Article Text
PDF
Abstract
Objective Infant and toddler subdural haemorrhages (SDH) are often considered indicative of abuse or major trauma. However, accidental impact events, such as falls, cause contact extra-axial haemorrhages (EAHs). The current study sought to determine frequency and clinical behaviour of EAHs with infant and toddler accidental and abusive skull fractures.
Patients and methods Children aged <4 years with accidental skull fractures and abusive fractures identified by CT at two paediatric tertiary care centres. Clinical data were abstracted by child abuse paediatricians and images were reviewed by paediatric radiologists. Data were analysed using univariate and multivariate logistic regression as well as descriptive statistics.
Results Among 227 subjects, 86 (37.9%) had EAHs. EAH was present in 73 (34.8%) accidental and 13 (76.5%) of the abusive injuries. Intracranial haemorrhage rates were not different for children with major or minor accidents but were fewer than abused. EAH was equally common with falls <4 and >4 ft. EAH depths did not differ by mechanism, but 69% of accidental EAHs were localised solely at fracture sites vs 38% abuse. Widespread and multifocal EAHs were more common with abuse. Children with abuse or major accidental injuries presented with lower initial Glasgow Coma Scales than those with minor accidents. Abused children had initial loss of consciousness more often than those with either minor or major accidents.
Conclusions Simple contact EAHs were common among children with minor and major accidental skull fractures. Accidental EAHs were more localised with less neurological dysfunction than abusive.
- child abuse
- neurosurgery
Data availability statement
Data are available upon request.
Statistics from Altmetric.com
What is already known on this topic?
The presence of subdural haemorrhages in young children poses higher odds of being from abuse than from other head injuries.
Accidental head injuries in children are common and can present with a spectrum of findings, from mild cephalohematoma to skull fracture with intracranial haemorrhage.
What this study adds?
Small contact extra-axial haemorrhages associated with skull fractures due to minor injury events are typically thin, unifocal, subjacent to skull fractures and associated with overlying soft tissue swelling.
Initial loss of consciousness is more frequent with abuse than with either minor or major accidental injury mechanisms.
Initial Glasgow Coma Scale in children with fractures due to minor accidents is higher than for those with abuse or major accidental mechanisms.
Introduction
Head trauma is the leading cause of infant morbidity and mortality among abused children.1 About 30% of children presenting with abusive head trauma (AHT) are missed because symptoms of intracranial injury are often subtle and confused with other common conditions.2 3 Consequently, clinicians must be attuned to subtle signs and symptoms. They should remain sceptical of implausible caretaker histories and obtain relevant imaging.2 3 Of children who are initially missed, 28% are reinjured and 40.7% experience medical complications of subsequent trauma.2 Traditionally, subdural haemorrhages (SDH), absent predisposing diseases, have been considered a strong indicator of AHT or other major traumatic causes of head injury.4–7 A meta-analysis of literature published in 2009 found SDH conveyed more than eight times the odds of abuse, compared with other head injuries.6 However, we know accidental head injuries in children are common and can present with a spectrum of findings, from mild cephalohematoma to skull fracture with intracranial haemorrhage (ICH).4 5 8 9 With advanced head imaging, we are less apt to miss small fractures or ICH, which often lack clinical significance and do not require intervention.5 10–13 Cranial CT including three-dimensional reconstructions identifies skull fractures with high sensitivity.14 Extra-axial haemorrhages (EAH) have been associated with skull fractures.8 10 15 Radiological distinctions between small SDHs and epidural haemorrhages (EDHs) can be difficult.16 It is unclear how often small contact SDHs and EDHs accompany skull fractures result from common accidental injuries of childhood. The Paediatric Emergency Care Applied Research Network showed that relatively minor head injury histories can cause skull fractures and/or EAH, generally not requiring intervention, but the authors did not report how many had both skull fractures and small contact EAHs.12 Likewise, clinical patterns of these injuries are incompletely described. It is important to know whether such small EAHs commonly accompany normal accidental infant and toddler injury mechanisms, as with short falls. Abuse versus accident implications of SDHs might differ based on the haemorrhage’s and child’s clinical characteristics. The current study evaluated young children with accidental and abusive skull fractures to better characterise frequency, attributes and clinical behaviour of associated EAHs.
