Retinal Hemorrhages: Abusive Head Trauma or Not?

Ophthalmologic findings in abusive head trauma

A leading cause of death and disability in infancy is abusive head trauma (AHT) and there are common clinical signs that help to establish this diagnosis. Children diagnosed with AHT can have many ophthalmologic findings, including retinal hemorrhages, retinoschisis, subconjunctival hemorrhages, corneal injury, and globe rupture. If any such injuries are suspected, an ophthalmologic consultation, with indirect ophthalmoscopy, should be completed. In addition to a complete physical exam, a thorough history imaging, and lab work, should be obtained to investigate the etiology of ophthalmic pathology including accidental and systemic causes. In general, studies show that retinal hemorrhages that are multilayered, too numerous to count, and located from the posterior pole to the ora serrata are highly suspicious for abusive head trauma.

Retinal hemorrhages due to intracranial hypertension, is it possible to differentiate them from those associated with abuse?

BACKGROUND

Retinal hemorrhages (RHs) are suggestive of abusive head trauma (AHT). Even so, controversy persists about other possible causes, a fact that hinders the diagnosis of abuse.

OBJECTIVE

To determine the prevalence and patterns of RHs associated with increased intracranial pressure (ICP) in young children.

PARTICIPANTS AND METHODS

Prospective, single center study of children aged 29 days-3 years undergoing a lumbar puncture with opening pressure (OP) measured during routine clinical care in the emergency department, over a 4-year period. Children with known causes of RHs were excluded. All the children underwent a detailed ophthalmological examination within 72 h of admission. For children with RHs, an in-depth investigation was carried out to rule out AHT.

RESULTS

Thirty-four patients were included: 20(58.8 %) were boys, median age 11.7 months (range 1-33 months). Overall, 29(85.3 %) had clinical findings of increased ICP. The duration of symptoms was ≤4 days in all patients except one, in whom it was 1.5 months. The median OP was 27cmH2O (range 20-60cmH2O). One of the children was found to have RHs, with evaluation resulting in a diagnosis of AHT. No RHs were found in any of the others studied. Using the Wilson method, we can be confident to an upper limit of 95 % that the probability of RHs occurring secondary to increased ICP alone is at most 0.1.

CONCLUSIONS

RHs were not detected in patients with isolated nontraumatic increased ICP, measured via OP and diagnosed in the ED. Therefore, if RHs are detected, investigation into the possibility of AHT is warranted.

Utility of balanced steady-state field precession sequence in the evaluation of retinal and subdural hemorrhages in patients with abusive head trauma

Abusive head trauma is the leading cause of physical child abuse deaths in children under 5 years of age in the United States. To evaluate suspected child abuse, radiologic studies are typically the first to identify hallmark findings of abusive head trauma including intracranial hemorrhage, cerebral edema, and ischemic injury. Prompt evaluation and diagnosis are necessary as findings may change rapidly. Current imaging recommendations include brain magnetic resonance imaging with the addition of a susceptibility weighted imaging (SWI) sequence which can detect additional findings that suggest abusive head trauma including cortical venous injury and retinal hemorrhages. However, SWI is limited due to blooming artifacts and artifacts from the adjacent skull vault or retroorbital fat, which can affect the evaluation of retinal, subdural, and subarachnoid hemorrhages. This work explores the utility of the high-resolution, heavily T2 weighted balanced steady-state field precession (bSSFP) sequence to identify and characterize retinal hemorrhage and cerebral cortical venous injury in children with abusive head trauma. The bSSFP sequence provides distinct anatomical images to improve the identification of retinal hemorrhage and cortical venous injury.

Development of a Deep Learning Model for Retinal Hemorrhage Detection on Head Computed Tomography in Young Children

Importance

Abusive head trauma (AHT) in children is often missed in medical encounters, and retinal hemorrhage (RH) is considered strong evidence for AHT. Although head computed tomography (CT) is obtained routinely, all but exceptionally large RHs are undetectable on CT images in children.

Objective

To examine whether deep learning-based image analysis can detect RH on pediatric head CT.

Design, Setting, and Participants

This diagnostic study included 301 patients diagnosed with AHT who underwent head CT and dilated fundoscopic examinations at a quaternary care children's hospital. The study assessed a deep learning model using axial slices from 218 segmented globes with RH and 384 globes without RH between May 1, 2007, and March 31, 2021. Two additional light gradient boosting machine (GBM) models were assessed: one that used demographic characteristics and common brain findings in AHT and another that combined the deep learning model's risk prediction plus the same demographic characteristics and brain findings.

Main Outcomes and Measures

Sensitivity (recall), specificity, precision, accuracy, F1 score, and area under the curve (AUC) for each model predicting the presence or absence of RH in globes were assessed. Globe regions that influenced the deep learning model predictions were visualized in saliency maps. The contributions of demographic and standard CT features were assessed by Shapley additive explanation.

