A 10-Year Review of Family Preservation Research: Building the Evidence Base

Background: Intensive family preservation services (IFPS) have received considerable public and research scrutiny. In 1997, Fraser, Nelson, and Rivard conducted an in-depth analysis of IFPS studies from the previous 10 years, and discussed methodological challenges the researchers encountered. Among other things, they suggested that studies are needed that implement consistent service models with homogeneous samples of children at genuine risk of placement.

Methods: This study replicates and updates the work of Fraser and his colleagues by examining studies of IFPS published since 1997. The studies that met our criteria and aligned most closely with HomebuildersŪ standards include the Family Enhancement Program in Oregon, the Intensive Family Preservation Services Program in North Carolina, and the Families First Program in Michigan. A three-state study conducted by Westat also was reviewed. The most common structural features of the IFPS programs include immediate response to referrals (within 24 hours), worker accessibility 24 hours a day, 7 days a week, intensity (12-15 hours of services per week), brief services, and low caseloads (2 families per worker or 5 families per worker with paraprofessional assistance). Effect sizes were calculated to compare the magnitude of differences between intervention and comparison/control groups.

Findings: Although a range of effects was found, the only large effect size (.77) was in Michigan with 93% of the intervention group families intact at the 12-month follow-up compared to 63% of control group families. Medium effect sizes (.33 to .55) were found mostly in sub-samples with clear risk factors for placement. In North Carolina, a medium effect (.55) was found in a small sub-sample of cases with prior placements. Over 80% of the intervention group avoided placement in contrast to 56% of the comparison cases. The only other sub-sample to show a medium effect (.35) involved cases in Kentucky that had family court petitions filed. Over 80% of the intervention cases avoided placement compared to 66% of the control group. Small (.00 to .20) or negative effect sizes were found for the remaining studies.

Conclusions: In some ways, these findings replicate those of Fraser, Nelson and Rivard (1997) and highlight the same challenges: targeting, treatment fidelity, and methodological issues. In other ways, however, they present promising new findings: the replicability of the intervention, the need to consider and control for risk factors, and additional evidence of effectiveness with child welfare populations.

Although some of the studies in this review report promising results, the field still needs additional evidence that IFPS programs prevent unnecessary child placement and about which types of programs are most effective with different client sub-populations including racial and ethnic minorities, children of different ages, and those involved in physical abuse, neglect, and other areas. We also need a better understanding of effectiveness of different program components with different families. Nonetheless, as state and county child welfare systems look ever more carefully at reducing their foster care populations and reinvesting their savings in high quality services, we believe that IFPS programs should be one part of that array of interventions.