BACKGROUND: Parenting capacity assessment (PCA) in the context of maltreatment is complex and the stakes are high for children and families. Among parents where a formal PCA is undertaken due to child abuse or neglect, maltreatment is more pervasive and chronic than among other families involved with child protection services (CPS). Several methodological approaches to PCA have been published, but evidence is lacking for the predictive validity of these for subsequent harmful parenting among this population. In-depth assessments have not been evaluated, and actuarial methods are insufficiently specific for application to individual cases. Among parents involved with CPS where harmful parenting is ongoing, some are likely to become ‘good enough’ parents, and some are unlikely to significantly improve within developmentally-appropriate timeframes. Accurately identifying parents who can and cannot change is not currently possible, leading to a situation whereby each case becomes a natural experiment over time. Children consequently grow up in families who do not sufficiently improve their parenting in response to support, exposing these children to continued suffering and adversity. Improving the predictive accuracy of PCA is therefore essential to providing child welfare interventions that reduce suffering, interrupt intergenerational cycles of maltreatment and optimise children’s chances to develop according to their potential.

AIM: This study firstly aimed to examine the evidence for parenting interventions that reduce child maltreatment among a population of parents who were known to have already harmed their children. The second aim was to conduct a scoping review of the published models of PCA for the child protection context. Finally, this study aimed to identify psychological characteristics that were predictive of subsequent notifications of harm among a population of CPS-involved parents whose youngest child remained in their day-to-day care.

METHODS: Participants were eligible for inclusion into the study if they were the parent of at least one child under the age of 11 years who had never been taken into out-of-home care (OOHC). They must also have reported several past notifications of harm, and had recent or ongoing involvement with CPS. Participants were recruited over an eighteen-month period via agencies providing parenting support services in collaboration with CPS in Christchurch, Aotearoa/New Zealand (NZ). This study measured participants’ mood, emotional regulation, developmental histories, parenting self-efficacy, representations of the parent-child relationship (using the Working Model of the Child Interview) and parental reflective functioning during a three-hour, two- to four-session baseline assessment. Outcome measures were subsequent notifications of harm to CPS occurring during the eight months following the baseline assessment, entered as the ‘total number’ and ‘presence or absence’ of notifications. These notifications were tied to participants’ CPS files rather than children’s files, and applied to notifications regarding all children of the participant. ‘Entry into care’ was also a main outcome variable. Psychological constructs were examined as potential predictors. The primary methods for data analysis were bivariable correlations and regression analyses.

RESULTS: Twenty-nine participants out of an initial sample of 41 referred parents were recruited to the study, with complete information collected for 26 parents. The average age of the youngest child was 26 months, and the average age of the participants was 31 years. All but one parent (97%) reported at least one form of maltreatment during their own childhood, and half reported two or more. Approximately one-third of participants received one or more CPS notification during the eight months following assessment. No individual psychological constructs, nor cumulative risk scores, were significantly predictive of subsequent notifications of harm among the study sample. Parents with a higher number of children, and parents reporting experiencing neglect were more likely to receive subsequent notifications

(OR =2.7 and 6.5, respectively). A predictive model derived from latent discriminant analysis and stepwise logistic regression yielded three variables that, when combined, accurately classified 92% of parents as having a subsequent notification or not. These variables were having a higher number of children, reporting a history of neglect, and parents’ poor awareness of their own and their child’s emotional state.

CONCLUSION: Parental reflective functioning, attachment representations and other psychological characteristics related to parenting were not predictive of subsequent harm among the CPS-involved parents in this study, and no single psychological measure differentiated between parents with and without notifications of harm. Existing quantitative measures of some psychological constructs are likely to be invalid or unreliable for use in PCA with CPS-involved parents. Attachment and reflective functioning-related constructs may interact with other risk factors for individual parents in ways that are not yet clearly delineated by the evidence to date. CPS records, while currently the most robust indicator of actual harm occurring towards children, are blunt indicators of child maltreatment and more sensitive measures are needed to supplement these. Those conducting PCA in the context of child maltreatment may need to examine other methods for establishing risk for future harmful parenting, and should also be aware of the limitations of currently-available measurement tools for this population.