a b s t r a c t

Introduction: Blunt trauma is a leading cause of pediatric morbidity. We compared injuries, interventions and out- comes of acute pediatric blunt torso trauma based on intent.

Methods: We analyzed de-identified data from a prospective, multi-center emergency department (ED)-based observational cohort of children under age eighteen. Injuries were classified based on intent (unintentional/ inflicted). We compared demographic, physical and laboratory findings, ED disposition, hospitalization, need for surgery, 30-day mortality, and cause of death between groups using Chi-squared or Fisher’s test for categor- ical variables, and Mann-Whitney test for non-normal continuous factors comparing median values and inter- quartile ranges (IQR).

Results: There were 12,044 children who sustained blunt torso trauma: Inflicted = 720 (6%); Unintentional = 9563 (79.4%); Indeterminate = 148 (1.2%); Missing = 1613 (13.4%). Patients with unintentional torso injuries significantly differed from those with inflicted injuries in median age in years (IQR) [10 (5, 15) vs. 14 (8, 16); p-value b 0.001], race, presence of pelvic fractures, hospitalization and need for non-abdominal surgery. Mortal- ity rates did not differ based on intent. Further adjustment using binary, logistic regression revealed that the risk of pelvic fractures in the inflicted group was 96% less than the unintentional group (OR: 0.04; 95%CI: 0.01-0.26; p-value = 0.001).

Conclusions: Children who sustain acute blunt torso trauma due to unintentional causes have a significantly higher risk of pelvic fractures and are more likely to be hospitalized compared to those with inflicted injuries.

(C) 2017

Introduction

Injuries are a leading cause of morbidity and mortality in US children aged 1-18 years [1]. Most of these injuries are due to blunt trauma. Non- fatal unintentional injuries are almost 39 times higher than injuries sustained due to assaults among US children aged 18 years and under [2]. motor vehicle crashes account for a majority of unintention- al injury death in children [3] whereas non-self-inflicted injuries occur as a result of child maltreatment in infants and young children [4] or due to peer-violence in adolescents [5,6].

The mechanism of injury and the anatomy of the child determine their type of injuries [7]. An important pitfall in diagnosing injuries to the chest and abdomen in children is the relative lack of external signs of trauma. The flexibility and resilience of the pediatric skeleton and surrounding structures allows external forces to be transmitted to the

? The abstract of the article was presented at the Pediatric Academic Societies meeting in San Diego, CA in May, 2015.

* Corresponding author at: Section of Emergency Medicine, Baylor College of Medicine, 6621 Fannin, Suite A 2210, Houston, TX, United States.

E-mail addresses: [email protected] (R.P. Shenoi), [email protected]

(E.A. Camp), [email protected] (D.M. Rubalcava), [email protected] (A.T. Cruz).

deeper structures in the body [7]. Additionally, in young children the horizontal insertion of the diaphragm permits the abdominal organs to be more exposed and less protected by the ribs and muscle further increasing the risk for injury. In victims of inflicted injuries, diagnostic challenges may occur because of a delay in presentation or the absence of a reported mechanism [8-10]. Clinical prediction rules [11] and cer- tain patterns of recognizable injury such as the Waddell’s triad in pedes- trian-motor vehicle crashes or the seat-belt sign [7,12] in inappropriately restrained child passengers are helpful in diagnosing and managing children with certain types of blunt traumatic injuries. However, an accurate description of the mechanism of injury and being aware of the different injuries based on intent is also important in diagnosing and treating injuries in children who sustain blunt torso trauma.

Lane et al. described abusive abdominal trauma leads to more severe injuries, longer hospitalizations, higher charges, and higher mortality rates compared to those suffering accidental trauma injuries in a cohort of children less than nine years of age but called for studies to explain these differences [13]. Further, the authors were unable to describe the mechanisms of injuries in their cohort. The types of injuries to the chest, abdomen and pelvis and clinical outcomes of children who present to the ED with blunt torso trauma based on the mechanisms

http://dx.doi.org/10.1016/j.ajem.2017.05.053

0735-6757/(C) 2017

of injury and intent have not been described. We hypothesize that in pe- diatric blunt torso trauma that is severe enough to require treatment in the ED, children who sustain unintentional injuries have worse clinical outcomes than those with inflicted injuries. This is because children with unintentional injuries may be exposed to higher energy mecha- nisms such as motor vehicle crashes and falls than children with inflicted injuries. The aim of the study was to compare injuries, inter- ventions and clinical outcomes based on injury intent among children who present to the pediatric emergency department with blunt torso trauma.

