Diagnostic Imaging of Child Abuse

The concept of child abuse as a medical entity has an origin in the studies of pediatric radiologist John Caffey, MD,¹  as well as many other specialists in the field of diagnostic imaging.^2–4  Kempe et al⁵  relied heavily on the work of Caffey and his protégé, Frederick Silverman, MD,⁶  when developing the now-familiar concept of the “battered-child syndrome.”

When all cases of child abuse and neglect are studied, the incidence of physical evidence documented by diagnostic imaging studies is relatively small. However, imaging studies are often critical, particularly in the assessment of the infant and young child with evidence of physical injury. Imaging alterations may also be the first indication of abuse in a child who is seen with an apparent natural illness. When viewed in conjunction with clinical and laboratory studies, imaging findings commonly provide additional objective evidence in the evaluation of possible inflicted injury or abuse.⁷  For severely abused infants, the imaging findings alone may form the basis for a diagnosis of the inflicted injury. The role of imaging in cases of suspected abuse is not only to identify the extent of physical injury when abuse has occurred but also to elucidate all imaging findings that point to alternative diagnoses.^8,9  All imaging studies involving use of ionizing radiation should be performed in accordance with the ALARA ([using an exposure] as low as reasonably achievable) principle.¹⁰  Because the detection of inflicted skeletal injury depends on the technical quality of the radiographs and the imaging protocol, recommendations regarding imaging should focus on examinations that provide the highest diagnostic yield at acceptable patient risk and cost.¹¹  Certain diagnostic imaging studies entail additional risks associated with sedation, and these risks should be weighed against the benefit of the study.¹² 

Although skeletal injuries rarely pose a threat to the life of the abused child, they are often the strongest radiologic indicators of abuse. In fact, in an otherwise normal infant, certain patterns of injury are sufficiently characteristic to permit a firm diagnosis of inflicted injury in the absence of clinical information.¹³  Furthermore, dating of skeletal injuries may provide investigators with critical temporal data, which may help in identifying potential assailants. These facts mandate that imaging surveys performed to identify skeletal injury be acquired with at least the same level of technical excellence routinely used to evaluate accidental injuries. The “baby gram” (a study that encompasses the entire infant or young child on 1 or 2 radiographic exposures) or abbreviated skeletal surveys have no role in the imaging of these subtle but highly specific bony abnormalities.¹⁴ 

In general, the radiographic skeletal survey is the method of choice for global skeletal imaging in cases of suspected abuse.¹⁴  General-purpose (medium-speed) pediatric imaging systems provide insufficient anatomic detail to image the skeleton of the abused infant or young child. The American College of Radiology has published standards for skeletal survey imaging in cases of suspected abuse. Modern pediatric imaging systems commonly use special film cassettes and intensifying screens to minimize exposure. Although these low-dose systems are adequate for chest and abdominal imaging, they fail to provide the necessary detail (contrast and spatial resolution) to image subtle metaphyseal, rib, and other high-specificity injuries that are characteristic of abuse. According to the American College of Radiology, high-detail imaging systems should be used for suspected abuse in infancy.¹⁴  These systems should be used without an antiscatter grid. Beyond infancy, faster general-purpose systems are required for thicker body regions (eg, skull, lateral lumbar spine).

Digital or filmless radiography has replaced screen/film imaging in most pediatric health care facilities in the United States.¹⁵  A wide array of digital systems are available, including direct and indirect technologies. Data suggest that despite its lower spatial resolution, digital radiography, with its wide dynamic range and high contrast capabilities, can provide diagnostic performance comparable to high-detail screen/film systems in the evaluation of skeletal injury.^16,17  High-detail digital systems are being used successfully in the rigorous field of mammography, and efforts are currently underway to apply similar techniques to demanding skeletal applications.¹⁸  Unfortunately, most digital systems currently used for skeletal surveys in the United States operate at relatively low spatial resolutions, and they are not routinely optimized for the demanding practice of imaging for suspected child abuse.¹⁵  Imaging departments should use digital systems with sufficient spatial resolution and signal-to-noise characteristics to detect subtle skeletal injuries. Image detail and diagnostic performance can be improved by optimizing exposure to diminish system noise. Acceptable diagnostic accuracy with traditional screen/film imaging has required higher radiation exposures than for routine pediatric radiography, and similarly, an increase in exposure over standard digital radiography can be expected if maximal diagnostic accuracy is to be preserved with the digital medium. However, recent data have suggested that a performance comparable to the high-detail screen/film gold standard may be achievable with high-detail digital imaging at a substantially lower patient dose.¹⁷  Digital facilities should optimize their acquisition and display parameters for high diagnostic performance,¹⁹  and if the resultant images provide insufficient bony detail, an acceptable alternative should be sought. A radiologist should monitor the skeletal survey to ensure that appropriate high-quality images are obtained and to determine if additional views are required to fully define the pathologic alterations.