Results

The final study population included 301 patients (187 [62.1%] male; median [range] age, 4.6 [0.1-35.8] months). A total of 120 patients (39.9%) had RH on fundoscopic examinations. The deep learning model performed as follows: sensitivity, 79.6%; specificity, 79.2%; positive predictive value (precision), 68.6%; negative predictive value, 87.1%; accuracy, 79.3%; F1 score, 73.7%; and AUC, 0.83 (95% CI, 0.75-0.91). The AUCs were 0.80 (95% CI, 0.69-0.91) for the general light GBM model and 0.86 (95% CI, 0.79-0.93) for the combined light GBM model. Sensitivities of all models were similar, whereas the specificities of the deep learning and combined light GBM models were higher than those of the light GBM model.

Conclusions and Relevance

The findings of this diagnostic study indicate that a deep learning-based image analysis of globes on pediatric head CTs can predict the presence of RH. After prospective external validation, a deep learning model incorporated into CT image analysis software could calibrate clinical suspicion for AHT and provide decision support for which patients urgently need fundoscopic examinations.

Assessing retinal hemorrhages with non-invasive post-mortem fundus photographs in sudden unexpected death in infancy

INTRODUCTION

In the case of sudden unexpected death in infancy (SUDI), eye examination is systematic to detect retinal hemorrhages (RH) that are a crucial hallmark for abusive head trauma (AHT). The aim of this study is to assess the ability of non-invasive post-mortem fundus photographs (PMFP) to detect RH in case of SUDI.

METHODS

Bicentric retrospective analysis of consecutive cases of SUDI under 2 years of age were managed by two French SUDI referral centers with PMFP by RetCam (Clarity Medical Systems USA). PMFP were reviewed randomly, twice, by three independent ophthalmologists blinded for clinical data.

RESULTS

Thirty cases (60 eyes) were included. Median age was 3.5 months (interquartile [1.6; 6.0]). No child died of AHT. Image quality was sufficient to assert presence or absence of RH in 50 eyes (83%). Sufficient quality rate was significantly higher when the post-mortem interval was inferior to 18 h (91%, 42/46) as opposed to over 18 h (57%, 8/14, p=0.0096). RH were found in six eyes (10%), four children (13%), with excellent inter and intra-raters' concordance (Cohen's Kappa from 0.81 [0.56-1.00] to 1.00 [1.00-1.00]).

CONCLUSION

PMFP can detect RH in case of SUDI and is a relevant systematic screening test to be carried out as soon as the deceased child arrives in the hospital. It can decrease the need of eye removal for pathological examination, but further studies are needed to define the best decision algorithm.

Ophthalmological lesions in shaken baby syndrome: a retrospective analysis of 133 consecutive cases (1992-2018)

INTRODUCTION

Shaken Baby Syndrome (SBS) is a non-accidental head trauma in which shaking causes cranio-cerebral lesions. Shaking can lead to ophthalmologic lesions such as retinal hemorrhage (RH). The aim of the present study was to compare our long-term results in to the literature data.

PATIENTS AND METHODS

This study was a single-center retrospective descriptive analysis of 133 consecutive SBS cases (1992 - 2018). Only seniors in ophthalmology were authorized to perform these examinations. We studied type of lesion (retinal, intra-vitreal, papilledema), location (uni- or bi-lateral), and correlation with gender and age. Infants with a traumatic context without suspicion of child abuse were excluded.

RESULTS

Mean age at diagnosis was 131 days (range, 14 days - 10 months). Boys accounted for 72.2% of the population. The prevalence of ophthalmologic lesions was 70.3%. 94.4% were RH; intra-vitreous hemorrhage (6.7%) and papilledema (11.1%) were less frequent. Lesions were bilateral in 81.1% of cases. Retinal lesions were classified in terms of location. Macular involvement was diagnosed in 8.2% of cases. 18.8% of retinal lesions could not be classified because of lack of precision in the ophthalmology report. The prevalence of ophthalmic lesions was higher for children aged over 6 months: 80%.

CONCLUSION

This series highlighted a high rate of ophthalmic lesions in SBS, with a high rate of bilateral involvement. RH was the most frequent lesion. RH in a context of subdural hematoma is a strong argument in favor of SBS. The forensic implications are that rigorous ophthalmologic examination by a senior practitioner is mandatory.

Which Clinical Features of Children on Initial Presentation to the Emergency Department With Head Injury Are Associated With Clinically Important Traumatic Brain Injury, Classification as Abuse, and Poor Prognosis?

BACKGROUND

Pediatric traumatic brain injury (TBI) and abusive head trauma (AHT) are leading causes of morbidity and mortality. Clinicians may not be aware of AHT at presentation to the emergency department (ED).