Methods

We undertook a secondary analysis of de-identified data obtained from a prospective, multi-center observational cohort of children youn- ger than 18 years old, presenting within 24 h of blunt torso trauma to 20 participating EDs of the Pediatric Emergency Care Applied Research Network from 2007 to 2010 [11]. The full inclusion criteria are described in the study [11]. The exclusion criteria, based on the pri- mary study, were children in whose injury occurred N 24 h before pre- sentation, penetrating trauma, preexisting neurologic disorders precluding reliable examination, known pregnancy, transfer from an- other hospital with previous abdominal computed tomogram (CT) or Diagnostic peritoneal lavage.

The treating physician collected the patient history, physical exami- nation and injury variables on a structured data collection form. Imag- ing, including radiography and Computerized tomography and laboratory investigations were performed at the discretion of the treating physician. Data collection included demographic information, mechanism of injury, symptoms and physical examination (including abdominal and thoracic examination), laboratory and imaging informa- tion and outcomes (ED disposition, Need for hospitalization, need and anatomic location for surgery, 30-day mortality, and cause of death). thoracic injury was diagnosed based on the chest radiograph. Abdomi- nal and pelvic injuries were diagnosed based on CT. The decision to hos- pitalize the child was made by the treating physician. Subjects were classified based on injury intent (unintentional, inflicted or indetermi- nate) using the mechanism of injury and additional narrative informa- tion obtained from the electronic case report form (ECRF4) [14]. The injury intent was indeterminate when the mechanism was unknown and the narrative on the ECRF4 did not specify an intent. Intentional in- jury was coded as inflicted injury. The method of data collection of sub- jects, the investigations performed and the follow up procedures have been described elsewhere [11]. The study was granted exempt status by the Baylor College of Medicine Institutional Review Board (IRB).

Outcome measures and definitions

We defined torso injuries to those that included the chest, abdomen and pelvis. The outcomes investigated included ED disposition, need for hospitalization, need and anatomic location for surgery, 30-day mortal- ity, and cause of death. Intra-Abdominal injury (IAI) was defined as any radiographically or surgically apparent injury to the following struc- tures: spleen, liver, urinary tract (from the kidney to the urinary blad- der), gastrointestinal tract (including the bowel or associated mesentery from the stomach to the sigmoid colon), pancreas, gallblad- der, adrenal gland, intraabdominal vascular structure, or traumatic fas- cial defect (traumatic abdominal wall hernia). thoracic injuries consisted of pneumothorax, hemothorax, rib fracture, pulmonary con- tusion, pneumomediastinum, pneumopericardium and diaphragmatic injury. Additionally, pelvic injuries consisted of fractures to the pubis, ilium, sacrum, ischium, acetabulum and arterial bleeding from a pelvic fracture. Cases where the site investigator could not make a definitive determination were adjudicated by a 5-member study panel for final determination [11].

Primary data analysis

We compared demographic, physical examination, laboratory and clinical outcomes between inflicted and unintentional injuries. Descrip- tive comparisons were made using the Pearson Chi-Square test for cat- egorical variables or Fisher’s Exact test for variables with a cell value less than five. The Mann-Whitney test was utilized for non-normally contin- uous cofactors.

Unadjusted odds ratios were calculated for statistically signifi- cant variables between patients with inflicted and unintentional in- juries. Binary logistic regression modeling, using a backward-step approach was utilized to further adjust clinical associations. Factors that had a p-value b 0.20 were included in the regression model. The highest p-values were removed one-by-one until only variables with p-value(s) b 0.05 remained. Statistical significance was defined as a p-value b 0.05. All statistics were computed using the Statistical Package for the Social Sciences (SPSS) version 23 software (IBM Corp., Armonk, NY).

Results

The initial study cohort with a history blunt torso trauma included 12,044 subjects with the following injuries by intent: unintentional: 9563 (79.4%); inflicted: 720 (6%), indeterminate: 148 (1.2%) and miss- ing intent: 1613 (13.4%) (Fig. 1). The differences between children who had a recorded intent of injury and those with a missing intent of injury are shown in Table 2S in the Supplement. Patients with a known injury mechanism were twice as likely to be African-American

Fig. 1. Consort diagram of PECARN data showing victims of blunt torso injury by intent in analysis dataset. ATV: all-terrain vehicle; MVC: motor vehicle collision PECARN: Pediatric Emergency Care Applied Research Network.

(OR = 2.21 (95% CI: 1.95-2.51); p-value b 0.001) and have a pelvic fracture (OR = 2.02 (95% CI: 1.43-2.84); p-value b 0.001) and less likely to be male (OR = 0.48 (95% CI: 0.43-0.54)) than patients with a missing injury mechanism.

