Optimizing Advanced Imaging of the Pediatric Patient in the Emergency Department: Technical Report

Diagnostic imaging is an important tool frequently used in the emergency department (ED), and 1 out of every 4 children evaluated in the ED receives some type of imaging.¹ Advanced imaging (ie, computed tomography [CT], ultrasonography, and magnetic resonance imaging [MRI]) has revolutionized the evaluation and management of illness and injury. Pediatric patients are responsible for only 20% of total ED volumes² but do require unique considerations with respect to both their care and imaging strategies in the ED. Point-of care ultrasonography use by emergency physicians is not addressed as it is outside the scope of this document.

Imaging for children evaluated in the ED has increased over time without improvements in patient outcomes.^1,3–7 These patterns have called into question the appropriateness of use of imaging, and many studies have demonstrated that much of the advanced imaging performed in EDs in children is low value, with the risks outweighing the net patient benefit.^1,8–16 Risks specific to imaging include those from radiation exposure,^17–19 false-positive and incidental findings and the downstream testing that may result,^20–24 increases in ED length of stay,^25,26 sedation,²⁷ transport away from the ED, and overall health care costs.¹⁴ In addition, there is the risk that a study will need to be repeated if not optimally performed, thus compounding the aforementioned risks. Many emergency physicians agree that a significant proportion of advanced imaging studies performed in the ED is medically unnecessary.²⁸ 

In an effort to curb low-value imaging as well as optimize imaging strategies, initiatives, including Choosing Wisely²⁹ (Table 1) and Image Gently,³⁰ seek to educate those who care for patients about imaging that may not be necessary and the associated risks. The American College of Radiology (ACR) has developed a repository of imaging guidelines for many pediatric conditions.^31–40 The ACR Appropriateness Criteria are derived from up-to-date, available evidence with periodic review by expert multidisciplinary panels and provides appropriateness ratings and relative radiation risks from imaging studies. The ACR Pediatric Rapid Response Committee has developed additional pediatric clinical scenarios and imaging appropriateness ratings.⁴¹ The appropriateness rating of an imaging study is solely based on the study accuracy and ratings are not determined with consideration of radiation risk, cost, and other factors associated with imaging decisions. The ACR Appropriateness Criteria have also been incorporated into clinical decision support to aid clinicians at the point of care.⁴² 

Inherent in the decision to perform a diagnostic test is the balance between benefits and risks to the patient. Advanced imaging carries risks, including false-positive and incidental findings.^20–24 These may lead to subsequent unnecessary tests, treatments, procedures, or visits and create undue caregiver anxiety. There are risks associated with the sedation that may be necessary for some children to obtain an optimal study.²⁷ There are risks associated with intravenous (IV) contrast from CT⁴³ and MRI.⁴⁴ Finally, advanced imaging results in increased ED lengths of stay^4,25,26 and increased financial costs to the patient and the health care system.¹⁴ 

Historically, CT was the imaging modality of choice for the evaluation of many conditions because of its speed, accuracy, and convenience, with many EDs having a dedicated CT scanner. Imaging requiring CT has come under increasing scrutiny given the exposure to ionizing radiation. Radiation risks are more pertinent in pediatric patients, given their relatively greater body surface-to-weight ratio and the inverse relationship between patient age and risk of future malignancy from ionizing radiation exposure.⁴⁵ Although randomized clinical trials are not feasible, several large observational cohort studies have found a small, but significant, increase in the risk of malignancy from childhood exposure to CT imaging.^17–19 

Given the aforementioned risks, it is important for emergency physicians, physician assistants, and nurse practitioners to ensure that imaging strategies are optimized for children in the ED. Such strategies include seeking radiation-reduction strategies, including minimizing the radiation from CT (Box 1) as well as alternatives to CT imaging, when feasible. The ACR recommends that radiation doses from CT be “as low as reasonably achievable.”⁴⁶ Radiology departments are able to become accredited by the ACR if they achieve certain quantitative benchmarks, the description of which are outside the scope of this document. If CT is determined to be the optimal study because of the patient’s clinical status or based on a lack of alternative, nonradiating modalities available, it is important that the radiation dose include patient weight or size-based parameter adjustments and that lower-dose parameters are used when appropriate for the indication.⁴⁴ Instructions on how to “child-size” CT protocols using diagnostic reference levels are available for institutions seeking to improve their pediatric CT imaging strategies.⁴⁷ Strategies that can be employed by the individual ordering the study include minimizing potentially unnecessary CT examinations and collaborating with the radiologist who will be interpreting the imaging, which may reduce unnecessary scanning, and also ensure that an optimal study is performed.⁴⁸ Although it is important to minimize unnecessary imaging studies, it is also important to ensure that appropriate imaging studies are performed when indicated to avoid missed diagnoses.⁴⁹ In some cases where observation in lieu of CT imaging is an option, CT may be the more appropriate strategy when taking into account length of stay, parental preferences,^50,51 and ability to follow-up in a timely fashion.

As an alternative to CT imaging, ultrasonography is being increasingly used as a nonradiating modality to diagnose many pediatric conditions.¹ Ultrasonography can be performed at the bedside to avoid transport for critically ill or injured patients. However, ultrasonography is operator dependent,⁵² and many institutions that do not routinely care for children may lack the experience, protocols, and/or equipment for pediatric indications. With the relatively recent introduction of ultrasonographic contrast agents to the United States, the variety of indications for ultrasonography has expanded, resulting in even more alternatives to CT or MRI.^53,54 

MRI is another attractive alternative in children, but its utility can be limited by examination duration, potential need for sedation, cost, and reduced availability. In the last 2 decades, faster, better tolerated, “rapid” MRI (rMRI) protocols have emerged as an alternative to CT for many pediatric indications.^55–59 Prior work has found rMRI to be both accurate and feasible, and implementation of rMRI protocols can safely reduce the use of CT.^60,61 There are also strategies that have been developed to assist patients with tolerating MRI that may help younger patients tolerate the examination.⁶² rMRI may not be available in some settings and remains a longer examination than CT, which are important barriers, particularly for critically ill and injured patients. Use of MRI may also result in higher health care costs.

Racial and ethnic inequities in health care delivery are well recognized, and pediatric emergency care is no exception.⁶³ Specific to emergency imaging, studies have found that non-Hispanic white children are more likely to undergo imaging than are children of other racial groups,^{8–10,15,64–69} although some of this imaging is of low value.^11,64 The etiology of this differential care is likely multifactorial and includes parental, clinician, and system-level influences. It is critical for those who care for children to understand that these disparities exist, to recognize their potential in real time, and to work toward mitigating them. Evidence-based clinical guidelines (see Supplemental File) and clinical decision support tools can reduce variability in practice patterns for specific conditions.^70–73 Internal quality assurance audits may assist clinicians with a better understanding of their own practices and highlight possible unintended disparities in care delivery.

Clinical decision support mechanisms (CDSMs) communicate appropriate use criteria to ordering physicians, physician assistants, and nurse practitioners to assist with the selection of appropriate imaging studies. Appropriate use criteria are developed and/or endorsed by national professional medical specialty societies or other clinician-led entities to assist in making appropriate clinical decisions.⁴¹ CDSMs may incorporate data from validated clinical decision rules and can be embedded into order entry systems and electronic health records so they are readily available at the point of care. Implementation of CDSMs into clinical practice has shown potential to increase the frequency of indicated studies, reduce the frequency of low-value studies, and improve positive yield of imaging.^74–84 The Protecting Access to Medicare Act of 2014 requires clinicians to consult evidence-based appropriate use criteria to be eligible for payment and this can be documented through use of Centers for Medicare and Medicaid Services-qualified CDSMs.^85,86 Although this requirement does not extend to the management of pediatric patients presently, it has driven more widespread point-of-care availability of both adult and pediatric CDSMs, including those based on the ACR Appropriateness Criteria.⁴² 

Introduced in 1993, the ACR Appropriateness Criteria provide rigorous, evidence-based guidelines for those who order imaging testing. When scientifically based data are not sufficiently available, consensus is obtained from experts in the field with input from both radiologists and nonradiologists. As of 2021, there are more than 200 diagnostic and interventional radiology topics available, 21 of which are pediatric-specific.⁸⁷ The relatively limited set of criteria available for pediatric conditions is due, in part, to the general scarcity of clinical trials and data pertaining to children relative to adults. The Pediatric Rapid Response Committee, a joint committee of the ACR and the Society for Pediatric Radiology, was created to develop guidance for appropriate imaging of an expanded set of pediatric clinical scenarios, approximately 300 more than are currently available through the ACR Appropriateness Criteria. Although the limited availability of high-quality data does not allow these documents to meet the standard for ACR Appropriateness Criteria, they have been developed with similar transparent methodology and reliance on published data and expert consensus.⁴¹ 

