Prevalence and Prognosis of Traumatic IntraventricularHemorrhage in Patients with Blunt Head TraumaClare Atzema, MD, William R. Mower, MD, PhD, Jerome R. Hoffman, MA, MD, James F. Holmes, MD,Anthony J. Killian, RN, and Allan B. Wolfson, MD

For the National Emergency X-Radiography Utilization Study (NEXUS) II Group

Background: Most studies of trau-matic intraventricular hemorrhage (tIVH)contain fewer than 25 subjects and areretrospective in design, providing minimalinformation about the entity and its clini-cal significance.

Methods: We prospectively enrolledtrauma patients from 18 centers in NorthAmerica in the National Emergency X-Radiography Utilization Study (NEXUS)II if they received an emergent headcomputed tomography (CT) scan, as de-termined by the managing physician.Clinical data were collected at the time ofenrollment and CT reports were compiledat least 1 month later. We calculated prev-alence and demographics of tIVH from

the 18 sites, while outcome data were gath-ered from medical records of patients withtIVH who were seen at any of six sites thatparticipated in the follow-up portion ofthe study. We considered patients who un-derwent a neurosurgical intervention orwho had a “poor outcome” (Glasgow Out-come Scale score of 1 to 3, death, persis-tent vegetative state, or severe disability)to have suffered a “combined outcome.”

Results: Prevalence of tIVH amongall trauma patients who received a headCT was 118 in 8,374, or 1.41%. AmongtIVH patients, 70% had a “poor outcome”and 76% had a “combined outcome.” Apoor outcome appeared to be associatedwith an abnormal presenting Glasgow

Coma Scale score and involvement of thethird or fourth ventricle, whereas age ap-peared to be unrelated. Patients withtIVH and no major associated injury onCT tended to do well; only one patientwith isolated tIVH had a poor outcome.

Conclusions: Traumatic IVH is rareand is associated with poor outcomes thatseem to be the consequence of associatedinjuries. Isolated tIVH patients who areclinically well appear to have a functionaloutcome; we were unable to identify a caseof isolated tIVH, combined with a normalneurologic examination, resulting in apoor or combined outcome.

Key Words: Traumatic intraventric-ular hemorrhage; Head injury.

J Trauma. 2006;60:1010–1017.

Existing data on traumatic intraventricular hemorrhage(tIVH) is very limited; studies in the medical literatureare restricted by their age, design, and sample sizes.

Most are small retrospective or case series studies, with fewerthan 25 patients.1–16 To date, the largest study of tIVH pa-tients consists of 43 patients.17 Prevalence of tIVH in all headtrauma patients who receive computed tomography (CT)scanning ranges from 0.4% to 4%.3,4,6,10,11,13,15–22 Prognosisof tIVH was considered very poor before the advent of CTscanning, probably because the entity was only discoveredpostmortem or in patients sick enough to require a neurosur-

gical intervention. Some studies conducted since the advent ofCT scanning also suggest that tIVH is associated with a poorprognosis, with functional outcomes ranging from 12% to47%,3–6,11,15,17 but other studies have reported no difference inrecovery rates between patients with and without tIVH whenthey are matched for associated CT findings.1,8 Several trialshave suggested that isolated tIVH is associated with a functionaloutcome,1,8,11 although this entity is not well studied because itis rare.3,13,15,18 In sum, it is not known with any certainty whatthe significance of tIVH is, in and of itself; we do not know whatoutcomes are associated with only tIVH. Keeping in mind that anormal CT scan is compatible with major brain injury (i.e.,diffuse axonal injury), we do not know what an otherwisenormal CT scan and a normal clinical examination portend inthe presence of only tIVH.

During the derivation phase of National Emergency X-Radiography Utilization Study (NEXUS) II, a prospectiveobservational study that seeks to develop a clinical decisioninstrument for the use of head CT scanning in patients withminor head injury (MHI), we encountered some CT scanswith only minimal tIVH and either minimal or no associatedfindings. Like many research groups who have attempted todevelop a clinical decision instrument for the use of head CTscanning in patients with MHI,23–41 we classified abnormalhead CT scans into a clinically relevant, positive group and aclinically irrelevant, negative group.42 Because patients with

Submitted for publication March 13, 2004.Accepted for publication January 19, 2006.Copyright © 2006 by Lippincott Williams & Wilkins, Inc.From the Emergency Medicine Center (C.A., W.R.M., J.R.H.), David

Geffen School of Medicine at UCLA, Los Angeles, California; Division ofEmergency Medicine (J.F.H.), University of California, Davis School ofMedicine, Sacramento, California; Cooper University Hospital (A.J.K.),Camden, New Jersey; and the Department of Emergency Medicine(A.B.W.), University of Pittsburgh School of Medicine, Pittsburgh, Pennsyl-vania.

This work is funded in part by Grant RO1 HS09699 from the Agencyfor Healthcare Research and Quality.

Address for reprints: William R. Mower, MD, PhD, UCLA EmergencyMedicine Center, 924 Westwood Blvd., Suite 300, Los Angeles, CA 90024;email: nexusgrp@ucla.edu.

DOI: 10.1097/01.ta.0000218038.28064.9d

The Journal of TRAUMA� Injury, Infection, and Critical Care

1010 May 2006

CT findings that are not classified as relevant positives maynot receive a head CT under the derived clinical decisioninstrument, it is imperative that a study of the “irrelevant,negative” findings be conducted to verify their inconse-quence. When the medical literature did not supply sufficientinformation to classify CT scans with isolated tIVH or tIVHwith minimal associated findings, we initiated this substudyof NEXUS II.

