Objectives. 1) To determine whether clinical signs of brain injury are sensitive indicators of intracranial injury (ICI) in head-injured infants. 2) To determine whether radiographic imaging of otherwise asymptomatic infants with scalp hematoma is a useful means of detecting cases of ICI. 3) To determine whether head-injured infants without signs of brain injury or scalp hematoma may be safely managed without radiographic imaging.
Methods. We performed a 1-year prospective study of all infants younger than 2 years of age presenting to a pediatric emergency department with head trauma. Data were collected on historical features, physical findings, radiographic findings, and hospital course. Follow-up telephone calls were made 2 weeks after discharge to assess for any late deterioration.
Results. Of 608 study subjects, 30 (5%) had ICI; 12/92 (13%) infants 0 to 2 months of age had ICI, compared with 13/224 (6%) infants 3 to 11 months of age, and 5/292 (2%) infants 12 months of age or older.
Only 16/30 (52%) subjects with ICI had at least one of the following clinical symptoms or signs of brain injury: loss of consciousness, history of behavior change, seizures, emesis, depressed mental status, irritability, bulging fontanel, focal neurologic findings, or vital signs indicating increased intracranial pressure.
Of the 14 asymptomatic subjects with ICI, 13 (93%) had significant scalp hematoma. Among subjects who had head computed tomography, significant scalp hematoma had an odds ratio of 2.78 (95% confidence interval: 1.15,6.70) for association with ICI.
A total of 265 subjects (43%) were asymptomatic and had no significant scalp hematoma. None (95% confidence interval: 0,1.2%) required specific therapy or had any subsequent clinical deterioration.
Conclusions. Clinical signs of brain injury are insensitive indicators of ICI in infants. A substantial fraction of infants with ICI will be detected through radiographic imaging of otherwise asymptomatic infants with significant scalp hematomas. Asymptomatic infants older than 3 months of age who have no significant scalp hematoma may be safely managed without radiographic imaging.
- ICI =
- intracranial injury •
- SF =
- skull fracture •
- ED =
- emergency department •
- CT =
- computed tomography •
- CI =
- confidence interval
Traditionally, authors have recommended a conservative approach to the diagnostic evaluation of head-injured infants, arguing that infants are at increased risk of intracranial injury (ICI) and that symptoms or signs of brain injury may not be reliably present in those with ICI.1–4 A number of previous studies have reported that a significant fraction of ICIs in infants occur in patients with a normal neurologic status and with no symptoms or signs of brain injury.2 ,4 These studies have suggested that skull fracture (SF), or abnormalities of scalp examination associated with underlying SF, may be a marker for ICI in otherwise asymptomatic infants.2 ,4 Past studies have been limited, however, by small numbers of patients or by a retrospective study design, which has led to questions about the completeness and accuracy of documentation of symptoms and signs.
We undertook this study to collect prospective data on the historical features and physical findings of head-injured infants, so that we could determine better how well clinical features predict ICI in infants. Our goal was to determine which head-injured infants require radiographic imaging and whether there are low-risk clinical criteria that describe infants who do not require radiographic imaging after head trauma. Our study hypotheses were 1) clinical symptoms and signs of brain injury are insensitive indicators of ICI in infants; 2) a substantial fraction of ICIs will be diagnosed in otherwise asymptomatic infants who have abnormalities on physical examination of the scalp, indicating possible underlying SF; and 3) infants who are asymptomatic and who have no scalp abnormalities may be safely managed without radiographic imaging.
The study was performed in the emergency department (ED) of Children's Hospital, Boston, a tertiary care pediatric hospital. The ED receives ∼50 000 patients per year and is a certified level I Pediatric Trauma Center. Attending physicians, who are specialists in pediatric emergency medicine, directly provide or supervise all patient care.
All patients younger than 2 years of age seen in the ED between March 11, 1997 and March 10, 1998, with a complaint or a diagnosis of head trauma were included in the study.
Treating physicians were asked to complete a data sheet at the time of the initial evaluation, before obtaining any radiographic studies. The data sheet included information about mechanism of injury, symptoms reported by parents, and signs of head injury or neurologic abnormalities on physical examination.
The ED registration log was reviewed each day by a research assistant to ensure that all eligible patients were enrolled. When an eligible patient was noted not to be enrolled, the treating clinician was contacted immediately and data were collected at that time.
