Nonaccidental trauma (NAT) is one of the leading causes of serious injury and death among young children in the United States.^1,2 The incidence of traumatic brain injury and fractures attributable to abuse is significant, with 21.9 cases per 100,000 children younger than 36 months, and 50 cases per 100,000 children during their 1st year of life.³ Abusive head trauma accounts for an estimated 80% of childhood maltreatment deaths, making it the leading cause of death in this population.^2,4 Children in their 1st year of life are at the greatest risk of experiencing abusive head trauma and are especially susceptible to subsequent negative health outcomes such as vision problems, developmental delays, physical disabilities, and hearing loss.⁵ Despite the clinical implications of a failed diagnosis, studies have shown that a large number of children presenting with nonaccidental skull fractures are not identified as being victims of abuse in initial presentation.^6,7 Although there are certainly patterns of injury attributable to inflicted head trauma including diffuse axonal injury and subdural hematoma, there is no clear method of identification of abusive head trauma.⁸ Isolated skull fractures (ISFs), which we are defining as skull fractures without intracranial findings on cranial imaging, are difficult to attribute to NAT given the frequency with which they occur in nonabusive injuries.⁹ Therefore, it is of utmost importance to improve identification of NAT in high-risk children through optimization of resources in order to ensure child protection and positive health outcomes for this population going forward.

Increasing evidence suggests that children with ISFs rarely require intervention warranting admission, and they can safely be discharged home with a low risk of clinical decline.^10–13 Several institutions have developed consensus guidelines for safe disposition of asymptomatic children with ISFs from the emergency department (ED).^14,15 Where some studies have noted that younger age more often led to hospitalization, there is a general consensus that children for whom there is a concern for NAT may require admission to ensure safe disposition.^13–16

Even the most experienced pediatric trauma centers face challenges in balancing the need to identify an abused child with timely assessment and discharge of nonserious skull fractures. For example, prior to this protocol, a determination for additional workup to evaluate for NAT was based on provider concern, imaging findings, and a description of the injury. Additionally, the workup itself varied depending on the initiating team. To better serve this patient population, our institution’s trauma committee (comprising physician members from pediatric surgery, neurosurgery, pediatric intensive care, orthopedic surgery, emergency medicine, a child protection team [CPT], and anesthesiology) developed a standard protocol for evaluation of infants younger than 12 months presenting with skull fractures. The workup included evaluations by trauma surgery, neurosurgery, social work, the CPT (pediatricians with training in child abuse), ophthalmology, and a skeletal survey, and was carried out as a quality improvement initiative. In this study, we aimed to evaluate our adherence and subsequent barriers to this protocol and to identify the highest-yield components in order to tailor our evaluation with consideration to clinical yield and cost.

Methods

A retrospective chart review was conducted at our regional level I pediatric trauma center with IRB approval through University Hospitals Cleveland Medical Center and Case Western Reserve University. The Rainbow Babies & Children’s Hospital pediatric trauma database was queried for all children diagnosed with a skull fracture who presented between August 1, 2014, and July 31, 2016. These charts were reviewed for potential inclusion. All infants younger than 12 months were included in the study (Fig. 1). A total of 45 patients were included in our analysis. Subgroup evaluation was performed in patients with ISFs.

Flow diagram for selection of patients for inclusion. RCB = Rainbow Babies & Children’s Hospital. *Benign extraaxial collections were not considered a pathologic intracranial finding.

Assessed data points included age, sex, comorbidities, diagnostic imaging, location of the skull fracture, originating hospital, fracture morphology, concurrent intracranial injury, and mechanism of injury (Table 1). Our hospital standardly obtains a neurosurgery consult for all patients with radiographic evidence of injury inclusive of skull fractures, and, as such, compliance with the neurosurgical consultation was universal. To assess adherence to our protocol, we noted whether social work evaluations, pediatric surgery consultation, CPT consultation, skeletal surveys, and ophthalmology evaluations were performed and their results assessed.

All patients with intracranial hematomas were admitted to the pediatric intensive care unit (PICU) for neurological monitoring. Patients with ISFs who were admitted were candidates for ward admission unless other injuries or concerns warranted closer neurological or hemodynamic monitoring. Admission was determined based on clinical readiness for discharge and completion of this evaluation so that some patients may have been admitted for completion of this workup.