Methods
We conducted a retrospective chart and imaging review for children aged <4 years with a skull fracture (International Classification of Diseases (ICD)-9 code 800-804) at two paediatric tertiary care centres in Seattle, Washington, Seattle Children’s Hospital and Harborview Medical Center. We compared children who had sustained accidental and abusive skull fractures between 2011 and 2014 in a case-control design. Only children who had a cranial CT revealing a cranial vault fracture were included. Children with only maxillofacial fractures were excluded. We excluded infants with birth-related fractures. Subjects with insufficient clinical data, unavailable images or injuries indeterminate for abuse versus accident were excluded. CTs were reread by paediatric radiologists. Radiologists were blinded to abuse determination but had access to earlier radiological reads. Radiology data included fracture type, soft tissue swelling (STS) and location and presence, type and depth of EAHs.
Study radiologists measured EAH depth from the skull’s inner table to the subjacent inner EAH margin using INFINITT PACS (V.G3, Infinitt Heathcare, Seoul, South Korea). If depth was previously measured by the initial radiologist, that measurement was used. Radiological STS and EAH were evaluated for proximity to skull fractures. STS was directly associated with skull fractures if any STS region overlaid the fracture. Radiologists analysed EAHs for location and size, using three groups: (1) localised, (2) widespread and (3) multifocal. Localised EAHs were small and solely at fracture site. Widespread EAHs were contiguously, extensively distributed. Multifocal EAHs were present at multiple disparate sites. Some children had both widespread and multifocal EAHs. EAHs subjacent to fractures, and pooling in nearby regions within the same hemisphere due to CT positioning, were considered directly associated with overlying fractures. Only children with three-dimensional reconstructions were reviewed for skull fracture complexity. Based on prior studies, we defined complex skull fractures as having one or more characteristic: (1) greater than one fracture of any single skull bone, (2) greater than one separate fracture, (3) fractures crossing suture lines or fractures involving two or more skull bones, (4) non-linear fractures/branching fractures, (5) fracture diastasis >3 mm or (6) depressed fractures.17
Radiologists identified EAHs as EDHs when they failed to cross sutures and assumed lenticular shapes. Haemorrhages with tapered margins, often crossing sutures, were considered SDHs. For very small haemorrhages, this distinction became problematic. Thus, although we present data on SDHs and EDHs, we also combined subjects with either haemorrhage into EAHs for analysis.
Child abuse pediatricians (CAPs) reviewed medical records for presence and duration of initial loss of consciousness (LOC), initial Glasgow Coma Scale (GCS), scalp STS, mechanism of injury and abuse signs. Abuse was determined based on the entire history, clinical presentation and multidisciplinary team review during hospital admission. Medical findings were often supplemented by protective services and police investigations.
We defined ‘short falls’ based on caretaker histories as <4 ft. We further classified subjects into: (1) ‘minor’ injury mechanisms, such as short falls, toddlers running into or bumping objects and children struck with low mass objects and (2) ‘major’ injury mechanisms, including falls from heights >4 ft, motor vehicle trauma, severe crush events and children hit with high mass or force objects.
Data were entered into REDCap (Vanderbilt University, Nashville, Tennessee, USA). A stepwise approach was used to assess clinical and radiographic factors associated with injury type. First, 2×2 contingency tables were analysed and ORs calculated using Fisher’s exact test. Factors with more than two levels were assessed using univariate logistic regression. Second, factors identified on univariate analysis to be statistically significantly associated with accidental injuries compared with abuse (p<0.05) were further evaluated with univariate and multivariable logistic regression. Using an all-subset model parameter selection approach, the optimised multivariable logistic regression model with minimum Akaike information criterion (AIC) and Bayes information criterion (BIC) was identified. Each statistical analysis was performed using complete cases with all available data without imputation. P values are reported without correction for multiple testing. Statistical analyses were performed using the R Statistical Programming Language (V.3.6.1).
