and other financial principles are present across some but not all initiatives
The Wayne County Pilot Initiative • A performance-based contracting approach to pay for foster care system which uses incentive payments to encourage better outcomes for children in the foster care system • Restructures existing resources in Wayne County to • Attempt to provide for the service needs of the family and the child • Increase the percentage of youth who reach permanent placement • Shorten the time needed to achieve permanence
Rationale for Pilot Payment Structure • Initial and incentive payments are financed through savings achieved by diminishing administrative per diem • Cost neutrality not assumed; state willing to have increased costs should that be necessary to achieve better and more timely permanency plans for children
Evaluation Plan – Three Simultaneous Studies • Process Evaluation • Qualitative interviews with key informants (complete) • Quantitative data on services (analysis complete through 300 days; data now being processed through 570 days; data collection expected through 930 days) • Output Evaluation • Quantitative data on children’s movement through the system (analysis complete through 300 days; data now being processed through 570 days; data collection expected through 930 days)
Participatory Evaluation in Action • Control of the evaluation project was jointly shared by researchers and participating agencies • Meetings with cooperating agencies occurred every two weeks for 9 months to plan the project and its instrumentation; every month thereafter to iron out problems, receive data, disseminate findings • All decisions regarding all phases of the evaluation were discussed and approved by the cooperating agencies (major research questions, design, sampling, data collection instruments and techniques, etc.) • All meetings are archived for future organizational learning • Community Zero website for discussion, postings of instruments, data tracking, etc.
Qualitative Process Evaluation:Research Questions • What organizational accommodations did private nonprofits make when they shifted to a specific managed care contracting environment? • What organizational accommodations did private nonprofits expect to make as they anticipated shifting to a specific managed care contracting environment? • Were there differences between the pilots and the non-pilots in their expectations/realities? • Were there differences between people at various levels and in various positions within the agencies in terms of their expectations/realities?
Method • Telephone interviews were conducted with personnel from pilot and non-pilot agencies • Administrators (N = 45) • Foster care supervisors (N = 19) • Foster care line staff (N = 20) • Agencies were asked to provide the research team with a list of respondents who would best be able to answer questions • Sample chosen to maximize knowledge of foster care and the pilot process • This sampling strategy allowed for the selection of respondents that could best speak to the organizational adaptations that nonprofits made (or were expecting to make) in implementing the pilot process
Sample Characteristics • Pilot and non-pilot respondents had roughly 14 years of experience in the social services, 13 of which were in the field of child welfare • Pilot respondents had been at their agencies for nearly 10 years, six of which had been in their present position • Non-pilot respondents had been at their agencies for about six years, with four years of experience in their present positions • 60% of respondents were Caucasian and 35% were African American • Non-pilot administrators were twice as likely to be African American as were pilot administrators (40% vs. 20%) • Non-pilot line staff were much more likely to be Caucasian than were pilot line staff (71% vs. 39%) • Half of the administrators at pilot and non-pilots agencies were female, while nearly all supervisors and line staff were female • Pilot administrators were more likely to have an advanced degree (76%) than were non-pilot administrators (47%)
The Interview • Interviews were structured and contained mostly open-ended questions • Interviews lasted between an hour and an hour and a half on average, and ranged in length from 30 minutes to five hours • Respondents were asked a series of questions about • Service delivery • Interdepartmental and interorganizational relations • Staffing patterns • Staff roles and responsibilities • Financial management • Technology use • Due to the nature of people’s positions and responsibilities, not all respondents were able to answer all questions
Data Analysis – Qualitative Data • A cross-case matrix was constructed for each question. Each full data matrix contained all responses to the identified question, organized by respondent type and agency type • Two staff independently read through the matrix and each constructed a preliminary list of codes that described common themes across individual responses. • Staff compared their preliminary lists of themes and agreed upon a set of codes • Staff independently applied the codes to the responses, with each staff member noting where a response contained one of the agreed upon themes • Coders compared their judgments and resolved any differences regarding whether a theme was present or absent (initial reliability @ 75%) • Percentages were calculated for the presence of each theme across respondent type and agency type, thus allowing for comparisons between pilot and non-pilot respondents at various organizational levels.