Shared decision-making may be another strategy to optimize imaging in children, particularly in low-risk patients.⁸⁸ Defined as the process of health care professionals making decisions in conjunction with patients or families through the mutual assessment of options and risk and consideration of family preferences,⁸⁹ shared decision-making has mounting evidence supporting its integration into decisions about diagnostic tools in both adults and children.⁹⁰ Several organizations and professional societies support the use of shared decision-making, including the American College of Emergency Physicians⁹¹ and the American Academy of Pediatrics.⁹² With respect to diagnostic imaging, some have opined that it is the ethical responsibility of those who order imaging to engage patients and families in shared decision-making.^93–95 In emergency medicine, shared decision-making may be used in cases where the decision to perform advanced imaging may not be absolute. For example, according to the Pediatric Emergency Care Applied Research Network (PECARN) head injury clinical decision rules,⁵⁰ in children with minor head trauma at intermediate risk of clinically important head injury, the decision to perform CT may be guided by clinician experience, parental preference, and/or symptom progression. As such, these patients represent an excellent opportunity for incorporating shared decision-making. Understanding the risks of clinically important injuries and having the opportunity to participate in the interpretation of these risks can be a helpful step in the provision of family and patient-centered care.⁹⁶ The use of decision aids, patient-centered tools to assist in applying the latest evidence and patients’ values and preferences into care decisions,⁹⁷ may also be incorporated into the shared decision-making approach. A decision aid used for patients with intermediate risk of clinically important traumatic brain injury was found to safely reduce health care utilization 1 week after injury.⁹⁸ Although the tool did not reduce the rate of CT in the ED, it did increase parental knowledge, decrease decisional conflict, and increase involvement in decision-making. These findings were consistent with prior studies of decision aids in medicine.⁹⁷ Another potentially important condition for which shared decision-making may be applied, although not specifically studied to date, is the evaluation of children with abdominal pain who are at low risk for appendicitis based on clinical risk scores.^99–101 Decisions regarding CT imaging, transfer to a pediatric center for ultrasonography, or discharge with return precautions may be facilitated through the use of shared decision-making. Although data on shared decision-making in emergency medicine continue to emerge, surveys have found that both patients¹⁰² and emergency physicians^103,104 are supportive of this approach in emergency care. Shared decision-making may also be used as an aid for improving equity and bringing historically marginalized and/or vulnerable populations into discussions regarding their child’s health.^105,106 

When possible, the goal is for children to receive definitive care at the originating ED. Most children (approximately 80% of ED visits by pediatric patients) seek care in general EDs^107,108 rather than specialized children’s hospitals, and there is a decreasing ability of hospitals to provide inpatient pediatric care.¹⁰⁹ Protocols and processes are available to ensure timely transfer of those patients who would benefit from specialized care and may minimize delays in definitive care, reduce radiation exposure, and limit redundancy of imaging.^110,111 

It is important for physicians, physician assistants, and nurse practitioners who work in EDs without pediatric inpatient or subspecialist capabilities to consider what tests are important at their institution if the patient will eventually require transfer to a pediatric facility. Imaging that proves suboptimal because of lack of pediatric equipment, excess movement by the patient, inadequate scanning protocols, or lack of contrast when needed is often repeated at the receiving institution, further increasing risks to the child and costs. Imaging typically falls into 3 categories: imaging that may determine whether transfer is needed; imaging that determines the need for emergent intervention that can be provided by the originating ED; and nonemergent imaging in a patient who will be transferred regardless of the imaging findings. Imaging pediatric patients who fall into the third category only delays definitive care insofar as the time needed to obtain imaging at the referring facility as well as the time to access the imaging by the receiving facility. Therefore, such imaging is best deferred to the receiving facility. Hospitals with higher numbers of pediatric visits often have special equipment to support the pediatric patient, technologists who may be more experienced in imaging children, and radiologists with expertise in pediatric radiology. Pediatric hospitals, specifically, are more likely to use CT protocols that impart significantly less radiation dose compared with nonpediatric hospitals,^112–114 and nonradiating modalities may be preferable and more readily available. It is important for physicians, physician assistants, and nurse practitioners to be familiar with the protocols in place at their institution. In patients for whom imaging may be able to prevent transfer, however, it is reasonable to obtain that imaging at the originating ED. This may be particularly important in rural locations or other settings in which transfer creates a burden for families and the health care system.

One key resource available to assist with pediatric readiness and the development of transfer agreements is the Emergency Medical Services for Children Innovation and Improvement Center.¹¹⁵ This federally funded center seeks to improve pediatric emergency care with a particular focus on quality improvement. There are available resources on pediatric readiness¹¹⁶ as well as a toolkit to help develop transfer agreements between general EDs and children’s hospitals.¹¹⁵ In the event of a transfer, it is important to ensure proper transfer of any imaging and radiology reports performed at the index hospital. For hospitals without easy mechanisms for transfer, imaging practices can be made as pediatric-friendly as possible, with a reduction in CT-related radiation dosing when feasible and increased use of nonradiating modalities, such as ultrasonography and MRI.

Teleradiology allows for the provision of efficient access to radiologic services and subspecialty pediatric radiology expertise in settings where such interpretations would be limited.^117–119 Image-sharing technologies allow for second opinion interpretations of imaging studies by pediatric radiologists as well as review by the treating team at receiving institutions where pediatric patients are transferred for subspecialty care. Pediatric radiologist-interpretation of pediatric imaging studies is important to ensure diagnostic accuracy and appropriate patient management.^120,121 For example, up to 42% of all imaging studies submitted for secondary interpretation to a tertiary pediatric care facility were found to have discrepancies from the initial interpretation, with nearly 22% of all secondary interpretations being deemed “major” discrepancies.¹²⁰ The ACR, American Association of Physicists in Medicine, and Society for Imaging Informatics in Medicine advocate for each referring facility to, at a minimum, have either the capability to transmit images securely over the internet or mechanisms for sharing on physical media to account for those unable to access a network¹²²; one or the other are recommended at a minimum. The full set of digital images and a subset of prerendered images of appropriate quality are needed by the receiving institution, as is a mechanism to ensure accurate association and reconciliation of foreign patient identifiers, accession numbers, and procedure descriptions or codes. It is important that images are accompanied by an interpretation of the images from the referring facility. The standards and practice parameters also specifically recommend against proprietary formatted images or embedded viewers that depend on proprietary image formats.¹²² The ACR outlines guiding principles that support teleradiology activities, including training, certification, licensing, and medical staff inclusion where radiologists are engaged in patient care.¹¹⁷ 

Patients often receive advanced imaging and are subsequently transferred to have their diagnostic imaging studies reviewed by a pediatric radiologist or other pediatric medical subspecialist (eg, plastic surgery, otolaryngology). Because many imaging findings may not be clinically significant, referring facilities could partner with pediatric hospitals to allow real-time review of outside images, as such review may preclude the need for the patient to be transferred.

The ACR, in conjunction with other organizations, periodically defines new practice guidelines and technical standards for radiologic practice.^122,123 These documents outline the procedures and minimum system requirements needed for the protection, privacy, security, and integrity of patient information while ensuring appropriate access for patient care activities. Adherence to these standards and Electronic Protected Health Information regulations are necessary to ensure compliance with the Health Insurance Portability and Accountability Act,¹²⁴ and the Health Information Technology for Economic and Clinical Health Act.¹²³ Clinicians frequently request new interpretations from pediatric radiologists of outside imaging studies. To demonstrate the medical necessity for payer consideration for payment, the report would state the purpose of the interpretation and the name of the physician requesting the new interpretation.¹²⁵ 

Approximately 1% of ED visits by children are for seizures.¹²⁶ ED management of a child with an afebrile seizure involves both cessation of seizure activity and evaluation for seizure etiology. Although many children presenting to the ED with a seizure may ultimately require neuroimaging, such imaging is typically not emergent and may be deferred to the outpatient setting.^127,128 Recommendations for deferred imaging are based on exceedingly low rates (<1%) of emergent or urgent findings that acutely change management.^129–133 This is in contrast to adult patients who present to the ED with seizures and have higher rates of clinically significant pathology.^{127,134–138} Importantly, most children with abnormal neuroimaging have significant findings on history and/or physical examination, present in status epilepticus, or are critically ill.^{129–131,139} The ACR Appropriateness Criteria for seizures in children includes several clinical scenarios with corresponding imaging strategies.³¹ 

The decision regarding the need for emergent neuroimaging is based on multiple factors. Children with a first-time afebrile seizure are more likely to have abnormal imaging that changes clinical management versus children with a history of prior seizure.¹²⁹ Therefore, it is important for children with history of prior seizure or epilepsy to be evaluated for other causes of breakthrough seizure, as the rate of acute imaging abnormalities is low.^140,141 Among children with a first-time afebrile seizure, emergent neuroimaging may be considered in children who are critically ill or present with status epilepticus, infants younger than 6 months, children for whom a reassuring neurologic examination cannot be obtained (abnormal neurologic examination, intubated), children with focal onset of seizure, children with comorbid conditions or histories that would increase the likelihood of abnormal findings (eg, ventriculoperitoneal shunt, bleeding diathesis, congenital cardiac disease, immunocompromised status, history of trauma, signs or symptoms concerning for abuse), and children who do not return to their neurologic baseline while in the ED.^{127–131,133,139,142–144} Imaging in the ED may also be considered for patients with epilepsy who present with new seizure semiology or patients for whom follow-up may be unreliable.