In this descriptive study, we first define the prevalence,demographics, and mechanism of tIVH among a large groupof prospectively enrolled patients with blunt head trauma.Next, in a subgroup of tIVH patients for whom we collectedfollow-up information, we examine the relationship betweenthree potential predictor variables (age, mental status at pre-sentation, and anatomic location of hemorrhage) and outcomein tIVH. Finally, we evaluate the outcome of patients with“isolated tIVH,” and tIVH in the presence of “minor associ-ated CT findings,” in comparison to tIVH patients with “sig-nificant associated CT findings” (Fig. 1).

PATIENTS AND METHODSThis is a substudy of NEXUS II, a multicenter, prospec-

tive, observational study of blunt head injury patients for

whom head CT scanning was ordered at the discretion of themanaging physician. The protocols and methodology used inthe study were reviewed and approved by the human subjectsreview boards of the participating institutions.

Eighteen participating centers enrolled subjects fromApril 1999 to December 2000 (Fig. 2). These institutionsrepresent a wide variety of facilities, including universityhospitals, community hospitals with and without teachingprograms, public hospitals and private hospitals, and hospi-tals with all levels of trauma categorization.

All persons with blunt traumatic head injury who pre-sented to a participating study center and received a head CTwere included in the study. Patients without blunt trauma(including those with penetrating head trauma) and thoseundergoing head CT imaging for other reasons were noteligible for inclusion. Patients were excluded from the studyif their injury occurred more than 24 hours before theirassessment, or if they were transferred to a participatingcenter from an outside facility. There were no other exclusioncriteria. The decision to obtain head CT was made by theexamining emergency physician and was not directed in anyway by the study.

Data CollectionAt the time of enrollment, and before the release of CT

information, managing clinicians collected data on age, sex,and the presence or absence of specific signs and symptomsassociated with head injury (Fig. 3). Guidelines regarding thedefinition for each of the signs and symptoms of head injurywere available to clinicians involved in the study. If thepatient was unstable on presentation, incomplete enrollmentdata could be entered into the dedicated study computer andcompleted at the earliest appropriate opportunity.

We collected initial and follow-up CT reports for eachenrolled patient at least 1 month after the date of injury,according to NEXUS II protocol. The radiologists at the 18participating centers were not given special instructionsregarding CT interpretation; although this may introduce cer-tain limitations to the data, it replicates current clinical prac-tice and therefore increases the external validity of the study.Because one of the goals of our study was to identify theclinical significance of initial CT findings, to apply the pre-dicted outcomes in clinical practice the initial CT interpreta-

Isolated tIVH: Traumatic IVH with no other findings on initial head CT, with the exception of a contusion in the basal ganglia which directly caused the tIVH by rupturing into the intraventricular space.

True isolated tIVH: Traumatic IVH with no other findings on initial and follow-up CT scans, with the exception of a contusion in the basal ganglia which directly caused the tIVH by rupturing into the intraventricular space.

Minor associated CT findings: CT scan with an acute traumatic finding in addition to the tIVH that does not include any “significant associated CT findings.”

Significant associated CT findings: CT scan with an acute traumatic finding in addition to the tIVH that includes one of the following: mass effect (any sulcal effacement), basal cistern compression or midline shift, an EDH or SDH larger than one centimetre in width (or causing mass effect), pneumocephalus, extensive SAH (more than two sites, not including scattered sulcal SAH as a site), bilateral hemorrhage of any type (not including scattered sulcal SAH as a site), a contusion of more than one centimetre diameter or more than one contusion, signs of herniation, hemorrhage in the posterior fossa, a depressed or diastatic skull fracture, diffuse cerebral edema, or diffuse axonal injury (DAI).

Abnormal head CT: CT scans with any acute traumatic finding, no matter how minor.

Poor outcome: Glasgow outcome scale (GOS) [44] score of one, two, or three, corresponding to death (secondary to neurosurgical causes), persistent vegetative state, and severe disability.

Functional outcome: GOS [44] score of four or five, corresponding to moderate disability and minimal or no deficits.

Combined outcome: The performance of a directed intervention (defined as any one of intubation for the purposes of hyperventilation, intracranial pressure [ICP] monitoring, or craniotomy, within five days of injury), and / or a poor outcome, defined by a GOS score of one, two, or three (death, persistent vegetative state, and severe disability, respectively).

Directed intervention: Intubation for the purposes of hyperventilation, intracranial pressure (ICP) monitoring, and craniotomy within five days of injury.

Study cohort: All patients from 18 study sites with head CT scan evidence of acute tIVH. Group from which prevalence data was collected.

Outcome subset: From six study sites selected for the follow-up portion of the study, all patients with head CT scan evidence of acute tIVH. Group from which outcome data was collected.Diffuse Axonal Injury (DAI): Hemorrhage in the dorsolateral brainstem, the corpus callosum, or the grey/white border at least two days post-injury.

Fig. 1. Study definitions.

