Systematic review of instruments designed to predict child maltreatment during the antenatal and postnatal periods

The purpose of this review was to identify instruments designed to predict future child maltreatment and to evaluate their predictive accuracy. A number of electronic databases were searched to identify published studies. Only studies that had utilized a prospective methodology and tested at least one standardized instrument designed to identify families during the antenatal or postnatal periods who are at increased risk of maltreatment were included in the review. Eight studies were included that had prospectively tested a standardized instrument to predict child maltreatment during the antenatal or postnatal periods. Only six of the included studies followed up the entire sample, and of these only two did not provide an intervention for some or all of the high-risk group. While many of the included instruments obtained sensitivity and specificity above 80%, only two of the included instruments combined specificity over 80% with a positive predictive value above 25%. Currently available instruments have limited accuracy, suggesting that they should only be used to focus non-punitive interventions. These findings, in conjunction with other issues such as the stigma attached to the use of such instruments and difficulties in their application within a clinical and research context, suggest the need for a new approach to the identification of parents in need of higher levels of intervention. Copyright © 2003 John Wiley & Sons, Ltd.