Demographics

After exclusion of children whose mechanism was indeterminate or missing, the study population was 10,283 children; of these, 9563 (93%) were unintentional and 720 (7%) were intentional (Table 1). Males con- stituted 59% of all cases. Baseline characteristics of the study subjects and mechanisms of injury by intent are described in Table 1. Children with inflicted trauma were significantly older and African-American than those who sustained unintentional trauma. There were no differ- ences in gender and ethnicity by group. The common mechanisms of unintentional torso trauma were occupants of motor vehicle involved in crashes, falls and pedestrian or bicyclist struck by a motor vehicle. The most common mechanism for inflicted injuries was an object strik- ing the abdomen.

Injuries

There were significantly more Intraabdominal injuries and pelvic fractures in the unintentional group compared to the inflicted group. There were no differences in the frequency and severity of injuries to the liver, spleen, kidney and gastro-intestinal tract based on intent. However, Pancreatic injuries more commonly occurred in inflicted torso trauma. Pelvic fractures commonly resulted from motor vehicle crashes and falls. There were no differences in thoracic injuries by injury intent though victims of unintentional trauma were intubated in the ED more often than the inflicted group. (Table 2).

Table 1 Demographics and injury mechanisms in pediatric victims of blunt torso trauma based on intent of injury (N = 10,283)

Outcomes

Several significant differences were noted regarding the clinical out- come of patients with blunt torso trauma based on intent (Table 3). A greater number of patients with unintentional injuries required hospi- talization and intensive care compared to those with inflicted trauma. A higher percentage required of orthopedic surgery compared to chil- dren with inflicted injuries. Children with inflicted injuries required neurosurgery more often. The overall 30-day mortality was similar be- tween groups. Death was most commonly due to Traumatic brain injury in both groups.

Significant unadjusted associations between the two groups includ- ed age, race (African-American), the presence of pelvic fractures, hospi- talization and need for non-abdominal surgery with pelvic fractures having the greatest effect estimate (Table 4). Patients with an intention- al injury were 96% less likely to have a pelvic fracture than patients with an acciDental injury after adjusting for age and race (OR = 0.04 (95% CI 0.01-0.26); p-value = 0.001). (Table 5).

No differences in the rate of thoracic, intra-abdominal or pelvic inju- ries and the 30-day mortality rates based on intent were observed in pa- tients who were discharged from the ED. (Results not shown).

Discussion

In our study, inflicted trauma accounted for 7% of children who presented to the ED with blunt torso trauma. This rate is similar to that reported by the National Trauma Data Bank for children

[15] where assault-related injuries accounted for 8.6% of children up to 21 years of age who were treated at US pediatric trauma centers. Our results are also in agreement with Lane et al. who reported that abusive abdominal trauma accounted for 4.8% -5.9% of all abdominal trauma hospitalizations in the US [16]. We observed that in our cohort, children who sustained inflicted torso trauma were significantly older and minority compared to those who were injured unintentionally. Peer related violence is common during adolescence [5].

We also observed that in children who presented to the ED with

Inflicted 720 (7%) N (%)

Median age in years (IQR) 14.0 (8.0, 16.0)

Unintentional 9563 (93%)

N (%)

10.0 (5.0,

14.0)

p-Value

b0.001

blunt torso trauma, unintentional causes such as motor vehicle crashes (MVC) and falls contributed to a greater number of intra-ab- dominal injuries and pelvic fractures than children with inflicted injuries. Falls are commonest in children under age 6 years; whereas, adolescents are most often injured in MVCs or by being struck by or against an object [15]. In addition, in our cohort the rate and severity of injuries to the abdominal solid organs and viscera did not differ based on intent except for pancreatic injuries which more commonly resulted from inflicted trauma in agreement with Lane [13,16]. The mechanism for the latter is reported as being caused by a direct epigastric blow that compresses the organ against the vertebral column together with a thinner abdominal musculature in children [7]. Furthermore, no differences in thoracic injuries were observed based on intent.