Similar Perceptions Between Pilot and Non-pilot Personnel • Service delivery-related barriers • Accessing community services • Staff-related barriers • Altering staff’s philosophical beliefs concerning the appropriate goals of foster care and the difficulty in moving to a service delivery system emphasizing speed and immediacy of service provision • Need for increased communication, cooperation, and/or meetings with FIA, the court, and other private agencies • Understaffing and/or staff turnover
Similar Perceptions Between Pilot and Non-pilot Personnel • Agency Functions • Need for changes in the area of budgeting and financial management • Need for changes in MIS and the way they tracked children in care • Administrators mentioned that cost centers and/or new accounts receivable and payable had to be (or would need to be) created
Different Perceptions Between Pilot and Non-pilot Personnel • Pilot respondents were more likely to see internal, systemic, operational changes • More meetings, trainings, and communication involving different departments within the agency • Conflict between units within the agency • Particularly between foster care and clinical units • New staff positions had to be created • Need for communication about MIS, and communication between MIS and other agency departments • Changes both within the foster care department and in interactions with other agency department • Changes that would have to occur in • Budgeting • Forecasting • Tracking • Accounting systems
Different Perceptions Between Pilot and Non-pilot Personnel (con’t) • Non-pilot respondents did not anticipate the scope of operational changes in service delivery and agency administration it would take to move to the pilot • Less likely to anticipate that their agency would have to alter their procedures when dealing with foster care cases • Less likely to anticipate changes in roles and responsibilities • Did not see that new linkages within the agency would have to be created • Personnel in non-pilot agencies were less optimistic about the impact of the pilot on the clients it served • Those in non-pilot agencies tended to see the move to this system of reimbursement as more difficult than it had proven to be in the pilot agencies • Those in non-pilot agencies had a fear of revenue loss that was not experienced by the pilot agencies
Similar Perceptions Among Staff at Various Organizational Levels • Agencies would have to do business differently in order to succeed in the managed care environment • Inter-agency interactions increased as a result of entering the pilot process • These increased interactions were not easy • The court was seen as a significant, and possibly deliberate, obstacle to progress • Problems with interactions with FIA were seen as being generally less severe and difficult to overcome • Supervisors’ roles did not (would not) change regarding foster families • Supervisors’ roles would change in terms of responsibilities regarding children and families
Different Perceptions of Staff at Various Organizational Levels • Administrators had a wider ‘lens’ on the movement toward managed care, and were more concerned about the fundamental changes that would have to take place • More concerned with how the pilot required fundamental transformations in how staff viewed the goals of foster care and the primary target of foster care services • More likely to mention that there were few avenues for interdepartmental communication • More concerned about conflict between various departments • More concerned about the substantial changes in how services were delivered
Different Perceptions of Staff at Various Organizational Levels • Supervisors and line workers tended to focus more on service delivery issues and the ability to provide adequate service • More likely to mention that there were changes in how the agency interacted with FIA, the court, and/or other agencies • More aware that there were problems with accessing community resources and services • More concerned about issues of efficiency and speed in assessment and service delivery
Implications • Managed care contracting requires service providers to quickly diagnose clients’ needs, focus additional resources on collecting client and service information in databases, and pool funds to address the multiple needs of clients systematically and simultaneously • New types of positions within foster care departments often need to be created — case aides, family engagement workers, and relative assessors —whose jobs are specifically designed to provide a set of services to move children through the system more efficiently • Changes in financial management and accounting procedures are needed to succeed under managed care
Implications (con’t) • Collaboration is needed between foster care and accounting departments, as well as with other agencies, in order to provide information, secure collateral services, settle financial issues, and resolve case-related concerns • Regular communication between departments reduces payment complications with FIA and maintains accurate client status and service information within the agency • Knowledge sharing, resource sharing, and common problem solving reduce the need for new agencies entering this arena to reinvent the wheel • Technical assistance can help agencies to install systems that smooth the transition to managed care
Research Questions – Quantitative Data • What was the effect of the shift to this managed care contracting environment on the services provided to children and families in the first 300 days of foster care? • Which variables explain differential service patterns between the pilot and non-pilot agencies? • Which children reached the first pilot performance point of placement with a parent, relative, guardian, or in independent living within 290 days? • Which variables explain differential outcomes between the pilot and non-pilot agencies?
