Can child maltreatment be predicted during the perinatal period: Evidence from longitudinal cohort studies?

Abstract

The early identification of children who are at high risk for maltreatment would help clinicians direct scarce preventive resources towards families with the greatest needs. This report reviews 11 longitudinal studies in which subjects were categorized during the perinatal period into high-risk and non-high-risk children and then followed over time to determine the occurrence of at least one adverse outcome: maltreatment, non-organic failure-to-thrive, or relinquishment of the child by the parent. Three aspects of these studies are presented: (1) methodological issues in design, (2) statistical terms used in predictive studies, and (3) details of the 11 studies and a meta-analysis to determine whether accurate predictions of adverse outcomes were made. The results of the meta-analysis suggest that prediction is feasible; improvements, however, in the approach to the assessment of risk are necessary, including further development and use of a standardized clinical assessment.