1) Los Angeles, CA (University of California, Los Angeles Medical Centre) 2) Sacramento, CA (University of California, Davis) 3) Camden, NJ (Cooper Hospital/University Medical Centre) 4) Phoenix, AZ (University of Arizona, Maricopa Medical Centre) 5) San Diego, CA (University of California, San Diego Medical Centre)

7) New Orleans, LA (Charity Hospital) 8) Irvine, CA (University of California, Irvine Medical Centre) 9) Pittsburgh, P A (University of Pittsburgh Medical Centre Presbyterian Hospital) 10) Nashville, TE (Vanderbilt University Medical Centre) 11) Stoney Brook, NY (Stony Brook Hospital and Medical Centre) 12) Stanford, CA (Stanford University School of Medicine) 13) Calgary, AB (Foothills Medical Centre) 14) Pittsburgh, P A (Pennsylvania Mercy Hospital) 15) Fresno, CA (Valley Medical Center) 16) Boston, MA (Boston University, Boston Medical Centre) 17) Minneapolis, MN (Hennepin County Medical Centre) 18) Cincinnati, OH (University of Cincinnati)

6) Kansas City, MO (Truman Medical Centre)

Fig. 2. Eighteen study centers.

Prognosis of Traumatic Intraventricular Hemorrhage

Volume 60 • Number 5 1011

tion in the study mirrored the initial CT interpretation inclinical practice. Summaries of the CT reports were enteredinto a NEXUS database, and patients with tIVH were iden-tified by a search using the keywords “IVH” and “intra-ventricular hemorrhage” and combining this search with alltrauma patients who had not yet undergone any neurosurgery.These patients constituted the “study cohort,” or the groupfrom which prevalence data were collected.

Discharge summaries, notes from follow-up appoint-ments, and extra CT reports were collected from six sitesparticipating in the follow-up portion of the study (the “out-come subset”). Mechanism of injury, presenting GlasgowComa Scale (GCS) score,43 and neurosurgical status wereextracted; the same records were used to calculate the GlasgowOutcome Scale (GOS) score44 at the time of the final, chartedevaluation of each patient. To assess interobserver agreement,several sites were selected for assessment by a second trainedreviewer, who evaluated the presence of neurosurgical inter-vention for each patient, calculated a GOS score, and alsograded associated CT findings as either “minor” or “signifi-cant.” Disagreement was resolved by third-party review bya coinvestigator.

Outcome MeasuresWhen comparing the outcomes in tIVH patients with

varying degrees of associated brain injury, as well as poten-tial predictor variables (age, mental status at presentation, andanatomic location of hemorrhage), our primary outcome mea-sure was “functional outcome,” as defined by the GOS score.

We considered a GOS score of 5 or 4 (minimal or no deficitsand moderate disability, respectively) to be a “functionaloutcome,” and a GOS score of 1, 2, or 3 to be a “pooroutcome” (death, persistent vegetative state, and severe dis-ability, respectively).44 To evaluate those patients who mayhave reached a functional outcome only because they attaineda neurosurgical intervention, a secondary outcome measurewas the “combined outcome,” which we defined as the per-formance of a directed intervention (defined as any one ofintubation for the purposes of hyperventilation, intracranialpressure [ICP] monitoring, or craniotomy, within five days ofinjury), and/or a “poor outcome.”

Data AnalysisGiven that the study is descriptive in design, formal

hypothesis testing was not utilized. Data were described us-ing proportions and relative risk (RR) ratios with 95% con-fidence intervals. Interrater agreement was calculated usingraw percentage scores.

RESULTSFrom April 1999 to December 2000, 18 NEXUS II sites

enrolled 8,374 patients with blunt head trauma for whom CTwas ordered. Abnormal head CT findings of an acute, trau-matic etiology were reported on 863 scans (10.3%; 95%confidence interval, 9.7% to 11.0%) including 118 with tIVH(13.7% of abnormal scans; 95% confidence interval, 11.5%to 16.1%). Among all trauma patients who received a headCT during the study period, the prevalence of tIVH was1.41% (95% confidence interval, 1.17% to 1.69%).

The age distribution in the 118 patients with tIVHdemonstrated a bimodal pattern (median 67; range 1 to 95),with peaks in the 21- to 30-year and 71- to 80-year agegroups (Fig. 4). Males represented 71% of patients. Bothage and sex distribution were similar to both the group ofsubjects with an abnormal head CT (n � 863) and thelarger group of all study subjects who received head CTscanning (n � 8374). Mechanism of injury was dominatedby motor vehicle accidents in the 11- to 30-year age group,and falls accounted for the largest proportion of injuries inthe 71- to 90-year age group (Fig. 5).

There were 75 patients with tIVH enrolled at the six sitesincluded in the follow-up portion of the study. Of thosepatients who lived, median follow-up was 5 weeks (range, 2days to 104 weeks), and of those patients who died themedian was 2 days (range, 2 hours to 2 weeks). Three patientsdied of a nonneurologic causes between 6 days and 3 weekspostinjury; because we were unable to assess the neurologicconsequences of their tIVH, these three patients were ex-cluded them from our final analyses (Table 1), leaving 72patients in our outcome subset. In addition there were twopatients who died 6 to 7 days postinjury in whom head injurycould not conclusively be ruled out as the cause of death(Table 1). To be conservative, we categorized these twopatients as death secondary to head injury and included them

Fig. 3. Blunt head trauma questionnaire.

The Journal of TRAUMA� Injury, Infection, and Critical Care

1012 May 2006

in the final analyses. Two patients who did not have sufficientfollow-up information to assign an accurate GOS score wereexcluded from the final analysis, further reducing the out-come subset to 70 patients. Finally, one patient was includedin the analysis despite having less than 24 hours of follow-upinformation available. This patient was “found down,” had ablood alcohol level of .304%, and on head CT had a smallpunctate hemorrhage in the right posterior temporal lobe anda small tIVH in one lateral ventricle. Only one CT was doneand when the patient became sober it is unclear if he wasdischarged or left against medical advice, but he was given“no restrictions” to his activity, and he was lost to follow-up.We assigned him a GOS score of 5 based on his last knownneurologic status.

