Home > Programs & Funding > Profiles of the Title IV-E Child Welfare Waiver Demonstration Projects > North Carolina - Flexible Funding/Assisted Guardianship - Phase II

North Carolina

Demonstration Type:	Flexible Funding/Assisted Guardianship - Phase II1
Approval Date:	June 18, 2004
Implementation Date:	January 1, 2005
Completion Date:	Terminated early on February 28, 2008
Interim Evaluation Report Received:	July 2, 2007
Termination Summary Received:	April 23, 2008

Background

On June 18, 2004, the Children's Bureau approved a five-year extension (Phase II) of North Carolina's Child Welfare Demonstration Project. Although the five-year extension officially began on July 1, 2004, although implementation did not occur until January 1, 2005 following approval of the State's updated evaluation plan in October 2004. Due to problems maintaining cost neutrality, North Carolina terminated its waiver demonstration in May 2007 retroactive to December 2006 for claiming purposes. The State considered reactivating its waiver later in 2007, but withdrew its request to the Children's Bureau for reactivation on February 28, 2008 due to a lack of interest among counties to restart waiver activity.

Target Population

Eligible children included both title IV-E-eligible and non-IV-E eligible children residing in experimental group counties who were at imminent risk of foster care placement or who were already in placement. Each participating county could choose to implement initiatives aimed at improving child welfare outcomes for eligible children.

Jurisdiction

Nineteen of 100 counties in the State participated in Phase I of the demonstration. For Phase II, the State expanded demonstration services to 38 counties, including 17 original Phase I counties and 21 new counties. Two of the original Phase I counties elected not to participate in the Phase II demonstration.

Intervention

Phase II of North Carolina's demonstration enabled participating counties to use Federal title IV-E foster care funds to develop and implement strategies to reduce costs for out-of-home placement while improving or maintaining safety, permanency, and well-being outcomes for children. Each participating county developed a unique set of services and interventions that were outlined in a detailed proposal and approved by the State prior to implementation. All participating counties sought to use title IV-E funds flexibly to institute new contracted services that ranged from substance abuse and mental health services to respite care, intensive family preservation services, and parenting classes. In addition, experimental group counties could use flexible IV-E funds to pay for assisted guardianship, make one-time payments for services to children and families (e.g., payments for rent deposits, utility bills, day care fees), and to facilitate child and family team meetings.

Evaluation Design

The evaluation included process and outcome components, as well as a cost analysis. The State utilized a comparison group Evaluation Design in which changes in child welfare outcomes for 34 selected "comparison counties" that did not receive flexible IV-E funds were compared with changes in outcomes for the 38 "experimental counties" that received flexible IV-E dollars. Factors used to select comparison group counties included (1) total title IV-E maintenance expenditures, (2) total administrative expenditures, and (3) number of children initially entering out-of-home placement during the past three State Fiscal Years (SFY).

Sampling Plan

The State's evaluators used two samples of cases selected from large and medium-sized counties to assess changes in service utilization and child welfare outcomes. To study placement prevention outcomes, the State's evaluators randomly selected 840 children from 24 large and medium-sized experimental counties who had experienced a first substantiated report of abuse or neglect during the baseline year (SFY 2002-2003). They then matched these children to 840 children from 23 large and medium-sized comparison counties using propensity score matching. This sample is referred to as the "prevention sample." For the second sample (referred to as the "permanency" sample), the State randomly selected 400 children from the 24 large and medium-sized experimental counties who initially entered placement during the baseline year and matched these children to 400 children from the 23 large and medium-sized comparison counties using propensity score matching.

Process Evaluation

The foundation of the State's process evaluation rested on the 38 logic models submitted by individual experimental group counties. The logic models identified changes in service availability and utilization that were expected to occur as a result of the waiver. These expected changes were then evaluated using two web-based surveys administered to county child welfare managers in 2005 and 2006, as well as through data submitted by participating counties in quarterly reports from 2005 through 2006. Baseline data on service utilization and availability were obtained through an analysis of case records during the summer of 2005.

Outcome Evaluation

The outcome evaluation compared experimental and comparison group counties for significant differences in the following measures: rate of entry into out-of-home care; length of stay in out-of-home care; exits to permanency; rate of re-entry into out-of-home placement; and maltreatment recurrence rates.

One challenge faced by the State in tracking maltreatment recurrence involved North Carolina's adoption of the Multiple Response System (MRS) for Child Protection Services (CPS). Through MRS, most reports of maltreatment are addressed through a family needs assessment and do not result in a formal determination of abuse or neglect. To accommodate this change, the State's evaluation team modified its definition of maltreatment to include the MRS family assessment category of "services needed." This change allowed the evaluators to track children who received or were recommended for services and follow their interactions with CPS.

Cox non-proportional hazard modeling and time series analyses were used to test for differences between children in experimental and comparison counties entering the child welfare system before and after waiver implementation. Survival analyses modeled waiver outcomes to control for differences in the demographic characteristics of children (age, gender, race, etc.), child welfare agency characteristics (e.g., level of participation in the waiver), and community characteristics (e.g., urbanicity and region).

