One of the key functions of child protection agencies is to ensure that children who have been maltreated are not subjected to further maltreatment. Strategies for enhancing the capacity of agencies to meet this basic requirement should be based on a clear understanding of the local patterns and predictors of maltreatment recurrence. A comprehensive examination of relevant published maltreatment recurrence literature revealed that a firm empirical foundation regarding the extent of this phenomenon in New South Wales has not yet been established. Additionally, actuarial models for predicting maltreatment recurrence in New South Wales do not appear to have been developed.
The purpose of this thesis was: (a) to investigate the extent of maltreatment recurrence amongst confirmed neglect and physical abuse cases referred to child protection services in New South Wales (Study 1), and (b) to develop models for predicting maltreatment recurrence (Study 2). Extensive literature reviews of key child maltreatment and maltreatment recurrence research were conducted to highlight commonly encountered methodological issues in the field and to identify key potential predictors of maltreatment recurrence. An ecological model of maltreatment recurrence was used as a framework for integrating findings pertaining to key potential predictors.
A unique focus of the research was the examination of four previously uninvestigated child protection subgroups, defined by index maltreatment type (neglect, physical abuse) and CPS referral history status (CPS referral history, no CPS referral history). Specifically, four mutually exclusive subgroups were investigated, that is, those characterised by: (a) a neglect incident and a CPS referral history ('Neglect-Prior'), (b) a neglect incident and no CPS referral history ('Neglect-No Prior'), (c) a physical abuse incident and a CPS referral history ('Physical-Prior'), and (d) a physical abuse incident and no CPS referral history ('Physical-No Prior').
The sample comprised 331 families from four prominent cultural groups in New South Wales (Anglo-Australian, Greek, Tongan and Vietnamese) who were involved in a confirmed neglect or physical abuse incident that was notified to the Department of Community Services during the 1991 through 1994 calendar year period. All families from the three minority cultural groups and a stratified random sample of Anglo-Australian families who each fulfilled a number of selection criteria were selected for investigation. The case files pertaining to these families were searched and pertinent information recorded. For the purposes of this investigation, maltreatment recurrence was defined as any type of further confirmed maltreatment involving one or more children from a participant family within a standardised two-year period.
Two studies were conducted. The first was an investigation of the extent and distribution of maltreatment recurrence. Results of the first study indicated that maltreatment recurrence contributed substantially to child protection rates within the sample, and that it was highly concentrated among a minority of families. Overall, 24.8 percent of participant families were involved in a further confirmed incident within the two-year follow-up period. These families between them accounted for 46.8 percent of all incidents within the follow-up period. Significant differences in the rate and distribution of recurrence were also observed between the four-child protection subgroups described above, as well as between the four cultural groups and the two CPS referral history status groups (that is, those with a CPS referral history and those without). The most marked variability in the rate and distribution of maltreatment recurrence was observed between the four-child protection subgroups (i.e., 'Neglect-Prior', 'Neglect-No Prior', 'Physical-Prior' and 'Physical-No Prior'), suggesting the need for further investigation of these groups.
The second study sought to develop multiple models, comprising different combinations of predictors, for collectively predicting maltreatment recurrence within each of the subgroups. The predictive accuracy results for each of the resultant groups of models were equal to, or better, than those of actuarial models described elsewhere in the literature, ranging from 80 to 97 percent across the four subgroups. A crossvalidation study, using the standard split-half method with the largest of these subgroups, showed that extremely high levels of sensitivity were maintained, together with a relatively lower, but adequate level of overall predictive ability. The combinations of risk factors that were important for predicting maltreatment recurrence were clearly different between each of the subgroups examined.
The results suggest several directions for future research as well as a number of policy and practice implications for enhancing child protection agency response to recurrent maltreatment. Among the strengths of the research are the detailed examination of methodological issues that guided the design, the examination of previously uninvestigated subgroups of the child protection population, the inclusion of culture as a key variable and the development and validation of a new analytic approach to predicting maltreatment recurrence. The research provides a firm foundation for further investigations aimed at reducing child maltreatment through preventing maltreatment recurrence.