Febrile seizures represent one of the most common causes of seizure in children who present to the ED.¹⁴⁵ A febrile seizure is defined as a seizure occurring in a child 6 months to 6 years of age that is associated with a fever greater than or equal to 100.4°F (38°C) in a child without a history of afebrile seizures and that is not caused by a central nervous system infection or inflammation or other metabolic abnormality.^145,146 A simple febrile seizure is further defined as a primary generalized seizure, lasting <15 minutes and without a recurrence within 24 hours.^145,146 Any febrile seizure not meeting the criteria for a simple febrile seizure is considered complex. The yield of emergent neuroimaging is usually very low for either simple or complex febrile seizures as the incidence of emergent and/or significant intracranial findings is very low (0% to 1.5%).^{145,147–149} In those children with findings on emergent neuroimaging, most will have other significant findings on examination including nystagmus, persistent vomiting, hemiparesis, bruising (suggestive of child abuse), or altered mental status that persists after a period of observation.¹⁴⁹ In children with developmental delay or complex febrile seizures that are prolonged or associated with focality, MRI may be obtained to evaluate for risk factors for future epilepsy; however, this need not be performed on an emergency basis in the ED.¹⁵⁰ 

For children requiring neuroimaging for these seizures, MRI is the preferred imaging modality.^128,146 If MRI cannot be obtained in a timely manner or if a child is too unstable to undergo MRI, CT may be used as an initial imaging examination with potential need for subsequent MRI. Noncontrast head CT may be used to rapidly rule out intracranial hemorrhage, hydrocephalus, and some masses or tumors and strokes.¹⁵¹ However, CT is less sensitive than MRI for identifying other urgent or emergent etiologies for seizure.¹⁵² The addition of contrast-enhanced imaging to the noncontrast head CT has not been shown to change management in limited studies of children with seizures.¹⁵³ Although MRI has the advantage of higher sensitivity compared with CT for identifying seizure etiology, these benefits can be weighed against its challenges.¹⁵⁴ Image acquisition time for MRI is significantly longer than CT, leading to the potential need for sedation, prolonged periods away from emergency care clinicians, and more remote transport from the ED in many care settings.¹⁵⁵ A 7-minute rapid brain MRI with limited sequences has been proposed as an initial screening examination for children presenting with neurologic complaints, including seizure, and has been shown to reduce CT use without missed intracranial findings.⁶⁰ 

Nontraumatic headache is a common reason for children to present to the ED, and visit rates have increased over time.^156,157 Given the numerous etiologies of headache, emergency physicians, physician assistants, and nurse practitioners are faced with the challenge of determining when to order neuroimaging. Less than 3% of children who present to the ED with headache are diagnosed with serious neurologic disorders, and most children with significant intracranial pathology, such as a tumor, will have additional neurologic signs or symptoms.^158–161 Up to one-third of children presenting to the ED with headache undergo neuroimaging,^{156,158,161–164} but clinically significant findings are rare^{6,158,162,165,166} (Box 2). Studies have found that neuroimaging affects management in <6% of children with headache.^{159,161,167,168} The purpose of emergent neuroimaging is to identify those rare but significant and time sensitive causes of headache that lead to morbidity or mortality, such as intracranial hemorrhage and mass lesions. Evidence-based guidelines exist to assist clinicians in determining which patients have a higher likelihood of underlying pathology as a cause of their headache and would benefit from emergent neuroimaging.^32,169,170 The American Academy of Neurology recommends that physicians, physician assistants, and nurse practitioners consider neuroimaging in children with an abnormal neurologic examination and/or the coexistence of seizures and in those in whom the history indicates the recent onset of severe headache, change in type of headache, or associated features that suggest neurologic dysfunction.¹⁶⁹ 

Once neuroimaging is deemed necessary, there are 2 options: CT or MRI. If there is concern for acute subarachnoid hemorrhage, the ACR Appropriateness Criteria for pediatric headache recommends noncontrast CT as the best imaging choice; however, lumbar puncture or CT angiogram (CTA) may be needed to definitively exclude the diagnosis. For children with signs of increased intracranial pressure or focal neurologic findings, MRI (without contrast or with and without contrast) is preferred over CT.³² If there is clinical concern for ischemic stroke, which remains a rare diagnosis in children,¹⁷¹ the use of MRI is more sensitive for detection of acute ischemia.¹⁷² If cerebral sinovenous thrombosis is suspected as the etiology of the headache, a normal head CT does not rule out the diagnosis, and definitive imaging such as CT venography or magnetic resonance venography is preferred.⁴⁴ Similarly, CT imaging may be normal in posterior reversible encephalopathy syndrome and MRI is recommended.⁴⁴ 

Although contrast-enhanced MRI may be used to evaluate for specific diagnoses (eg, infection such as brain abscess), noncontrast MRI is the study of choice for most children with headaches when imaging is being obtained.³² In light of recent studies demonstrating gadolinium deposition in various pediatric organs, including the brain,¹⁷³ radiologists recommend judicious use of gadolinium-based contrast agents until more data are available.⁴⁴ Proposed noncontrast rapid brain MRI sequences require less time (eg, 5–11 min)^44,174 than full brain MRI protocols and may be a reasonable alternative to CT for children with headaches.⁶⁰ Depending on the sequences omitted for the sake of time, certain diagnoses may be missed. The ordering physician, physician assistant, or nurse practitioner may consider consultation with a radiologist in the absence of already established protocols to discuss the optimal imaging modality.

Ventricular shunts carry up to a nearly 90% risk of failure by 10 years.¹⁷⁵ Symptoms associated with shunt malfunction are often nonspecific, overlapping with many commonly seen diagnoses, such as viral gastritis and migraine.¹⁷⁶ Because of the risk of a missed shunt malfunction,¹⁷⁷ the threshold to perform neuroimaging is often low and children with ventricular shunts may, therefore, undergo frequent neuroimaging.^178–180 In addition to neuroimaging, the entire shunt tubing is often imaged with plain radiographs (ie, shunt radiographs or shunt series), adding to the burden of ionizing radiation imposed on these patients.¹⁷⁸ Rapid (∼7 minute) MRI protocols⁶⁰ for the evaluation of shunt malfunctions have been shown to be accurate and feasible for assessment of changes in ventriculomegaly.^181,182 As such, rapid MRIs are being increasingly used in the EDs at children’s hospitals for the evaluation of pediatric shunt malfunctions,¹ with a reduction in CT for this population.⁶¹ Although rMRI provides many benefits over CT, such imaging results in increased time to neuroimaging.⁶¹ Also, if the shunt is a programmable shunt, there is a risk that the shunt setting will be altered from the MRI and require reprogramming. Therefore, CT remains an important tool in the evaluation of patients who are critically ill (eg, altered mental status, Cushing’s triad) and in patients for whom reprogramming by a neurosurgeon is not feasible. Some hospitals have developed ultra-low-dose CT protocols specifically for ventricular shunt evaluations, which overcomes some of the barriers to MRI.¹⁸³ 

The availability of rMRI for the evaluation of ventricular shunt malfunctions may not be the same at nonchildren’s hospitals. There are several aspects worthy of consideration. Most nonchildren’s hospitals do not have pediatric neurosurgical capabilities, and if a shunt malfunction is diagnosed, the patient will require transfer. Similarly, the absence of ventriculomegaly on neuroimaging may not be sufficient to diagnose a shunt malfunction, and a “normal” study may still require transfer for further neurosurgical evaluation and testing (eg, nuclear medicine study, shunt tap). A diagnosis of ventriculomegaly from a shunt malfunction often requires comparison and, therefore, access to the patient’s prior imaging, and the patient may need to be transferred. In these cases, imaging at the initial ED could result in unnecessary radiation exposure, added cost associated with the encounter, and a potential delay in transferring the patient for definitive care. General EDs may consider consulting with local pediatric hospitals to determine optimal imaging strategies for this patient population before ordering imaging so that any on-site imaging will be goal-directed.

Stroke is rare in children^184–189 but associated with significant morbidity and mortality.^190–195 There are multiple risk factors associated with ischemic stroke in children, including cardiac lesions, extracranial and intracranial arteriopathies, thrombophilia, sickle cell disease, and systemic causes, such as systemic lupus erythematosus.¹⁹⁶ In neonates, additional risk factors include infection, trauma, and asphyxia.¹⁹⁶ The diagnosis of stroke in children relies on the appropriate clinical presentation and supportive diagnostic neuroimaging.¹⁹⁶ Identifying children with stroke-like symptoms can be challenging, especially in preverbal children, with failure to recognize stroke-like symptoms leading to delays in diagnosis.^197–199 Importantly, stroke mimics are more common in children than in adults, accounting for up to 93% of children with acute stroke-like symptoms.^{196,197,200,201} Mimics include conditions such as migraine with aura, Bell’s palsy, Todd’s paralysis, syncope, and psychogenic disorders as well as more severe conditions, such as posterior reversible encephalopathy syndrome, demyelinating disease, cerebellitis, encephalitis, intoxication, metabolic disorders, brain tumors, and epilepsy.²⁰² The high rate of stroke mimics has important implications for choice of neuroimaging and treatment.

Rapid diagnostic neuroimaging is key to evaluating for stroke, to determine its potential cause, and to identify potential stroke mimics.²⁰³ Compared with adults, in whom 85% of acute strokes are ischemic, rates of hemorrhagic and ischemic strokes are approximately equal in children.¹⁹⁶ Unlike in adults, a child with stroke-like symptoms and a normal noncontrast head CT may not be at sufficiently high risk of acute ischemic stroke to warrant reperfusion therapy.¹⁹⁶ In very young infants, the onset of symptoms is often not known and because stroke therapies, including thrombolytics and mechanical thrombectomy, are rarely used, neonates can be given appropriate supportive care until definitive imaging can be performed. In fact, presently, based on lack of appropriate data, thrombolysis should not be routinely used in pediatric patients with evidence of an ischemic stroke, and treatment decisions should be made in conjunction with neurologists with expertise in the treatment of pediatric stroke.¹⁹⁶ Decisions regarding appropriate stroke imaging and treatment can be made in consultation with a pediatric neurologist with experience in evaluating and managing pediatric stroke. Various approaches to the neuroimaging evaluation of children with suspected stroke have been proposed (Table 2).^{196,203–206} The ACR Appropriateness Criteria for cerebrovascular disease in children also rates strategies based on several clinical scenarios.³³ 

Limited studies of MRI in children demonstrate high sensitivity (100%, 95% confidence interval [CI], 86%–100%) for acute ischemic stroke.⁵⁶ In addition to its ability to detect stroke, it can aid in the identification of stroke mimics, including potentially life-threatening conditions.²⁰² Although a diagnostically superior imaging modality, MRI availability is often limited when imaging needs to be performed on an emergency basis²⁰⁷ as sedation is often required because of the time required to obtain full MRI sequences.¹⁵⁵ This challenge can be overcome with rapid brain MRI protocols,^{56,60,196,204} with completion of full MRI stroke imaging if a stroke is identified.⁵⁶ An exhaustive evaluation for ischemic stroke, hemorrhagic stroke, and stroke mimics may include numerous sequences.