Definitions of Study Variables
Subjects were considered to be symptomatic if they had any of the following symptoms or signs of possible brain injury: history of loss of consciousness; history of lethargy or irritability; seizures; two or more episodes of emesis; irritability, or depressed mental status on physical examination; bulging fontanel; abnormal vital signs indicating possible increased intracranial pressure; or focal neurologic findings. Mental status was rated by the treating physicians in one of the following four categories: 1) normal mental status (“alert and interactive”); 2) mildly depressed mental status (“awake but quiet” or “asleep but arouses to voice or light touch”); 3) moderately depressed mental status (“asleep but arouses to vigorous tactile stimuli”); or 4) severely depressed mental status (“arousable to painful stimuli” or “unarousable”).
Treating physicians rated scalp hematomas as small (“barely perceptible”), moderate, or large (“obvious swelling and/or boggy consistency”). Subjects were considered to have significant scalp hematoma if they were younger than 1 year of age with any scalp hematoma or older than 1 year of age with a moderate or large scalp hematoma. Subjects not meeting these criteria were considered to have no significant scalp hematoma.
Subjects who were asymptomatic and who had no significant scalp hematoma were considered to meet our low-risk criteria.
Head computed tomography (CT) and/or skull radiography was performed at the discretion of the treating physician. In making their decisions about imaging, physicians were encouraged to follow guidelines, which previously had been developed in our ED, for the imaging of head-injured infants. These guidelines recommend 1) head CT as the initial study for any infant who is symptomatic; 2) skull radiography as the initial study for any asymptomatic patient who has a significant scalp hematoma; and 3) head CT for any patient found to have an SF on initial skull radiographs. In addition, physicians were encouraged to have a low threshold for radiographic imaging in any infant younger than 1 year of age with a nontrivial mechanism of injury.
Results from all radiographic studies were recorded. The official report, dictated by the attending radiologist, was considered to be the definitive interpretation of the study. All subjects with any acute intracranial hematoma, cerebral contusion, and/or diffuse brain swelling evident on head CT were considered to have ICI.
For all study subjects, the need for any interventions in the ED or during subsequent hospital admission was determined from a review of the medical record. For all patients admitted, any clinical deterioration and the condition on discharge were recorded. Follow-up telephone calls were made to all study subjects 2 weeks after the initial ED visit.
For all statistical calculations, the Stata (Stata Corp, College Station, TX) statistical package for the Macintosh computer was used.
χ2 Testing was performed and odds ratios were calculated for univariate analysis of the relationship between dichotomous clinical predictor variables and the outcome variables of interest. The unpaired t test was used for univariate analysis of the relationship between continuous clinical predictor variables and the outcome variables of interest. Ninety-five percent confidence intervals (CI) were calculated for rates and proportions and are presented where relevant for frequency data.
The study was approved by the Children's Hospital institutional review board. Because our study consisted solely of data collection, with no alteration in clinical practice and no risk for dissemination of confidential information, informed consent was not required.
The Study Sample
A total of 608 study subjects were enrolled, including 344 (57%) males, and 264 (43%) females. The mean age was 11.2 ± 6.8 months.
Falls were reported for 508 (84%) of the study subjects, including 266 (44%) who reportedly fell directly to the ground from a height of <3 feet, 169 (28%) who fell ≥3 feet, and 73 (12%) who fell down stairs. Of the 73 stair falls, 26 involved a walker.
Sixty-six (11%) subjects reportedly suffered blows to the head, and 17 (3%) were reported to have been passengers in a motor vehicle collision. For 17 (3%) subjects, the caretakers reported no history of trauma.
Of the 608 subjects, 177 (29%) were symptomatic and 431 (71%) were asymptomatic. Of these 431 asymptomatic subjects, 166 (39%) had significant scalp hematomas and 265 (61%) had no significant scalp hematomas. These 265 subjects (44% of our entire study sample) met our low-risk criteria.
Rates of Diagnostic Imaging
Of the 608 subjects, 188 (31%) underwent CT, 122 (20%) underwent skull radiography but no CT, and 298 (49%) underwent no imaging studies. The rates of imaging for symptomatic subjects and for asymptomatic subjects, with or without significant scalp hematomas, are shown in Fig 1.