OBJECTIVE

The objective of this study was to determine which clinical features associated with head injury in children on initial presentation to the ED trauma bay predict 3 outcomes including clinically important TBI (CiTBI), classification as confirmed abuse by Child Protection Team (CPT), and poor neurologic status on hospital discharge.

PARTICIPANTS AND SETTING

Inclusion for this study were children 3 years or younger, presenting to the ED with significant TBI. In addition, presentations where the mechanism of injury was not verifiable such as with falls, being struck by object, or no mechanism of injury reported by caregiver were included.

METHODS

Researchers used 3 sources of information for this analysis: a regional trauma registry, hospital records, and the CPT database. Clinical features included demographics, mechanisms of injury, physical, radiological findings, and CPT classification.

RESULTS

On pairwise analysis, seizures, apnea, and no mechanism of injury reported by caregiver were the only clinical features related to all 3 outcomes (P < 0.001). Rib fractures (relative risk [RR], 3.3; P < 0.001), long bone fractures (RR, 3.1; P < 0.001), retinal hemorrhages (RR, 3.0; P < 0.001), seizures (RR, 3.6; P < 0.001), apnea (RR, 4.4; P < 0.001), and younger than 6 months (RR, 1.8; P < 0.001) were related to AHT. On multivariable logistic regression, no mechanism of injury reported by caregiver and seizures remained significantly related to CiTBI; seizures and retinal hemorrhage remained significantly related to classification as abuse by CPT, and no mechanism of injury by the caregiver, apnea, and seizures were significantly related to poor outcome on hospital discharge.

CONCLUSIONS

No mechanism of injury reported by the caregiver, seizures, and apnea at the time of presentation to the ED are important features associated with CiTBI, classification as AHT, and poor prognosis. In addition, younger age, retinal hemorrhage, rib, and long bone fractures were found to be important clinical features associated with AHT.

Unusual eye injury related to abusive head trauma

Ophthalmological examination is an essential component in clinically diagnosing abusive head trauma (AHT). Typical of AHT injuries is retinal bleeding, with other parts of the eye sometimes also being affected. Visual impairment and blindness are long-term complications in this context. We present a case with unusual eye injuries after a diagnostically confirmed AHT.

The ophthalmological findings on admission day showed a massive subhyaloid hemorrhage at the posterior pole and extensive flame-shaped intraretinal hemorrhages accented on the right side., A centrally localized vitreous hemorrhage appeared on both sides 3 weeks later, obscuring the optic disc and macula most probably due to a breakthrough of a sub-internal limiting membrane (ILM) hemorrhage into the vitreous cavity. Follow-up 4 weeks later revealed a dense and organized vitreous hemorrhage occupying the optic disc and macula so that a pars plana vitrectomy was performed on both eyes.

The bilateral bleeding in different retinal layers, the vitreous hemorrhage and the proliferative vitreoretinal reaction (PVR) indicated massive damage caused by a significant acceleration-deceleration trauma. A two-phase vitreous hemorrhage in a child with AHT does not yet appear to have been described in the literature.

Catching the red eye: A retrospective review of factors associated with retinal hemorrhage in child physical abuse

BACKGROUND/PURPOSE

Accurate identification of child physical abuse is crucial during the evaluation of injured children. Retinal hemorrhages (RH) are used for diagnosis, but clear criteria for screening with direct fundoscopic exam are lacking. We sought to identify key factors associated with RH to guide evaluations.

METHODS

Electronic medical records for patients <1 year of age presenting to a Level I Pediatric Trauma Center with unwitnessed head injury from January 2015 to December 2018 were retrospectively reviewed. Multivariable logistic regression was used to identify factors associated with RH.

RESULTS

Two hundred and seventy-six patients were included; 63% underwent direct fundoscopic examination, of which 23% were positive and 77% were negative for RH. Unscreened patients tended to be older and have isolated skull fractures. Multivariable regression analysis revealed that abnormal GCS and subdural hemorrhage were positively associated with a diagnosis of retinal hemorrhage, while isolated skull fracture was negatively associated.

CONCLUSIONS

Children under 1 year of age with subdural hemorrhage have a greater risk of associated RH and should undergo routine screening with direct fundoscopic examination. Conversely, those with isolated skull fractures may not require an ophthalmology consultation. Standardized screening protocols may help reduce the risk of missing child physical abuse.

LEVEL OF EVIDENCE

III (Diagnostic Test).

Bilateral Parietal Skull Fractures in Infants Attributable to Accidental Falls

INTRODUCTION

Multiple skull fractures, including bilateral parietal skull fractures (BPSFs) in infants are considered to be suspicious for abusive head trauma (AHT). The aim of this report is to describe a series of BPSF cases in infants which occurred due to accidental falls.