The addition of magnetic resonance angiography (MRA) of the head and neck without contrast can aid in identification of clot and risk factors for stroke^196,203 but requires longer scanning times and may require sedation.¹⁵⁵ The addition of magnetic resonance venography (MRV) to brain MRI improves diagnostic imaging of the venous drainage of the brain.

Noncontrast head CT is the first-line imaging modality for adults with stroke symptoms. Although noncontrast head CT has a high sensitivity for hemorrhagic stroke, its sensitivity for ischemic stroke is only 27% to 53%.^56,198 Therefore, although CT imaging may be used in the initial evaluation, because of its low sensitivity for both ischemic stroke and stroke mimics, additional neuroimaging (MRI or MRA or CTA) is often needed to confirm the diagnosis.

CTA improves diagnostic accuracy of CT for stroke.^196,208 It has the advantage of more rapid vessel imaging than MRA and may be less likely to require sedation. CT venography (CTV) allows visualization of the brain venous and cerebrosinus drainage system. Limited data exist on the test characteristics of CTV in children with suspected cerebrosinus thrombosis. Notably, noncontrast head CT is reported to have a sensitivity of only 30% for cerebrosinus venous thrombosis in adult patients.²⁰⁹ 

More than 6 million children in the United States experience unintentional trauma annually requiring medical attention in EDs.²¹⁰ With more than 10 000 deaths annually, trauma is the leading cause of death in children 1 to 18 years of age.²¹⁰ Trauma is one of the most frequent reasons for transfers from general EDs without trauma center designation to dedicated trauma centers, including pediatric trauma centers. Outcomes for younger and more severely injured children are optimized when care is delivered at a dedicated trauma center.²¹¹ The American College of Surgeons Advanced Trauma Life Support guidelines recommend immediate transfer to definitive care when a patient’s needs exceed the capabilities of the institution.²¹² Delays in care can result in poorer outcomes, which include performing unnecessary diagnostic tests, particularly CT.⁵² 

When imaging is performed before transfer, studies may need to be duplicated at the receiving ED, with one study finding that 27% of imaging studies were duplicated at the trauma center.¹¹⁰ The reasons for this include inadequate studies at the transferring facility (eg, noncontrast abdominal or pelvic CT scan), failure to send the studies with the patient, and when imaging is sent with the patient, delays in processing or successfully accessing the images for review.¹¹¹ Many studies indicate referring facilities use higher radiation doses for CT imaging than children’s hospitals.^{113,114,213–216} There is wide variability in CT imaging practices for pediatric injuries, with pediatric trauma centers providing the most judicious rates.^{67,214,217–219} 

Advanced imaging is indicated for injured patients if it will allow the patient to be discharged from the ED or remain at the originating facility. The newest American College of Surgeons Field Triage Guidelines provides guidance on which patients are best triaged to a trauma center.^220–222 For injured children who have indications for transfer to a pediatric trauma center, advanced imaging at the originating facility is best performed in consultation with the receiving pediatric trauma center or deferred until after transfer.

Children have different patterns of injury compared with adults, and some of the more common examples follow. Cervical spine injuries are far less common in children than adults, occurring in approximately 1.8% of blunt trauma patients,²²³ and differ in mechanism, anatomic location, and radiographic findings until patients reach skeletal maturity around 14 years of age. Sports injuries and child abuse are important considerations in childhood spinal trauma. Most neck fractures in infants and toddlers occur in the upper cervical spine because of the large size of a young child’s head. Children younger than 8 years are susceptible to spinal cord injury without radiographic or CT abnormality, accounting for approximately 5% of cervical spine injuries.²²⁴ Aortic injuries are rare (as low as 0.018%), with a sevenfold lower rate of thoracic aortic injury in children compared with adults.^225–227 Vascular injuries are less common in children younger than 16 years, with the upper extremity most commonly affected.²²⁸ In children, the compliance of the chest wall leads to pneumothoraces and pulmonary contusions without rib fractures.^229–231 In blunt abdominal trauma, the spleen is the most commonly injured organ, and nonoperative management of the vast majority of abdominal solid organ injuries of the abdomen is now standard care.²³² Bruising on the abdomen from a seat belt or bicycle handle bar is strongly associated with intra-abdominal injury in children.^233,234 Hollow viscus injuries and pancreatic or duodenal hematomas may be associated with abuse in the setting of an unusual mechanism, requiring clinician awareness, and appropriate workup and investigation by social services, child protective service, and/or the institutional child abuse or forensics team in addition to medical treatment.

Physicians, physician assistants, and nurse practitioners who order advanced imaging are often faced with the issue of identifying any injuries versus those that are clinically relevant or actionable. For example, chest CT scans in trauma patients may uncover a tiny “occult pneumothorax” for which surgical management is not beneficial at any age.^230,231,235 Identification of such injuries is of limited benefit in cases of unintentional injury, as the focus becomes the injury itself rather than patient-oriented outcomes. Evaluation of child abuse poses special circumstances, and consultation with a child abuse specialist is helpful in delineating the appropriate imaging studies, as specific imaging may be required to document all injuries, even those that may not be clinically significant. The Choosing Wisely Campaign encourages consideration of outcomes related to diagnosis.²⁹ Patients with diminished Glasgow Coma Scale, neck pain, and/or abdominal tenderness are unlikely to be discharged from the ED, and imaging is best performed at a pediatric trauma center. For screening of asymptomatic patients with high-energy mechanisms, several large pediatric studies have focused on use of clinically important outcomes in their development of clinical decision tools used to determine the need for advanced imaging.^52,236 

Pediatric traumatic brain injury (TBI) is responsible for 500 000 US ED visits annually.²³⁷ The vast majority will not have clinically important TBI; however, those who do will benefit from timely and appropriate diagnosis and management. The most challenging patients are those with minor head injury as defined by a Glasgow Coma Scale score of 14 or 15, as just under 1% will have clinically important intracranial injuries.²³⁶ Several clinical decision rules have been developed to risk stratify patients and determine who is at very low risk of clinically important TBI such that CT imaging can be avoided.^236,238,239 These clinical decision rules cannot be applied to those with suspected abuse, as these patients were excluded from these studies. In a prospective study comparing the PECARN clinical decision rule, Canadian Assessment of Tomography for Childhood Head Injury Rule, Children’s Head Injury Algorithm for the Prediction of Important Clinical Events Rule, and physician judgment, the PECARN rule had the highest sensitivity and slightly better specificity.²⁴⁰ When imaging is needed, CT is a fast and effective imaging tool and the current gold standard.^241,242 Serial imaging, even for moderate to severe traumatic brain injury, by any modality is rarely management altering.²⁴³ The ACR Appropriateness Criteria for pediatric head trauma describes imaging strategies for several clinical scenarios.³⁴ 

The incidence of cervical spine injury in pediatric patients is very low.^219,223,244 Consequently, the American Academy of Pediatrics recommends against routine advanced imaging of the cervical spine in children after injury.²⁴⁵ Clinical decision rules for cervical spine clearance can assist with determining which children should receive imaging.^223,246 Plain radiographs are sensitive for detecting cervical spine fracture in children²⁴⁷ and often obviate the need for advanced imaging.²⁴⁸ Definitive advanced imaging such as cervical spine CT or MRI is seldom indicated acutely as a screening study even if it is needed later. For example, according to a 2019 consensus statement on pediatric cervical spine clearance,²⁴⁹ if a child has normal cervical spine radiographs, is not otherwise injured, and the cervical spine cannot be cleared because of neck pain or tenderness, the child can be discharged with the collar in place and a plan for follow-up with a spine specialist (eg, neurosurgery, orthopedics) with flexion or extension films or advanced imaging as indicated at the time of follow-up. The ACR Appropriateness Criteria for suspected pediatric spine trauma describes imaging strategies for several clinical scenarios.³⁵ 

The diagnosis of intra-abdominal injury in children can be challenging, and injury mechanism, physical findings, laboratory evaluation, and imaging all have a role in determining management. Although predictions rules have been developed, not all have been validated.^236,250 The American College of Surgeons Trauma Quality Improvement Program²²² provides guidance for advanced imaging in both children and adults.

Although commonly used in adult trauma patients, most studies have found that the focused assessment with sonography in trauma (FAST) has inadequate sensitivity for ruling out intra-abdominal injury in children.^251,252 However, there may be utility to the FAST examination when used in combination with hepatic enzymes,²⁵³ in hypotensive patients²⁵⁴ and when performed serially²⁵⁵ in admitted patients. Although CT remains the most accurate for abdominal imaging,^245,246,256 CT is less reliable in blunt bowel and mesenteric injuries, and these injuries are often defined by mechanism and evolving physical findings.²⁵⁷ A period of observation can be diagnostic for children at high risk. The use of contrast-enhanced ultrasonography has emerged in recent years as an alternative to CT imaging and is more accurate than the FAST examination for identifying intra-abdominal injury.^258–261 Indications for contrast-enhanced ultrasonography in the setting of trauma remain under clinical investigation, and these studies require sufficient experience and knowledge of the technique and interpretation and may not be available at all institutions.