As shown in Fig 1, 73 (41%) symptomatic subjects underwent no head CT. Of these 73 subjects, 55 (75%) had only one symptom or sign. In 8 of the subjects, vomiting was the only symptom. In another 19, parental report describing the child as “alert but less active than usual” or “crying more than usual” was the only symptom or sign.
All 21 children whose mental status was described as “asleep but arouses to voice or light touch” or worse underwent head CT.
Results of Diagnostic Imaging
Of the 608 study subjects, 30 (5%) were diagnosed with an ICI and 63 (10%) were diagnosed with an isolated SF, with no associated ICI. A total of 515 (85%) had no diagnosis of ICI or SF.
Types of ICI diagnosed included subdural hematoma (13 cases), epidural hematoma (12), cerebral contusion (7), diffuse brain swelling (1), and subarachnoid hemorrhage (1). Some subjects had more than one type of ICI. Of the 30 subjects with ICI, 23 (77%) also were diagnosed with SF.
Fifteen subjects with SF had at least some depression of the fracture fragments, including 7 of the 63 subjects with isolated SF.
Relationship of Age to ICI
There was a clear relationship between age and ICI, with younger patients being more likely to have ICI (Fig 2). The mean age of subjects who had ICI was 6.4 ± 6.1 months, compared with 11.4 ± 6.8 months for those subjects without ICI (P < .001). Of 92 infants 0 to 2 months of age, 12 (13%) had ICI, compared with 13/224 (6%) infants 3 to 11 months of age (P = .03) and 5/292 (2%) infants 12 months of age or older (P < .001).
When only the 188 subjects who underwent head CT were considered, there still was a relationship between young age and increased risk of ICI. Among this subgroup, 12/52 (23%) subjects 0 to 2 months of age had ICI, compared with 13/76 (17%) subjects 3 to 11 months of age (P = .40), and 5/60 (8%) subjects 12 months of age or older (P = .03).
Relationship of Mechanism of Injury to ICI
Four (24%) of 17 subjects with no reported history of trauma had ICI. All 4 of these subjects were suspected victims of child abuse. Patients with no reported history of trauma were significantly more likely to have ICI than were subjects with any other mechanism of injury (P < .05 for all comparisons).
Fifteen (9%) of the 169 subjects who fell ≥3 feet had ICI, compared with 5 (2%) of the 266 subjects who fell <3 feet (P< .05). Five (7%) of 73 subjects who fell down stairs suffered ICI (P < .05 for comparison with falls <3 feet).
None of the 17 subjects who were passengers in motor vehicle collisions and 1 (2%) of 66 subjects who suffered direct blows to the head had ICI.
When the subgroup of 188 subjects who underwent head CT were analyzed separately, there was no statistically significant relationship between mechanism of injury and likelihood of ICI.
Relationship of Clinical Signs of Brain Injury to ICI
Table 1 shows the frequency of clinical signs and symptoms of brain injury among patients with ICI. Odds ratios are calculated to represent the odds of ICI given the presence of each of the clinical signs and symptoms of interest. Two odds ratios are shown for each study variable: one is calculated from the sample of all subjects and the other from the subset of subjects who underwent head CT.
Among the group of all study subjects, history of lethargy or irritability, depressed mental status on examination, irritability on examination, full or bulging anterior fontanel, and abnormal vital signs indicating increased intracranial pressure were all associated with ICI (P < .05). Loss of consciousness, seizures, and vomiting were not associated with ICI (P = not significant). No significant association was found between focal neurologic findings and ICI, but it is important to note that only one subject in the study had focal findings.
Among the subgroup of study subjects who underwent head CT, only history of lethargy and bulging fontanel were found to be associated with ICI (P < .05).
All the clinical symptoms or signs of brain injury were insensitive indicators of ICI in this population. Only 16 (52%) of the 30 subjects with ICI had at least one of these symptoms or signs of brain injury and were considered symptomatic; the other 14 (48%) were asymptomatic. The frequency of symptomatic and asymptomatic cases of ICI by age is presented in Fig 2.
Among the 14 asymptomatic cases of ICI, there were 8 cases of epidural hematoma, 3 cases of subdural hematoma, 2 cases of cerebral contusion, and 1 case of subarachnoid hemorrhage.
