Subdural hematoma, retinal hemorrhage, and fracture triad as a clinical predictor for the diagnosis of child abuse

Abstract

OBJECTIVE Nonaccidental trauma (NAT) is a major cause of traumatic death during infancy and early childhood. Several findings are known to raise the index of clinical suspicion: subdural hematoma (SDH), retinal hemorrhage (RH), fracture, and external trauma. Combinations of certain injury types, determined via statistical frequency associations, may assist clinical diagnostic tools when child abuse is suspected. The present study sought to assess the statistical validity of the clinical triad (SDH + RH + fracture) in the diagnosis of child abuse and by extension pediatric NAT. METHODS A retrospective review of The University of Arizona Trauma Database was performed. All patients were evaluated for the presence or absence of the components of the clinical triad according to specific International Classification of Diseases (ICD)–10 codes. Injury type combinations included some variation of SDH, RH, all fractures, noncranial fracture, and cranial fracture. Each injury type was then correlated with the ICD-10 codes for child abuse or injury comment keywords. Statistical analysis via contingency tables was then conducted for test characteristics such as sensitivity, specificity, positive predictive value, and negative predictive value. RESULTS There were 3149 patients younger than 18 years of age included in the quantitative analysis, all of whom had at least one component of the clinical triad. From these, 372 patients (11.8%) had a diagnosis of child abuse. When compared to a single diagnosis of either SDH, RH, all fractures, noncranial fracture, or cranial fracture, the clinical triad had a significantly greater correlation with the diagnosis of child abuse (100% of cases) (p < 0.0001). The dyad of SDH + RH also had a significantly greater correlation with a child abuse diagnosis compared to single diagnoses (88.9%) (p < 0.0001). The clinical triad of SDH + RH + fracture had a sensitivity of 88.8% (95% CI 87.6%–89.9%), specificity of 100% (95% CI 83.9%–100%), and positive predictive value of 100% (95% CI 99.9%–100%). The dyad of SDH + RH had a sensitivity of 89.1% (95% CI 87.9%–90.1%), specificity of 88.9% (95% CI 74.7%–95.6%), and positive predictive value of 99.9% (95% CI 99.6%–100%). All patients with the clinical triad were younger than 3 years of age. CONCLUSIONS When SDH, RH, and fracture were present together, child abuse and by extension pediatric NAT were highly likely to have occurred.