Outcomes were very poor: only 21 (30%) patients madea “functional ” recovery, four of whom had neurosurgicalintervention, leaving 17 of 70 (24%) patients who avoided the“combined outcome.” Outcomes in patients with various po-tential predictor variables are summarized in Table 2. At leasttwo-thirds of both the older patients (age 70 years or older)and their younger counterparts had a poor outcome and thecombined outcome; there was no apparent difference betweenthe age groups. Patients who presented with a low GCS scoreshowed a trend toward worse outcomes relative to patientswho presented with a GCS �8. Although the majority ofpatients with GCS �8 did have a poor or combined outcome,this was true in over 80% of patients with a presenting GCS�8. Patients with hemorrhage identified in the third or fourthventricle fared worse than subjects without involvement ofthese ventricles, as almost all of these patients had a poor orcombined outcome, versus 62% in the group without hemor-rhage in these ventricles.

Outcomes were better among the 12 (17%) patients inour follow-up cohort who had isolated tIVH on initial CT:nine (75%) had a functional outcome and three fared poorly(GOS 3). None of the former nine patients required neuro-surgical intervention. Of the three patients with a poor out-come, all three presented with a GCS �8. In comparison tothe 12 patients with isolated tIVH, in the remaining 58 tIVHpatients (who had associated brain injury on CT), 12 (21%)had a functional outcome and only 8 (14%) avoided thecombined outcome.

Four of the 12 patients with isolated tIVH developedother findings on follow-up CT scanning, such as a contusionor subdural hygroma, leaving eight “true” isolated tIVHpatients. Seven (88%) of these patients had a functionaloutcome (with no neurosurgical interventions in the group)compared with 14 (23%) of 62 patients in the remainingnontrue isolated tIVH group (with only 10 [16%] avoidingthe combined outcome). Patients with minor CT findings inaddition to tIVH also fared better than those with majorassociated findings: 15 (65%) of 23 had a functional out-come, versus six (13%) of 47 with significant associated CTfindings. Nine (39%) of the 23 patients with minor associatedfindings suffered the combined outcome, as compared with

Fig. 4. Age and sex of 118 tIVH patients.

Fig. 5. Mechanism of injury in 118 tIVH patients. tIVH, traumaticintraventricular hemorrhage; MVA, motor vehicle crash; Other,assault, fall from horse or bicycle; “trauma,” reason for head CTlisted only as “trauma.”

Table 1 Cases in which Non-neurological ProblemsContributed to Death

PatientGroup Age/Sex Day of Death

Postinjury Summary of Final Charted Note

1 73/F Day 6 Fall down 16 stairs, Pmedhxascites & cirrhosis, malnutrition,chronic renal failure. Died ofacute on chronic renal failure,unable to extubate

84/M Day 5 Fall, suffered C4 on C5 subluxationwith central cord syndrome. Diedof cardiac arrest duringpreparation for surgery in OR

88/M Day 20 MVA, multiple trauma with multipleorthopedic injuries. Givencomfort measures in ICU, diedof “respiratory failure”

2 77/M Day 6 Died secondary to “respiratorydistress” & had a DNR

86/M Day 7 Fall, 2 episodes of bradycardiawith asystole in ICU. Familydesignated a DNR at this pointand died after third episode

Group 1: patients who died �48 hours postinjury of nonneuro-logical causes and were excluded from data analysis. Group 2: pa-tients who died of causes that could not be ruled out as neurologicaland were included in data analysis.

Pmedhx, past medical history; C: cervical; OR: operating room;MVA: motor vehicle accident; ICU: intensive care unit; DNR: do notresuscitate.

Prognosis of Traumatic Intraventricular Hemorrhage

Volume 60 • Number 5 1013

44 (94%) of 47 with significant associated CT findings.Relative risk ratios are included in Table 2.

DISCUSSIONThe prevalence of tIVH among our large cohort of patients

with blunt head trauma for whom a head CT was ordered was1.41%, which is concordant with prior reports (of between 0.4 to4%) in the medical literature.3,4,6,10,11,13,15–22 The variation inprevalence rate is likely to be partly a reflection of criteriaused for performing a head CT: older studies only utilized CTtechnology on patients with severe head injury, thus increas-ing their prevalence rates to between two and threepercent.3,16,13,19,20 Several large studies (5,000 to 7,075 pa-tients scanned) reported somewhat lower prevalence rates of0.4% and 0.5%, but they did not delineate their CT selectioncriteria, if there were any.4,11 We believe our estimate, de-rived from a large “real-world” cohort at many differentcenters, accurately reflects the prevalence of tIVH amongpatients with head trauma of a sufficient degree to require CTscanning.

Demographics of tIVH were similar to trauma patients ingeneral, with more males than females and peaks in theyoung adult and elderly age groups, suggesting that tIVHdoes not have a predilection for certain patient groups. Re-corded mechanism of tIVH also mirrored that of traumapatients in general, with a preponderance of motor vehicleaccidents among young adults and falls in the elderly, al-though it must be acknowledged that the mechanism wasrecorded simply as “trauma” in a substantial proportion ofsubjects. Of note, three elderly patients apparently sufferedtIVH after a fall from a standing position; it has traditionallybeen hypothesized that significant force is required to pro-duce tIVH.6,15 There is, however, another study that includedseveral patients with tIVH after a fall from standing position.4

