Share this article on:

Computed Tomography–Estimated Specific Gravity at Hospital Admission Predicts 6-Month Outcome in Mild-to-Moderate Traumatic Brain Injury Patients Admitted to the Intensive Care Unit

Degos, Vincent MD, PhD*,†; Lescot, Thomas MD, PhD*; Icke, Christian MD*; Le Manach, Yannick MD*; Fero, Katherin BS; Sanchez, Paola MD, PhD*; Hadiji, Bassem MD*; Zouaoui, Abederrezak MD, PhD; Boch, Anne-Laure MD§; Abdennour, Lamine MD*; Apfel, Christian C. MD, PhD; Puybasset, Louis MD, PhD*

doi: 10.1213/ANE.0b013e318249fe7a
Neuroscience in Anesthesiology and Perioperative Medicine: Research Reports

BACKGROUND: It is clear that patients with a severe traumatic brain injury (TBI) develop secondary, potentially lethal neurological deterioration. However, it is difficult to predict which patients with mild-to-moderate TBI (MM-TBI), even after intensive care unit (ICU) admission, will experience poor outcome at 6 months. Standard computed tomography (CT) imaging scans provide information that can be used to estimate specific gravity (eSG). We have previously demonstrated that higher eSG measurements in the standard CT reading were associated with poor outcomes after severe TBI. The aim of this study was to determine whether eSG of the intracranial content predicts 6-month outcome in MM-TBI.

METHODS: We analyzed admission clinical and CT scan data (including eSG) of 66 patients with MM-TBI subsequently admitted to our neurosurgical ICU. Primary outcome was defined as a Glasgow Outcome Scale score of 1 to 3 after 6 months. Discriminating power (area under the receiver operating characteristic curve [ROC-AUC], 95% confidence interval) of eSG to predict 6-month poor outcome was calculated. The correlation of eSG with the main ICU characteristics was then compared.

RESULTS: Univariate and stepwise multivariate analyses showed an independent association between eSG and 6-month poor outcome (P = 0.001). ROC-AUC of eSG for the prediction of 6-month outcomes was 0.87 (confidence interval: 0.77–0.96). Admission eSG values were correlated with the main ICU characteristics, specifically 14-day mortality (P = 0.004), length of mechanical ventilation (P = 0.01), length of ICU stay (P = 0.045), and ICU procedures such as intracranial pressure monitoring (P < 0.001).

CONCLUSIONS: In this MM-TBI cohort admitted to the ICU, eSG of routine CT scans was correlated with mortality, ICU severity, and predicted 6-month poor outcome. An external validation with studies that include the spectrum of TBI severities is warranted to confirm our results.

Published ahead of print February 24, 2012 Supplemental Digital Content is available in the text.

From the Departments of *Anesthesiology and Critical Care, Neuroradiology, and §Neurosurgery, Groupe Hospitalier Pitié-Salpêtrière, Paris, France; and Perioperative Clinical Research Core, Department of Anesthesia and Perioperative Care, University of California at San Francisco, San Francisco, California.

Supported by Assistance Publique des Hopitaux de Paris and the nonprofit organization Fondation des Gueules Cassées, Paris, France.

The authors declare no conflicts of interest.

Reprints will not be available from the authors.

Address correspondence to Vincent Degos, MD, PhD, Perioperative Clinical Research Core, Department of Anesthesia and Perioperative Care, University of California at San Francisco, 1600 Divisadero St., San Francisco, CA 94115. Address e-mail to degosv@anesthesia.ucsf.edu.

Accepted December 21, 2011

Published ahead of print February 24, 2012

Traumatic brain injury (TBI) is the leading cause of death for those younger than 45 years in Europe,1 and every year, >1 million patients experience TBI.2 The vast majority of these cases are classified as “mild-to-moderate” TBI (MM-TBI) as measured by Glasgow Coma Scale (GCS) with a score of ≥8, whereas a smaller number are deemed “severe” (GCS score ≤7). Patients with MM-TBI typically have a better prognosis than those with severe TBI; however, a small subset of MM-TBI patients experience early neurological deterioration and, consequently, poor outcome. TBI patients have an improved outcome when transferred immediately to a neurosurgical intensive care unit (ICU).35 What is currently lacking is a clinical tool that can be used to stratify MM-TBI patients transferred to the neurosurgical ICU and define those at high risk for developing neurological deterioration. The development of such a resource would serve to better judge which of the MM-TBI patients are at high risk for complications.

