Abstract:
This report summarizes research completed for National Institute of Justice (NIJ) grant (No. 2008-IJ-CX-0025), which funded a joint effort between New Hampshire’s Bureau of Elderly and Adult Services (BEAS) and the National Council on Crime and Delinquency (NCCD) to construct and validate an actuarial risk assessment. The resulting assessment is completed by Adult Protective Services (APS) workers to help inform decisions related to the likelihood of future maltreatment or self-neglect. Other public service agencies, including child protection and corrections, have implemented simple, objective, and reliable actuarial risk assessment instruments to help workers identify high-risk clients and prioritize them for service intervention at the close of an investigation. The research question was whether an actuarial risk assessment could serve a similar purpose for APS agencies by enhancing their abilities to reduce maltreatment of the most vulnerable clients. The goal of the grant was to determine whether an actuarial risk assessment could be constructed that validly and reliably classified individuals reported for adult maltreatment or self-neglect, and whether workers and other agency staff would find the risk assessment useful in practice. Initiated in 2008, Phase I involved a longitudinal study of 763 individuals investigated for allegations of self-neglect or maltreatment between March 1 and September 30, 2009, then observed for a standardized six-month period to measure subsequent reports to BEAS. Phase I resulted in an actuarial risk assessment that BEAS workers began completing in 2009 to help inform decisions regarding an individual’s likelihood of future harm. 1 Workers scored the risk assessment composed of items related to self-neglect and/or abuse to obtain a risk classification related to the likelihood of future reports to APS. Phase II, completed in 2010, consisted of inter-rater reliability testing and a process evaluation to assess implementation fidelity. Reliability testing involved 24 caseworkers (of 35 total) who scored the risk assessment items for three vignettes, and resulted in high inter-rater agreement across items. The process evaluation showed that implementation fidelity and workers’ perceived utility of the actuarial tool varied by region. As a result of the process evaluation and reliability studies, BEAS managers and supervisors initiated a number of practice-improvement efforts. Phase III involved a second, prospective validation study of the risk assessment conducted with a larger client sample (n=1,064) and observation of harm recurrence for a longer, one-year, standardized follow-up period. The self-neglect index of the risk assessment more accurately classified individuals by risk than did the risk of abuse/neglect index. This could be the result of low base outcome rates for re-maltreatment, varied implementation fidelity, and/or missing risk or protective factors not measured in existing data sources. Overall, the risk assessment validly classified individuals reported for allegations of harm by the likelihood of future harm of any type. Findings suggest the possibility of constructing a valid and reliable actuarial assessment to classify individuals reported to APS by the likelihood of future maltreatment; however, additional research is needed to improve the classification abilities of the assessment. Results from the process evaluation suggest that completing a validated actuarial risk assessment could be helpful to APS workers. For example, supervisors reported they were beginning to consider risk in decisions about attempting to engage individuals refusing APS involvement (i.e., more varied efforts at re-engaging if high risk). This research indicates a strong need to continue developing research-based assessments for APS field staff and managers—specifically, to improve on the classification abilities of the risk assessment, identify additional empirical risk factors if possible, and to study how an actuarial risk assessment and other decision-support tools can help improve the accuracy and consistency of decisions made by APS caseworkers. (Author Abstract)