Relationship of Significant Scalp Hematomas to ICI
Significant scalp hematomas were a more sensitive indicator of ICI than were the symptoms or signs of brain injury discussed above. Significant scalp hematomas were noted in 23/30 (77%) subjects with ICI.
Among all study subjects, significant scalp hematoma had an odds ratio of 4.65 (95% CI: 2.00,10.79) for association with ICI (P < .001). Among those subjects who underwent head CT, the odds ratio was 2.78 (95% CI: 1.15,6.70; P = .02).
Among asymptomatic subjects, there was an even stronger association between significant scalp hematoma and ICI. In this subgroup, significant scalp hematoma had an odds ratio of 22.41 (95% CI: 2.90,172.9) for association with ICI (P < .001). Among the subgroup of asymptomatic subjects who underwent head CT, the odds ratio was 7.22 (95% CI: 0.89,58.50; P = .05). The age distribution of cases of ICI in asymptomatic subjects with scalp hematoma is shown in Fig 2.
Of 23 subjects with significant scalp hematoma and ICI, 22 (96%) also had SF. Of 75 subjects with significant scalp hematomas and SF, 22 (29%) had ICI. In contrast, of 164 subjects with significant scalp hematomas and no SF, only 1 (0.6%) had ICI (P < .001).
A total of 265 subjects met our low-risk criteria of being asymptomatic and having no significant scalp hematoma. One (0.4%) (95% CI: 0.1, 1.9%) of these subjects had ICI. This one subject was a 2-month-old child with a small epidural hematoma who required no intervention.
Hospital Course and Relationship to Symptoms
Three (19%) of the 16 symptomatic subjects with ICI required surgical evacuation of an intracranial hematoma. Indications for surgery were severely depressed mental status in 2 cases, and emesis, decreased activity, and a large subdural hematoma exerting mass effect in a third case. In addition, 2 (12%) symptomatic subjects with ICI required intubation for airway control, and 4 (25%) received therapy with anticonvulsant medication. One symptomatic subject with ICI (6%) had clinical deterioration while admitted to the hospital. This 1 subject had a small subdural hematoma that was found to have enlarged on a subsequent head CT. Three symptomatic subjects with ICI (19%) had persistent neurologic abnormalities on discharge from the hospital.
Of the 14 asymptomatic subjects with ICI, 9 (64%) were admitted to the hospital. Two (14%) of these subjects received specific therapy for the head injuries. One was an 11-month-old girl who had evacuation of an epidural hematoma. Head CT showed a 1.5 cm epidural hematoma associated with midline shift and effacement of the lateral ventricle. The decision was made to perform an evacuation of this hematoma because of its size and mass effect, despite the absence of clinical symptoms.
The other asymptomatic child with ICI who received specific therapy was a 7-month-old boy with a cerebral contusion who was begun on prophylactic anticonvulsants. No asymptomatic subjects received any other form of specific therapy for their head injuries. No asymptomatic subject with ICI (95% CI: 0, 21%) had any subsequent clinical deterioration.
Of 63 subjects with isolated SF, 24 (38%) were admitted to the hospital. None had any complications or clinical deterioration. Only 1 of the 63 subjects with isolated SF required any form of therapy, a 2-month-old girl who required elevation of a depressed SF.
Of 515 subjects with no SF or ICI, 24 (5%) were admitted, most for reasons unrelated to the head trauma. No subject required any specific therapy or had any complications related to the head trauma while in the hospital.
Follow-up calls were completed successfully for 595 (98%) of the 608 study subjects, 2 weeks after the initial visit. None of the 420 subjects who did not undergo head CT (95% CI: 0,0.7%) had any complications by the time of the follow-up call.
Of the 265 subjects who met our low-risk criteria, none (95% CI: 0,1.2%) had any subsequent clinical deterioration or required any form of therapy for their head injuries.
None (95% CI: 0,3%) of the 104 subjects who had normal head CT findings subsequently had any clinical deterioration or new diagnoses of ICI. Of those subjects with ICI or isolated SF, none had any new complications after discharge from the hospital.