Outcomes of patients with tIVH were very poor, asonly 30% of patients in this series made a “functional”

recovery and only 24% avoided the “combined outcome.”This is consistent with most previous studies, consisting of9 to 43 patients, where between 12% and 47% of tIVHpatients have a similarly defined “functional”recovery,3– 6,11,15,17 with most rates falling between 37%and 47%. The sample size in our cohort is substantiallylarger, but we were only able to obtain limited length offollow-up in this study. Up to a third of patients with aparticular GOS score may improve by one point in 1 yearand 80% of those patients will make the improvement by6 months.44 Because follow-up data in this study rangedfrom several days to 2 years, with a median of 5 weeks inpatients who survived, our early evaluation might overes-timate the severity of outcome in this series; the length offollow-up in previous studies, although not always re-ported, was most often 6 months. However, the groupswith a score of 1 (death) and 2 (persistent vegetative state)will not improve within 1 year, nor will patients with aGOS 5, and patients with GOS 4 will not improve beyondour definition of a “functional outcome.” Therefore if 7 ofthe 22 patients (one-third) with a GOS score of 3 improvedone point, the final percent of patients with a “functionaloutcome” (GOS 4 or 5) would be 40% at 1 year (39% at 6months). We also used a conservative definition of etiol-ogy of death, including several patients with multisystemtrauma in whom it was not clear whether the head injurycaused or contributed to the death was in patients, perhapsfalsely worsening our outcomes (Fig. 6), and we did notcollect other important predictors of outcome in trauma,such as hypotension and hypoxia, which may have ac-counted for some of the poor outcomes. Indeed, in ourstudy, 52% of patients presented with a GCS �8, in compar-ison to previous studies that had between 6% and 33% of tIVHpatients present with a GCS �8. We would expect this toimprove our outcomes relative to these studies,4,6,15,17 but it wasnot the case.

Table 2 Outcomes in Relation to Potential Predictor Variables and Associated CT Findings

Predictor VariableAge GCS Third or Fourth

Ventricle tIVHNo tIVH 3rd or 4th

VentricleAge 70� Age �70 GCS �8 GCS �8

n 27 43 30 32 19 39Poor outcome 21 (78%) 28 (65%) 25 (83%) 17 (53%) 18 (95%) 22 (56%)Combined outcome 21 (78%) 32 (74%) 25 (83%) 21 (66%) 19 (100%) 24 (62%)Relative risk

Poor outcome 1.2 (95% CI, 0.89 to 1.6) 1.6 (95% CI, 1.1 to 2.3) 1.7 (95% CI, 1.2 to 2.3)Combined outcome 1.0 (95% CI, 0.80 to 1.4) 1.3 (95% CI, 0.94 to 1.7) 1.6 (95% CI, 1.3 to 2.1)

Associated CT FindingsIsolated tIVH True Isolated tIVH Minor Associated CT Findings

Isolated Not Isolated True Isolated Not True Isolated Only Minor Major

n 12 58 8 62 23 47Poor outcome 3 (25%) 46 (79%) 1 (13%) 48 (77%) 8 (35%) 41 (87%)Combined outcome 3 (25%) 50 (86%) 1 (13%) 52 (84%) 9 (39%) 44 (94%)Relative risk

Poor outcome 0.32 (95% CI, 0.12 to 0.85) 0.16 (95% CI, 0.03 to 1.0) 0.40 (95% CI, 0.23 to 0.71)Combined outcome 0.29 (95% CI, 0.11 to 0.78) 0.15 (95% CI, 0.02 to 0.94) 0.42 (95% CI, 0.25 to 0.70)

The Journal of TRAUMA� Injury, Infection, and Critical Care

1014 May 2006

Although this is the largest prospective cohort of patientswith tIVH reported to date, our numbers remain small, sothere would be little utility to traditional probabilistic statis-tical analysis of any of our outcomes. In contrast to previousstudies,11,45 we did not find any trend toward worse outcomesin older patients with tIVH, although those authors studieddifferent age groups (age greater than 40 years)11 or werelooking at age in relation to head trauma in general (notspecifically tIVH).45 We did find a trend toward worse out-comes among patients with a worse mental status (lower GCSscores) at presentation, as in previous studies.6,11,17 Hemor-rhage in the third or fourth ventricle was predictive of a pooroutcome in this and another study.17

Isolated tIVH is rare: French et al. had no cases,3

whereas most tIVH studies include only rare cases of isolatedtIVH.4,10,12–15,17,18 One series, which sampled 5,000 head-in-jured patients who received a head CT, included six cases,11 thelargest group in current medical literature. In our cohort, therewere 12 (17%) patients with isolated tIVH on initial CT scan,among whom two went on to develop a subdural hygroma(and DAI in one of them), and another two patients developeda contusion (visualized on follow-up CT scan), leaving a finalgroup of eight patients with true isolated tIVH.

Although isolated tIVH is usually associated with a func-tional outcome,2,4,11,15 several studies have reported caseswith a poor outcome,10,11,46 suggesting that tIVH can, byitself, cause significant morbidity. We found a similar pat-tern: outcomes in this small group were far better than inpatients with tIVH and associated brain injury, but there werethree patients with isolated tIVH on the initial CT scan, ofwhom one did not develop any associated injury on anyfollow-up CT scan, who nevertheless had a poor outcome.However, this patient presented with a GCS score of 6; testedpositive for cocaine, barbiturates, and alcohol; and had com-plications of pneumonia and sepsis during his hospital course,which may explain his poor outcome better than his tIVH.The other two isolated tIVH patients (who developed con-current brain injuries on follow-up CT scanning) who faredpoorly included one who developed subdural hygromas withmass effect seen on MRI more than 2 months postinjury, andthe other who developed DAI on CT scan 6 days later. Bothpresented with a GCS score of less than 8, allowing them tobe readily identified. Thus, although it appears that isolatedtIVH is capable of causing poor outcomes in conjunction withan abnormal neurologic examination, we were unable to finda case of tIVH where the patient had no neurologic abnor-malities clinically, had a normal head CT save the tIVH, andstill went on to a poor or combined outcome. A recent studysuggests that serial head CTs in blunt head injury patientswho do not have neurologic changes are not warranted47; webelieve this is the case in neurologically normal tIVH patientsas well.