Although major extremity vascular trauma has a high mortality rate, such injuries are uncommon (<1% of children who experience trauma).^228,262 Extremity CTA has replaced conventional angiography for vascular trauma. Arterial injuries are most often associated with bony fracture or penetrating injury in proximity to major vessels, although they can occur with blunt trauma, and iliac occlusions have been reported with blunt handlebar trauma to the inguinal area.²⁶³ Children’s vessels can be less protected by soft tissue compared with adults, their collateral circulation may be less well developed, and the smaller-sized vessels and higher rate of traumatic vasospasm make these injuries more challenging to identify.²⁶⁴ Open repair has traditionally been the standard for pediatric vascular injuries, and outcomes are optimized by treatment at a trauma center.²⁶⁴ Although not specific to children, the Eastern Association for the Surgery of Trauma guideline for the evaluation and management of penetrating lower extremity arterial trauma recommends against routine imaging.²⁶⁵ The guideline recommends CTA in patients with an abnormal physical examination or ankle-brachial index <0.9 and immediate operative intervention for patients with hard signs of vascular compromise (eg, pulse deficit, pulsatile bleeding, bruit, thrill, expanding hematoma). If there is concern for vascular compromise, the guideline recommends immediate vascular surgery consultation and surgical intervention, without delay for imaging studies. CT for bony injuries is often less emergent in the resuscitation phase of injury. An exception is a significant knee injury, such as posterior knee dislocation, in which vascular injury occurs in up to 30% of patients and CTA of the knee is appropriate in children older than 5 years.²⁶⁶ 

Whole-body CT has become commonplace in the evaluation of adult patients with blunt trauma but is seldom indicated in children.^{222,267–269} Literature on the routine use of whole-body imaging in children indicates no difference in mortality compared with selective CT scanning.^270,271 Additionally, there is no specific variable that predicts the need for whole-body CT imaging.²⁷² Despite these data, one study found that children are 1.8 times more likely to undergo whole-body CT imaging at adult trauma centers versus pediatric trauma centers.²⁷³ Performance of any advanced imaging, including whole body CT, increases the potential for identification of incidental findings. One study of 57 adult and pediatric trauma patients who underwent whole-body CT found that more than half had incidental findings.²⁷⁴ Whole-body CT also delivers significantly more radiation than selective CT imaging, with dosing estimated to be between 10 and 30 mSv, similar to hundreds or even thousands of chest radiographs.²⁷⁵ The National Institute for Health and Care Excellence recommends clinical judgment over routine whole-body CT for children younger than 16 years.²⁷⁶ Selective region-specific scanning based on clinical prediction models is preferred, unless the patient has an unreliable physical examination because of severe neurotrauma or intoxication and a high-energy mechanism of injury. If necessary, whole-body CT is best performed with single-phase contrast to avoid scanning body regions multiple times.

Imaging plays an important role in the diagnosis, management, and documentation of suspected child abuse, including abusive head injury (AHI) (also referred to as abusive head trauma), and several resources exist to help guide clinicians in choosing the most appropriate imaging strategies.³⁶ Use of clinical decision support rules or algorithms can improve standardization and compliance with image ordering guidelines.^277–279 The ACR Appropriateness Criteria can be used to assist with imaging decisions for children with suspected physical abuse.³⁶ 

Misdiagnosis of child abuse can have significant consequences as failure to identify injuries suspicious for abuse can lead to return of the infant or child to the perpetrator and risk of repeat trauma and potentially death. Alternatively, misinterpreting findings as suspicious or consistent with child abuse can result in parental anxiety and loss of faith in the clinicians as well as legal and financial implications for the family. Many normal developmental changes and anatomic variants can be seen when imaging infants and children. Some of these can be mistaken for injuries associated with abuse.²⁸¹ Referral or transfer of patients to a pediatric ED may minimize these errors. In a study of secondary interpretations of skeletal surveys by a pediatric radiologist, the discrepancy rate was 16%, with 78% of those deemed “significant,” including missed injuries and falsely identified injuries.²⁸¹ 

Skeletal injuries are one of the most common injuries in child abuse.²⁸² An appropriately performed skeletal survey (Box 3) is recommended as the first-line imaging modality when physical abuse is suspected in children younger than 2 years and, in very select cases, children 2 years or older.^36,283,284 Although the number of radiographs in a complete skeletal survey may appear high, it is important that parents and caregivers understand that the overall radiation dose when performed appropriately is small (∼0.2 mSv in one study).²⁸⁵ Authors recommend against a “babygram” consisting of a limited number of large field-of-view images of the patient in the evaluation for suspected child abuse because of the poor sensitivity for detecting fractures of abuse.²⁸⁶ If a child is going to be transferred to a pediatric institution regardless of imaging findings, the skeletal survey may be delayed and performed at the receiving institution.

Ultrasonography may play an adjunctive role in some suspected cases of fracture evaluation. Ultrasonography has the benefit of relatively high spatial resolution and real-time imaging and is particularly useful for assessing unossified portions of the bone, including the unossified growth plates in younger children. Ultrasonography can be helpful in exclusion or confirmation of indeterminate classic metaphyseal lesions suspected on skeletal survey.²⁸⁷ However, such imaging is best performed at institutions with experienced pediatric radiologists, as these studies require sufficient expertise with the technique and interpretation.

The use of chest CT is not typically indicated in the ED setting; however, there are a few exceptions. Fractures that are particularly difficult to identify with skeletal survey, including certain rib fractures, scapula fractures, and some spinal fracture types are more easily identified with CT.^282,288,289 Costochondral fractures, which would raise the possibility for abuse in young children, are notoriously difficult to identify on skeletal survey and are more likely to be seen on CT.²⁹⁰ These chest CT studies can be performed with a relatively low radiation dose and very quickly, thus obviating the need for sedation.²⁸⁹ If there is concern for these fractures, advanced imaging is best performed at a pediatric trauma center with a child abuse specialist available for consultation.

AHI is the leading cause of fatal injuries in young children, and the diagnosis is frequently missed or delayed.^291,292 Current recommendations for imaging advocate for neuroimaging in patients younger than 6 months if any abuse is suspected, given the high risk of AHI in this age group.^293–294 Because of concerns about cost and radiation exposure, there has been considerable work in developing decision tools to aid in selection of appropriate imaging.²⁹⁶ Of the available decision tools, the Pittsburgh Infant Brain Injury Score²⁹⁹ (Box 4) has the most potential relevance in the ED setting. This rule, although not externally validated to date, seeks to guide clinicians in the decision to obtain neuroimaging in well-appearing children between the age of 30 and 364 days who present with potential but nonspecific symptoms of AHI, such as vomiting, fussiness, and poor feeding (among others) and no history of trauma. Current evidence is limited on the use of this rule for universal screening (for example, all children with vomiting), and it also requires a laboratory test that may be challenging to perform as part of a screening protocol when suspicion is low; point-of-care capillary hemoglobin tests may minimize this barrier. It is important that protocols be applied universally to all patients with relevant historical or physical findings²⁹⁸ to increase identification of patients with child abuse, reduce missed cases, and also to decrease bias in who undergoes evaluation.²⁹⁹ One quality improvement project was able to demonstrate increased adherence to age-related guidelines from 47% to 69% with institution of universal protocols.²⁷⁹ 

Neuroimaging for AHI can be performed with either CT or MRI. Ultrasonography is rarely used or recommended in most children suspected of AHI as it lacks sufficient sensitivity.²⁹² For symptomatic children, CT will identify findings that require emergency care and is preferred over MRI for identifying skull fractures from blunt trauma.³⁰⁰ MRI is preferred for assessing intracranial injury.³⁰⁰ The ACR Appropriateness Criteria recommends either CT of the head for patients who are critically ill or MRI of the brain.³⁶ In stable patients, the decision to obtain a CT versus MRI depends on availability and institutional practice. A 16-minute rapid MRI protocol has been shown to be well tolerated by young infants and has high sensitivity such that it can be used as a screening tool for suspected AHI and obviate the need for CT imaging.^57,60 

Historically, spinal injuries were not believed to be significant injuries resulting from child abuse. However, mounting evidence indicates a high rate of concomitant cervical spine injuries in patients with AHI.^301–303 Bony spine injuries in children who have experienced abuse are uncommon, and these patients are more likely to suffer from ligamentous injuries.³⁶ Imaging may also reveal secondary signs suggestive of abuse, such as spine subdural hematomas.^294–297 The ACR Appropriateness Criteria recommends that MRI of the cervical spine be considered at the time of complete head MRI, as injuries may be present in more than one-third of children with AHI.³⁶ New retrospective data suggest whole-spine MRI may identify additional thoracolumbar injuries.³⁰⁴ Such comprehensive imaging should be deferred to a pediatric hospital with child abuse specialists; spinal immobilization is sufficient before transfer.