Whereas several studies have focused on patients with severe brain damage, few markers have been specifically evaluated as predictors of long-term outcome in patients with MM-TBI.69 It has been demonstrated that pulsatility index measurement using transcranial Doppler in the emergency room is able to predict neurological deterioration.10 However, transcranial Doppler is an operator-dependent method that is not widely available in emergency rooms. Additionally, the serum marker S100β protein improved outcome prediction after severe TBI11 but not after mild TBI.12,13 Although there is a computed tomography (CT)-based Traumatic Coma Data Bank classification scheme for patients with severe TBI,14 no similar tool is available for MM-TBI. Currently, in addition to the GCS, one of the most recent models to predict outcome in TBI is the CRASH score. The CRASH score was developed by the Medical Research Council CRASH Trial group. It combines clinical and CT findings at admission to predict death within 14 days and poor outcome defined by a Glasgow Outcome Scale (GOS) score of 1 to 3 after 6 months.8

A different approach uses a quantitative assessment of a standard CT scan to estimate the specific gravity of the brain.15,16 The key point of this technique is that the radiological attenuation is linearly correlated with the specific gravity in the range of human tissue densities.17,18 In patients with severe TBI, a diffuse increase in estimated specific gravity (eSG) of the brain has been reported.15,19 The eSG increase was evenly distributed between the white and gray matter and was observed in the initial CT scan immediately after the trauma. Significantly higher eSG in the noncontused areas was illustrated in 120 patients with severe TBI. Additionally, high eSG values have been associated with brain swelling at admission and with greater therapeutic intensity level in the ICU.19 Interestingly, eSG values of the noncontused hemisphere area were found to be highly correlated with the overall intracranial content area,19 which suggests the potential to make a semiautomatic measurement of eSG. As of yet, however, eSG values have not been evaluated for their ability to predict outcome in MM-TBI.

We hypothesized that the eSG value of overall intracranial content area from an initial CT scan can be used as an outcome predictor for MM-TBI patients admitted to the ICU. In this study, we compared eSG values with clinical and admission CT scan findings, using 6-month poor outcome (defined as a GOS score of 1–3) as the primary end point.

Back to Top | Article Outline

METHODS

Patients

This study was approved by our IRB and was performed in accordance with the ethical standards in the Declaration of Helsinki. Because the study data were collected through standard care practices, the requirement to obtain written informed consent was waived. However, patients and families were informed that CT imaging scans and clinical data would be used for this study. None of the patients or controls was included in our previous studies.15,16,19 For this study, we searched our neurosurgical ICU's prospectively established database of TBI patients to identify those who met the following inclusion criteria: initial GCS score ≥8 and an initial CT scan performed at our hospital (with High Speed Siemens Sensation 16; Siemens, New York, NY) within 10 hours of the injury and before any neurosurgical intervention or any artificial device had been inserted into the brain. All patients were admitted to our neurosurgical ICU for clinical follow-up and treated there according to our ICU's previously described treatment algorithm.15,20 Patients with major extracranial injuries (hemorrhagic injuries with hemostatic surgical procedures) were not admitted to our neurosurgical ICU and thus were not included in this study.

Back to Top | Article Outline

Control CT for Standardization

To standardize the CT scan data, control values of eSG were obtained from randomly selected patients who had normal findings on CT scans performed to evaluate headaches. The control CT measurements were performed with the same CT scan apparatus as that used for the TBI patients. All CT scans were assessed by a neuroradiologist (AZ). Controls were matched with TBI patients for age (40 ± 10 years) and gender (20 men and 10 women).

Back to Top | Article Outline

Clinical and Therapeutic Characteristics Collection

The clinical characteristics of each patient were extracted from the neurosurgical ICU database. The following variables were recorded upon each patient's arrival to the hospital and just before the CT scan: mean GCS score, pupil reaction to light, and mechanism and kinetics of the accident. To characterize the ICU treatment intensity, the following items concerning intracranial hypertension therapies were recorded: continuous norepinephrine infusion ≥0.5 mg/h to maintain cerebral perfusion pressure between 65 and 70 mm Hg, mechanical ventilation, external ventricular drainage device, continuous midazolam infusion ≥10 mg/h, continuous propofol infusion ≥100 mg/h, thiopental in any dose, continuous neuromuscular blocking drug infusion at any dose level, therapeutic hypothermia defined as use of a cooling blanket to maintain the core temperature <37°C, and the sum of the hypertonic sodium chloride boluses. A single physician (LP) determined the GOS score (ranging from 1 meaning death to 5 meaning full recovery) 6 months after the injury.