METHODS

We searched our neuroradiology database for BPSF in infants (<1 year old) diagnosed between 2006 and 2019; we reviewed initial presentation, mechanisms of injury, clinical course, head imaging, skeletal survey X-rays, ophthalmology, social work and child abuse physicians (CAP) assessments, and long-term follow-up. "Confirmed accidental BPSF" were strictly defined as having negative skeletal survey and ophthalmology evaluation and a CAP conclusion of accidental injury.

RESULTS

Twelve cases of BPSF were found; 3 were confirmed to be accidental, with a mean age at presentation of 3 months. Two infants had single-impact falls, and 1 had a compression injury; all 3 had small intracranial hemorrhages. None had bruises or other injuries, and all remained clinically well. A literature search found 10 similar cases and further biomechanical evidence that these fractures can occur from accidental falls.

CONCLUSION

While AHT should be kept in the differential diagnosis whenever BPSFs are seen, these injuries can occur as a result of accidental falls.

[Unilateral retinal hemorrhage in infants-two cases of shaken baby syndrome?]

Ein 2,5 Monate alter Junge und ein 2 Monate altes Mädchen wurden wegen schwerer Bewusstseinstrübung pädiatrisch behandelt. Bei beiden Kindern fanden sich Subduralhämatome. Bei Verdacht auf nichtakzidentelles Schädel-Hirn-Trauma (NAHI) erfolgte eine Untersuchung des Augenhintergrundes, bei der sich bei beiden Säuglingen unilaterale Netzhautblutungen zeigten. Nach intensiver Differenzialdiagnostik wurde in beiden Fällen der Verdacht auf ein NAHI gestellt und eine rechtsmedizinische Begutachtung initiiert. Wichtig an dieser Fallserie ist, dass die Einseitigkeit von Netzhautblutungen ein NAHI nicht ausschließt.

Nonaccidental trauma in pediatric patients: evidence-based screening criteria for ophthalmologic examination

BACKGROUND

Ophthalmologic examination is included in the work-up for pediatric nonaccidental trauma (NAT) when there is concern for retinal hemorrhage. However, dilated fundus examination entails patient discomfort and prohibition of assessment of pupillary response. Previous studies have suggested that patients without neuroimaging abnormalities are unlikely to have retinal hemorrhage. The purpose of the current study was to analyze the findings in patients who received NAT evaluation with eye examination at our institution, and to propose screening criteria for inclusion of ophthalmologic examination in NAT evaluation.

METHODS

The medical records of patients who received NAT evaluation with ophthalmologic examination at The Johns Hopkins Children's Center Pediatric Emergency Department from August 2014 to July 2018 were reviewed retrospectively. Data collected included demographics, presenting symptoms, imaging findings, and ophthalmologic examination findings. The main outcome measure was presence of retinal hemorrhage.

RESULTS

A total of 192 evaluations with ophthalmologic examination were included, representing 190 unique individuals of mean age 8.4 ± 9.5 months at presentation. In approximately half (54%) of the evaluations, there were abnormal findings on neuroimaging. Fifteen children (8%) had retinal hemorrhage, all of whom also had abnormal neuroimaging. Abnormal neuroimaging was associated with presence of retinal hemorrhage, with an odds ratio of 21.0 (95% CI, 3.47-∞; P < 0.001). Of the 15 children with retinal hemorrhage, 14 had subdural hemorrhage.

CONCLUSIONS

When neuroimaging abnormalities are present, ophthalmologic examination should be performed as part of the pediatric NAT evaluation. When there is no evidence of head injury on neuroimaging, ophthalmologic examination should not be routine.

New grading criterion for retinal haemorrhages in term newborns based on deep convolutional neural networks

BACKGROUND

To define a new quantitative grading criterion for retinal haemorrhages in term newborns based on the segmentation results of a deep convolutional neural network.

METHODS

We constructed a dataset of 1543 retina images acquired from 847 term newborns, and developed a deep convolutional neural network to segment retinal haemorrhages, blood vessels and optic discs and locate the macular region. Based on the ratio of areas of retinal haemorrhage to optic disc, and the location of retinal haemorrhages relative to the macular region, we defined a new criterion to grade the degree of retinal haemorrhages in term newborns.

RESULTS

The F1 scores of the proposed network for segmenting retinal haemorrhages, blood vessels and optic discs were 0.84, 0.73 and 0.94, respectively. Compared with two commonly used retinal haemorrhage grading criteria, this new method is more accurate, objective and quantitative, with the relative location of the retinal haemorrhages to the macula as an important factor.

CONCLUSIONS

Based on a deep convolutional neural network, we can segment retinal haemorrhages, blood vessels and optic disc with high accuracy. The proposed grading criterion considers not only the area of the haemorrhages but also the locations relative to the macular region. It provides a more objective and comprehensive evaluation criterion. The developed deep convolutional neural network offers an end-to-end solution that can assist doctors to grade retinal haemorrhages in term newborns.