Hospitalized Cases of Child Abuse in America

BACKGROUND: Orthopaedic surgeons need to recognize features of child abuse. The purpose of this study was to identify common characteristics of child abuse and to delineate features of physical injury cases that would raise concern for child abuse. METHODS: A nationally representative sample of all pediatric inpatient discharges for calendar years 1997, 2000, and 2003 was queried for physical injury diagnoses. Cases were identified as those with a diagnosis of child abuse and controls were those without the diagnosis of child abuse. Incidence of coded child abuse and a comparison of rates across time, demographics, and injury pattern were determined. Cases were compared with controls, and relative risks for coded child abuse were computed using the generalized linear model specifying the Poisson distribution and a log link. RESULTS: There were 665,195 physical injury cases identified. Of these, 11,554 (1.74%) had a diagnosis of child abuse. Fracture requiring orthopaedic management was present in 28% of the child abuse cases. The 3 strongest demographic predictors of coded child abuse were age younger than 1 year, age 1 to younger than 2 years, and Medicaid as primary payer with adjusted relative risks of 11.46, 3.07, and 1.99, respectively. Winter and weekday presentation were significantly higher for coded child abuse. Fractures with the greatest adjusted relative risks for coded child abuse were rib or sternum (5.34) and scapula (3.22). Pelvic fracture was the only fracture significant for lowered adjusted relative risk of coded child abuse. CONCLUSIONS: This study shows that young age continues to be a strong predictor of child abuse in the setting of physical injury. Supporting features include Medicaid as primary payer and winter or weekday presentation. Orthopaedists should be particularly aware of child abuse as 28% of cases had a fracture requiring orthopaedic management. LEVEL OF EVIDENCE: Prognostic Study, Level III (case-control study).