We have found clinical symptoms and signs of brain injury to be insensitive indicators of intracranial injury in infants, because 48% of our subjects with ICI were asymptomatic. This finding is in keeping with a previous report from our institution, which found that 19% of infants admitted to the hospital with ICI were asymptomatic.4 The higher rates of asymptomatic ICI in the current study may reflect a few differences in study design. First, the current study included all ED patients, rather than only those subjects who were admitted to the hospital. Five subjects who had small asymptomatic ICIs were deemed stable for discharge to home from the ED. Second, the prospective nature of data collection in our study may have provided a more accurate assessment of what clinical findings were evident in these patients before the head CT results were known. Finally, we believe that the attention given to the diagnosis of ICI in infants because of our study, as well as our dissemination of guidelines for radiographic imaging of head-injured infants, may have increased the frequency with which we diagnosed asymptomatic ICI during our study.
Relatively high frequencies of asymptomatic ICIs in infants have been reported in previous investigations as well. Quayle et al2found that 5/37 (14%) cases of ICI in children occurred in infants who had no symptoms or signs except for hematoma of the scalp. Schunk et al5 found that 1 of 13 cases of ICI in children occurred in a 19-day-old patient who had no symptoms.
Given that infants with ICI may be asymptomatic, how can the clinician direct radiographic screening at those asymptomatic infants most likely to have ICI? Our data indicate that even among children in the first 2 years of life, younger age is a somewhat useful criterion for distinguishing patients at higher risk for ICI. Patients in the first 3 months of life appeared to be at an especially high risk of ICI, because 13% of subjects in this group had intracranial lesions noted on head CT. It seems reasonable, therefore, for clinicians to have a lower threshold for imaging when assessing the youngest head-injured infants.
Our data also show a relationship between mechanism of injury and likelihood of ICI, with direct falls from heights of ≥3 feet and stair falls being more likely than falls <3 feet to result in ICI. Nonetheless, even among the group of subjects who fell <3 feet, there was a 2% risk of ICI. Previous investigations also have shown that even short falls may result in ICI.4 ,6 ,7 It seems reasonable for clinicians to pay special attention to those head-injured infants who have higher impact mechanisms of injury, but to recognize that those infants who have suffered low-impact mechanisms of injury also are at some risk for ICI.
Our data indicate that scalp hematomas probably are the most useful clinical indicator, among those we studied, of ICI in asymptomatic infants. Significant scalp hematomas are a surrogate marker for SF, which, in itself, is frequently associated with ICI. Previous studies also have shown that scalp hematomas are highly associated with SF in head-injured infants, and that SF is highly associated with ICI.2 8–12
These findings suggest that radiologic screening of head-injured infants should be directed at two groups of patients: 1) those with symptoms and signs of brain injury, and 2) those without symptoms or signs of brain injury but with significant scalp hematomas, which may indicate underlying SF. Data from a previous study in our institution have shown that infants younger than 1 year of age have an increased risk of SF.9 Based on these data, we recommend radiographic screening for all asymptomatic infants younger than 1 year of age with any scalp hematoma, and screening of asymptomatic infants who are older than 1 year only if they have a large, boggy scalp hematoma. In our institution, skull radiography is used as the initial study for these asymptomatic subjects, with follow-up head CT for those patients who have SF. Other clinicians may favor head CT as the initial study for asymptomatic infants with significant scalp hematomas.
As we discuss the frequency of asymptomatic ICI in infants, it is reasonable to question the clinical significance of these lesions. If these patients are truly asymptomatic, is it important that their ICIs be diagnosed? Our data indicate that the majority of asymptomatic children with ICI do well with conservative management. In 1 of the 14 asymptomatic cases, however, a large epidural hematoma exerting mass effect was present. Although we cannot know what would have happened had this case not been diagnosed, it is certainly plausible that the patient might have had complications. Previous data from our institution also indicate that although most asymptomatic cases of ICI do not require specific therapy, there is a small risk of subsequent complications.4 All the data currently available are limited by small sample sizes, which makes it impossible to exclude a small risk of subsequent clinical deterioration in initially well-appearing subjects with ICI. Future studies involving larger numbers of asymptomatic infants with ICI would be useful to determine better the clinical significance of these lesions and to analyze the costs and benefits of screening for these cases. Based on the limited data currently available, we believe it is worthwhile to continue to screen for ICI in children even when symptoms or signs of brain injury are absent.