Back to Top | Article Outline

Radiological Characteristics Collection

CT scan findings were taken from our central neuroradiological storage system. All CT scans were performed with the same apparatus, and manufacturer technicians performed complete calibration of the machine monthly as a routine procedure. An internal check was also performed once a week for the absence of drift. CT images were obtained as 5-mm-thick contiguous slices including posterior fossa (corresponding to an average of 30 slices/patient). CT images were analyzed using a custom-designed software package (BrainView) as we have described and validated elsewhere for the human brain.15,19 BrainView provides semiautomatic tools for brain analysis and quantification from cerebral CT scan images. For each CT image, BrainView inputs a series of continuous axial scans of the brain. It then automatically excludes extracranial compartments on each CT section by means of a mathematical morphology-based algorithm. The overall intracranial content was delineated after automatic delineation of the intradural content and manual exclusion of any extradural hematoma (EDH). Subdural hematomas (SDHs), ventricles, and intraparenchymal brain hematomas (IBHs) were not excluded from the intracranial content measurements. For each section of a known number of voxels, the volume, weight, and eSG were computed using the following equations:

(1) Volume of the voxel (mL) = Surface (mm2) × Section thickness (mm)

(2) Weight of the voxel (g) = (1 + CT/1000) × Volume of the voxel (mL)

where CT is the attenuation coefficient (expressed in Hounsfield unit).

(3) Volume of the compartment = Number of voxels × Volume of the voxel

(4) Weight of the compartment (g) = summation of the weight of each individual voxel included in the compartment

(5) eSG of the compartment (g/mL) = Weight of the compartment (g)/Volume of the compartment (mL).

The eSG is expressed as a physical density in grams per milliliter.

A neuroradiologist (AZ), who was blinded to patient data, recorded retrospectively and in a blinded manner CT scan findings as brain petechiae, EDH, subarachnoid hemorrhage (SAH), SDH, IBH, and brain edema.

Back to Top | Article Outline

Potential Predictors of Outcome

The ability of eSG to predict 2 major outcome measures was assessed: death within 14 days and poor 6-month outcome as described by a GOS score between 1 and 3 at 6 months. We compared the predictive performance of eSG with the GCS and the CRASH radio-clinical score8 and the other CT scan findings. The CRASH score (calculation online, www.crash2.lshtm.ac.uk/Risk%20calculator),8 which considers 6 items (country, age, GCS, pupil modification, major intracranial injury, and CT scan findings), was used to calculate probabilities of 14-day mortality and 6-month poor outcomes. The country status corresponded to the “high income countries” status, the GCS was coded 14 for the patients who did not present any GCS deterioration, and the CT scan findings were coded according to a neuroradiologist reading as previously described.

Back to Top | Article Outline

Statistical Analysis

Continuous data were expressed as mean ± SD, integer variables as median, and categorical data as frequency (percentage). Significance of categorical data was tested using the Fisher exact method, and significance of continuous data was tested using the Student t test after testing whether variances were equal (F test).

To test the null hypothesis that admission characteristics were not associated with 6-month outcome, univariate comparison for the different admission findings (GOS score 4 or 5 vs scores 1–3) at 6 months was followed by multivariate logistic regression stepwise analyses. Variances were equal (F test) for all admission characteristics. All variables with a P < 0.20 in the univariate analyses were entered into the multivariate logistic regression. Linearity of eSG in the final model was assessed using a 4-quartile distribution according to the eSG values. Calibration and discrimination of the logistic models were assessed using the Hosmer-Lemeshow statistics and the receiver operating characteristic (ROC) curves, respectively. The area under the ROC curve (ROC-AUC) of eSG (continuous values) was compared with the ROC-AUC of the CRASH score (continuous values), the GCS admission values, and the presence of brain edema on the admission CT scan with 2-sided P values with and without adjustments for multiple comparisons. The ROC-AUC were expressed with 95% confidence intervals (CIs) according to Delong et al.21

To test the null hypothesis that eSG was independent of other admission CT scan findings, we compared the eSG mean values according to the presence of brain petechial, EDH, SDH, IBH, and brain edema using a t test with equal variance adjusted for multiple comparisons. Variances were equal (F test) for all CT scan findings group.