Lending further support to the notion that it is the asso-ciated injuries, rather than a small tIVH that is responsible forpoor outcomes, is the decreasing proportion of tIVH patients

who fared well when worse associated brain injuries wereconsidered. When patients with isolated tIVH and those withminor associated findings are considered as a group, morethan half of these 23 patients had a functional outcome,compared with less than 15% in the patients with major headCT findings. Eight of 23 patients with minor associatedfindings still fared poorly and nine had a combined outcome.Although on CT scan these injuries appeared minor, clini-cally these patients did not have minor injuries, which wouldhave allowed them to be identified for a high level of care; allof these patients presented with an abnormal GCS score saveone of them, an individual who had a seizure during hisevaluation. In addition, these patients often had complicatinginjuries, including resuscitation from cardiac arrest, DAI,interval development of subdural hygromas with mass effect,and interval development of hydrocephalus.

For developers of a clinical decision instrument that willidentify patients who need a head CT, it is important toevaluate not just outcomes of tIVH but whether there wereany signs or symptoms that would have identified each pa-tient as needing a head CT. Of the three patients with initiallyisolated tIVH who fared poorly, all three presented with aGCS �8, making the need for a head CT obvious. Three ofthe five poor outcome patients with isolated tIVH plus minorassociated CT findings presented with a GCS less than 15,while the other two, who were marked “unknown” on severalsections of the data collection sheet, had an unknown GCSbut one was noted to have an altered level of alertness whilethe other patient seized. Therefore, no patient would bemissed if the investigators of a clinical decision instrumentrelied on an abnormal GCS score, and potentially an alteredlevel of alertness and seizure to identify these patients for ahead CT.

Although most of our data were collected from a pro-spective observational cohort, outcome data were derived bychart review. Thus, although we rigorously followed acceptedcriteria for chart review methodology,48 we were still faced,not surprisingly, with a substantial amount of informationthat was unavailable from the patient charts, which limits ourresults to some extent. However, two independent reviewersused a standardized data collection form to procure follow-upinformation, with raw agreement of 97%, which should limitsignificant bias in one direction. Despite the very large size ofour overall cohort, our conclusions about tIVH itself are ofnecessity somewhat limited by a small sample size of thisgroup. Given the rarity of tIVH, however, it is unlikely that alarger, prospective study will be conducted on tIVH.

CONCLUSIONThe findings in this study suggest that tIVH is rare, and

that prognosis is determined by associated brain injuriesrather than by the tIVH itself. One case of truly isolated tIVHdid have a poor outcome, but this patient had significantmedical problems and a presenting GCS �8. Therefore, if aphysician does encounter a rare case of tIVH, the manage-

Prognosis of Traumatic Intraventricular Hemorrhage

Volume 60 • Number 5 1015

ment should be determined by the patient’s clinical presen-tation and associated injuries, and in the case of isolatedtIVH, in the presence of a normal neurologic examination, itis unlikely to warrant further evaluation. A clinical decisioninstrument that included abnormal GCS and level of alertnessshould correctly identify tIVH patients who need a CT scan.

ACKNOWLEDGMENTSThe authors wish to thank Guy Merchant, Haneefa Wawda, Marsha

Lecovey, Cindi Kirk, and Joni Fuller for their outstanding contributions tothe project, as well as the House Officers and Attending Physicians at eachof the study sites, without whose cooperation and hard work the study wouldnot have been possible.

The following centers and investigators collaborated in this study.Principal Investigators: C. Atzema, W. Mower, J. Hoffman. Site Inves-tigators: Boston University, Boston Medical Center (Boston, MA): N.Rathlev and G. Barest; Cooper Hospital, University Medical Center(Camden, NJ): C. Terregino, A. Nyce and S. Ross; Hennepin CountyMedical Center (Minneapolis, MN): B. Mahoney and J. Hollerman;Louisiana State University Medical Center, Charity Hospital (New Or-leans, LA): M. Haydel and E. Blaudeau; Maricopa Medical Center (Phoe-nix, AZ): C. Pollack and M. Connell; Stanford University Medical Center(Stanford, CA): S.V. Mahadevan and G. Arabit; State University of NewYork at Stonybrook (Stonybrook, NY): P. Viccellio and S. Fuchs; Uni-versity of Calgary, Foothills Hospital (Calgary, Canada): G. Lazarenkoand C. Fong; University of California, Davis, Medical Center (Sacra-mento, CA): J. Holmes and R.A. McFall; University of California, Irvine(Irvine, CA): J. Oman; University of California, Los Angeles, Center forthe Health Sciences (Los Angeles, CA): J. Hoffman and M. Zucker;University of California San Diego Medical Center Hillcrest (San Diego,CA): D. Guss and D. Meltzer; University of California, San Francisco atFresno, University Medical Center (Fresno, CA): G. Hendey and R.Lesperance; University of Cincinnati, University Hospital (Cincinnati,OH): G.J. Fermann and H.H. Hawkins; University of Missouri, KansasCity, Truman Medical Center (Kansas City, MO): E. Westdorp and S. Go;University of Pennsylvania Medical Center (Philadelphia, PA): J Hol-lander; University of Pittsburgh Medical Center (Pittsburgh, PA): A.Wolfson and J. Towers; Vanderbilt University Medical Center (Nash-ville, TN): V. Norton.

