The multi-institutional validation of the new screening index for physical child abuse.

BACKGROUND/
PURPOSE: There is currently no evidence-based screening instrument to assist in the detection of physical child abuse patients. The screening index for physical child abuse (SIPCA) was previously developed as a potentially new tool for this need. It is a scale that assigns point values, on the basis of variable weights from logistic regression models, to age and patterns of injuries (including fracture of base or vault of skull, contusion of eye, rib fracture, intracranial bleeding, multiple burns), with higher scores indicating greater suspicion for abuse. The purpose of this study is to validate this new tool in another independent data set.
METHODS: A cross-sectional hospital discharge database from 1961 hospitals in 17 states is used (n = 58558). Children aged 14 years or younger with International Classification of Diseases, Ninth Revision, Clinical Modification codes 800 to 959 are included for analysis. Child abuse cases are identified by E codes and certain International Classification of Diseases, Ninth Revision, Clinical Modification codes in the 995.5x range. Screening index for physical child abuse performance is evaluated by discrimination (receiver operating characteristic) and goodness of fit (pseudo r2).
RESULTS: A total of 447 abused patients (0.76%) was identified. The receiver operating characteristic of SIPCA in this data set is 0.89 as compared with 0.86 in the development data set. The pseudo r 2 of SIPCA in this data set is 0.26 as compared with 0.28 in the development data set. A SIPCA score of 3 has a sensitivity of 86.6% and a specificity of 80.5% for detecting physical abuse; raising the threshold to a score of 4 improves the specificity to 93.1% but at a loss of sensitivity to 71.8%.
CONCLUSIONS: The validity of the new SIPCA instrument is supported by its performance in an independently derived data set. A score of 3 on SIPCA represents a balanced trade off in the sensitivity and specificity of the instrument in detecting physical abuse and is an optimal threshold above which to begin considering abuse in differential diagnosis. Application of the instrument could assist clinicians in detecting physical child abuse cases among pediatric trauma patients.