The higher propensity of pediatric victims of unintentional blunt torso trauma in our study who required admission and intensive care likely reflects their more severe injury mechanisms. This is suggested by the increased need for them to be intubated in the ED. Since our data did not allow the calculation of the Injury Severity Score, we were unable to stratify patients on the severity of their injuries. Mortality rates were similar among the groups, with traumatic brain injury being the most likely cause of death. Our results differ from Lane [13,16] who observed that abusive abdominal trauma had higher mortality rates (9% vs. 3.4%) compared to unintentional trauma among hospitalized children aged 0-9 years. The possible explanations are that our cohort included patients who were presented to the ED whereas Lane et al. only investigated

Gender Female	271 (37.6)	0.056 3947 (41.3)
Male	449 (62.4)	5616 (58.7)
Race	b0.001
US Indian or Alaska	5 (0.8)	71 (0.8)
Asian	8 (1.2)	195 (2.2)
African American	389 (58.9)	3025 (34.5)
Nat. Hawaiian/Pacific	0 (0.0)	32 (0.4)
White	223 (33.8)	5059 (57.7)
Other	35 (5.3)	382 (4.4)
Missing	60	799
Ethnicity			0.12
Hispanic	92 (17.3)	1061 (14.8)
Non-Hispanic	439 (82.7)	6089 (85.2)
Missing	189	2413
Mechanism of injury			b0.001
Occupant in MVC	1 (0.1)	3741 (39.1)
Fall from elevation	13 (1.8)	2093 (21.9)
Fall down stairs	3 (0.4)	263 (2.8)
Pedestrian/bicyclist struck by motor	1 (0.1)	2279 (23.8)
vehicle
Bicycle collision or fall from riding	0 (0.0)	24 (0.3)
Motorcycle/ATVa/motorized scooter	0 (0.0)	86 (0.9)
collision
Object struck abdomen	649 (90.1)	829 (8.7)
Unknown mechanism	33 (4.6)	15 (0.2)
Other mechanism of injury	20 (2.8)	233 (2.4)

ATV = all-terrain vehicle; IQR = interquartile range; MVC = motor vehicle crash; Nat = native; US = United States.

patients who were admitted to the hospital for abdominal trauma. Additionally, there could be differences in the severity of co-existing

Table 2 Comparison of children with blunt torso trauma based on intent (N = 10,283)
	Inflicted	Unintentional	p-Value
	720 (7.0%)	9563 (93.0%)
	N (%)	N (%)
Intraabdominal injurya,f			b0.001
Yes	17 (2.4)	603 (6.3)
No	703 (97.7)	8960 (93.7)
First AST N 200 IU/L			0.01
No	416 (96.5)	5369 (93.1)
Yes	15 (3.5)	395 (6.9)
Not done	289	3799
First ALT N 125 IU/L			0.68
No	385 (93.2)	5090 (93.7)
Yes	28 (6.8)	340 (6.3)
Not done Liver injury by CTb	307	4133	0.72
I	4 (44.4)	56 (25.2)
II	2 (22.2)	65 (29.3)
III	1 (11.1)	54 (24.3)
IV	2 (22.2)	33 (14.9)
V	0 (0.0)	13 (5.9)
VI Spleen injury by CTb	0 (0.0)	1 (0.5)	0.30
I	0 (0.0)	54 (22.3)
II	2 (50.0)	59 (24.4)
III	0 (0.0)	67 (27.7)
IV	2 (50.0)	50 (20.7)
V	0 (0.0)	12 (5.0)
VI Kidney injury by CTb	0 (0.0)	0 (0.0)	0.80
I	1 (50.0)	44 (36.4)
II	0 (0.0)	26 (21.5)
III	1 (50.0)	26 (21.5)
IV	0 (0.0)	18 (14.9)
V	0 (0.0)	7 (5.8)
VI	0 (0.0)	0 (0.0)
GI tract injury			0.32
Yes	4 (23.5)	93 (15.4)
No	13 (76.5)	510 (84.6)
Pancreatic injury			b0.001
Yes	5 (29.4)	33 (5.5)
No Intubated in EDf	12 (70.6)	570 (94.5)	0.045
Yes	21 (2.9)	430 (4.5)
No Thoracic trauma on examf,c	698 (97.1)	9104 (95.5)	0.11
Yes	134 (18.7)	1560 (16.4)
No	583 (81.3)	7950 (83.6)
Chest radiograph done in ED			b0.001
Yes	384 (53.3)	6513 (68.1)
No Trauma on chest radiographd	336 (46.7)	3050 (31.9)	0.2
Present	16 (4.2)	373 (5.7)
Absent	368 (95.8)	6140 (94.3)
Pelvic radiograph performed			b0.001
Yes	158 (21.9)	4233 (44.3)
No Fracture pelvis on radiographe	562 (78.1)	5330 (55.7)	0.0015c
Yes	1 (0.6)	276 (6.5)
No Pelvic fracture diagnosed at anytime	157 (99.4)	3957 (93.5)	b0.001a
Yes	1 (2.8)	414 (31.4)
No	35 (97.2)	903 (68.6)
Pelvic fracture MOI			b0.001
Occupant in MVC	0 (0.0)	172 (41.5)
Fall from elevation	0 (0.0)	40 (9.7)
Pedestrian/bicyclist struck by moving vehicle	0 (0.0)	181 (43.7)
Bicycle collision or fall from riding	0 (0.0)	1 (0.2)
Motorcycle/ATV/motorized scooter collision	0 (0.0)	3 (0.7)
Object struck abdomen	1 (100.0)	12 (2.9)
Other MOI	0 (0.0)	5 (1.2)

Note: ALT = alanine aminotransferase; AST = aspartate aminotransferase; ATV = all-terrain vehicle; ED = emergency department; IU/L = international units per liter; MOI = mecha- nism of injury; MVC = motor vehicle collision.