Results
Subjects
Between 2011 and 2012, 288 subjects with skull fractures were identified by ICD-9 codes; 62 lacking a cranial vault fracture, or lacking CTs or sufficient clinical records were excluded (figure 1). Six children were excluded because the determination of accidental versus abusive cause was inconclusive (table 1). Two children with birth fractures were excluded. Between 2013 and 2014, 36 children evaluated by CAPs were AHT victims, contributing 9 additional cases with abusive skull fractures. This resulted in 227 subjects, 210 (92.5%) with accidental and 17 (7.4%) with abusive skull fractures.
Study cohort. *Between 2013 and 2014, 36 children evaluated by CAPs were abusive head trauma victims, contributing 9 additional cases with abusive skull fractures. EAH, extra-axial haemorrhage; ICD, International Classification of Diseases.
Indeterminate cases
Children who had known accidental event histories
Twenty-eight of 210 (13.3%) accidental trauma subjects were not included in mechanism of injury analysis due to insufficient trauma history or unknown fall height. Twelve were early ambulatory children with skull fractures who lacked specific trauma histories. Among included subjects, 56% were male. Average age was 13.6 months and median and IQR 8.4 (3.6–19.9) months (table 2). Frequency of EAHs did not differ for children with minor versus major accidental mechanisms (table 3). Height was not documented for 22 accidental fall subjects. Of 93 children with short falls, 42 (45.2%) had EAHs. This did not differ from 31 of 67 (46.3%) high falls (OR 0.31 (95% CI 0.17 to 0.56)). The percentage of EAHs in children with simple skull fractures (20/68, 29.4%) was less than those with complex fractures (30/65, 46.1%) OR 0.49 (95% CI 0.24 to 0.99).
Demographics data for 227 patients (2011–2014)
Univariate analysis of clinical and radiographic findings and injury type using Fisher’s exact test
Extra-axial haemorrhages, accident versus abuse
Eighty-six subjects with skull fractures (37.9%) had at least one EAH. Of 210 children with accidental injuries, 73 had EAHs (34.8%); 27 EDHs and 54 SDHs. Among the 17 abused subjects were 13 (76.5%) with EAHs; 2 EDHs and 13 SDHs. More abused children had EAH compared with children with accidental fractures. Similarly, SDHs were present in more abused (76.5%) than accidental (25.7%) children. The OR for accident versus abuse in a child whose skull fracture was accompanied by SDH was 0.11 (95% CI 0.002 to 0.37). For EAH it was 0.16 (95% CI 0.004 to 0.56).
Comparison of accidentally injured children and abused children
CAPs consulted, by phone or in person, on 80 (38%) accidental cases. Skeletal surveys were completed in 46 (21.9%) and formal eye exams in 29 (13.8%) accidental cases. Of 182 children with documented falls, 31 (17.0%) had bruising at sites other than the skull fracture. On detailed review, all such bruising was consistent with the history or at normal age-appropriate sites. No retinal haemorrhages, unexplained bruises, oral injuries, burns or non-skull fractures were identified in accidentally injured children. No accidentally injured child with a minor trauma history died or required surgery acutely. One child subsequently required surgery for a growing fracture. By contrast, all but one abused child had other abusive injuries beyond skull fractures and intracranial injuries.
Mean EAH depth with accidental skull fractures was 5.2 mm (n=71), EDH 6.9 mm (n=25) and SDH 4.3 mm (n=46) (online supplemental material 1). Mean EAH depth with abusive fractures was 5.3 mm (n=15), EDH 4.4 mm (n=2) and SDH 5.4 mm (n=13). There was sufficient overlap of EAH depth to not statistically distinguish accidents from abuse.