Once the appropriate imaging system is chosen, a precise protocol for skeletal imaging must be developed to ensure consistent quality. In routine skeletal imaging, an accepted principle is that film must be coned or restricted to the specific anatomic area of interest. It is common practice to encompass larger anatomic regions when skeletal surveys are performed, which results in areas of underexposure and overexposure as well as loss of resolution resulting from geometric distortion and other technical factors. The standard skeletal survey imaging protocol that has been developed by the American College of Radiology¹⁴  is provided in Table 1. Of special note is the inclusion of lateral views of the spine to assess for vertebral fractures and dislocations and separate views of the hands and feet to identify subtle digital injuries. Anteroposterior and lateral views of the skull are mandatory even when cranial computed tomography (CT) has been performed, because skull fractures coursing in the axial plane may be missed with axial CT.¹¹  A full 4-view examination of the skull (right and left lateral, Townes, and anteroposterior) should be obtained when head injury is present. Skeletal injuries, especially those requiring orthopedic management, necessitate at least 2 radiographic projections. Oblique views of the thorax increase the yield for the detection of rib fractures.²⁰  At a minimum, right and left oblique views of the thorax should be obtained when rib fractures are evident, and consideration should be given to including obliques in the standard survey protocol. A follow-up skeletal survey approximately 2 weeks after the initial study increases the diagnostic yield^21,22  and should be performed when abnormal or equivocal findings are found on the initial study and when abuse is suspected on clinical grounds. A repeat study may permit more precise determination of the age of individual injuries. Lack of interval change may indicate that the initial radiographic finding is a normal anatomic variant or is related to a bone dysplasia. Views of the skull can be omitted from the follow-up study.

When performed by staff experienced with pediatric nuclear imaging, skeletal scintigraphy may offer an alternative or adjunct to the radiographic skeletal survey in selected cases, particularly for children older than 1 year. Scintigraphy can provide increased sensitivity for detecting rib fractures, subtle shaft fractures, and areas of early periosteal elevation. However, scintigraphy is less sensitive than radiography for detection of classic metaphyseal lesions, which are fractures that carry a high specificity for abuse in infants.^23–25  Skeletal scintigraphy usually requires sedation and is generally more expensive than radiographic surveys. Bone scans are used to supplement radiographic skeletal surveys in the acute-care setting,²⁴  but for the child who is placed in a “safe” environment, a follow-up skeletal survey is an attractive alternative to initial scintigraphy. If radionuclide bone scans are performed as the initial study, all positive areas must be evaluated further with radiography. Because scintigraphy is insensitive for detecting cranial injuries, skull radiography in at least 2 projections must supplement the bone scan.

The skeletal survey is mandatory in all cases of suspected physical abuse in children younger than 2 years; its utility diminishes thereafter.^8,26  The screening skeletal survey or bone scan has little value in children older than 5 years. Decisions about which types of imaging to perform for patients in the 2- to 5-year-old age group must be made individually on the basis of the specific clinical indicators of abuse. At any age, when clinical findings point to a specific site of injury, the customary radiographic protocol for imaging that anatomic region should be used. MRI and ultrasonography may be indicated when epiphyseal separations are suspected on the basis of plain-film radiography.^2,27  Evidence suggests that if 1 infant twin is injured, the other is at risk and should also undergo a skeletal survey.²⁸  Although there seems to be an association between physical and sexual abuse, the prevalence of fractures in sexually assaulted children is low; thus, skeletal surveys should be performed only in selected cases on the basis of specific clinical indications.^29,30  Although the high-quality skeletal survey is essential in the evaluation of suspected physical abuse in infants and toddlers, it may not be available in the emergency department setting during evening hours.¹⁵  The American Academy of Pediatrics has stated that hospitalization of the abused child may be medically necessary for diagnosis,³¹  and in some instances it may be advisable to place a child in a safe haven until a proper skeletal survey can be performed.³¹  A skeletal survey may be difficult to obtain in a critically injured child on life support. Efforts should be directed at performing an adequate examination in a timely manner, because the results may have an effect on the early investigation of the case.