^a Diagnosed in Emergency Department (ED) or Hospital (any method).

^b CT Grade: Injury Grade based on American Association for the Surgery of Trauma (AAST) Classification.

^c Includes erythema, abrasion, ecchymosis, subcutaneous air, laceration, other.

^d Includes pneumothorax, hemothorax, rib fracture, pulmonary contusion, pneumomediastinum, pneumopericardium, diaphragmatic injury.

^e Includes fractures to the pubis, ilium, sacrum, ischium, acetabulum and arterial bleeding from a pelvic fracture.

^f Cells may not add up because missing values are not shown.

traumatic brain injury [17,18], differing traumatic brain injury man- agement at non-pediatric community hospitals or delays in care. Multisystem trauma occurs with both unintentional and inflicted in- juries. In these situations, head injury contributes highest to mortal- ity. Rib fractures and pelvic fractures are also considered markers for severe injuries [19]. case fatality is highest in collisions between pe- destrian and motor vehicles and occupants in MVCs [15].

About half of the children who sustained acute blunt torso trauma were hospitalized. Children with unintentional torso trauma more commonly required non-abdominal surgery (orthopedic) compared to those with inflicted torso trauma (neurosurgical). There are two possible explanations. First, this could reflect the preponderance of limb injuries in unintentional torso injuries. Second, though liver [13,16,20] and splenic injuries are the most common types of injuries encountered in blunt abdominal trauma in children, clinically stable children with isolated grades I through IV liver and spleen injuries receive non-operative management [21] regardless of intent. Blood transfusion, while uncommon, is the most frequent therapeutic intervention; angiographic embolization and laparotomy are uncommon [22]. Around 6% of patients in our cohort had intraabdominal injury that required acute interventions. Landau

[23] reported only 4% of pediatric liver injuries due to blunt trauma require laparotomy. Though the mechanisms were similar in both studies, Landau et al. reported a preponderance of pedestrian related

Table 3

Comparison of clinical outcome of children with blunt torso trauma based on intent (N

= 10,283)

injuries, whereas the mechanism of injuries in our dataset was dis- tributed between motor vehicle crashes, pedestrian injuries and falls. Rates for surgical interventions for bowel, mesenteric and pan- creatic injuries due to abusive abdominal trauma can approach 26% [24].

In our study, we observed that the risk of pelvic fractures children who sustained inflicted torso trauma was 96% less than the unintentional group after controlling for age and race. Minor pelvic fractures such as those due to avulsion or iliac wing fractures are commonly sports-related. More severe pelvic fractures are sustained in motor vehicle crashes [25]. In children with multiple blunt trauma, those without pain or abnormal pelvis or hip examination and those without a femur deformity, hematuria, abdominal pain or tenderness, Glasgow Coma Score of 13 or lower, or hemodynamic instability have a low-risk for pelvic fracture (b 0.5% risk rate) [26]. Mortality of Pediatric trauma patients with pelvic fractures is correlated with the severity of the fractures often due to associated pelvic organ injury, but is not associated with the presence of extra-pelvic abdominal injuries, except for kidney injuries [27].

Our results differ from Lane [13] and Trokel [28] in that we observed that unintentional pediatric torso trauma leads to a significantly higher risk of pelvic fractures compared to those due to inflicted causes and that these children are more often hospitalized and require intensive care. The reasons for this are that our cohort included all children where adolescent peer violence and unintentional injuries from MVCs and falls are more common. Lane utilized an administrative dataset de- veloped by the Healthcare Cost and Utilization Project (HCUP), which is a Representative sample of US hospitals. Hospital admission as an out-

Inflicted 720 (7.0%) N (%)

Unintentional 9563 (93.0%) N (%)

p-Value

come can be influenced by factors that are independent of injury such as bed availability, access to care and professional variations in practice which can all lead to potential bias [29]. In addition, bias may occur be-

ED disposition b0.001

Home 483 (67.1) 4804 (50.2)

AMA 3 (0.4) 6 (0.1)

ED Death	0 (0.0)	1 (0.0)
Admit general patient	138 (19.2)	2807 (29.4)