REFERENCES1. Abraszko RA, Zurynski YA, Dorsch NW. The significance of

traumatic intraventricular haemorrhage in severe head injury.Br J Neurosurg. 1995;9:769–773.

2. Christie M, Marks P, Liddington M. Post-traumatic intraventricularhaemorrhage: A reappraisal. Br J Neurosurg. 1988;2:343–350.

3. French BN, Dublin AB. The value of computerized tomography inthe management of 1000 consecutive head injuries. Surg Neurol.1977;7:171–183.

4. Fujitsu K, Kuwabara T, Muramoto M, Hirata K, Mochimatsu Y.Traumatic intraventricular hemorrhage: Report of twenty-six casesand consideration of the pathogenic mechanism. Neurosurg. 1988;23:423–430.

5. Graeb DA, Robertson WD, Lapointe JS, Nugent RA, Harrison PB.Computed tomographic diagnosis of intraventricular hemorrhage.Radiology. 1982;143:91–96.

6. Jayakumar PN, Sastry Kolluri VR, Basavakumar DG, Arya BYT,Das BS. Prognosis in traumatic intraventricular haemorrhage. ActaNeurochir. 1990;106:48–51.

7. Karavelis A, Sirmos Ch. Primary post-traumatic intraventricularhemorrhage. J Neurosurg Sci. 1995;39:253–256.

8. Kerr P, Iansek R, Helme RD, Rosengarten A. The prognosticsignificance of intraventricular haemorrhage. Clin Exp Neurol. 1985;

21:123–128.9. Kim CH, Tanaka R, Kawakami K, Ito J. Traumatic primary

intraventricular hemorrhage. Surg Neurol. 1981;16:415–417.10. Kobayashi S, Nakazawa S, Otsuka T. Acute traumatic intraventricular

hemorrhage in children. Child Nerv Syst. 1985;1:18–23.11. Lee JP, Lui TN, Chang CN. Acute post-traumatic intraventricular

hemorrhage analysis of 25 patients with emphasis on final outcome.Acta Neurol Scand. 1991;84:85–90.

12. Muller H, Brock M. Primary intraventricular traumatic hemorrhage.Surg Neurol. 1987;27:398–402.

13. Oliff M, Fried AM, Young AB. Intraventricular hemorrhage in blunthead trauma. J Comp Assist Tomography. 1978;2:625–629.

14. Piek J, Bock WJ. Secondary intraventricular haemorrhage in blunthead trauma. Acta Neurochir. 1986;83:105–107.

15. Sato M, Tanaka S, Kohama A, Fujii C. Traumatic intraventricularhaemorrhage. Acta Neurochir. 1987;88:95–103.

16. Zuccarello M, Iavicoli R, Pardatscher K, et al. Posttraumaticintraventricular haemorrhages. Acta Neurochir. 1981;55:283–293.

17. LeRoux PD, Haglund MM, Newell DW, Grady MS, Winn HR.Intraventricular hemorrhage in blunt head trauma: An analysis of 43cases. Neurosurg. 1992;31:678–684.

18. Cordobes F, de la Fuente M, Lobato RD, et al. Intraventricularhemorrhage in severe head injury. J Neurosurg. 1983;58:217–222.

19. Danzinger A, Price H. The evaluation of head trauma by computedtomography. J Trauma. 1979;19:1–5.

20. Dublin AB, French BN, Rennick JM. Computed tomography in headtrauma. Radiology. 1977;122:365–369.

21. Merino-de Villasante J, Taveras JM. Computerized tomography (CT)in acute head trauma. Am J Roentol. 1976;126:765–778.

22. Weisberg LA. CT and acute head trauma. Comput Tomography.1979;3:15–28.

23. Arienta C, Caroli M, Balbi S. Management of head-injured patientsin the emergency department: A practical protocol. Surg Neurol.1997;48:213–219.

24. Borczuk P. Predictors of intracranial injury in patients with mildhead trauma. Ann Emerg Med. 1995;25:731–736.

25. Duus B, Boesen T, Kruse K, Nielsen K. Prognostic signs in theevaluation of patients with minor head injury. Br J Surg. 1993;80:988–991.

26. Feuerman T, Wackym PA, Gade GF, Becker DP. Value of skullradiography, head computed tomographic scanning, and admissionfor observation in cases of minor head injury. Neurosurg. 1988;22:449–453.

27. Harad FT, Kerstein MD. Inadequacy of bedside clinical indicatorsin identifying significant intracranial injury in trauma patients.J Trauma. 1992;32:359–361.

28. Haydel MJ, Preston CA, Mills TJ, et al. Indications for computedtomography in patients with minor head injury. N Engl J Med. 2000;343:100–105.

29. Jeret JS, Mandell M, Anziska B, et al. Clinical predictors ofabnormality disclosed by computed tomography after mild headtrauma. Neurosurg. 1993;32:9–15.

30. Madden C, Witzke DB, Sanders AB, Valente J, Fritz M. High-yieldselection criteria for cranial computed tomography after acutetrauma. Acad Emerg Med. 1995;2:248–253.

31. Mikhail MG, Levitt MA, Christopher TA, Sutton MC. Intracranialinjury following minor head trauma. Am J Emerg Med. 1992;10:24–26.

