Ct Hypodensity Accid Ped Inj Neurosurg Steinbok Poskitt

Hypodensities on computed tomo-graphic (CT) brain scans after headtrauma have traditionally been attrib-

uted to cerebral ischemia. Previously pub-lished reports have suggested that it is diffi-cult to precisely determine how much timemust elapse between the injury and the devel-opment of hypodensities, particularly diffusehypodensities (6, 20). As commented on byDias et al. (9), and in keeping with our experi-ence, the standard opinion is that CT scan lowdensities do not become readily apparent untilat least 6 hours after the injury. Typically, theybecome progressively more obvious during the

next 3 to 5 days and then regress during thesubsequent few days. Thus, the presence ofobvious CT scan hypodensities within 6 hoursof a reported head injury may lead physiciansto call into question the timing of the injury orraise the possibility of an etiology other thanthe reported head injury. In young childrenand infants, the apparent inconsistencybetween the CT scan findings and the reportedhistory of the trauma may lead to a concernregarding non-accidental inflicted injury.

There are a number of reports in which theoccurrence of parenchymal hypodensities onthe initial CT scan after non-accidentally

NEUROSURGERY VOLUME 60 | NUMBER 4 | APRIL 2007 | 689

CLINICAL STUDIES

Paul Steinbok, M.B.B.S.Division of Pediatric Neurosurgery,Department of Surgery,BC Children’s Hospital,and University of British Columbia,Vancouver, Canada

Ashutosh Singhal, M.Sc., M.D.Division of Pediatric Neurosurgery,Department of Surgery,BC Children’s Hospital,and University of British Columbia,Vancouver, Canada

Ken Poskitt, M.D.C.M.Department of Radiology,BC Children’s Hospitaland University of British Columbia,Vancouver, Canada

D. Douglas Cochrane, M.D.Division of Pediatric Neurosurgery,Department of Surgery,BC Children’s Hospital,and University of British Columbia,Vancouver, Canada

Reprint requests:Paul Steinbok, M.B.B.S.,Division of Pediatric Neurosurgery,Department of Surgery,BC Children’s Hospital,4480 Oak Street, #K3–159,Vancouver, BC, Canada V6H 3V4.Email: [email protected]

Received, June 7, 2006.

Accepted, December 8, 2006.

EARLY HYPODENSITY ON COMPUTEDTOMOGRAPHIC SCAN OF THE BRAIN INAN ACCIDENTAL PEDIATRIC HEAD INJURY

OBJECTIVE: Hypodensities on computed tomographic (CT) brain scans are thought totake at least 6 hours to become apparent after blunt head trauma. This finding, in con-junction with the later evolution of the hypodensities, is used in timing the injury inchildren with suspected non-accidental brain injury, in whom the history may be inac-curate. The purpose of this study is to report the occurrence of diffuse cerebral parenchy-mal hypodensities on CT scans performed within 5 hours of a well-defined accidentalhead injury.METHODS: A retrospective review was performed of five patients admitted to BritishColumbia Children’s Hospital who had accidental head injury and who were identi-fied as having diffuse cerebral hemispheric hypodensities on early CT scans.RESULTS: We present five patients (age range, 4 mo–14 yr) with well-documented acci-dental head injuries who demonstrated obvious and extensive CT brain scan cerebralhemispheric hypodensity from 60 minutes to 4.5 hours after trauma. All five patients pre-sented with severe head injuries and immediate, unremitting coma, and all five pro-gressed rapidly to brain death within 48 hours.CONCLUSION: It is unusual, but possible, to develop CT hypodensities as early as1 hour after accidental head injury. In our small series of cerebral hemispheric hypo-density occurring less than 5 hours after trauma, all five patients had a uniformly fataloutcome. These observations may be important medicolegally in the assessment of thetiming of head injury when the history of the trauma is not clear, as in children withsuspected non-accidentally inflicted injury. It is inappropriate to generalize these find-ings to patients who are not unconscious immediately after a head injury, who regainconsciousness after an injury before deteriorating, or who do not progress rapidly tobrain death.