Admit short stay 38 (5.3) 560 (5.9)

OR 5 (0.7) 276 (2.9)

ICU 43 (6.0) 1079 (11.3)

Transferred 6 (0.8) 24 (0.3)

Other 4 (0.6) 6 (0.1)

Missing 0 0

Patient hospitalized b0.001

Yes	230 (31.9)	4750 (49.7)	lapping injuries to the chest, abdomen and pelvis or multi-system
No	490 (68.1)	4813 (50.3)	trauma. Our data consisted of detailed trauma information and outcome

cause the data cannot differentiate repeat hospitalizations for the same injury (which can happen with victims of child maltreatment), and does not include children who died prior to reaching a hospital, or those who died in a hospital ED. Further, because their patients were chosen from all hospitals in the US (including community hospitals), it is possible that a greater proportion of their patients may have been less severely injured compared to our cohort.

The strengths of our study is that we studied blunt torso trauma as opposed to the other studies which have only studied abdominal inju- ries [13,16,28]. Young children who have blunt trauma may have over-

Non-abdominal surgery b0.001

Yes	28 (4.2)	1358 (15.6)
No	631 (95.8)	7360 (84.4)
Missing	61	845

Location of first surgery 0.03

Orthopedic 9 (45.0) 666 (68.4)

Facial 3 (15.0) 128 (13.1)

Neurosurgical 7 (35.0) 123 (12.6)

Ophthalmologic 0 (0.0) 5 (0.5)

Vascular 0 (0.0) 14 (1.4)

Urological 1 (5.0) 8 (0.8)

Other 0 (0.0) 30 (3.1)

Missing 8 384

30-day mortality 0.59

beginning from ED presentation to hospital discharge including the mechanism of injury. Trauma registry data is considered the best source of in-depth data based on the anatomic nature and physiologic conse- quences of injuries [30,31]. Further, our study utilized rigorous methods to obtain patient outcomes after discharge by using telephone calls,

Table 4 Unadjusted odds ratios for significantly selected Demographic and clinical factors between unintentional and inflicted injuries (unintentional = 0; inflicted = 1)

Factor OR 95% CI p-Value

Yes	7 (1.0)	75 (0.8)		Age (continuous)	1.06	1.04-1.08	b 0.001
No	713 (99.0)	9488 (99.2)		Gender (male)	1.16	1.00-1.36	0.06
Cause of death			0.97	Race (Black vs. all others)	2.72	2.32-3.20	b 0.001
intra-abdominal injury	0 (0.0)	1 (1.3)		Non-abdominal surgery (yes)	0.24	0.16-0.35	b 0.001
Multiple blunt force	1 (14.3)	10 (13.3)		Patient hospitalized (yes)	0.48	0.41-0.56	b 0.001
Traumatic brain injury	5 (71.4)	57 (76.0)		Pelvis XR results (yes)	0.09	0.01-0.66	0.02
Thoracic injury	0 (0.0)	1 (1.3)		Abdomen/pelvis CT results (abnormal)	0.41	0.30-0.55	b 0.001
Other	1 (14.3)	6 (8.0)		Pelvic fracture (yes)	0.03	0.004-0.22	0.001

AMA: Against medical advice; ED: emergency department; ICU: intensive care unit; OR: operating room.

Chest XR findings (trauma) 0.72 0.43-1.19 0.20

CI: confidence interval; CT: computed tomography; OR: odds ratio; XR: radiograph.

Table 5 Adjusted logistic regression model for pelvic fractures between intentional and accidental injuries in children 0-17 (unintentional = 0; inflicted = 1; n = 9424)

Factor	OR	95% CI	p-Value
Pelvic fracture (yes)	0.04	0.01-0.26	0.001
Age (continuous)	1.07	1.05-1.08	b0.001
Race (African American vs. all other races)	2.69	2.29-3.16	b0.001

Note: CI = confidence interval; OR = odds ratio.

mailing letters and reviewing medical and local morgue records to min- imize missing outcomes [11]. Finally, our data are more recent than the other two studies and we selected children of all ages.