High-energy forces associated with impact or violent shaking result in a variety of central nervous system injuries that can be detected by modern neuroimaging techniques. The evolution of these injuries, as well as processes that develop secondary to the original insult, are often effectively displayed on serial imaging studies.^2,32–35 

All infants and children with suspected intracranial injury must undergo cranial CT, MRI, or both. Strategies should be directed toward the detection of all intracranial sequelae of abuse and neglect with a thorough characterization of the extent and age of the abnormalities.³⁶  In the acute-care setting, efforts are directed toward rapid detection of treatable conditions. Subsequent studies are designed to more fully delineate all abnormalities, determine the timing of the injuries, and monitor their evolution.

CT without intravenous contrast should be performed as part of the initial evaluation for suspected acute inflicted head injury. CT has high sensitivity and specificity for diagnosing acute intraparenchymal, subarachnoid, subdural, and epidural hemorrhage. Abnormalities that require emergency surgical intervention generally are well demonstrated. CT is readily available and rapidly performed for critically ill patients and is generally better than MRI for evaluation of acute hemorrhage. Skull fractures, associated soft tissue swelling, and facial fractures also can be diagnosed on CT images with appropriate bone window and level settings. With modern multidetector CT scanners, the brain can be imaged in a few seconds, usually obviating the need for sedation.

Ultrasonography via the anterior fontanelle in young infants has gained an important role in clarifying the nature of extra-axial fluid collections in infancy. Because ultrasonography reliably differentiates convexity subdural from subarachnoid collections, it is particularly useful for the infant with macrocephaly or any infant with large hypodense cerebral convexity collections demonstrated by CT.³⁷  Subcortical white matter tears in the frontal and anterior parietal parasagittal regions can be demonstrated with ultrasonography.³⁸  These lesions are less well defined by axial CT, and ultrasonography provides the advantage of a bedside technique. Because ultrasonography is insensitive for detecting small acute subdural hematomas, particularly within the interhemispheric fissure, and many other acute intracranial injuries, it must be performed in conjunction with CT or MRI when traumatic injury is suspected.^29,39 

Although cranial MRI has some limitations in the acute-care setting, it remains the best modality for fully assessing intracranial injury, including extra-axial collections, intraparenchymal hemorrhages, contusions, shear injuries, and brain swelling or edema. A strong argument can be made for MRI in all cases with positive cranial CT findings and in selected cases with normal CT results but strong clinical concerns. MRI should be performed with T₁ and T₂ weighting with proton-density or inversion-recovery sequences to differentiate cerebrospinal fluid collections from other water-containing lesions. Gradient echo sequences should be included to detect hemorrhage or mineralization not demonstrable by other MRI techniques. Although the specific type and order of pulse sequences may vary, imaging must be performed at least in the axial and coronal planes. Acute subarachnoid and subdural hemorrhage may be inconspicuous on MRI, and consideration should be given to delaying the examination for 5 to 7 days, permitting subacute blood to become hyperintense on T1-weighted sequences.

Diffusion-weighted imaging (DWI) is a relatively new and valuable technique for the evaluation of stroke and is gaining a role in the assessment of inflicted cerebral injury. When performed early in the critically injured infant, DWI may provide information regarding cerebral injury before parenchymal abnormalities are visible on CT or conventional MRI sequences.^40,41  The value of the potential findings revealed with DWI must be weighed against the lower sensitivity of MRI in the detection of acute extra-axial blood collections and the practical problems encountered when performing MRI in severely ill infants.