32. Miller EC, Holmes JF, Derlet RW. Utilizing clinical factors toreduce head CT scan ordering for minor head trauma patients.J Emerg Med. 1997;15:453–457.

33. Mohanty SK, Thompson W, Rakower S. Are CT scans for headinjury patients always necessary? J Trauma. 1991;31:801–804.

34. Moran SG, McCarthy MC, Uddin DE, Poelstra RJ. Predictors ofpositive CT scans in the trauma patient with minor head injury.

The Journal of TRAUMA� Injury, Infection, and Critical Care

1016 May 2006

Am Surg. 1994;60:533–535.35. Nagy KK, Joseph KT, Krosner SM, et al. The utility of head

computed tomography after minimal head injury. J Trauma. 1999;46:268–270.

36. Reinus WR, Wippold FJ 2nd, Erickson KK. Practical selectioncriteria for noncontrast cranial computed tomography in patients withhead trauma. Ann Emerg Med. 1993;22:1148–1155.

37. Schynoll W, Overton D, Krome R, et al. A prospective study toidentify high-yield criteria associated with acute intracranialcomputed tomography findings in head-injured patients. Am J EmergMed. 1993;11:321–326.

38. Stein SC, Ross SE. Mild head injury: A plea for routine early CTscanning. J Trauma. 1992;33:11–13.

39. Stiell IG, Wells GA, Vandemheen K, et al. The Canadian CTHead Rule for patients with minor head injury. Lancet. 2001;357:1391–1396.

40. Taheri PA, Karamanoukian H, Gibbons K, et al. Can patients withminor head injuries be safely discharged home? Arch Surg. 1993;128:289–292.

41. Vilke GM, Chan TC, Guss DA. Use of a complete neurologicalexamination to screen for significant intracranial abnormalities inminor head injury. Am J Emerg Med. 2000;18:159–163.

42. Stiell I, Lesiuk H, Vandemheen K, et al. Obtaining consensus for thedefinition of “clinically important” brain injury in the CCC study.Acad Emerg Med. 2000;7:572b.

43. Teasdale G, Jennett B. Assessment of coma and impairedconsciousness. A practical scale. Lancet. 1974;7:81–83.

44. Jennett B, Bond M. Assessment of outcome after severe braindamage. A practical Scale. Lancet. 1975;1:480–484.

45. Pentland B, Jones PA, Roy CW, Miller JD. Head injury in theelderly. Age Ageing. 1986;15:193–202.

46. Inamasu J. Hori S. Aikawa N. Traumatic intraventricular hemorrhagecausing talk and deteriorate syndrome. Am J Emerg Med. 2001;19:167–168.

46. Brown CV, Weng J, Oh D, et al. Does routine serial computedtomography of the head influence management of traumatic braininjury? A prospective evaluation. J Trauma. 2004;57:939–943.

47. Gilbert EH, Lowenstein SR, Koziol-McLain J, Barta DC, Steiner J.Chart reviews in emergency medicine research: Where are themethods? Ann Emerg Med. 1996;27:305–308.

EDITORIAL COMMENTWith the event of computerized axial tomography (CT),

the diagnosing of intraventricular hemorrhage resulting fromblunt head trauma came “into its own.” Until that time, it wasconsidered a neuropathological condition that was lethal andinvariably a diagnosis made at necropsy.

This paper argues that the prevalence and prognosis ofintraventricular hemorrhage in blunt head trauma has notbeen well documented because of the small number of cases

analyzed to date. As a consequence, the authors employingthe National X-Radiolow Utilization Study (NEXUS #2 group)were able to obtain (from April, 1999 to December) 118 casesof intraventricular hemorrhage from an enrolled population of8,374 blunt head injuries from 18 sites. What this elaboratedatabase has provided was further substantiation of what wasalready known in the traumatic brain literature. Their studiesmerely showed that intraventricular hemorrhage is rare inthese patients and that the progress is determined by associ-ated brain injuries rather than by the presence of intraven-tricular hemorrhage.

One has to seriously wonder that if these authors are soconcerned about the number of intraventricular hemorrhagesthat have so far been reported, and feel that their review of anadditional 118 cases is so important, why didn’t they examineseveral of their references in which the total number of casesof intraventricular hemorrhage would have exceeded theirnumber. As a matter of fact these authors in handling thecrucial issue of clinical outcome provides the reader an out-come sub-set of only 70 patients from 6 of the NEXUS #2study sites. Without explanation, the other 12 sites and theirpatients appear to have been lost to follow-up and, thus, theirresults are not part of this paper. Interestingly, the patientoutcomes have apparently been examined over a 5 weekperiod of study.

Unfortunately, efforts to improve their report by provid-ing: a) study definitions (Fig. 1) and, b) a blunt head traumaquestionnaire, prove difficult to interpret. Table 2 entitled,“Outcomes to Potential Variables and Associated CT” isexcessive, complicated and confusing.

I would like to recommend to the authors, and to theirreaders, a classic publication that they have already refer-enced (ref. 17) in the area of Neurotramatology by Winn, etal, and his colleagues. While it is true, they followed only 43patients with intraventricular hemorrhage out of a cohort of1,856 blunt head trauma persons in 1992, they established,once and for all, the etiologies, clinical presentations, andneurologic courses of these patients during a 6-month periodto obtain their outcome statistics.

Robert J. White, MD, PhDMetrohealth Medical CenterCleveland, Ohio

Prognosis of Traumatic Intraventricular Hemorrhage

Volume 60 • Number 5 1017