Research Design • Data were collected on a quarterly basis on a sample of 244 children that came into foster care between May 1, 2001 and October 1, 2001 • 175 children from the pilot agencies and 69 children from the non-pilot agencies • This phase of the study took advantage of the conditions for a natural experiment in order to isolate the effect of managed care from other possible influences on service delivery and child outcomes • Differences in service patterns and placement status between the pilot and non-pilot groups, where found, can be attributed to the type of agency responsible for the child since the design is the equivalent of a true experiment • Data collected at 30 days after child’s entry into care, and then for each 90 days thereafter
Sample Identification • Children were admitted into the study sample if they met the following criteria: • The child was a resident of Wayne County • The child was under the age of 13 or was part of a sibling group in which there was at least one child under the age of 13 • The child was assigned to agencies through the Family Assignment System or was a “must take” for the agency • At least 365 days had passed since the child’s last non-relative out of home placement (if the child had previously been in care and had come back to the agency) • The child was assigned to the agency that had supervisory responsibility of “first origin” • The child was ordered into relative or foster care at the preliminary court hearing
Mean Age = 6 51% female Race 78% African American 18% Caucasian Previously Placed = 8% Drug/Alcohol Pre-natal Exposure = 34% Assessment at Intake (% with needs) Behavior/Coping Skills = 30% Education = 29% Family Relational Problems = 19% Non-Family Relational Problems = 18% Health = 17% Life Skills/Development = 16% Sexual Adjustment/Victimization = 12% Cultural Identity = 8% Child Characteristics(no significant differences between pilot and non-pilot children)
Age = 31 years Gender = 94% female Single Female Headed Families = 64% (P>NP)** Marital Status = 84% Not Currently Married Mother Married at First Birth = 11% yes Age of Mother at First Birth = 21 (large SD) Formal Education < 9yrs = 10% non-HS grad = 44% HS grad = 31% Total Income = $643/mo Relation to Child = 88% biological mother Number of Children Now in Home = 2.7 (NP>P)* Average Number of Children Removed from Household = 3.1 Prior CPS Involvement = 53% Involvement in Criminal Justice System = 26% Hospitalization in Adult Mental Health System = 16% History of Being Maltreated = 43% (NP>P)* History of Domestic Violence = 51% Primary Caregiver Characteristics
300-day Child and Family Service Delivery Patterns • Two major sets of analyses were conducted to examine patterns of service delivery to sample children and their families: • Analyses of in-agency services to foster children and families • In-person, non-therapeutic services to the child and/or the foster family over the first 300 days of care • In-agency therapeutic services to the child during this time period • Phone calls to FIA and other collateral contacts during this time period • Analyses of out-of-agency services to foster children and families • The total number of referrals made over the first 300 days in care • The total number of out-of-agency services that had begun during this time period. • For both sets of analyses, descriptive analyses were followed by multivariate statistical testing in order to determine what factors were associated with service delivery patterns
Services at 300 Days • On average, non-pilot agencies provided significantly more in-agency services than did pilot agencies • Non-pilot agencies made an average of 85 in-person non-therapeutic service contacts for each foster child and/or foster family, as opposed to 51 such contacts per child by pilot agencies • Non-pilot agencies had roughly six therapeutic service contacts with each foster child, as opposed to one therapeutic service contact per child made by pilot agencies • Non-pilot agencies made roughly 23 successful phone calls and other collateral contacts per child, compared to 17 such contacts made per child in pilot agencies.
Services at 300 Days • There were no statistically significant differences between pilot and non-pilot agencies in terms of the amount of out-of-agency services provided • On average, pilot and non-pilot agencies made 13 referrals for each foster child over the first 300 days in care. • On average, foster children and their families were involved in seven out-of-agency services at the 300-day mark
Factors Associated with Service Provision at 300 days: Regression Analyses • Pilot status was a powerful and consistently negative predictor of receipt of in-agency services over the child’s first 300 days in foster care, controlling for other factors. • Very few other variables (including reason for removal, strengths/ needs at time of removal, child, family, and worker demographics) were statistically related to any of the service utilization variables. • Client age was a significant and negative predictor of two categories of service receipt. Children with older primary caregivers received fewer in-person non-therapeutic services, and older children received fewer in-person therapeutic services. • Children whose primary caregivers had been diagnosed with chronic mental health problems, and children whose primary caregivers had previously been involved with child protective services, received more in-person non-therapeutic services and in-person therapeutic services, respectively. • More phone calls to FIA and other collaterals were completed on behalf of children whose primary caregivers scored low on the worker-completed assessment of primary caregiver needs and strengths.
