Assessing final outcomes: Predicting time to CPS report

BACKGROUND and PURPOSE. For a program designed to target families at high risk for child maltreatment, evaluating whether families experience future reports of child abuse or neglect following receipt of an intervention is an important marker of program effectiveness. Using a prevention science framework, the Family Connections (FC) program employs an individualized and tailored social work practice approach to focus on the reduction of risk factors and the enhancement of protective factors related to child maltreatment. The original program (DePanfilis & Dubowitz, 2005) documented a reduction of risk factors and enhancement of protective factors over time and despite the fact that more than 50% of families that received FC intervention had experienced prior reports of child abuse and neglect,, a very small number of families were known to CPS agencies within six months following the closure of services. The purpose of this paper was to compare the relative success of eight replicating agencies at preventing future child maltreatment as measured by CPS reports. METHOD. Replicating agencies agreed to follow fidelity criteria for recruiting families who met risk criteria but were not known to have CPS involvement within 12 months prior to recruitment into the FC program. As conditions of their cooperative agreements, agencies also agreed to randomly assign families either to FC intervention or a no treatment group or to two levels of intervention and were required to measure both the degree to which risk and protective factors changed over time (Symposium paper 2) and whether families experienced CPS reports following intervention. Three of the eight sites included a no-FC control condition. Seven sites included treatment conditions in which FC was provided for shorter or longer durations. A total of 802 families were recruited into the FC Replication studies. The average number of families per site was 100 with a minimum of 64 and a maximum of 130. Through agreements with state and local CPS agencies, replicating sites conducted searches to determine whether families in all groups experienced a CPS report prior, during, or following the service or control condition. Survival analysis procedures were employed to plot the time until CPS report following recruitment into the study; comparisons between sites were made with Kaplan Meier methods and the Cox Proportional Regression Model was used to predict the timing of report. RESULTS. Survival patterns indicated differences between treatment conditions and between some sites. Families with prior CPS histories also evidenced higher levels of risk factors at intake and the greatest likelihood to experience a CPS report following FC intervention. CONCLUSIONS and IMPLICATIONS. Results suggest that the FC program may have been effective preventing CPS reports in some localities, but because only three agencies employed a no-treatment control group, findings need to be interpreted with caution. Findings also point to differences in the timing of recruiting families as those families that had experienced prior CPS reports were less likely to function without future CPS reports over time. Implications for working with families at risk of maltreatment will be discussed.