To dichotomize the MM-TBI population according to their eSG value, a calculated cutoff value was defined as the mean eSG + 2 SE, corresponding to the upper bound of the 95% CI of eSG's mean. Using the calculated cutoff value, 2 groups were defined: low eSG and high eSG. The sensitivity and the specificity with 95% CIs of the CT scan findings (brain petechiae, EDH, SAH, SDH, IBH, edema, and high eSG) to predict 6-month outcome were calculated.

Finally, we tested the correlation of eSG continuous values with individual characteristics (patient characteristics, ICU characteristics, and outcome markers), using the Spearman test for continuous variables and the Wilcoxon rank sum (Mann-Whitney test) for binary variables.

With an expected mean eSG difference of 0.003 between the good and the poor 6-month outcome groups, an expected standard deviation of 0.002 (determinate from Ref. 19), and an α risk of 0.05, the number of patients to include was at least 10 patients per group to achieve a power >80%.

All P values were 2-tailed and P values <0.05 were considered to be significant. Statistical analyses were performed using Stata version 11.1 (StataCorp, College Station, TX).

Back to Top | Article Outline

RESULTS

Patient Cohort

From May 1, 2004 to November 1, 2006, 217 TBI patients were admitted to our neurosurgical ICU, including 95 patients with GCS values of ≥8 (95 of 217 = 44%). Admission to our neurosurgical ICU occurred when neurological degradation was anticipated based on clinical and CT scan findings. Twenty-nine patients (29 of 95 = 31%) were excluded because the initial CT scan was performed at another hospital (18 patients) or after a neurosurgical procedure (11 patients). The 66 remaining patients were included in the study.

Patients' characteristics are presented in Table 1. The mean patient age was 38 ± 17 years. Ten patients were younger than 20 years of age, 37 patients were between 20 and 49 years, 14 patients between 50 and 59 years, and 6 patients were older than 60 years. The median GCS score upon patient arrival to the hospital was 14 (95%CI: 11–15). Ten patients had a GCS score of 8 or 9, 21 of 10 to 13, 17 of 14, and 18 of 15. None of the patients had mydriasis or major extracranial injuries. None of the patients received hyperosmotic solutions (colloids or crystalloids) before CT scan acquisition. The mechanisms of injury were passenger motor vehicle accident (22 of 66 = 33%), fall (23 of 66 = 35%), assault (14 of 66 = 21%), and pedestrian motor vehicle accident (6 of 66 = 9%). Six patients (9%) died in the ICU within the first 14 days. The mean age of the deceased patients was 37 ± 16 years and all the deaths were related to the primary brain lesion deterioration with uncontrolled intracranial hypertension. The mean length of ICU stay was 17 ± 14 days. Six months after injury, 12 patients (18%) presented with a poor outcome (6 deceased patients GOS score 1, and 6 surviving patients GOS scores 2 and 3). The CRASH model predicted a mean 14-day mortality of 9% ± 9% and a mean risk of 6-month poor outcome (GOS scores 1–3) of 28% ± 17%. Twenty-five patients (38%) required neurosurgical procedures other than ventricular drain insertion, including 11 patients (17%) requiring surgery for EDH and 42 patients (64%) presented with IBH. During their ICU stay, 40 patients (60%) required mechanical ventilation and intracranial pressure (ICP) was measured in 25 patients (38%).

Table 1

Table 1

For the 66 patients included, CT scans were performed on average 4 ± 4 hours after TBI, all within 10 hours of the injury and before any neurosurgical intervention. The standardized CT scan values of the weight, the volume, and the eSG were respectively 1440 ± 103 g, 1398 ± 100 mL, and 1.0295 ± 0.0016 g/mL. As shown in Table 2, admission value for eSG, presence of brain edema, and presence of IBH were associated with 6-month poor outcome. The stepwise multivariate analyses showed that eSG was significantly associated with 6-month poor outcome (P = 0.001). Interestingly, eSG was also associated with 14-day mortality (P = 0.0017; Fig. 1). Among CT findings, the presence of EDH, SDH, SAH, brain edema, IBH, and brain petechiae was not significantly associated with an increase of eSG value after adjustment for multiple comparisons.