KEY WORDS: Cerebral edema, Child abuse, Computed tomographic hypodensity, Head injury, Medicolegal,Timing of injury

Neurosurgery 60:689–695, 2007 DOI: 10.1227/01.NEU.0000255398.00410.6B www.neurosurgery-online.com

vehicle collision in one case, and a bicycle accident (without ahelmet) in one case.

None of the patients was documented to have a cardiorespi-ratory arrest or any period of hypoxia or hypotension duringinitial transfer or during the trauma resuscitation before theinitial CT scan. None of the patients was identified to haveradiographic or clinical evidence of spine or spinal cord injury.In one of three patients in whom funduscopy was documented,retinal hemorrhages, consistent with non-accidental trauma,were identified (Table 1). This patient was one of three in thisseries investigated by our child protection team. On the basis ofcorroborating eyewitness accounts, it was concluded that theinjury was genuinely the result of a fall as described by thefamily. In another patient, there were bilateral retinal infarcts,which are not typical of non-accidental injury. In the thirdpatient, funduscopy was normal.

All five initial CT scans demonstrated diffuse, obvious, cere-bral hemisphere hypodensity, and, in some scans, there werehypodensities in more-central structures, such as the basal gan-glia and/or the brainstem (Table 2). All five patients had tento-rial and/or interhemispheric acute subdural blood. In twopatients, there were acute convexity subdural hematomas, bothof which were of mixed density; a right frontoparietal convex-ity hematoma, 7 mm in maximal thickness, and a right frontalhematoma, 7 mm in maximal thickness. All five initial CT scansshowed evidence of brain swelling with basal cistern efface-ment, and four of the five scans showed complete obliterationof the basal cisterns. Only one of five patients had a cranial frac-ture. The distributions of the hypodensities on the initial CTscan are presented in Table 2. Representative early CT scanimages of each patient are shown in Figure 1.

In Patients 2 and 4 (Table 1), carotid Doppler ultrasound per-formed 18 and 8 hours, respectively, after injury demonstratedreversed diastolic flow. In Patient 3, an intracranial pressuremonitor was inserted, which demonstrated pressures of greaterthan 65 mmHg. In Patient 4, xenon-CT and magnetic resonanceangiography at approximately 6.5 hours after injury confirmedthe absence of cerebral blood flow (CBF). A diffusion-weightedmagnetic resonance imaging scan performed at the same timeas the magnetic resonance angiography showed markedrestriction of diffusion in the cerebral hemispheres. In Patient 5,a craniotomy was performed to evacuate a 1-cm-thick cerebralconvexity acute subdural hematoma, and postoperativeintracranial pressure monitoring demonstrated an intracranialpressure of greater than 40 mmHg.

Autopsies were performed in two patients (Patients 2 and4). In both patients, there was marked cerebral edema withuncal and tonsillar herniation. In Patient 2, there was a thinsubdural hematoma, with an associated contusion and lacera-tion of the right cerebral hemisphere, diffuse subarachnoidblood, and diffuse hypoxic-ischemic changes, with no definitehistopathological evidence of acceleration/deceleration braininjury. No abnormalities were noted in the brainstem. InPatient 4, there were thin diffuse subdural hemorrhages overthe cerebral hemispheres and neuroaxonal spheroids at the cer-vicomedullary junction, but not in the remainder of the brain,

inflicted injury has been mentioned (8–10, 15). Because the diag-nosis of inflicted shaking injury is associated with inaccuraciesin the history, it may be argued that the timing of the trauma insuch reports may be unreliable. There are well-documentedreports of brain “swelling” within 6 hours of accidental headtrauma, but no reports of early parenchymal CT scan hypoden-sities in documented accidental injuries in children or adults.

We present this case series to demonstrate that diffuse andobvious CT scan parenchymal hypodensities occur within 5hours of well-documented accidental pediatric head injury andto discuss the prognostic and medicolegal implications of thesefindings.

