The 1996 evaluation by Fraser and colleagues included 57 children whose families received experimental family reunification services (FRS), based on the HOMEBUILDERS model, compared with a group of 53 children whose families received routine reunification services as a component of an overall out-of-home care plan. Families participating in the study were randomly selected from a computer-generated list of foster care caseloads in four child welfare districts across the State of Utah. Eligibility criteria consisted of the following:
- The child was in placement for more than 30 days.
- Reunification was not imminent.
- Reunification was part of an overarching case plan.
- The child could be returned home to at least one parent.
- The child was not thought to be at risk of harm if returned home.
- The child was not in a specialized treatment program.
Overall, 120 families were randomly selected and invited to participate (however 10 families were lost during the service period). They were randomly assigned to either the experimental or control condition. Children in the intervention group were primarily white (82.7 percent) and female (57.9 percent), with an average age of 10.8 years. On average, children had been in 2.8 previous placements, with a range from 1 to 7 prior placements. The length of prior placements ranged from 1 to 85 months, with an average of 10.7 months. The most frequent reason for placement was child behavior (28.1 percent), followed by neglect (24.6 percent), parent–child conflict (14 percent), sexual abuse (10.5 percent), and physical abuse (8.8 percent). No significant differences were found on demographic variables between the groups.
Outcomes measured included length of time to reunification and length of stay at home before any future out-of-home placement. Proportional hazards analysis was used to investigate outcomes. The observation period started with the beginning of the 90-day FRS experimental condition (or the corresponding date for a child selected for the control condition) and ended 1 year after the termination of that 90-day service period.
The evaluation by Kirk and Griffith (2003) used a retrospective, population-based design to examine the ability of intensive family preservation services (IFPS) to prevent out-of-home placements of children in abusive or neglectful families. The study sites included 51 of North Carolina’s 100 counties. The treatment group included all families from counties offering IFPS with children who received their first IFPS intervention between July 1, 1994, and March 31, 2001, and their first substantiated report after July 1, 1993, but before March 31, 2001. The comparison group, which did not receive IFPS, included all other families in the same counties with children who experienced their first substantiated report after July 1, 1993, but before March 31, 2001.
An operational definition of “imminent risk of placement” was imposed retrospectively, using ratings on the standardized child protective services (CPS) risk assessment instrument. A risk rating of “high” mandated removal of the child, unless an approved alternative plan was immediately implemented that ensured the child’s safety. Only families with high-risk ratings were including in the study, whether or not they received IFPS. The final study sample included 542 high-risk children who received IFPS and 25,772 high-risk children who did not receive IFPS but resided in the same counties.
The treatment group was 51 percent female, 59 percent white, and 41 percent nonwhite. Seventy-five percent of the treatment group were from 0 to 10 years old at the time of their referral to IFPS, and 25 percent were 11 or older. The comparison group was 50 percent female, 54 percent white, and 46 percent nonwhite. Seventy-nine percent of the comparison group was 0 to 10 at the time of the CPS report, and 21 percent were 11 or older. There were several significant differences between the groups with respect to race, age, and type of maltreatment. Children in the treatment group were more likely to be white and younger. IFPS cases were also more likely to be substantiated for injurious environment, whereas non–IFPS cases were more likely to be substantiated for general neglect. IFPS cases were also more likely to have experienced open or more prior substantiated reports of maltreatment and one or more prior high-risk substantiated reports.
The outcome measure of interest was placement prevention, defined as the absence of out-of-home placement 1 year from the beginning of IFPS for the IFPS treatment group or 1 year from the date of the substantiated report of abuse and neglect for children in the comparison group. Data was obtained from several sources, including the IFPS–specific Management Information System, the CPS risk assessment information from the North Carolina Child Abuse and Neglect System, and child placement data from the State information system used to track children entering out-of-home placements.
Event history analysis was used to assess differences in placement rates and patterns for children in the study. Event history analysis computes the relative risk of placement over time. The Cox proportional hazards regression model (the Cox model) was also used to examine associations between each independent variable and the hazard rate for placement while holding all other independent variables constant. A Cox regression model with time-dependent covariates was also used to examine time-related interactions.