Comparison of Multivariate Analyses of Service Data at 120 and 300 Days • The effect of pilot status had a comparable negative effect on service provision at both the 120-day and 300-day marks • Certain child and primary caregiver variables were significant predictors of service receipt at 120 days but not at 300 days. • Older and African American children received more services in certain areas, including in-person non-therapeutic services as well as therapy, at 120 days. • Children from drug-impacted homes, and children whose caregivers had abandoned them, received fewer non-therapeutic services at 120 days.
Implications of the Service Findings • There are a number of explanations for these results which cannot be dismissed at this stage of the research • Pilot agencies, either formally or informally, choose to bring services to bear only as performance deadlines approach, which would explain the narrowing of the gap between the two groups in later time periods • Pilot agencies provide only those services that are seen as absolutely necessary to achieve performance payments and/or improve client outcomes • Pilot agencies choose to save the initial lump-sum payments rather than spend them on out-of-agency services • Pilot contracting leads to service rationing • The fact that service provision cannot, in general, be predicted based on case characteristics suggests that case assessment and planning is not as refined as it should be • Some families (drug impacted, those that abandon) are difficult to engage, and therefore may be at jeopardy for TPR given current ASFA timelines. Such families may face double jeopardy in a managed care system.
Child Outcomes at 300 Days • In general, there were no statistically significant differences in the placement status of children from pilot and non-pilot agencies • 36% of the full sample was returned to a relative, 21% was returned home to a parent, six foster children (3%) were placed with a guardian during this time, and no child was placed into independent living • 55% of the children served by pilot agencies and 45% of children served by non-pilot agencies reached this first pilot payment milestone -- this trend is not statistically significant
Child Outcomes at the 300 Days • Pilot agencies were significantly more likely than non-pilot agencies to have their agency supervision terminated (15% as opposed to 3%) • Children from pilot agencies were more likely than those from non-pilot agencies to have their court supervision terminated (19% versus 4%) • Only two children reentered care over this time • 30 children (12%) experienced more than one placement with parents, relatives, or guardians by the end of their 300th day in care • While children from pilot agencies were roughly twice as likely as children from non-pilot agencies to have experienced multiple placements (14% versus 7%), this trend was not statistically significant
Factors Associated with the Achievement of the First Incentive • Pilot status was not a statistically significant predictor of the achievement of the first pilot milestone • Only a few variables were statistically associated with the achievement of the first pilot performance payment • Children that had experienced prenatal drug and/or alcohol exposure were less likely to reach the first performance point • Children that had been neglected or abandoned were also less likely to achieve the first pilot milestone • Children whose primary caregiver scored highly on the worker-completed assessment of needs and strengths were more likely to reach the first performance point
Factors Associated with Other Child Outcomes at 300 Days • Children were less likely to be returned home over this time period if they had been neglected, removed from a drug-abusing household, had been abandoned, or if their primary caregivers scored high on the worker-administered assessment of primary caregiver needs and strengths • Older children, and children whose primary caregivers had previously been involved with child protective services, were more likely to be placed with a relative • Court supervision was less likely to be terminated if the foster child had been removed from a drug abusing household or a household with inadequate housing. Additionally, the more in-person non-therapeutic services that were given to children and their families, the less likely it was that court supervision was terminated. Court supervision was more likely to be terminated if children had primary caregivers scored highly on the worker-administered assessment of primary caregiver needs and strengths.
Implications of the Output Analysis • Children from non-pilot agencies were being placed with parents, relatives, or guardians at rates comparable to that of children from pilot agencies controlling for other factors • Possible explanations for these results include: • Non-pilot agencies made unusually intense efforts to reunify children with parents and assess potential relative placements • The courts were unable or unwilling to rule expeditiously on petitions brought by pilot agency staff • Pilot agencies were unable to effectively coordinate the necessary resources to place children with parents and relatives • Neglectful, drug-impacted families, and families who abandoned their children need more careful attention if a permanent plan with family or relatives is to be achieved within the acceptable timeframe
Comparison of Findings To Those From Other Systems • Findings and Implications Conform With What is Known From Behavioral Health and Health Care Managed Care Studies • Requires agencies to recognize that “the meter is running” and to initiate quick and accurate diagnostic processes that result in immediate service provision • Risk-averse managed care agencies may avoid providing services or serving clients that they deem unprofitable, particularly those with special needs • Leads to delivery of fewer services in general • Outcomes are mixed – generally works well but some fall through the cracks or are denied needed services
Comparison of Findings To Those From Other Systems (con’t) • Associated with a greater focus on the management of service delivery through utilization review techniques, and thus requires a heavy IT investment • May initially strain public-private relationships, but such relationships may ultimately improve as a result of increased inter-agency communication • Requires agencies to invest in staff training in order to respond to staff turnover and “managed care phobia,” particularly among clinical departments • Places a heavy emphasis on financial management, particularly in terms of data collection, actuarial analyses of agency caseloads, and prospective budgeting
Conclusions To Date: Service Delivery • Contract incentives directly influence the work environment, employee behavior, job satisfaction, and employee retention • Performance deadlines must be set up with enough time to allow employees to conduct thorough assessments, provide necessary in-agency services, access community resources, and place children in appropriate living situations • Monitor in-house service delivery to ensure that decreased client contact, if it occurs, is warranted by efficiencies achieved in service provision rather than cost considerations • Consider formalizing relationships with collateral service providers in order to bring necessary services to bear in a timely way • Assess the availability of high-demand, out-of-agency services. If such services are unavailable, document this shortage and convey this information to court officials and the public agency so that the managed care contract can reflect this reality.