PATIENTS AND METHODS

British Columbia’s Children’s Hospital is a tertiary referralcenter for pediatric head trauma, with a catchment area includ-ing approximately 4.2 million residents of British Columbia,Canada. This is a retrospective review of a series of childrentreated at British Columbia’s Children’s Hospital and identifiedas having early CT scan hypodensities after accidental headinjury. The medical records of these patients were reviewed toreconfirm the timing and accidental nature of the injury. Allinformation in the medical record, including ambulancereports, emergency department reports, medical and nursesnotes, as well as any social worker notes and discharge sum-maries were reviewed. The time of injury was estimated fromthese records and only patients with well-corroborated traumahistories were included in this review. The timing and find-ings on the CT scans performed after the trauma were ascer-tained with the assistance of a pediatric neuroradiologist (KP).The clinical course of the patients and the results of any addi-tional radiological investigations were also determined. Thestudy was approved by the Ethics Committee of the Universityof British Columbia and the Research Review Committee ofBritish Columbia’s Children’s Hospital.

RESULTS

Five patients were identified with well-documented acciden-tal head trauma and early posttraumatic hypodensities; allwere noted on CT scans performed less than 5 hours after theinjury. In three of the five patients, the possibility of non-accidental inflicted injury had been raised because of the CTscan findings and the child protection team had investigatedthe circumstances of the injury. In all cases, there was incontro-vertible evidence that the injury was a bona fide accident. Thedemographic data, mechanism of injury, initial Glasgow ComaScale (GCS) score, pupillary reaction at initial assessment, CTscan findings, and outcome are summarized in Table 1. All fivepatients had severe blunt head injuries and were immediatelyunconscious, with a GCS score of less than 8. The age rangewas 4 months to 14 years; four of the five patients wereyounger than 2 years of age at the time of injury. There weretwo girls and three boys. The mechanism of injury was a fall inthree cases, one of which was from a moving vehicle, a motor

690 | VOLUME 60 | NUMBER 4 | APRIL 2007 www.neurosurgery-online.com

STEINBOK ET AL.


COMPUTED TOMOGRAPHIC SCAN HYPODENSITIES AFTER HEAD INJURY

and extensive bilateral retinal hemorrhages. The vertebral andcarotid arteries were normal.

DISCUSSION

The patients reported in this study all experienced acciden-tal severe traumatic brain injuries, characterized on CT scanswithin 5 hours by severe diffuse brain swelling and parenchy-mal hypodensities and resulting in rapid brain death. Themechanisms of injury varied from a relatively mild type oftrauma, a fall from a stool, to a high-energy impact, a motorvehicle collision. There was nothing special regarding themechanism of injury in any of the patients that would havesuggested the rapid progression to death that occurred there-after. The patients in this study did not necessarily experiencesevere diffuse axonal injury because neither of the two autop-sied brains showed evidence of acceleration and decelerationinjury in the cerebral hemispheres.

It is well documented that diffuse brain swelling can be iden-tified on CT scans performed within hours of a severe trau-matic brain injury in both children and adults (1, 5, 12, 18).However, the establishment of early and extensive hypo-densities in association with the brain swelling after well-documented accidental pediatric head injury, as noted in thecases presented in this series, is, to our knowledge, a novelobservation in the literature.

The radiological diagnosis of diffuse brain swelling aftertrauma has typically been based on the findings of obliteratedor compressed basal cisterns, with or without small lateral andthird ventricles, in the absence of significant midline shift (�3–6mm, depending on the study) or significant intracranial focalmass lesion (1, 12). In one study by Willman et al. (18), “poorgray-white differentiation” was also included as one of the CTscan criteria of brain swelling. None of the studies includedcomments regarding the presence of hypodensities on the CTscans in the patients with a diagnosis of early diffuse brainswelling. It can be argued that the presence of poor gray-whitedifferentiation, as in the study by Willman et al. (18), implies

hypodensity of the gray matter, but the authors did not com-ment specifically regarding hypodensities.

Early CT scan hypodensities after head injury have beendemonstrated previously in non-accidental injuries (8–10, 15).However, only Dias et al. (9) tried to time the first CT scanafter the injury; in their report, parenchymal CT scan hypo-densities were noted on scans performed an average of 3.2hours after the “suspected” time of the injury. The authors triedto pinpoint the time of the injury but, in some cases, had to datethe injury to the time of “an apneic spell, seizure, abrupt coma,or other significant and immediate event,” which might notrepresent the precise timing of the actual injury itself. In thecases reported herein, the injuries were accidental and the timeof injury was precisely documented.