There are several limitations. First, 13% of cases had a missing intent of injury. Additionally, some cases may have been misclassified and we were unable to determine whether injures were self-inflicted. However the most common mechanism of self-inflicted injury (SII) in adoles- cents is cut/pierce and not blunt trauma and SII constitute a small per- centage of visits to the ED [32]. Second, the history in child abuse victims may be vague and symptoms may not point towards abdominal injury which leads to corresponding delays in presentation [8-10]. Also, screening for occult abdominal trauma is not a routine practice in hospi- tals and cases of occult abdominal trauma may be missed [10]. These may reduce the likelihood of detection of intra-abdominal injury and contribute to underascertainment of victims of child abuse in our co- hort. Wood [20] found no significant differences in the proportions of abused and non-abused children with delays in seeking care. The group of children in whom delays in care were witnessed was small and involved a single center [33]. Moreover, delay in seeking care is not specific for inflicted injury and occurs in some children with low-ve- locity accidental abdominal trauma [20]. Third, under-ascertainment of cases may have resulted from us not including transfers from other hos- pitals. However, we believe that our study identified most cases of sig- nificant pediatric abdominal trauma which presented to the ED because we included patients based on the mechanism of trauma, the presence of blunt torso trauma in association with suspected head inju- ry or extremity injury and where there existed a physician concern for abdominal trauma based on the ED physician performing diagnostic or Screening tests. Fourth, we analyzed data of children who presented to the EDs of 20 children’s hospitals nationwide which may limit the gen- eralizability of our results to community hospitals. Nevertheless, the re- sults are important for emergency physicians, trauma surgeons and child protection teams working at US children’s hospitals. Children’s hospitals account for 5% of all U.S. hospitals and 11% of inpatient admis- sions to children’s hospitals are transfers from other community hospi- tals [34]. Fifth, we did not obtain socio-economic data and information about rural or urban injury burdens. Finally, the need for hospitalization was at the discretion of the treating physician and could vary by site. However, it is likely that this practice variation would be less in pediat- ric trauma hospitals compared to all US hospitals.

Conclusions

Children who sustain acute blunt torso trauma due to unintentional causes have a significantly higher risk of pelvic fractures and are more likely to be hospitalized compared to those with inflicted injuries.

Sources of support

None.

Acknowledgement

The authors wish to acknowledge the Pediatric Emergency Care Applied Research Network for making the dataset available to the authors for analysis.

References

1. Centers for Disease Control and Prevention, National Center for Injury Prevention and Control, 2016. Web-based Injury Statistics Query and Reporting System (WISQARS) [online]. leading causes of death reports, national and regional, 1999- 2014 fatal injury reports. [Available from:] http://webappa.cdc.gov/cgi-bin/broker. exe. [Accessed on October 22, 2016].
2. Centers for Disease Control and Prevention, National Center for Injury Prevention and Control, 2016. Web-based Injury Statistics Query and Reporting System (WISQARS) [online]. http://webappa.cdc.gov/sasweb/ncipc/nfirates2001.html. [Accessed on October 22, 2016].
3. 10 leading causes of death by age group highlighting violence-related injury deaths, United States. [Available from:] http://www.cdc.gov/injury/images/lc-charts/lead- ing_causes_of_injury_deaths_violence_2014_1040w760h.gif; 2014. [Accessed on October 22, 2016].
4. Sedlak AJ, Mettenburg J, Basena M, Petta I, McPherson K, Greene A, et al. Fourth national incidence study of child abuse and neglect (NIS-4): report to congress, executive summary. Washington, DC: U.S. Dept. of Health and Human Services, Administration for Children and Families; 2010.
5. Understanding Youth violence fact sheet. [Available from:] https://www.cdc. gov/violenceprevention/pdf/yv-factsheet-a.pdf; 2015. [Accessed on October 22, 2016].
6. Centers for Disease Control (CDC), 2016. Injury prevention and control: division of Violence prevention. [Available from:] https://www.cdc.gov/violenceprevention/ childmaltreatment/index.html. [Accessed on October 22, 2016].
7. Pediatric trauma in advanced trauma life support for doctors atls. 8th ed. Chicago, IL: American College of Surgeons Committee on Trauma; 2008 238-9.
8. Lindberg D, Makaroff K, Harper N, Laskey A, Bechtel K, Deye K, et al. Utility of hepatic transaminases to recognize abuse in children. Pediatrics 2009;124: 509-16.
9. Coant PN, Kornberg AE, Brody AS, Edwards-Holmes K. Markers for occult liver injury in cases of Physical abuse in children. Pediatrics 1992;89:274-8.
10. Lane WG, Dubowitz H, Langenberg P. Screening for occult abdominal trauma among children with suspected physical abuse. Pediatrics 2009;124: 1595-602.
11. Holmes JF, Lillis K, Monroe M, et al. Identifying children at Very low risk of clinically important blunt abdominal injuries. Ann Emerg Med 2013;62:107-16.
12. Borgialli DA, Ellison AM, Ehrlich P, et al. Association between the seat belt sign and Intra-abdominal injuries in children with blunt torso trauma in motor vehicle collisions. Acad Emerg Med 2014 Nov;21(11):1240-8.
13. Lane WG, Lotwin I, Dubowitz H, Langenberg P, Dischinger P. Outcomes for children hospitalized with abusive versus noninflicted abdominal trauma. Pediatrics 2011; 127(6):e1400-5.
14. Identifying children at very low risk of clinically important blunt abdominal injuries

    – annotated ECRF for public use datasets. [Available from:] http://www.pecarn.org/ studyDatasets/IntraabdominalTrauma.html. [Accessed on December 10, 2015].

