Studies repeatedly show that double-reading screening mammograms increases interpretation accuracy. Computed tomography (CT) exams of pediatric head trauma should also be double-read, particular when abuse is suspected, say radiologists at the Indiana University School of Medicine in Indianapolis.
A study of second readings of 184 head CT scans of children under age five years with known or suspected head trauma found a high incidence of major errors in the primary reports. Most primary interpretation discrepancies included failure to detect hemorrhage and fractures, as well as mischaracterization of subdural versus enlarged subarachnoid spaces. The possibility of abusive head trauma was also reported more frequently in second interpretations.
Principal investigator Stephen F. Kralik, MD, assistant professor of radiology and imaging sciences, conducted the study to identify and evaluate radiologic interpretation errors made by radiologists at community hospitals. He wanted to determine the extent to which abusive head trauma was overlooked by radiologists who were not specialists in pediatric radiology.
Injured children requiring specialized treatment are transferred from hospitals throughout the state to Riley Hospital for Children. The initial CT scan reports in this study originated from 60 different community hospitals over a four-year period.
Two pediatric neuroradiologists blinded to the original reports as well as the patients’ medical records each independently reported the head CTs. Any discrepancies in the findings of these two neuroradiologists were recorded to identify an inter-reader discrepancy rate. The radiologists then worked together to reach a consensus opinion.
RADPEER™ scores were assigned for each primary and secondary interpretation. Reports were also compared to determine if they contained language expressing concern about potential abuse. Clinical records were checked for all patients to identify those children whose injuries were caused by abuse.
Both primary and secondary interpretations had statistically significant overall discrepancy rates (28% versus 6%) and major discrepancy rates compared to inter-reader discrepancies (16% to 1%). Discrepancy rates in the identification of head trauma caused by abuse also differed significantly. Most worrisome: of 15 patients identified in the second interpretation as potentially the victim of abuse, only one was identified in the primary interpretation. Ultimately, clinical records showed that 9 of these 15 patients were victims of abuse.
The authors stated that the high percentage of missed intracranial pathologies and the lack of documentation of potential abuse in the initial interpretation emphasize the importance of pediatric neuroradiologists providing a second interpretation. They wrote that “radiology error in head CT interpretations may contribute to misdiagnosis or underdiagnosis of abusive head trauma. It has been estimated that up to 28% of children with missed abusive head trauma diagnosis may be reinjured, leading to permanent neurological damage or death.”
The authors suggest that the high error rates may be the result of:
“It cannot be overstated that the radiologist performs a significant role in the diagnosis of abusive head trauma….Beyond being able to detect the findings, radiologists must be aware of the imaging findings that are significantly associated with abusive head trauma. This will provide an important role in their potentially being the first physicians to suspect abusive head trauma, possibly leading to the recommendation of a skeletal survey to assess for additional findings of trauma and a child protective services referral,” they concluded.
The value of double-reading pediatric trauma head CTs. Appl Radiol.
Cynthia E. Keen is a freelance writer based in Sanibel Island, Florida.