The CT scan hypodensities in our cases were diffuse, severe,and very obvious. The hypodensities affected primarily thecerebral hemispheres (Table 2), which were uniformly hypo-dense, and the findings were similar to the cases described byHan et al. (11) as having the so-called “reversal sign.” Han et al.used the term reversal sign to indicate diffuse cerebral edemaon CT scan, with decreased density of the cortical gray andwhite matter, and relatively increased density of the thalami,brainstem, and cerebellum. In their study, there were eight chil-dren with the reversal sign present at the time of the first CT

TABLE 1. Patient summarya

Patient no. Age/sex Mechanism GCS Pupils Funduscopy CT scan Outcome

1 4 mo/F MVC 3 2 fixed Not performed 3.5 hr: SAH, IVH, tentorial �24 hr: brain deathASDH, PH

2 7 mo/M Fall down stairs 3 1 fixed Bilateral retinal and 2 hr: falx, tentorial and con- �24 hr: brain deathpreretinal hemorrhages vexity ASDH, mild IVH, PH

3 14 yr/M Bike accident 4 2 fixed Not performed 4.5 hr: DAI, mild falx ASDH, �36 hr: brain deathmild SAH, mild IVH, PH

4 12 mo/F Fall from fast 5 2 fixed Bilateral retinal infarcts 3.5 hr: mild SAH, mild falx �24 hr: brain deathmoving hay wagon and tentorial ASDH, PH

5 2 yr/M Fall from stool 4 2 fixed Normal 1 hr: tentorial and convexity 48 hr: deathASDH, PH

TABLE 2. Distribution of low densities on initial computed tomographicscana

BilateralPatient

cerebral ThalamusLentiform

CaudateBrain- Cere-

no.hemispheres

nucleus stem bellum

1 � �

2 � �

3 � � � �

4 � � � �

5 � � �

a GCS, Glasgow coma scale; CT, computed tomographic; MVC, motor vehicle collision; SAH, subarachnoid hemorrhage; IVH, intraventricular hemorrhage; ASDH, subduralhematoma; PH, parenchymal hypodensities; DAI, diffuse axonal injury.

a �, hypodensity was present in that anatomical location.

to the occurrence of the reversal sign in the list of parenchymalabnormalities among the 15 patients with well-defined times ofthe initial CT scan relative to the injury. However, they note intheir discussion that “among infants with the reversal sign,50% were evident on the initial scans performed an average of3.5 hours after the injury was reported.” The findings in ourpatients with accidental head injury support the contention ofDias et al. (9) with respect to the abused child, namely thathypodensities may not take 6 to 48 hours to develop. In ourseries, there was no doubt regarding the accidental nature orthe timing of the trauma and all patients had extensive cerebralhemisphere hypodensity, similar to the reversal sign, between1 and 5 hours after the head injury.

The pathophysiology of the parenchymal hypodensitiesnoted in the children in this series is not clear. These CT scanfindings indicate the presence of cerebral edema, as was iden-tified in the two patients who had autopsies. However, thepathophysiology of such extensive edema within hours of thetrauma is not clear. In the series by Han et al. (11) of childrenwith reversal sign, when trauma was the etiology, child abusewas the most common cause. The authors opined that in thisgroup of children with abuse, the posttraumatic reversal signmay have been the result of repeated trauma to the brain,resulting in repeated “edema induced hypoperfusion of thebrain, in turn resulting in stress induced hyperglycemic cere-bral patterns of anoxic injury.” However, in our series, repeatedtraumatic brain injury cannot be implicated.