Table 2

Table 2

Figure 1

Figure 1

Back to Top | Article Outline

Comparison of the Admission Predictors for 6-Month Outcome

The discrimination abilities of the admission characteristics were compared with their ROC-AUC. The ROC-AUC of eSG as a continuous value was 0.87 (CI: 0.77–0.96). The ROC-AUC of eSG was compared with the ROC-AUCs of the GCS, the CRASH score, and the presence of brain edema at admission (Table 3, Fig. 2). After adjustment for multiple comparisons for these 4 predictors (adjusted significance level at 0.017), the ROC-AUC of eSG was significantly superior to the CRASH score and the presence of brain edema for predicting 6-month outcome (Table 3).

Table 3

Table 3

Figure 2

Figure 2

Back to Top | Article Outline

Correlation Between eSG and Patient Characteristics

As mean eSG was 1.03056 g/mL with a 95% CI of 1.0299 to 1.0312 g/mL, the calculated cutoff value was 1.0313 g/mL. Twenty-five patients (29% of the MM-TBI group) were above the cutoff. Sensitivity and specificity of the CT scan findings for the MM-TBI patients admitted to the ICU were calculated using this cutoff. Sensitivity was 83% and specificity was 67% to predict 14-day mortality and 92% and 74%, respectively, to predict 6-month poor outcome (Table 4). All the other CT scan findings predicted 6-month outcome with lower sensitivity.

Table 4

Table 4

The correlations between eSG values and the main characteristics of the MM-TBI patients are presented in Table 5. Among the patient characteristics, eSG was not correlated with age, gender, or the GCS score. In addition, the kinetics of the trauma (P = 0.11) and glucose concentration at admission (P = 0.83) were also not correlated with eSG values. Interestingly, admission eSG values were correlated with the main ICU characteristics, specifically length of mechanical ventilation (P = 0.01), length of ICU stay (P = 0.045), and more frequently required ICU treatments and procedures such as ICP monitoring (P < 0.001) (Table 5). Admission eSG values were also associated with 14-day mortality (P = 0.004) and 6-month outcome (P = 0.0001). In Table 6, the association of the main ICU characteristics with high eSG values was compared according to different cutoff value definitions, considering the adjustment for multiple comparisons.

Table 5

Table 5

Table 6

Table 6

Back to Top | Article Outline

DISCUSSION

The goal of this study was to assess the effectiveness of eSG in predicting the risk of 6-month poor outcome in patients with MM-TBI who were admitted to the neurosurgical ICU. We found that eSG measured on the first routine CT scan (an average of 4 hours after injury) was a predictor for 6-month poor outcome. The eSG measure performed better than the well-established CRASH score for predicting 6-month poor outcome, and was correlated with the main ICU characteristics such as 14-day mortality, length of stay in ICU, and length of ventilation time. In addition to being a good outcome marker, the quantitative nature of eSG measured on a routine CT scan at hospital admission is advantageous over other scores.

In this study, the eSG of the total intracranial content was measured, as opposed to a particular region of interest, to make the measurement easier, and quicker, and to allow for the possibility of developing automatic measurement techniques. We have previously shown that the eSG of the total intracranial content is correlated with the eSG of the hemispheric areas, and that the 2 values are very close.19 Interestingly, eSG increase affects both gray and white matter.15 Delineation of the intracranial content, automatically excluding the skull, eye sockets, sinuses, and artificial devices, was performed using computer software. A comparison of the control groups used for standardization between the preceding study with a different CT scan apparatus (GE Medical Systems, Milwaukee, WI),19 and this study showed that the difference between eSG means of the total intracranial content differed by 0.002 g/mL, corresponding to 2 Hounsfield units. This difference is explained by the fact that the 2 studies were performed on 2 different CT machines from 2 different manufacturers. Each machine requires standardization with 1 group of controls and regular routine calibration before interpretation of eSG.

To show the ability of eSG to predict outcome, we chose to evaluate indicators that were validated and/or clinically relevant such as the GCS, the CRASH score, and the CT scan findings. To best compare eSG with the CRASH score,8 we evaluated the same indicators (14-day mortality and 6-month outcome). Additionally, the correlation between eSG and different ICU end points such as length of stay in the ICU, length of mechanical ventilation time, and ICU characteristics is supported by our previous study about severe TBI.19 Because ICP is controlled by therapeutic strategies, and because previous research has demonstrated no correlation between brain densities measured by quantitative CT scan and ICP value,22 we decided to use indirect indicators such as therapeutic strategies to characterize the severity, instead of ICP.

