G-3 CHILD ABUSE EXPERTS SHOW WIDE VARIABILITY IN SETTING THE THRESHOLD FOR REPORTING SUSPECTED CHILD ABUSE

Tuesday, October 21, 2008: 3:00 PM
Grand Ballroom D (Hyatt Regency Penns Landing)
Benjamin H. Levi, MD, PhD and Megan M. Whitehead, BA, Penn State College of Medicine, Hershey, PA
Purpose: Non-expert mandated reporters have demonstrated wide variability in interpreting the threshold for reporting suspected child abuse. The purpose of this study was to discover whether clinical and research experts in the field of child abuse have a uniform understanding of what reasonable suspicion means in the context of mandated reporting.

Methods: A survey was administered to members attending the annual meeting of the Helfer Society —an international honor society of clinical and research experts in the field of child abuse. Respondents were asked to imagine having encountered a child who has an injury (or condition) that might have been caused by abuse. Using a rank order scale (1=most likely explanation, 10=least likely explanation) and an estimated probability scale (mark an “x” along the line between 0-100% possibility), respondents were asked to indicate how probable abuse would need to be before they felt there was “reasonable suspicion” of child abuse. An open-ended item also asked individuals to describe what “reasonable suspicion” means.

Results. Completed surveys were returned by 81 Helfer Society members (69% response rate, average age = 48, 69% female). There was wide variation between individual responses, with no consensus on any measurement scale. Open-ended responses ranged from “any concern” that child abuse may have occurred to “the most likely explanation” to “a belief that abuse has occurred.” Some respondents reported that “child abuse” would need to be first on one's differential diagnosis to qualify as reasonable suspicion, while others responded that it could be as low as the tenth. Similarly, respondents variously identified reasonable suspicion as corresponding to an estimated probability as low as 10% probability that abuse had occurred, to as high as 100% probability. Moreover, in comparing individual responses for the different scales, 84% of respondents were internally inconsistent, demonstrating that framing effects resulted in significant differences.

Conclusion: There is wide variation in how clinical and research experts in the field of child abuse interpret reasonable suspicion that a child may have been abused. Different frameworks for explicating the meaning of reasonable suspicion elicited significantly different answers from respondents. This variability raises concern over the justifiability of current thresholds for mandated reporting, which are the impetus for over 2.5 million investigations of suspected child abuse in the United States annually.