Acute posttraumatic diffuse brain swelling without cerebralhypodensities has been attributed to cerebral vascular engorge-ment secondary to hyperemia, on the basis of the initial reportof Bruce et al. (7) on so-called “malignant brain swelling” inchildren. Studies that are more recent, with CT scan dynamicscanning (19) and CT scan xenon assessment of CBF within afew hours of head injury, have indicated that, in the most severebrain injuries, decreased CBF, sometimes to ischemic levels,rather than cerebral hyperemia may be present (4, 5, 14). Onserial CBF measurements, the decreased CBF may be replacedby cerebral hyperemia after 24 hours (4, 5, 14); thus, if CBF is notmeasured within hours of the injury, the period of decreasedCBF may be missed. It may, therefore, be that the reduced CBF,which can occur early after trauma, is enough to cause ischemiaand infarction in some patients. This could, in turn, lead to rap-idly apparent low densities and swelling on CT scans.

One of the concerns with the proposition that the hypoden-sities in our cases were caused by ischemia is that the timecourse of such extensive low densities, as noted in our patients,is not in keeping with the CT scan findings observed afterknown ischemic events. For example, CT scan hypodensitiesare usually not observed for the first 12 hours after neonatalasphyxial injures (3) and become more obvious and extensiveduring a 72-hour time course (13). Similarly, in typical cere-brovascular accidents in adults, CT scan hypodensities are dif-ficult to identify until at least 6 hours (16). Indeed, for entry intoone multicenter trial for stroke, CT scan evidence of theischemic event within 6 hours of a clinical middle cerebralartery territory stroke included loss of density contrast of the


STEINBOK ET AL.

scan after admission; of these, two had an accidental headinjury and three were thought to have experienced child abuse.In these patients with an acute reversal sign, the time of thescan relative to the time of the injury was not reported. In thestudy by Dias et al. (9), in which a concerted attempt was madeto time the CT scan abnormalities in infants after non-acciden-tal head injury, there were 15 patients out of a total of 33 inwhom the authors thought that the time of the alleged abusecould be pinpointed. Six of these 15 patients had parenchymalhypodensities on a CT scan performed an average of 3.2 hoursafter injury. The extent of the hypodensities in these six patientswas not reported but they were not extensive enough to becategorized as showing the reversal sign because those authorsindicated the presence of the reversal sign as a separate cate-gory. In their results, Dias et al. (9) noted that six patients exhib-ited the reversal sign on CT scans but they make no reference

FIGURE 1. A, Patient 1, CT scanobtained 3.5 hours after injury in amotor vehicle collision. B, Patient 2,CT scan obtained 2 hours after a falldown stairs. C, Patient 3, CT scanobtained 4.5 hours after a fall off afast-moving bicycle. D, Patient 4, CTscan obtained 3.5 hours after a fall offof a fast-moving wagon during a hayride. E, Patient 5, CT scan obtained 1hour after a fall from a stool.

A

C

B

D

E

accidentally inflicted injury was queried and the child protec-tion team was asked to investigate because the CT scan find-ings of severe low density changes were thought to be incom-patible with the time of the injury as indicated in the historytaken at the time of arrival to the hospital. The case series ofchildren with non-accidentally inflicted head injury reported byDias et al. (9) and this case series of children with accidentalhead injury indicate that the traditional expectation regardingthe timing of onset of CT scan brain hypodensities after headtrauma may not always be correct. We think the usual expectedevolution of low densities on CT scan is still what one wouldobserve in the vast majority of patients and that the five casesreported with early diffuse hypodensities are unusual. Hence,a question regarding possible child abuse in some of our caseswas, and still would be, appropriate. However, the knowledgethat obvious and diffuse intracerebral low densities on CT scancan occur, albeit rarely, as early as 1 hour after a documentedaccidental head injury may have allayed some of the concernsregarding the possibility of non-accidental injury in these chil-dren. Furthermore, in cases of suspected non-accidental fatalhead injury in which the time of the head injury is not clear, thepossibility of very early appearance of diffuse low densities onthe CT scan has to be considered in the estimation of the tim-ing of the severe head injury.