Among the 66 MM-TBI patients studied, 12 patients (18%) presented a 6-month poor outcome, including 6 patients (9%) who died within 14 days after injury. When the CRASH score was applied to the study population, 28% of the patients were expected to have a 6-month poor outcome and 9% were expected to die within 14 days, despite a median GCS score of 14 ± 2. Noteworthy is the surprisingly high morbidity-mortality of the MM-TBI study patients. It is important to clarify that in this cohort, patients were selected only after being transferred into the neurosurgical ICU. This selection criterion likely explains the high morbidity-mortality of these patients. This study does not reflect the evolution of the majority of the MM-TBI cases but does reinforce the fact that a part of the MM-TBI population is at a high risk of neurological deterioration.

Regarding the observed association between early eSG elevation and subsequent increase in mortality and disability, several hypotheses can be put forward. Studies have shown that the intracerebral blood volume is diminished in patients with TBI23,24 and eSG increase persists for 3 weeks after TBI19 refuting the hyperemia hypothesis. However, rapid and diffuse blood-brain barrier (BBB) opening lasting <30 minutes has been described in animal models of TBI.2528 Interestingly, increased specific gravity of the white matter measured using the microgravimetric method was already described in anatomopathological studies from TBI patients.29 Thus, it can be hypothesized that BBB opening occurs immediately after TBI in a subgroup of patients, leading to extravasation of an exudate into the extracellular space and therefore to increased eSG and prolonging of the increased ICP period. This extravasation could be associated with early leukocyte infiltration and microglia proliferation already described in TBI.30 Thus, eSG may reflect the early BBB disruption specifically associated with TBI31 and may explain why it is correlated with TBI outcome.

Finally, limitations of this study include the fact that the study was monocentric and the 66 patients' data were retrospectively extracted from a prospectively established database. The CRASH score results support the fact that the observed outcome values of this cohort were similar to the predicted outcome values. Because the calculated CRASH score considers the country in which the patient was treated, the estimates are based on 2 alternative sets of models (high-income countries or low- and middle-income countries). However, because this study was conducted in 1 center and 1 country only, it is necessary to point out that cultural factors in different countries could affect outcomes.32

After showing that eSG was independently linked with 6-month outcome, we determined the correlation of ICU characteristics with the eSG values (Table 5). The next step would be to define an optimal cutoff value for eSG. Because the choice of any cutoff value can be viewed as somewhat arbitrary, we calculated the P values according to 5 different cutoffs in our small cohort (1.0306, 1.0310, 1.0313, 1.0318, and 1.0322 g/mL) (Table 6). An external validation with larger cohorts will be necessary to define the optimal cutoff. Additionally, it is important to note that because a patient's eSG value is directly dependent on the CT scan machine and the anatomical part of the brain explored, the cutoff eSG value is specific to this study. Before eSG measurements can be widely implemented in clinical use, prospective data about eSG measurement will be needed.

Back to Top | Article Outline

CONCLUSION

In this population, eSG of routine initial CT scans was associated with 6-month poor outcome. In addition, eSG was also correlated with ICU indicators such as increased length of stay in the ICU and length of ventilation time. To facilitate eSG measurement immediately upon admission in clinical practice, our group is planning to develop systematic, automatic, and integrative CT scan analysis software. An external validation with a multicenter prospective study in a large population of emergency room patients with MM-TBI is warranted to validate eSG as a predictor of MM-TBI outcome.

Back to Top | Article Outline

DISCLOSURES

Name: Vincent Degos, MD, PhD.

Contribution: This author helped design the study, analyze the data, and prepare the manuscript.

Attestation: Vincent Degos attests to the integrity of the original data and the analysis.

Name: Thomas Lescot, MD, PhD.

Contribution: This author helped design the study and collect the data.

Attestation: Thomas Lescot attests to the integrity of the original data and the analysis.

Name: Christian Icke, MD.

Contribution: This author helped collect the data.

Name: Yannick Le Manach, MD.

Contribution: This author helped analyze the data.

Name: Katherin Fero, BS.

Contribution: This author helped prepare the manuscript.

Name: Paola Sanchez, MD, PhD.

Contribution: This author helped collect the data.

Name: Bassem Hadiji, MD.

Contribution: This author helped collect the data.

Name: Abederrezak Zouaoui, MD, PhD.

Contribution: This author helped conduct the study and collect the data.

Name: Anne-Laure Boch, MD.

Contribution: This author helped conduct the study.

Name: Lamine Abdennour, MD.

Contribution: This author helped conduct the study and collect the data.