Results

Discussion

Early identification of NAT is imperative for better outcomes from injuries associated with abuse and removal of children from ongoing risk.¹⁷ Recent literature has identified risk factors and common presentations of children presenting with NAT, improving the ability to recognize and treat these vulnerable patients.¹⁸ Numerous studies have identified age as a major risk factor, specifically children 1 year of age and younger.^3,19 Those children were also shown to have polytrauma to the abdomen/pelvis and higher rates of skeletal and visceral injuries.²⁰ Other risk factors include children who were previously hospitalized. Furthermore, children sustaining abuse are more likely to be enrolled in Medicaid or have no insurance at all, pointing to evidence that abused children are more likely to be from economically disadvantaged backgrounds.^3,19

Despite the presence of established risk factors and apparent signs of NAT, there is a discrepancy within the literature regarding standardized workup in the case of skull fractures. Only a limited number of hospitals routinely use standardized protocols when evaluating abuse.^21,22 Given the findings of a recent study demonstrating that skull fractures can remain radiographically visible for 2 to 18 weeks, the finding of a skull fracture on imaging does not indicate acuity of the injury.²³ Given the percentage of our population who had delayed presentation or no clear known injury, and in keeping with the concerns of the authors of that study, consideration of a skull fracture as a potential finding of abuse should not be ignored.^5,8,23

For this reason, the American Pediatric Surgical Association recommends the identification of suspicious injuries for abuse via a team of social workers, child protective services, law enforcement, emergency medical services, and fire responders, as well as child abuse–experienced pediatricians.²⁴ Standardizing protocols may improve identification of abuse and potentially improve utilization of hospital and ED resources by employing the highest-yield components of NAT workup to effectively and expediently identify suspected cases. Escobar et al. proposed a screening tool to assist with protocol standardization that includes “any fracture in a non-ambulating infant” as a red flag radiographic finding warranting additional evaluation.¹⁸ The implications of such optimization include reductions in cost by limiting inpatient stays when unnecessary, improvement in ED patient flow, and the appropriate leveraging of social work or the CPT.

Based on our initial analysis, achievement of 100% compliance with our protocol was difficult. In evaluating our experience to find barriers to compliance, our highest rate of overall protocol noncompliance occurred in patients who were discharged from the ED. For patients presenting outside of a trauma team activation, the initial neurosurgical consult often prompted further workup recommendations. Of these patients discharged from the ED, 3 patients were not seen by any other services; 2 of these patients had a negative skeletal survey, and 1 patient was evaluated by pediatric surgery and social work but received no other workup. Only 1 patient discharged from the ED received the complete workup, and 1 patient received all but evaluations from the ophthalmology team and the CPT. These patients presented throughout the series and thus would not represent a pure “learning curve” of the neurosurgical or emergency medicine services in obtaining these consults. Given the retrospective nature of the review, it is unknown whether there were other factors that negated the need for additional workup (i.e., if the patient had an accident witnessed by a bystander and reported by emergency medical services). However, potential factors that may indicate poor compliance could include ED providers being unaware of the protocol, failure of the ED to initiate the protocol, or failure of the neurosurgical service to recommend the protocol. This could represent human error or the potential impact of implicit bias as there was no mechanism by which to enforce compliance with this workup. Four patients who were admitted to the hospital did not receive either social work or CPT evaluation, although we were unable to assess why these patients did not receive these based on our chart review.

Given that a significantly higher compliance rate can be achieved if patients receive at least one consult to assess the home social situation, it is important to understand the varying capabilities between CPT and social work, and further determine the interchangeability of these services at any given institution. Both entities routinely evaluate the home and social situation of patients; however, a CPT physician is further able to integrate other medical issues into the context. Our evaluation of interrater reliability demonstrates good correlation between social work services and the CPT in identifying concern for NAT. It is likely that this is largely impacted by the overlap between what the two services evaluate, including the integration of consulting service recommendations into the final assessment. Although it may represent less of an “interrater reliability” than fundamental overlap, the correlation between conclusions drawn from the two services would suggest that evaluation by one service would adequately identify any areas of concern, thus reducing the resources necessary to identify children at risk.

The finding that both social work and CPTs can comparably evaluate patients within the NAT protocol is of particular importance. Within our institution, a single CPT physician provides both inpatient consultations and outpatient clinical services, which can, at times, limit availability. To avoid disposition delays, social work teams would determine the level of concern for NAT independently if they felt confident that no further input would be needed. Additionally, other medical teams would share input on their level of concern given their respective fields of evaluation. Given that this methodology was successful at our institution, there is an opportunity to formally adopt this model within the protocol and, importantly, to extrapolate this pathway in hospitals without robust CPT physician services.