The presence of abdominal injury in a child without a plausible mechanism raises suspicion for child abuse. Child abuse is the most common cause of blunt abdominal trauma in children 4 years and younger, and death from blunt abdominal injury is most commonly attributable to child abuse in this group.³⁰⁵ As with other abusive injuries, identification is important for documentation and child protective services assessment even if medical intervention is not required. Although almost any abdominal injury can occur from abuse,^306,307 pancreatic injuries and hollow viscus injuries are more common in children who have been physically abused than in children who experience accidental trauma.^305,306,308 Identification of costochondral fractures on either skeletal survey or chest CT may increase the risk of associated abdominal injuries.²⁸⁹ 

Clinical decision rules intended for the evaluation of patients with blunt abdominal trauma are not meant to be applied to patients with suspected child abuse.³⁰⁹ The outcome of interest in decision tools is clinically important injuries, and the relevant outcome in the evaluation for physical abuse also includes forensically important injuries.^235,308,309 Screening laboratory evaluation of aspartate transaminase and alanine transaminase may be a more useful tool in determination of whether imaging should be performed for both clinically relevant and forensically important injuries.^{308,310–312} Evidence supports screening for intra-abdominal injury from abuse in those patients with transaminases >80 IU/L.³¹¹ 

The first-line imaging strategy for the evaluation for abdominal injury is CT of the abdomen and pelvis with IV contrast.³⁶ Such imaging allows for assessment of parenchymal and vascular injuries in the axial, sagittal, and coronal planes and allows for screening of the visualized ribs, spine, and pelvis for fracture.

Grayscale ultrasonography for trauma, often performed as a FAST examination in the ED, has been shown to have poor sensitivity and specificity for organ injury and hemoperitoneum^261,313 and is not recommended for diagnosis or screening of abdominal injury in child abuse.^36,308 The addition of ultrasonography contrast agents to sonographic imaging for blunt abdominal trauma has shown promising results but is not currently recommended as the first-line imaging modality for patients with concern for child abuse.^261,314 

Evaluation of children with suspected appendicitis is one of the most common reasons for advanced imaging in the ED.^1,3 As part of the Choosing Wisely recommendations, many organizations advocate for the use of ultrasonography as the initial imaging study for appendicitis.^316–319 Similarly, the ACR appropriateness criteria recommends the use of nonionizing radiation imaging techniques as the first-line approach for abdominal pain suspected to be appendicitis.³⁷ Referring EDs are far more likely to perform CT compared with pediatric institutions in which ultrasonography is used preferentially.^320,321 Validated scoring systems, such as the Alvarado score,⁹⁹ Pediatric Appendicitis Score,¹⁰⁰ and Pediatric Appendicitis Risk Calculator¹⁰¹ (Table 3), can be helpful in risk-stratifying patients to determine which patients are at low risk of appendicitis and do not require urgent imaging. Kharbanda et al¹⁰¹ found that the Pediatric Appendicitis Risk Calculator (area under the curve: 0.85; 95% CI: 0.83–0.87) outperformed the Pediatric Appendicitis Score (area under the curve: 0.77; 95% CI: 0.75–0.80). Such scoring systems may also assist clinicians practicing in EDs without access to ultrasonography to help determine who needs to be transferred for pediatric specialty evaluation. If a patient is going to be transferred to another facility with pediatric surgery capabilities regardless of imaging findings (eg, because of ongoing pain, dehydration, or other need for admission), imaging at the transferring hospital can be deferred as it may result in delays to definitive care as well as more imaging studies.^110,111,322 

Ultrasonography alone for the evaluation of appendicitis, including point-of-care ultrasonography, has varying sensitivities and is subject to operator dependence.^323,324 Standardized reporting tools for ultrasonography examinations may be useful for ultrasonography technologists, radiologists, and the clinicians who order the study (Table 4).³²⁵ Ultrasonograms can be categorized as: no sonographic evidence of appendicitis, incompletely visualized appendix without secondary signs, incompletely visualized appendix with secondary signs, and acute appendicitis.³²⁵ In cases in which there is no sonographic evidence of appendicitis or the appendix is incompletely visualized, the need for subsequent imaging will be determined in the context of the patient’s clinical presentation, risk of appendicitis, and ability to observe the patient either in the ED, in the hospital, or as an outpatient by the primary care provider. Persistent concern for appendicitis after applying a validated scoring system and normal or equivocal ultrasonographic imaging may indicate the need for additional advanced imaging modalities or transfer where the child can be admitted for observation with serial abdominal and repeat ultrasonographic examinations as dictated by clinical course.³²⁶ Although CT imaging of the abdomen and pelvis demonstrates the highest sensitivity and specificity, many authors recommend restricting its use to children with inconclusive ultrasonographic findings who would not otherwise qualify for operative intervention based on clinical suspicion.^327–329 There are varying opinions regarding noncontrast imaging versus that with IV and/or oral contrast, and decisions regarding contrast are best made in consultation with the interpreting radiologist. In the scenarios described by the ACR Appropriateness Criteria, CT with IV contrast consistently rates greater than or equivalent to CT without IV contrast, and there is disagreement on the appropriateness of studies without IV contrast. Lack of IV contrast potentially limits sensitivity for and characterization of complicated appendicitis and decreases sensitivity for alternative diagnoses. Oral contrast runs the risk of obscuring an appendicolith, which can affect management (eg, presence of an appendicolith is a relative contraindication to considering antimicrobial treatment of early appendicitis).³³⁰ Additionally, oral contrast increases risk of emesis and lengthens time to study completion, and its impact on diagnostic accuracy is debated. Limited MRI protocols, which are typically less than 20 minutes in duration, are gaining popularity in children’s hospitals and replacing CT when ultrasonographic examinations are equivocal.^331–333 The accompanying Supplemental File contains examples of imaging strategies for the evaluation of pediatric appendicitis at several children’s hospitals.

Although not as commonly diagnosed in children compared with adults, the incidence of nephrolithiasis has increased over recent decades.^334,335 The adolescent population appears to be driving this increase, and female adolescents have the highest rates^334,336 among all pediatric patients. Anatomic or structural abnormalities (eg, ureteropelvic junction obstruction, posterior urethral valves), medical conditions (eg, inflammatory bowel disease, cystic fibrosis), and medications (eg, topiramate, furosemide) predispose patients to stone formation.³³⁷ Up to 50% of stones may be familial.^338,339 

The American Urological Association and the European Society for Pediatric Radiology recommend ultrasonography as the initial imaging study for the evaluation of suspected nephrolithiasis.^340,341 The ACR Appropriateness Criteria on the evaluation of hematuria also includes recommendations for patients in whom nephrolithiasis is a consideration.³⁸ Although the sensitivity (approximately 70%)^342,343 is lower than CT, one study found that ultrasonography was sufficient for diagnosing 89% of clinically significant stones requiring surgical intervention.³⁴⁴ The detection rate is much higher for stones in the kidney compared with the ureter.³⁴⁵ Use of both ultrasonography and plain radiography of the abdomen and pelvis may improve detection of ureteral calculi.³⁴⁴ Low-dose noncontrast CT studies (ie, “stone protocol”) significantly reduce the total radiation exposure compared with standard abdominal CT imaging protocols. Although not well studied in children specifically, one study of adult patients found that low-dose CT maintained a high sensitivity (96.6%) for diagnosing nephrolithiasis.³⁴⁶ Low-dose CT in children is typically reserved for indeterminate cases or if further clarification is needed, for example, surgical planning.^341,347,348 

In contrast to adults, pulmonary embolism (PE) in children is relatively rare with an annual incidence of 0.14 to 0.9 per 100 000 children.³⁴⁹ When a PE is present in a child, there is usually an identifiable risk factor.³⁵⁰ Risk factors primarily consist of medically complex devices or comorbidities such as indwelling catheters, malignancy, and any medical conditions that cause immobility, among others.³⁵¹ Pediatric PE has a bimodal distribution, with neonates who require intensive care and adolescents considered high-risk groups.^349,350 Although data are limited, it appears that PEs in children are associated with deep vein thrombosis (DVT), although the DVT itself may not be known at the time of PE diagnosis.³⁵¹ One study found that 72% of patients with a PE had a concurrent DVT³⁵¹ and another found that 11% of patients with DVT also had a PE.³⁵² Precise mortality rates are challenging to determine because of the frequent coexistence of medical complexity and difficulties in diagnosis and range from 8.9% to 22%.^349,351 

The D-dimer test is frequently used in adults as a screening test to determine the need for advanced imaging. Unfortunately, current evidence suggests inadequate test characteristics for the study to be used as either a rule-in or rule-out test in children. For example, a 10-year retrospective study of the use of D-dimer in the evaluation of PE in 526 children found a lower limit of sensitivity of 89% and specificity of 54%.³⁵³ To date, there are no published prospective studies using D-dimer as part of an evaluation process for DVT or PE in pediatrics. As a result, the current evidence does not support the routine use of D-dimer as a screening test to rule-out PE.

Because many pediatric PEs are associated with DVT, it may be reasonable to pursue lower extremity ultrasonography as part of the initial imaging process, although this should not be relied on to exclude the diagnosis.^351,354 Sensitivity is likely higher for lower limb DVT compared with upper extremity DVT, which poses a challenge in children, as many DVTs are located in the upper extremity because of the presence of indwelling catheters.³⁵⁴ Other options that can be considered include CTV and MRV, depending on the clinical concern and availability.³⁵⁵ 

CT pulmonary angiogram is generally considered the diagnostic test of choice when there is high clinical concern for PE, although data are limited in children. Data from studies of adults demonstrate a sensitivity of 83% and specificity of 95%.³⁴⁹ CT offers the benefit of providing additional information and alternative diagnoses if the study is negative for PE. Although precise numbers are lacking, ventilation or perfusion scans have lower sensitivity and are often non diagnostic and run the risk of false-positivity in patients with alternative diagnoses that can cause ventilation or perfusion mismatch, such as sickle cell disease and pneumonia, among others.^349,356 

Eighty to 90% of neck masses in pediatric patients are inflammatory, congenital, or benign neoplastic lesions but need to be distinguished from malignant neoplastic pathologies.^357,358 It is important to recognize and distinguish infections of the deep neck spaces including the retropharyngeal, parapharyngeal, and peritonsillar spaces as treatment varies widely based on location.