Previous studies have indicated liberal use of radiographic imaging when evaluating head-injured infants. Masters et al1recommended that age younger than 2 years in itself be considered a moderate risk factor for intracranial injury after head trauma. Both Lloyd et al13 and Ros and Cetta3 recommend skull radiography for all infants presenting after head trauma. We believe that our data allow us to narrow these recommendations by identifying a low-risk group of infants who may be managed safely without radiographic imaging. Of our entire study population, 43% met the low-risk criteria we defined, and none of the subjects in this low-risk group required any therapy or had any subsequent clinical deterioration. We believe, therefore, that it is reasonable not to pursue radiographic imaging on asymptomatic infants who have no significant scalp hematomas. However, we recommend that clinicians be especially cautious with head-injured infants in the first 3 months of life, recognizing both the high incidence of ICI in this age group and the fact that subtle changes in behavior may be especially difficult to detect in these patients. Clinicians should consider radiographic imaging in these young infants even when clinical findings are normal, especially if a higher force mechanism of injury was reported.
Some limitations of our study must be noted. One important limitation is that physicians did not adhere uniformly to our department's guidelines for radiographic imaging. Their decisions to vary from the recommended guidelines may have introduced their own biases into our data, affecting the apparent sensitivity and specificity of the clinical signs we studied.
Another important limitation is that we did not perform head CT on all study patients. We have likely underestimated the incidence of asymptomatic ICI, therefore, because there may have been some well-appearing children with ICI who were discharged to home without radiographic imaging. This bias also likely makes some of our clinical features appear to be more sensitive predictors of ICI than they truly are. Based on our follow-up data, however, we know that no ICIs were subsequently diagnosed in the 420 patients who did not have head CT at the initial ED visit. These data support further the notion that a large fraction of head-injured infants may be managed safely without radiographic imaging.
Finally, our study is limited in that we do not have long-term follow-up data about our patients. It would be interesting to know whether asymptomatic cases of ICI are associated with any long-term neurologic or developmental deficits. Future studies involving long-term follow-up of patients with asymptomatic ICI would be of interest.
It is important to remember that our study was performed in the ED of a tertiary care pediatric hospital, and that the epidemiology of head injuries we report might differ from that found in a community hospital or in a primary care clinic or urgent care center. Our recommendations for which patients require imaging may not be applicable to these different patient populations. In addition, we caution the reader that our recommendations should not be applied to patients for whom child abuse is suspected. In instances of suspected child abuse, head CT and/or skull radiography might be indicated even when clinical findings are normal.
We conclude that ICIs occur frequently in the absence of symptoms of brain injury in head-injured infants. The youngest head-injured infants—especially those younger than 3 months of age—are at the highest risk for ICI. Many cases of asymptomatic ICI will be detected by radiographic imaging of subjects with hematomas of the scalp, which may indicate underlying SF. Although most children with asymptomatic ICI do well with conservative management, a small fraction may require specific therapy.
We recommend that radiographic imaging of head-injured infants be directed not only at those subjects who are symptomatic, but also at those asymptomatic infants who have significant scalp hematomas. In addition, clinicians should have a low threshold for radiographic imaging in head-injured infants younger than 3 months of age. Asymptomatic infants older than 3 months of age who do not have significant scalp hematomas may be managed safely without radiographic imaging.
This study was funded in part by a grant from the David and Lucile Packard Foundation.
We thank Michael Wigotsky for his invaluable help in collecting and organizing our research data, and Dr Gary Fleisher for his careful review of the manuscript.
- Received October 22, 1998.
- Accepted March 25, 1999.
Reprint requests to (D.S.G.) Division of Emergency Medicine, Children's Hospital, 300 Longwood Ave, Boston, MA 02115. Email:
Data from this study were presented in abstract form at the Joint Emergency Medicine Trauma Symposium of the Pediatric Academic Societies' Annual Meeting; May 3, 1998; New Orleans, LA.
- ↵Quayle K, Jaffe D, Kupperman N, et al. Diagnostic testing for acute head injury in children: when are head computed tomography and skull radiographs indicated? Pediatrics. 1997;99(5). URL: http://www.pediatrics.org/cgi/content/full/99/5/e11
- ↵Kleinman P, Spevak M. Soft tissue swelling and acute skull fractures. J Pediatr. 1992;737–739
- Duhaime A,
- Alario J,
- Lewande J,
- et al.
- Copyright © 1999 American Academy of Pediatrics