Conclusions To Date: Information • Agency managers need to agree on what information is important, how this information is to be collected, whether agency staff have the resources and knowledge necessary to input such information in a timely and accurate manner, whether the requisite expertise exists to analyze trends in the data, and how the results of these analyses will be disseminated • Agencies should not pour effort into establishing “tickler” based systems—in which workers and accounting staff are reminded of impending performance and billing deadlines—without first carefully considering how staff behavior will respond to such a system. Tracking systems should not be a surrogate for careful case planning and the implementation of appropriate case plans for all clients
Conclusions To Date: Interdepartmental Relations • Managed care requires increased vertical communication; executives must listen more often and more closely to front line staff in all departments to ensure that managed care service delivery strategies are working as intended and are not jeopardizing the quality of service provided to the client • Managed care requires increased horizontal communication. Key departments—foster care, IT, and accounting in particular—cannot operate on “opposite sides of the hall” • Consider implementing interdepartmental meetings and other mechanisms so that staff can gain a holistic sense of how departments must balance competing pressures and responsibilities
Conclusions To Date: Public/Private Contracting • Managed care contracts should be negotiated rather than dictated. Agencies should perceive the playing field as level, and all parties’ voices should be heard in contract negotiations. • Public agencies, in conjunction with the nonprofit sector, should consider establishing different performance expectations for more difficult-to-serve children • Public agency managers must recognize that overdue reimbursements jeopardize private agencies’ cash flow, and possibly impact the type and amount of services that agencies are able to offer clients • Other systems that are essential to the provision of service (drug rehabilitation, domestic violence, metal health, etc.) and the movement of children thought the system (courts) must be aware of the new fiscal arrangements and be prepared to operate to accommodate them • Any future changes should be carefully considered and based on data
Conclusions To Date:Moving Systems to Managed Care • Joint strategic planning may help contracting agencies better understand the common problems they will face • Joint agency education and training programs at all levels might lead to efficiencies in the transition to managed care • Managed care may not promote a “one size fits all” strategy, where agencies respond uniformly to new fiscal and programmatic challenges. Agencies should recognize that the wholesale adoption of another agencies’ service delivery strategy, MIS, and training programs may result in both benefits and costs • Some smaller agencies are already signaling that they cannot survive in this managed care environment • Viability of agencies must be carefully considered, for it may impact the mix of services available in the system
Looking Forward… Research Analyses Underway – 570 Days • Continued analysis of service data • TPR at 515 days and factors related to it • Return to care after timely discharge to family or relatives and factors related to this phenomena • First payment for sustainability for children returned within 290 days and factors related to reaching this milestone • 290 + 180 days at home
Looking Forward… Future Analyses Year Three Data -- 935 Days • Continued analysis of service data • Positive outcomes and factors related to them • Second payment for sustainability for children returned home/relative within 290 days (655 days) • 6 and 12 month sustainability payments for children who did not return home/relative in a timely way • Termination of parental rights in 515 days and are adopted within seven months (725 days)
Looking Forward… Future Analyses Year Three Data -- 935 Days • Children who achieve TPR in a timely way but are not adopted within a year (880 days) and factors related to this phenomena • Children who reenter foster care after return home or adoption and factors related to this phenomena • Remain in the foster care system past all of the incentive deadlines and factors related to this phenomena