Supplemental material
All accidentally injured subjects with EAHs had some haemorrhage subjacent to their fracture (table 3). Thirty-eight (70.4%) of 54 accidental SDHs were localised to solely subjacent to fractures, 8 (14.8%) were subjacent and widespread and 8 (14.8%) were subjacent and multifocal. Thirteen of 17 (76.4%) abused children had some SDH subjacent to the fracture. However, only 5 (38.4%) of 13 abusive SDHs with a subjacent component were solely subjacently localised; 4 (30.8%) were widespread and 5 (38.5%) were multifocal (table 3). SDH located solely subjacent to the fracture was not significant for all accidental injury versus abuse (OR=0.53 (95% CI 0.16 to 2.04)) nor individually for minor or major accident versus abuse. Multifocal SDHs were less common for all accidents than for abuse (OR=0.10 (95% CI 0.02 to 0.44). Widespread SDHs were statistically less common with minor injury mechanisms versus child abuse (OR=0.03 (95% CI 0.00 to 0.38)) and versus major accidents (OR=0.11 (95% CI 0.00 to 0.88)). They did not differ for major accidents versus abuse (OR=0.31 (95% CI 0.07 to 1.64)). STS presence was not different for accidental versus abused children (table 3).
Abused children presented with lower GCS scores than children with accidental injuries. This included both minor and major accident mechanisms (table 4). Children with major mechanisms of injury had lower GCS scores that those with minor mechanisms of injury. Looking only at children with accidental skull fractures and associated intracranial haemorrhages, children with minor injury histories had higher GCS scores than those with major accidents or abuse.
Comparison of univariate and multivariable logistic regression analysis of statistically significant risk factors associated with accidental injury compared with abuse injury (n=212, 15 incomplete cases excluded)
More abused children experienced initial LOC than children with minor, major and all accidents (table 3). However, accidentally injured children had no difference in initial LOC frequency for minor versus major mechanisms. Three child victims of minor, three of major accidents and three abuse victims had documented LOC durations. Two minor accidents had LOC of <1 min and one 1–5 min. With major accidents, one had LOC of 1–5 min and two >10 min. One abused child had LOC <1 min, one 5–10 min and one >10 min. None of these comparisons was statistically significant.
Multivariable analysis
The main factors associated with a statistically significant modified risk for accidental injuries compared with abuse injuries identified on univariate contingency table analysis were evaluated using univariate logistic regression (table 4) with similar results. Due to the strong association between the presence of EAH and SDH, the presence of SDH was selected a priori for inclusion in the logistic regression models.
After simultaneously accounting for all these statistically significant factors using a multivariable logistic regression model, the presence of SDH remained statistically significantly associated with a lower risk of accidental injury compared with abusive injury among children with skull fractures (OR 0.14 (95% CI 0.03 to 0.63), p=0.01). Using an all-subset model selection strategy, the optimised multivariable logistic regression model with minimum AIC and BIC included the presence of SDH (OR 0.10 (95% CI 0.03 to 0.38), p=0.0006) and LOC (OR 0.22 (95% CI 0.06 to 0.82), p=0.02).
Discussion
Small contact EAHs, solely adjacent to fracture sites, were very common in young children with skull fractures sustained from accidental paediatric head trauma, including those with minor trauma mechanisms.18 Our 34.8% incidence of EAHs associated with accidental skull fractures is similar to the 29.2% and 42% rates reported in hospitalised infants with skull fractures from falls.19 Although EAHs and complex skull fractures have been considered an indication for a child abuse evaluation, contact EAHs alone may have low abuse specificity.20 Our study demonstrates that contact EAHs with skull fractures due to minor accidental head injury are typically thin, subjacent to skull fractures and associated with overlying STS. On 2×2 comparison, EAHs and SDHs due to both minor and major injury events were less often widespread or multifocal than with abuse. Although not reaching significance on multivariate analysis, EAHs due to minor accidental mechanisms were usually not widespread or multifocal. This is consistent with prior studies on accidental head trauma EAHs.11 13 18 19
Although EAHs were frequent with accidental skull fractures, they were much more frequent with abuse. As such, identifying EAHs accompanying a skull fracture raises the possibility of an abusive aetiology without confirming abuse. This relationship is stronger, when the EAH is SDH. Initial LOC was also more frequent with abuse than with either minor or major accidental injury mechanisms. Notably, more accidentally injured children with a reported initial LOC had EAH on imaging than those without LOC. Furthermore, initial GCS in children with fractures due to minor accidents was higher than for those with major accidental mechanisms or abuse. Most children with minor injury mechanisms were not concussed and presented with normal or near-normal GCS scores.