CONCLUSIONS

We have presented five children with documented and well-corroborated accidental head injuries in whom CT scans per-formed between 1 and 5 hours after injury showed diffuse andobvious cerebral hemispheric hypodensity. These findings callinto question the commonly held opinion that such extensiveparenchymal hypodensities take at least 6 hours to appear afterhead trauma. This may be of medicolegal importance in discus-sions regarding the timing of head injuries in children. It isimportant to recognize that, in our series, diffuse early hypo-densities on CT scan were associated with a severe traumaticbrain injury, immediate unconsciousness that never resolved,and rapid brain death in all patients. It is inappropriate to gen-eralize the findings of this case series to patients who are notunconscious immediately after a head injury, who regain con-sciousness after an injury before deteriorating, or who do notprogress rapidly to brain death.

REFERENCES

1. Aldrich EF, Eisenberg HM, Saydjari C, Luerssen TG, Foulkes MA, Jane JA,Marshall LF, Marmarou A, Young HF: Diffuse brain swelling in severelyhead-injured children. A report from the NIH Traumatic Coma Data Bank. JNeurosurg 76:450–454, 1992.

2. Anonymous: Thrombolytic therapy with streptokinase in acute ischemicstroke. The Multicenter Acute Stroke Trial—Europe Study Group. N Engl JMed 335:145–150, 1996.

3. Barkovich AJ, Sargent SK: Profound asphyxia in the premature infant:Imaging findings. AJNR Am J Neuroradiol 16:1837–1846, 1995.

4. Bouma GJ, Muizelaar JP, Choi SC, Newlon PG, Young HF: Cerebral circula-tion and metabolism after severe traumatic brain injury: The elusive role ofischemia. J Neurosurg 75:685–693, 1991.

lentiform nucleus and/or the insular ribbon and hemisphericsulcus effacement (2) but not hypodensity.

In support of the hypothesis that ischemia may be the causeof the early hypodensities observed in our patients, there isevidence that, in the most severe ischemic situations, CT scanhypodensities may be present earlier than in typical cases. Ourpatients, all of whom died, certainly qualify for being in thatmost severely affected group. In the study of hypoxic ischemicterm neonates by Lupton et al. (13), the most severely involvedchildren had patchy CT scan hypodensities on the first day oflife, although the specific time in hours was not reported. Inaddition, they had evidence of intracranial hypertension andwent on to die or have poor outcomes. Furthermore, CT scanhypodensities have been described within 5 hours of provenmiddle cerebral artery occlusions, and extensive hypodensitiesassociated with local brain swelling in the middle cerebralartery distribution predicted a fatal outcome (17). In two ofour five patients, absent CBF was demonstrated, and it may bethat extensive low densities on CT scan can occur rapidly, evenas early as 1 hour after trauma, in special and unusual situa-tions in which there is total ischemia with no blood flow intothe brain. If so, these findings may predict inevitable braindeath, as occurred in all of the patients reported in this article.The outcomes of our patients are consistent with those reportedby Han et al. (11), who noted that four out of eight children(five had head trauma) with acute reversal sign on CT scansdied; the other four children had poor outcomes.

The occurrence of acute obvious diffuse cerebral hypodensityis rare among head-injured patients, and, on the basis of the lackof reports in the literature, this is probably a rare phenomenon.It is difficult to understand why these five patients developedthis problem. The mechanism of injury, which included a falldown stairs and a fall from a stool, was less severe than in manyother head-injured patients. Nonetheless, all patients clearlyexperienced a severe brain injury with immediate coma and apoor GCS score and went on to die rapidly. It was somewhatunexpected that, in the two autopsied patients, despite theapparent clinical severity of the brain injury, neither showedhistopathological evidence of diffuse axonal injury and one hadfindings of axonal injury limited to the cervicomedullary junc-tion. It would be interesting to elucidate the factors that mayhave precipitated the disastrous outcome in these children.

The findings in this study of extensive CT scan intracerebralhypodensities within 5 hours of a well-documented head injuryare of medicolegal significance, particularly in the pediatricpopulation. Our experience and that of others, such as Diaset al. (9), are that, generally, CT scan intracerebral low densitiestake at least 6 hours (some think 24 h) to become apparent aftera traumatic brain injury. The low densities are expected tobecome progressively more obvious during the next 3 to 5 daysand then regress during the subsequent few days. The timecourse of evolution of such changes on CT scans may be usedas one of the criteria in determining the approximate time of ahead injury when the provided history is thought to be inaccu-rate, as might be the case in non-accidentally inflicted injuries.Indeed, in three of the five cases presented in this series, non-



versally poor prognosis, and, in this series, all of the patients died.This is of value to the pediatric neurosurgeon, who must make a deci-sion early on as to whether or not to perform aggressive measuressuch as intracranial pressure monitoring or hemicraniectomy.