Abused infants may not demonstrate neurologic signs and symptoms despite significant central nervous system injury.^2,42,43  MRI offers the highest sensitivity and specificity for diagnosing subacute and chronic injury and should be considered whenever typical skeletal injuries associated with shaking or impact are identified.^39,44 

Plain radiographs are often sufficient to evaluate vertebral compression and spinous process fractures. Complex fractures may require helical CT with multiplanar reformatted images. If a fracture or subluxation may compromise the spinal contents or if clinical findings indicate spinal cord or nerve root injury, MRI should be performed. Increasing attention has been directed toward a possible association of cervical spinal cord injury and extra-axial hemorrhage with inflicted head injury, and some centers include the cervical region in their cranial MRI trauma protocols.^45,46 

Blunt thoracoabdominal injury may occur in victims of child abuse. The evaluation and management of acute thoracoabdominal injury is the same as for children with accidental injuries.⁴⁷  However, when an infant or child sustains serious injury to the chest or abdomen without a known or observed mechanism or when the imaging findings are inconsistent with the purported history, investigation of potential child abuse is warranted. Failure to recognize inflicted blunt abdominal trauma contributes to higher morbidity and mortality rates than those seen with accidental abdominal injury.⁴⁸  Pancreatitis, duodenal hematomas, bowel perforation, and thoracoabdominal injury associated with rib fracture heighten the suspicion of child abuse. Chest, abdominal, and cervical spine radiographs often are obtained in the initial assessment of injured children. If internal chest or abdominal injury is suspected and the patient's condition is stable, a CT scan should be performed. A CT scan will best demonstrate many of the injuries associated with child abuse. The chest should be included if serious chest trauma is suspected.

The use of oral contrast is debatable. Oral contrast in the stomach and small bowel is useful to better define the lesser sac of the peritoneum, pancreas, duodenum, and jejunum. However, oral contrast may place the patient at greater risk of aspirating, especially if the patient is obtunded, sedated, or immobilized. If surgery or general anesthesia is likely, it is better to have an empty stomach.

Intravenous contrast is used routinely. Vascular injuries and injuries to the liver, spleen, pancreas, and kidneys are best demonstrated after administration of intravenous contrast material. Helical scanning with proper timing of the intravenous contrast bolus is important for accurate diagnosis. Given the heightened awareness of cancer risks associated with CT scanning in childhood, care should be taken to adjust technical factors to achieve diagnostic quality images at an exposure as low as reasonably achievable.^49,50  The only relative contraindications for intravenous contrast are a strong history of allergy to iodine, severe shock, and renal failure.

Abused children suffer some of the same injuries as children with accidental blunt trauma. In the chest, pulmonary contusion, pneumothorax, pleural effusion, rib fractures, and vascular or tracheobronchial injuries may occur. Abused children have an increased occurrence of pancreatic injuries and duodenal hematomas. Bowel injury should be suspected when there is peritoneal fluid without evidence of solid organ injury and assumed when free intraperitoneal air or extraluminal contrast is observed.⁵¹  Bone windows should be monitored not only for rib fractures but also for signs of pelvic or spine fractures.

Peritoneal lavage rarely is used in pediatric practice. If performed before CT, it may decrease the diagnostic usefulness. Peritoneal lavage sometimes is used when emergency surgery is required to treat a patient whose condition is not stable enough for a CT scan.⁴⁹ 

The use of ultrasonography in pediatric trauma is controversial. Some institutions have used ultrasonography successfully for a more detailed, comprehensive evaluation of organ injury. However, for seriously injured children and those with suspected child abuse, CT scanning is the preferred initial diagnostic modality in most institutions. Peritoneal fluid alone, which can be detected well with both ultrasonography and CT scanning, is a poor predictor of major trauma in children. An upper-gastrointestinal series sometimes is used to evaluate and follow-up duodenal hematomas.

Nonoperative management of injury to the liver, spleen, kidney, or pancreas is common in most pediatric centers. Follow-up imaging usually is limited but may be useful to help determine recommendations for the level of physical activity (Table 2).

The diagnostic imaging of suspected inflicted injury in infancy and childhood should be performed with at least the same rigor used in the imaging evaluation of accidental trauma and naturally occurring disease. To be confident that the imaging studies are acquired and interpreted in a thorough and informed manner, clinicians charged with reporting and providing evidence in cases of suspected abuse should work in close collaboration with radiologists experienced in pediatric imaging. This approach will help ensure that child abuse is accurately identified and reliably differentiated from conditions that may simulate abuse.¹⁴ 

Michael A. Di Pietro, MD, Chairperson

Alan S. Brody, MD

Christopher I. Cassady, MD

Paul K. Kleinman, MD^*

John B. Wyly, MD

Kimberly E. Applegate, MD

Beverly P. Wood, MD

Jeffrey M. Zerin, MD

Maria G. Mercado-Deane, MD

Joanna J. Seibert, MD

Aleksandra Stolic, MPH