    National Trauma Data Bank. Pediatric annual report. American College of Surgeons; 2015[Available from:] https://www.facs.org/~/media/files/quality%20programs/ trauma/ntdb/ntdb%20pediatric%20annual%20report%202015%20final.ashx[Accessed on October 22, 2016].

15. Lane WG, Dubowitz H, Langenberg P, Dischinger P. Epidemiology of abusive abdom- inal trauma hospitalizations in United States children. Child Abuse Negl 2012 Feb; 36(2):142-8.
16. Keenan HT, Runyan DK, Marshall SW, Nocera MA, Merten DF. A population- based comparison of clinical and outcome characteristics of young children with serious inflicted and noninflicted traumatic brain injury. Pediatrics 2004; 114(3):633-9.
17. Sills MR, Libby AM, Orton HD. Prehospital and in-hospital mortality: a comparison of intentional and unintentional Traumatic brain injuries in Colorado children. Arch Pediatr Adolesc Med 2005;159(7):665-70.
18. Naqvi G, Johansson G, Yip G, Rehm A, Carrothers A, Stohr K. Mechanisms, patterns and outcomes of paediatric polytrauma in a UK major trauma centre. Ann R Coll Surg Engl 2016 Aug;4:1-7.
19. Wood J, Rubin DM, Nance ML, Christian CW. Distinguishing inflicted versus accidental abdominal injuries in young children. J Trauma 2005;59(5): 1203-8.
20. Stylianos S. Evidence-based guidelines for resource utilization in children with iso- lated spleen or liver injury. J Pediatr Surg 2000;35:164.
21. Wisner DH, Kuppermann N, Cooper A, Menaker J, Ehrlich P, Kooistra J, et al. Manage- ment of children with solid Organ injuries after blunt torso trauma. J Trauma Acute Care Surg 2015 Aug;79(2):206-14.
22. Landau A, van As AB, Numanoglu A, Millar AJW, Rode H. Liver injuries in children: the role of selective non-operative management injury. Int J Care Injured 2006;37: 66-71.
23. Hilmes MA, Hernanz-Schulman M, Greeley CS, Piercey LM, Yu C, Kan JH. CT identifi- cation of abdominal injuries in abused pre-school-age children. Pediatr Radiol 2011; 41:643-51.
24. Ortega HW, Reid S, Velden HV, Truong W, Laine J, Weber L, et al. Patterns of injury and management of children with pelvic fractures at a non-trauma center. J Emerg Med 2014 Aug;47(2):140-6.
25. Haasz M, Simone LA, Wales PW, Stimec J, Stephens D, Beno S, et al. Which pediatric blunt trauma patients do not require pelvic imaging? J Trauma Acute Care Surg 2015 Nov;79(5):828-32.
26. Swaid F, Peleg K, Alfici R, Olsha O, Givon A, Kessel B. Does severity of pelvic fractures correlate with the incidence of associated intra-abdominal injuries in children? Eur J Pediatr Surg 2016 Jun;26(3):287-90.
27. Trokel M, Discala C, Terrin NC. Sege Rd. patient and injury characteristics in abusive abdominal injuries. Pediatr Emerg Care 2006;22(10):700-4.
28. Cryer C. Severity of injury measures and descriptive epidemiology. Inj Prev 2006; 12(2):67-8.
29. Rivara FP. Injury control: a guide to research and program evaluation. New York: Cambridge University Press; 2001.
30. Wynn A, Wise M, Wright MJ, et al. Accuracy of administrative and trauma registry databases. J Trauma 2001;51(3):464-8.
31. Cutler GJ, Flood A, Dreyfus J, Ortega HW, Kharbanda AB. Emergency department visits for self-inflicted injuries in adolescents. Pediatrics 2015;138:28-34.
32. Canty Sr TG, Canty Jr TG, Brown C. Injuries of the gastrointestinal tract from blunt

    trauma in children: a 12-year experience at a designated pediatric trauma center. J Trauma 1999;46(2):234-40.

    Children’s Hospital Association. [Available from:] https://www.childrenshospitals. org/About-Us/About-Childrens-Hospitals. [Accessed on February 9, 2016].