None of these children had spinal cord injury nor did they havedocumented hypoxic or hypotensive episodes. Because these have beentwo of the more commonly proposed mechanisims of the “black brain”on computed tomographic (CT) scan, the pathophysiology of this phe-nomenon remains a mystery.

Leslie N. SuttonPhiladelphia, Pennsylvania

1. Duhaime AC, Christian CW, Rorke LB, Zimmerman RA: Nonaccidental headinjury in infants—The “shaken-baby syndrome.” N Engl J Med 338:1822–1929,1998.

In this study, the authors have identified five patients who developedhypodensities on CT scans within 6 hours of accidental head injury.

Their findings of this early evolution of hypodensity raises the possi-bility that these hypodensities can occur early after injury, particularlyafter diffuse cerebral hypoxic damage. Most often, hypodensities arereported with nonaccidental trauma in infants and are thought to sec-ondarily occur owing to the injury and ischemic insults. In the presentseries, all of the patients had a rapid decline in function and with cor-responding findings on CT scans indicative of massive injury despitea number of instances that would seem to be a low impact trauma. Inall of these instances, the patients presented with brainstem involve-ment with fixed dilated pupils and had a fatal outcome. Although wehave the presentation, imaging characteristics, and a couple ofinstances of pathological findings, it is now necessary not only to fur-ther define these pathophysiological events but also their underlyingmechanisms in our attempt to improve the potential outcome for thisinjured pediatric population. Further study is necessary to determinethe similarities and differences in mechanism and response in patientswith nonaccidental trauma and accidental head injury with a goal tointerrupt the cascade of events that eventually lead to neurologicaldecline. This study also underlines the likely future need for early andaggressive intervention to attenuate the postinjury pathophysiologicalevents and the individual response of these patients.

P. David AdelsonPittsburgh, Pennsylvania

This article highlights the early changes on CT scans that may occurafter severe head trauma and appropriately points out that such

changes may occur earlier than has been touted by timing pundits.The authors make an important point that the low attenuation changescannot be used to establish that the onset of the damage had to occur6 to 12 hours before the changes but that they may occur within 1 to 2hours. Diffusion imaging and diffusion tensor magnetic resonanceimaging have established these facts. Now, in severe trauma cases, thesame has been established for CT scans.

Marvin NelsonPediatric NeuroradiologistLos Angeles, California

This study documents early hypodensity in accidental head injury infive children on CT scans obtained within 5 hours of the injury.

The children were very young (< 2 yr of age), had evidence of verypoor clinical grade (Glascow Coma Scale score, 3–5), and severely

5. Bouma GJ, Muizelaar JP, Stringer WA, Choi SC, Fatouros P, Young HF: Ultra-early evaluation of regional cerebral blood flow in severely head-injuredpatients using xenon-enhanced computerized tomography. J Neurosurg77:360–368, 1992.

6. Brown JK, Minns RA: Non-accidental head injury, with particular reference towhiplash shaking injury and medico-legal aspects. Dev Med Child Neurol35:849–869, 1993.

7. Bruce DA, Alavi A, Bilaniuk L, Dolinskas C, Obrist W, Uzzell B: Diffuse cere-bral swelling following head injuries in children: The syndrome of “malig-nant brain edema.” J Neurosurg 54:170–178, 1981.

8. Cohen RA, Kaufman RA, Myers PA, Towbin RB: Cranial computed tomogra-phy in the abused child with head injury. AJR Am J Roentgenol 146:97–102,1986.

9. Dias MS, Backstrom J, Falk M, Li V: Serial radiography in the infant shakenimpact syndrome. Pediatr Neurosurg 29:77–85, 1998.