We specifically found that the thoroughness of our workup appeared to be greater in patients for whom there was a suspicion of NAT; however, this relationship is only associative in nature. Given the methodology of this study, causal inference is difficult to achieve, specifically when attempting to determine whether an increasingly thorough workup truly is more effective at identifying NAT or that suspicion among providers leads to an increasing number of consultations and diagnostic studies. This latter phenomenon undoubtedly underscores the importance of understanding both implicit and explicit provider bias and its implications on the disparate use of social work or CPT physician consultation. Ultimately, an effective and standardized protocol would mitigate the effects on potentiating disparity in care.

Overall, among factors that prompted referral to the CPT, we found the highest yield in skeletal survey findings and evaluation of the social situation. This would suggest that a comprehensive social workup and thorough skeletal survey may provide the most high-yield aspects of the NAT evaluation. This is consistent with existing literature that has demonstrated that several long bone and skeletal fracture patterns have been associated with abuse.^25,26 Further evaluation and analysis of potential cost savings from streamlining our protocol will help to refine our strategy and will be the subsequent aim of our trauma committee.

Only 1 of the 23 patients flagged for concern of NAT was referred to the DCFS specifically because of a positive ophthalmology examination finding. Retinal hemorrhages have been demonstrated to be more prevalent in cases of abusive head trauma as opposed to accidental head trauma. Although there is no clear retinal sign that is unique to abusive head injuries, the presence of bilateral retinal hemorrhages is much more commonly noted in abusive head trauma, and the probability of abuse with the combination of intracranial injury and retinal hemorrhages is approximately 91%.²⁷ Although we did not have a preponderance of patients with retinal hemorrhages, this, combined with our cohort, would suggest that ophthalmology consults would be of low yield in patients without intracranial findings.

One major limitation in our study is that we did not have a mechanism to obtain final evaluation from the county social workers who were notified of potential concerns. As many cases of abuse are only officially determined following admission or conviction, it is impossible to assess whether the notification of potential abuse ultimately resulted in action taken. Importantly, it should be noted that our protocol was adopted to standardize a workup and evaluation for children at risk of abuse. Without a clear admission from a perpetrator, our profession is unable to identify these children with 100% sensitivity or specificity. We would expect the number of children warranting a report to the DCFS would be higher than the number of children who were victims of abuse. The goal of any such protocol should be to reduce the number of children who are missed while minimizing the number of accidental injuries that are reported, which both has psychosocial implications for the caregivers and can unintentionally overwhelm the social service system. One child evaluated and referred in this study ultimately died of abusive injuries several years after our initial evaluation for an ISF, which proved to be his sentinel injury. Despite our initial concern and notification of the county services, this child serves as a grim example of the difficulty in both identifying and protecting these children despite best efforts.

Conclusions

Our study suggests the rate of concern for NAT in infants presenting with skull fractures is substantial enough to warrant a standardized safety evaluation. Our comprehensive protocol allowed for evaluation of these patients; however, achievement of 100% compliance with the protocol was difficult. In our limited cohort, evaluation by social work or a CPT, as well as a skeletal survey, identified most children for whom there was a concern for risk. Ophthalmology evaluation in the setting of ISFs likely is of low yield but should be performed in patients with intracranial hemorrhage. Our study would also support the concept that appropriate utilization of social work in the absence of a physician CPT may adequately identify patients at risk. Further studies may be aimed at improving institutional adherence and consideration of cost implications.

Acknowledgments

We thank Hasina Momotaz, MS, biostatistician in the Case Western Reserve University Department of Population and Quantitative Health Sciences, for her statistical analysis support.

Disclosures

The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper.

Author Contributions

Conception and design: Tomei, Boruah, Potter, Hills, Dingeldein. Acquisition of data: all authors. Analysis and interpretation of data: all authors. Drafting the article: Tomei, Boruah, Potter, Shammassian. Critically revising the article: Tomei, Boruah. Reviewed submitted version of manuscript: Tomei, Boruah. Approved the final version of the manuscript on behalf of all authors: Tomei. Administrative/technical/material support: Tomei.

Supplemental Information