Although our data included few abuse cases, haemorrhages from AHT can present in similar locations as contact EAHs from accidental head trauma, with the main anatomic differences being their widespread and multifocal distributions. The higher frequency of low GCS scores and initial LOC with abuse compared with minor accidents is expected. Child victims of major accidental events behaved more like abused children with GCS scores. Too few children had reported initial LOC durations to allow statistical analysis, but the available data suggest both major accidental injuries and abuse are associated with longer LOC durations. While those attributes of children with abusive SDHs are well recognised, accidental contact SDHs also can occur.
Our study has several limitations. As we used retrospective data, documentation was occasionally incomplete or unclear regarding subject and injury event attributes. Fall heights may not have been accurately reported, recorded or interpreted. Distinctions between small EDHs and SDHs can be difficult, and the depth of EDHs might change quickly as blood moves through adjacent skull fractures into overlying cephalohematomas.11 16 As a result, we combined data for EAHs that our radiologists felt were EDHs or SDHs for analysis. Some rates that seemed consequential did not reach statistical significance; this may be the result of type 2 errors, occurring when we analysed smaller subject subgroups. Although our sample would have been larger if we had included subjects with skull fractures seen on plain X-rays, but without CTs, diagnosis of fracture complexity would have been less accurate and of intracranial injury impossible. Although problematic diagnostic cases were evaluated by CAPs and multidisciplinary teams, some skull fractures remained of indeterminate cause. A minority of children judged to have sustained accidental fractures had full child abuse evaluations, but their histories and/or developmental abilities provided plausible explanations for their fractures and they lacked unexplained extracalvarial injuries. Finally, it is possible that determination of abuse may have been influenced by the characteristics of the EAHs. However, almost all cases determined to be abusive had other, extracranial abusive injuries.
Conclusion
Small contact EAHs commonly occur subjacent to accidental skull fractures due to minor injury events. Most are of minor clinical import. It is important to recognise that accidental contact injuries can cause both skull fractures and EAHs. At the same time, EAHs accompanying skull fractures increase the possibility that a credible history is masking actual abuse. Additional findings that move this modest possibility to a significant probability may only be identified when a comprehensive evaluation for occult injury is completed. The presence of EAH underlying a well-explained skull fracture forms reasonable basis to further evaluate for medical signs of abuse, but a poor basis to diagnose abuse, absent those additional signs.
Data availability statement
Data are available upon request.
Ethics statements
Patient consent for publication
Ethics approval
This study was approved by Seattle Children’s Human Subjects Review Board (IRB #14947).
Acknowledgments
Part of this research was conducted as Jordan Wallace’s third year University of Washington School of Medicine medical student research project. Meghan Kelton contributed to the initial setup and design of the REDCap database and data extraction.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors JW: abstracted and analysed data, composed the initial manuscript draft. JBM: developed the parent study and database, assisted with data analysis, edited the paper. JO: assisted with development of the database, conducted radiology review, edited the paper. FAP: conducted radiology review and edited the paper. SD: conducted radiology review. ECBB: conducted clinical review and data abstraction, edited the paper. RTW: conducted clinical review and data abstraction. SCB: conceived the study purpose and analysis. SG: provided data analysis and manuscript review. KWF: conceived of and assisted with development of the parent study and the current subanalysis, had access to the data, participated in data analysis and study editing and is the guarantor for the study and controlled the decision to publish this manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests JBM, JO, SD, ECBB, RTW, SCB and KWF have provided medical legal consultation and testimony.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.