10. Giangiacomo J, Khan JA, Levine C, Thompson VM: Sequential cranial com-puted tomography in infants with retinal hemorrhages. Ophthalmology95:295–299, 1988.

11. Han BK, Towbin RB, De Courten-Myers G, McLaurin RL, Ball WS Jr: Reversalsign on CT: Effect of anoxic/ischemic cerebral injury in children. AJR Am JRoentgenol 154:361–368, 1990.

12. Lang DA, Teasdale GM, Macpherson P, Lawrence A: Diffuse brain swellingafter head injury: More often malignant in adults than children? J Neurosurg80:675–680, 1994.

13. Lupton BA, Hill A, Roland EH, Whitfield MF, Flodmark O: Brain swelling inthe asphyxiated term newborn: Pathogenesis and outcome. Pediatrics82:139–146, 1988.

14. Muizelaar JP, Marmarou A, DeSalles AA, Ward JD, Zimmerman RS, Li Z,Choi SC, Young HF: Cerebral blood flow and metabolism in severely head-injured children. Part 1: Relationship with GCS score, outcome, ICP, and PVI.J Neurosurg 71:63–71, 1989.

15. Sinal SH, Ball MR: Head trauma due to child abuse: Serial computerizedtomography in diagnosis and management. South Med J 80:1505–1512, 1987.

16. Steiner T, Bluhmki E, Kaste M, Toni D, Trouillas P, von Kummer R, Hacke W:The ECASS 3-hour cohort. Secondary analysis of ECASS data by time strati-fication. ECASS Study Group. European Cooperative Acute Stroke Study.Cerebrovasc Dis 8:198–203, 1998.

17. von Kummer R, Meyding-Lamade U, Forsting M, Rosin L, Rieke K, Hacke W,Sartor K: Sensitivity and prognostic value of early CT in occlusion of themiddle cerebral artery trunk. AJNR Am J Neuroradiol 15:9–18, 1994.

18. Willman KY, Bank DE, Senac M, Chadwick DL: Restricting the time of injuryin fatal inflicted head injuries. Child Abuse Negl 21:929–940, 1997.

19. Yoshino E, Yamaki T, Higuchi T, Horikawa Y, Hirakawa K: Acute brain edemain fatal head injury: Analysis by dynamic CT scanning. J Neurosurg63:830–839, 1985.

20. Zimmerman RA, Bilaniuk LT: Pediatric head trauma. Neuroimaging Clin NAm 4:349–366, 1994.

COMMENTS

This report describes five children who were found to have diffusehemispheric low density within 5 hours of well documented acci-

dental blunt head trauma. In the past, it has been more or less assumedthat this finding was diagnostic of child abuse and that it took 6 to 24hours to develop, although loss of the gray-white junction could beseen earlier.

This report is significant in several regards. The fact that all of thesechildren had clear blunt trauma rather than shaking lends further sup-port to the hypothesis of impact being required to develop this clinico-radiographic picture, as proposed by Duhaime (1). It is important thatthe finding of diffuse hemispheric low density is not diagnostic ofnonaccidental trauma, and this finding alone does not demand thisdiagnosis. The recognition that retinal hemorrhages are also not neces-sarily diagnostic of child abuse makes this diagnosis difficult at times.A thorough investigation of the circumstances of a childhood injury isneeded before accusations are made. Finally, this finding carries a uni-


STEINBOK ET AL.



edema would be unusual. Early low density has also been reported innonaccidental injury in which the timing issues are much harder todocument. When this issue was raised in this series, the patients wereappropriately investigated by the child protection team.

James M. DrakeToronto, Canada

increased intracranial pressure, and all died. The finding of “black”cerebral hemispheres on CT scans is always an ominous sign and, asthe authors discuss, probably represents irreversible global cerebralischemia. Distinguishing this radiological finding from that seen afternonaccidental injury can be difficult. Two of the children were reportedto have fallen down stairs (and had retinal hemorrhages) or from astool, both instances in which a rapidly fatal outcome with extensive

Documents

Ct Hypodensity Accid Ped Inj Neurosurg Steinbok Poskitt