Several algorithms for the evaluation of palpable inflammatory neck masses and deep neck space infections have been published, but there is a lack of consensus on a single best imaging strategy.^359–363 ACR Appropriateness Criteria for neck masses in children reflect the variability in imaging approach with ultrasonography, CT with IV contrast, MRI with and without IV contrast, and MRI without IV contrast all categorized as “usually appropriate.”³⁹ Imaging evaluation specific to the thyroid and parotid regions are not considered in the ACR Appropriateness Criteria for pediatric neck masses.

Lateral radiographs have high sensitivity and specificity for detection of soft-tissue thickening anterior to the spine but have limited ability to distinguish retropharyngeal abscess from other processes, such as edema, or to visualize infection in other deep neck spaces.^364,365 Given the limitation of radiographs, further imaging may be necessary when there is continued suspicion for deep neck space infection.³⁶⁴ 

Evidence and opinions regarding the ability of ultrasonography to be used as an initial or sole modality for the evaluation of pediatric neck infections vary. Proponents cite studies that demonstrate the high incidence of superficial reactive lymphadenopathy in this population and high diagnostic accuracy of ultrasonography in detecting fluid collections that require intervention.^{357,358,366–368} Use of ultrasonography as the primary diagnostic modality in pediatric patients with superficial soft-tissue infections of the neck has been shown to reduce CT utilization without significant difference in the rate of treatment failure.³⁶⁹ These studies are contradicted by others who have raised concern for limited ability of ultrasonography to visualize deep neck space collections with demonstrated sensitivity, specificity, and accuracy as low as 65%, 33%, and 65%, respectively.^{360,370–372} A recent study comparing primary use of ultrasonography versus CT in pediatric patients with lateral neck infections found no significant difference in outcomes between the groups but favored CT because of a significantly increased rate of repeat imaging and missed cases of retropharyngeal abscess in the ultrasound group.³⁷³ Proponents of CT cite its greater sensitivity and positive predictive value for diagnosis of deep neck space infections and ability to better delineate the anatomic extent of infection and congenital lesions.^{363,365,374–377} Further, CT is less user dependent compared with ultrasonography, and its wide field of view makes it useful for identifying complications including vascular thrombus, mediastinal extension, and airway compression.³⁷⁸ Size of deep neck space-associated abscesses as measured on CT has been shown to be associated with the likelihood of failure of medical therapy and need for incision and drainage.^379–381 Consequently, CT with IV contrast plays an important role in guiding the decision to pursue operative intervention.³⁶⁰ CT is limited in its ability to distinguish cellulitic or phlegmonous soft-tissue infection from abscess. This modest specificity of CT is well documented by multiple studies that report false-positive diagnoses of abscess, resulting in the patient being taken to the operating room even after clear CT identification of a rim enhancing lesion with central hypodensity and scalloping wall, findings classically associated with abscess.^{359,361,363,377,381–383} A recent retrospective study of pediatric patients with suspected deep neck space infections supports a limited role for CT.³⁵⁹ Most patients in the study were successfully managed medically, and the rate of successful medical management did not significantly differ between those with abscess versus without abscess identified by CT. This study supports others that have suggested deferral of CT for select patients pending empirical antibiotic treatment.^362,377,383 MRI has similar sensitivity and specificity compared with CT for diagnosis of deep neck space infections.^{359,360,372,384} MRI is limited by its availability and possible need for sedation³⁷²; however, it may be considered in certain patients (eg, older children with a stable airway).

Advanced imaging plays an important role in the diagnosis and management of musculoskeletal infections in pediatric patients. Guidance on the imaging evaluation is available through clinical practice guidelines developed by organizations such as the Infectious Diseases Society of America and the European Society for Pediatric Infectious Diseases, the ACR Appropriateness Criteria, as well as other published proposed algorithms.^40,385–388 Decisions on advanced imaging for diagnosis of musculoskeletal infection are driven by patient age, clinical suspicion for infection, physical examination, laboratory findings, and ability to localize symptoms.

Radiographs have low sensitivity for detection of acute osteomyelitis. Bone changes are typically not apparent for the first 7 to 14 days.³⁸⁹ However, radiographs are considered an important initial examination in the ED to evaluate for other potential pathologies, such as malignancy, foreign body, and fracture.^{40,87,387,390,391} One exception to radiography being the initial examination is in the specific setting of acute limp in children up to the age of 5 years with concern for osteomyelitis and nonlocalized symptoms for whom ACR Appropriateness Criteria categorize radiographs as “usually not appropriate” and MRI is advocated.⁴⁰ MRI has the ability to depict anatomic detail and high sensitivity for detecting soft-tissue and bone marrow pathology.^40,390–393 In addition to identifying sites of osteomyelitis, MRI is able to evaluate complications of infection that may impact management, including subperiosteal and soft-tissue abscesses, physeal involvement, deep venous thrombosis, and concomitant septic arthritis.^{390,394–396} Field of view could be focused on the area of interest or the entire extremity and is determined by patient age, ability to localize symptoms, and risk of multifocal disease. Whole-body MRI may be useful in neonates, infants, and children younger than 5 years for whom symptoms cannot be localized and risk of multifocal disease is greater.^391,397 Smaller fields of view of the affected areas with higher spatial resolution can be acquired once foci of infection are identified. Administration of contrast in older age groups increases reader confidence in diagnosis but does not significantly increase sensitivity or specificity for detection of osteomyelitis.^398,399 Contrast-enhanced images increase sensitivity for detection of osteomyelitis-related and soft-tissue abscesses, although some studies have reported that most of the additional abscesses detected are small and may not require surgical intervention.³⁹⁸ Consequently, acquisition of contrast-enhanced images may be of more value in the setting of abnormal noncontrast images.³⁹⁸ Use of contrast may be useful in young children and infants in whom bone marrow signal is variable and markedly improves detection of cartilaginous infections in children younger than 6 years.^400,401 Although CT has greater sensitivity for periosteal reaction, erosive changes, and necrosis compared with radiography, it is usually not appropriate for the diagnosis of acute osteomyelitis because of limited ability to characterize marrow abnormalities, poor soft-tissue contrast relative to MRI, and use of ionizing radiation.^392,402,403 CT with intravenous contrast can be valuable for depiction of abscess and guided procedures, such as aspiration or drainage.³⁹⁰ Although cross-sectional imaging is typically needed for the definitive diagnosis of osteomyelitis, it is not necessary that such testing be performed in the ED setting. Rather, such imaging is best performed at the institution where definitive care will be delivered unless negative imaging would allow for discharge from the ED.

Ultrasonography can be used for identification of joint effusion, synovitis, soft-tissue and periosteal abscesses, pyomyositis, and cellulitis^390,404 and is useful for image-guided procedures. Ultrasonography is unable to depict marrow abnormality and is used as an adjunctive modality to MRI for suspected acute osteomyelitis.

Septic arthritis most commonly affects the lower extremity in young children. The Kocher criteria are commonly used to assess the likelihood of septic arthritis and include fever, leukocytosis, inability to bear weight, and elevated inflammatory markers. If ≥3 criteria are positive, the probability of septic arthritis can be >90%.⁴⁰⁵ Such criteria can be used to risk stratify the limping child and determine whether imaging is necessary. Imaging cannot differentiate infected joint fluid of septic arthritis from sterile joint fluid seen with transient synovitis, for example. Definitive diagnosis requires synovial fluid analysis.⁴⁰⁶ 

Radiographs of the affected site may occasionally show evidence of large joint effusion, including widening of the joint space and displacement of fat pads, but are more valuable to exclude other pathologies. Ultrasonography is usually appropriate as a first-line imaging modality^40,407 and can identify joint effusions, synovial thickening, and hyperemia.^404,408 False-negative studies may occur within 24 hours of symptom onset when sufficient fluid for detection has not yet accumulated.⁴⁰⁹ Isolated septic arthritis is difficult to distinguish from disease with concomitant osteomyelitis, subperiosteal abscess, and soft-tissue abscess. MRI is also usually appropriate for first-line or next-step imaging as it is sensitive and specific for diagnosis of adjacent bone or soft-tissue infection, which may lead to treatment failure if unrecognized.⁴⁰⁷ Algorithms to predict which patients will benefit from further imaging with MRI to identify adjacent musculoskeletal infection have demonstrated variable sensitivity and specificity in validation studies.^410–412 The presentation of a psoas abscess may be similar to that for septic arthritis.⁴¹³ Several imaging modalities have been described for the evaluation of a psoas abscess, including ultrasonography, MRI, and CT.⁴¹³ If there is clinical concern for the diagnosis, consultation with a radiologist may be beneficial.

