Abstract
OBJECTIVE: Prior research on confessions of physical abuse (PA) has explored mechanisms of injury. Little is known about conditions supporting a confession of abuse or how confessions may influence case outcomes. METHODS: This cross-sectional study of suspected PA in children aged <10 years at 10 centers participating in CAPNET, a multicenter child PA research network, included children with in-person Child Abuse Pediatrics (CAP) consultations between February 2021 and December 2022 and excluded children without injury. Our focus was a confession of inflicted injury known to the CAP during clinical involvement. We compared child and clinical characteristics, out-of-home (OOH) placements, and arrests between cases with and without confessions. Multivariable models using generalized estimating equations (GEE) produced adjusted predicted probabilities (APP) clustering by CAPNET site. RESULTS: Confessions were known to CAPs in 115/4297 (2.7%) cases, with significant site variability. In a multivariable GEE model, confessions were more likely in cases with injuries with high specificity for abuse (APP 6.0% vs 1.6%, P < .001), near-fatality (APP 5.6% vs 2.3%, P < .001), and older children (APP 4.8% vs 2.5%, P = .025). OOH placements and arrests were more common in cases with confessions. CAP awareness of confessions, OOH placements, and arrests did not follow racial or ethnic patterns seen in the US child welfare system. Lower socioeconomic status was associated with OOH placements and arrests but not with CAP awareness of confessions. CONCLUSIONS: Our findings highlight differences in case characteristics and outcomes where a confession is known to the CAP and raise questions about agency response based on child characteristics.