Cost Analysis

The cost analysis focused on assessing how child welfare spending patterns changed as a result of the demonstration. Baseline cost indicators were compared with child welfare expenditures over time to identify changes in spending patterns across experimental and comparison counties.

Evaluation Findings

Process Evaluation

1. Availability and Utilization of Child Welfare Services:
   • The State's evaluation team developed a model to explore three dimensions of service provision: population focus (i.e., county-wide or child-specific services), service focus (e.g., child welfare services, treatment services, or basic needs), and level of service (low, medium, or high). Overall, a majority of experimental group counties focused on providing child welfare services (e.g., respite care, family team meetings) on a county-wide basis.
   • During the first two and a half years of the demonstration, 2,587 children in 28 experimental group counties benefited from some form of targeted, child-specific spending. The top six categories for targeted expenditures included housing, household utilities, child care, furniture purchases, legal expenses, and transportation.
   • In terms of service level (defined as the ratio of service episodes to the total number of children with an initial report of maltreatment or placement during the demonstration), 7 experimental counties (18 percent) were categorized as "high level" (37-148 episodes per 100 children), 24 counties (63 percent) were categorized as "mid level" (3-25 episodes per 100 children), and the remaining 7 counties were defined as "low level" (no waiver activity or fewer than 3 episodes per 100 children).
   • Overall, families in experimental group counties utilized child welfare and related services at higher levels than comparison group counties. In experimental group counties, 77 percent of families in the client-level "prevention sample" and 95 percent of families in the "permanency sample" used at least one service at a baseline measurement compared with 64 percent and 86 percent of families in comparison counties, respectively. In addition, 9 percent of clients in the prevention sample and 40 percent of clients in the permanency sample in experimental counties utilized six or more services compared with 1 percent and 10 percent of clients in comparison counties, respectively.
   • On average, families in experimental group counties accessed services more quickly than comparison group families. According to client-level baseline data, 67 percent of child-caregiver dyads in the experimental group prevention sample received services within one year of a maltreatment substantiation or prior to placement compared with 56 percent of dyads in the comparison group prevention sample. Moreover, 86 percent of child-caregiver dyads in the experimental group permanency sample received services within two years of placement or prior to placement exit compared with 78 percent of dyads in the comparison group permanency sample.
2. Use of Assisted Guardianship: A total of 209 children were identified as candidates for assisted guardianship in the first 10 quarters of the State's waiver extension. Of these, 105 children (54 percent) had exited to assisted guardianship by December 2006. Mecklenburg and New Hanover Counties accounted for 77 percent of children who exited to assisted guardianship. In addition, 55 percent of guardian placements were relatives, with grandparents accounting for 29 percent of exits, aunts/uncles for 16 percent, and other kin for 10 percent of exits.

Outcome Evaluation

The State's evaluation revealed differences in several outcomes that favored comparison counties:

• The number of children with a first report of abuse or neglect increased steadily in experimental group counties between SFY 2000-2001 and SFY 2004-2005. In comparison counties, this number decreased in SFY 2002-2003 and remained relatively unchanged afterwards.
• In general, experimental group counties had higher levels of entry into placement than comparison group counties, and children in experimental counties tended to enter placement more quickly. The State noted that these findings are not surprising given the presence of several large counties (e.g., Mecklenburg) in the experimental group.
• Median length of stay in out-of-home placement was generally longer for children in experimental group counties than in comparison counties, averaging 471 days between SFY 2003 and SFY 2006 compared with 357 days in comparison counties during the same period. In addition, Cox proportional hazard modeling revealed that children in experimental group counties were significantly more likely to exit placement after two years than children in comparison counties.
• Reunification was the most likely permanency outcome across all entry cohorts for children in both experimental and comparison group counties. Children in comparison group counties tended to achieve reunification more quickly (i.e., within one year) than children in experimental group counties. Exits to permanency via guardianship or another court-appointed custodial arrangement were slightly higher in comparison counties than experimental group counties, while adoption was more likely in experimental counties.

However, some positive trends in favor of the experimental group were discussed in the State's Termination Summary:

• Although total entries into placement were higher in experimental group counties, the probability of placement in experimental group counties decreased between SFY 2005 and 2006 while remaining unchanged in comparison counties. This trend was most pronounced among high level counties that made the most extensive use of IV-E funds. The State hypothesized that given the steady increase in maltreatment reports in experimental counties over time, it is likely that more children would have entered placement in experimental counties without the waiver.
• Although repeat maltreatment within six months of an initial incident of abuse or neglect decreased in both groups of counties through SFY 2005, the rate of maltreatment recurrence in experimental group counties continued to decline between SFY 2005 and SFY 2006 while it increased in comparison counties during the same period.
• In general, children in comparison group counties re-entered care at faster rates than children in experimental group counties, although not at statistically significant levels.

Lessons learned from the IV-E waiver demonstration have informed North Carolina's decision to pursue funding for statewide subsidized guardianship with the North Carolina General Assembly. This goal is outlined in North Carolina's 2007 Program Improvement Plan.

¹ Based on information submitted by the State as of April 2008. Back

Back to Table of Contents