Spondylodiscitis refers to infection of the pediatric vertebral disc and adjacent vertebral body endplates. This most commonly occurs secondary to hematogenous spread of infection, which is suspected to start in the disc and extend into the endplates through contiguous vascular channels.⁴¹⁴ Indicators of the need for imaging evaluation include constant pain, night pain, radicular pain, pain lasting >4 weeks, and an abnormal neurologic examination.⁴¹⁵ 

Initial investigation with radiography of the spine is usually appropriate for most cases,^414,415 with anteroposterior and lateral views being sufficient. Additional views increase radiation dose without proportional change in diagnostic yield.⁴¹⁵ Early in the disease process, radiography is insensitive for detection of inflammatory or infectious conditions. Mild disc-space narrowing can be seen, but bony changes of spondylodiscitis typically do not manifest until 2 to 3 weeks after onset.^390,416 Despite this limitation, initial imaging with radiography is frequently able to identify abnormalities of various etiologies in patients with confirmed pathology.^416–419 

Findings that lead to a diagnosis and etiology of the patient’s pain can be demonstrated in up to 24% of radiographs.⁴¹⁵ However, radiography is not sufficiently sensitive to exclude a pathologic diagnosis if results are normal.

MRI of the spine with IV contrast is used to follow-up both negative and positive radiographs in pediatric patients with back pain and at least 1 clinical indicator and concern for infection or inflammation.⁴¹⁶ MRI has high diagnostic accuracy for the identification of pediatric back pain pathology and, more specifically, spondylodiscitis.^{417,419–421} The superior soft-tissue contrast of MRI also allows for direct visualization of the spinal cord, nerve roots, and adjacent soft tissues, which is important as adjacent abscesses are reported in 17.8% of spondylodiscitis cases, usually in the paravertebral muscles or epidural space.⁴¹⁴ 

CT of the spine without IV contrast provides excellent visualization of bone detail, including destructive changes, sclerosis, and disc-space narrowing.⁴¹⁵ Compared with MRI, CT has limited ability to evaluate the spinal cord, marrow, and soft-tissue contrast. Consequently, CT is rarely used for the diagnosis of suspected spondylodiscitis and is not included in most imaging algorithms for the condition.⁴¹⁴ Limited CT of the spine has a role in guided procedures including aspiration or drainage of collections or biopsy.³⁹⁰ 

In general, ultrasonography, CT with contrast, and MRI with contrast are all useful in the detection and evaluation of soft tissue infections.³⁹² Pyomyositis is most commonly diagnosed in the pelvis and lower extremities.⁴²³ Radiographs are insensitive for the diagnosis with only subtle soft-tissue swelling occasionally seen.^423,424 Ultrasonography may demonstrate subtle increased volume and echogenicity of the muscle with hypoechoic septa and hyperemia during the invasive stage, which can progress to include a discrete abscess in the subsequent suppurative phase.^404,406,425 Early findings can be subtle and the diagnosis can be missed, particularly if deeper in the pelvis.^424,425 MRI is the preferred modality for evaluation of suspected pyomyositis given its high sensitivity for the diagnosis and ability to identify associated osteomyelitis or septic arthritis, which frequently occur concomitantly.^421,425 The Infectious Diseases Society of America and European Society for Pediatric Infectious Diseases specifically recommend MRI with IV contrast for evaluation of suspected pyomyositis.^390,424 CT and ultrasonography are considered helpful if MRI is not available.

Necrotizing fasciitis has a high rate of mortality,⁴²⁶ making early diagnosis critical. Unfortunately, diagnosis is challenging, with no specific criteria for diagnosis, and a high degree of clinical suspicion is paramount. Findings suggestive of the diagnosis include severe pain disproportionate to clinical findings, failure to respond to initial antibiotics, hard wooden feel of subcutaneous tissue, systemic toxicity, and crepitus.^424,426,427 Sensitivity and specificity of imaging is ill-defined and findings are not reliably present. The Infectious Diseases Society of North America notes that CT and MRI can demonstrate findings of necrotizing fasciitis but may delay definitive diagnosis and treatment.⁴²⁴ The ACR Appropriateness Criteria rate initial evaluation with radiography as “usually appropriate,” and radiographs may demonstrate soft tissue gas.³⁹⁴ Similarly, CT with or without contrast can detect soft tissue gas with high sensitivity and may be appropriate given the speed of the test.³⁹⁴ CT can also detect fascial thickening, fluid collections along the deep fascial planes, and intermuscular septal edema.³⁹⁴ Similar to CT, MRI with or without IV contrast can detect the aforementioned findings; however, MRI is less sensitive than CT in the detection of soft tissue gas.³⁹⁴ 

Advanced imaging is an important and frequently performed aspect of the ED evaluation. There are benefits and risks to advanced imaging. There are different imaging strategies for common pediatric conditions, and optimizing imaging for children is an important consideration as part of overall emergency care. Such optimization includes use of evidence-based guidelines, including those embedded as part of clinical decision support, to minimize low-value studies in patients who can be safely observed; reducing radiation exposure by using pediatric-specific dosing protocols for CT; providing alternatives to CT, specifically ultrasonography and/or “rapid” MRI, when appropriate; collaborating with radiologists, particularly those with expertise in pediatric imaging; and, shared decision-making when appropriate. Optimizing ED imaging for children is also an important aspect of pediatric readiness, as many EDs strive to improve their ability to care for acute medical conditions in children. Pediatric experts at referral institutions are important both for collaboration on imaging strategies and also for transfer, if indicated. In patients for whom imaging will not alter the decision to transfer or impact management before and/or during transport, deferring imaging to the receiving hospital can reduce delays to definitive care, radiation exposure, and the total number of imaging studies.

Jennifer R. Marin, MD, MSc, FAAP, FACEP, American Academy of Pediatrics

Todd W. Lyons, MD, MPH, FAAP, American Academy of Pediatrics

Ilene Claudius, MD, FACEP, American College of Emergency Physicians

Mary E. Fallat, MD, FAAP, FACS, American Academy of Pediatrics

Michael Aquino, MD, American College of Radiology

Timothy Ruttan, MD, FACEP, FAAP, American College of Emergency Physicians

Reza J. Daugherty, MD, FAAP, American Academy of Pediatrics

Gregory P. Conners, MD, MPH, MBA, FAAP, Chairperson

Sylvia Owusu-Ansah, MD, MPH, FAAP

Kerry S. Caperell, MD, FAAP

Jennifer Hoffmann, MD, FAAP

Benson Hsu, MD, MBA, FAAP

Deborah Hsu, MD, MEd, FAAP

Jennifer R. Marin, MD, MSc, FAAP

Jennifer E. McCain, MD, FAAP

Mohsen Saidinejad, MD, MS, MBA, FAAP

Muhammad Waseem, MBBS, FAAP

Sue Tellez, staff

Hansel J. Otero, MD, FAAP, Chairperson

Patricia Trinidad Acharya, MD, FAAP

Adina Lynn Alazraki, MD, FAAP

Ellen Benya, MD, FAAP

Brandon Patrick Brown, MD, MA, FAAP

Reza James Daugherty, MD, FAAP

Laura Laskosz, MPH, staff

Christopher S. Amato, MD, FACEP, Chairperson

Alexandria Georgadarellis, MD

Ann Marie Dietrich, MD, FACEP

Annalise Sorrentino, MD, FACEP

Ashley Foster, MD, FACEP, board liaison

Carmen D. Sulton, MD, FACEP

Cindy Chang, MD

Daniel Slubowski, MD

Dina Wallin, MD, FACEP

Donna Mendez, MD

Emily A. Rose, MD, FACEP

Erika Bishop Crawford, MD

Genevieve Santillanes, MD, FACEP

George Hsu, MD

Gwendolyn C. Hooley, MD

Ilene A. Claudius, MD, FACEP

Isabel Araujo Barata, MD, FACEP

James L. Homme, MD, FACEP

Jeffrey Michael Goodloe, MD, FACEP, board liaison

Jessica J. Wall, MD, MPH, MSCE, FACEP

Jonathan Harris Valente, MD, FACEP

Joshua Easter, MD

Joyce Li, MD

Kathleen Brown, MD, FACEP

Kathleen Theresa Berg, MD, FACEP

Kiyetta Hanan Alade, MD, MEd, RDMS

Lauren Rice, MD

Madeline Matar Joseph, MD, FACEP

Marc Auerbach, MD

Marianne Gausche-Hill, MD, FACEP

Melanie Heniff, MD, JD, MHA, FACEP

Michael J. Stoner, MD, FACEP

Michael Joseph Gerardi, MD, FACEP

Mohsen Saidinejad, MD, MBA, FACEP

Moon O. Lee, MD, FACEP

Muhammad Waseem, MD, MS, FACEP

Paul T. Ishimine, MD, FACEP

Samuel Hiu-Fung Lam, MD, MPH, FACEP

Sean M. Fox, MD, FAAP, FACEP

Shyam Mohan Sivasankar, MD, FACEP

Simone L. Lawson, MD, Med, FAAP, FACEP

Siraj Amanullah, MD, MPH

Sophia D. Lin, MD

Stephen M. Sandelich, MD

Tabitha Autumn Cheng, MD

Theresa Ann Walls, MD, MPH

Timothy Ruttan, MD, FACEP

Zachary Burroughs, MD

Sam Shahid, MBBS, MPH, staff

ACR

American College of Radiology

AHI

abusive head injury

CDSM

clinical decision support mechanism

CT

computed tomography

CTA

computed tomography angiogram

CTV

computed tomography venography

DVT

deep vein thrombosis

ED

emergency department

FAST

focused assessment with sonography in trauma

IV

intravenous

MRA

magnetic resonance angiography

MRV

magnetic resonance venography

rMRI

rapid magnetic resonance imaging

PE

pulmonary embolism

TBI

traumatic brain injury