Name: Christian C. Apfel, MD, PhD.

Contribution: This author helped analyze the data and prepare the manuscript.

Name: Louis Puybasset, MD, PhD.

Contribution: This author helped design the study, conduct the study, collect the data, analyze the data, and prepare the manuscript.

Attestation: Louis Puybasset attests to the integrity of the original data and the analysis.

This manuscript was handled by: Gregory J. Crosby, MD.

Back to Top | Article Outline

ACKNOWLEDGMENTS

We thank the nurses of the Neurosurgical Intensive Care Unit and the technicians of the Neuroradiology Department for their active participation in this study. We thank all of the members of the Perioperative Clinical Research Core at UCSF for their help. We also thank Rachel Whelan from Dr. Apfel's Clinical Research Core for her outstanding editorial assistance.

Back to Top | Article Outline

REFERENCES

1. Maas AI, Stocchetti N, Bullock R. Moderate and severe traumatic brain injury in adults. Lancet Neurol 2008;7:728–41
2. Tagliaferri F, Compagnone C, Korsic M, Servadei F, Kraus J. A systematic review of brain injury epidemiology in Europe. Acta Neurochir (Wien) 2006;148:255–68
3. Diringer MN, Edwards DF. Admission to a neurologic/ neurosurgical intensive care unit is associated with reduced mortality rate after intracerebral hemorrhage. Crit Care Med 2001;29:635–40
4. Nathens AB, Jurkovich GJ, Maier RV, Grossman DC, MacKenzie EJ, Moore M, Rivara FP. Relationship between trauma center volume and outcomes. JAMA 2001;285:1164–71
5. Patel HC, Menon DK, Tebbs S, Hawker R, Hutchinson PJ, Kirkpatrick PJ. Specialist neurocritical care and outcome from head injury. Intensive Care Med 2002;28:547–53
6. Wedekind C, Hesselmann V, Lippert-Gruner M, Ebel M. Trauma to the pontomesencephalic brainstem: a major clue to the prognosis of severe traumatic brain injury. Br J Neurosurg 2002;16:256–60
7. Hoelper BM, Reinert MM, Zauner A, Doppenberg E, Bullock R. rCBF in hemorrhagic, non-hemorrhagic and mixed contusions after severe head injury and its effect on perilesional cerebral blood flow. Acta Neurochir Suppl 2000;76:21–5
8. Perel P, Arango M, Clayton T, Edwards P, Komolafe E, Poccock S, Roberts I, Shakur H, Steyerberg E, Yutthakasemsunt S. Predicting outcome after traumatic brain injury: practical prognostic models based on large cohort of international patients. BMJ 2008;336:425–9
9. Yanagawa Y, Tsushima Y, Tokumaru A, Un-no Y, Sakamoto T, Okada Y, Nawashiro H, Shima K. A quantitative analysis of head injury using T2*-weighted gradient-echo imaging. J Trauma 2000;49:272–7
10. Jaffres P, Brun J, Declety P, Bosson JL, Fauvage B, Schleiermacher A, Kaddour A, Anglade D, Jacquot C, Payen JF. Transcranial Doppler to detect on admission patients at risk for neurological deterioration following mild and moderate brain trauma. Intensive Care Med 2005;31:785–90
11. Korfias S, Stranjalis G, Boviatsis E, Psachoulia C, Jullien G, Gregson B, Mendelow AD, Sakas DE. Serum S-100B protein monitoring in patients with severe traumatic brain injury. Intensive Care Med 2007;33:255–60
12. Stapert S, de Kruijk J, Houx P, Menheere P, Twijnstra A, Jolles J. S-100B concentration is not related to neurocognitive performance in the first month after mild traumatic brain injury. Eur Neurol 2005;53:22–6
13. Bazarian JJ, Zemlan FP, Mookerjee S, Stigbrand T. Serum S-100B and cleaved-tau are poor predictors of long-term outcome after mild traumatic brain injury. Brain Inj 2006;20:759–65
14. Marshall LF, Marshall SB, Klauber MR, Van Berkum Clark M, Eisenberg H, Jane JA, Luerssen TG, Marmarou A, Foulkes MA. The diagnosis of head injury requires a classification based on computed axial tomography. J Neurotrauma 1992;9:S287–92
15. Lescot T, Bonnet MP, Zouaoui A, Muller JC, Fetita C, Coriat P, Puybasset L. A quantitative computed tomography assessment of brain weight, volume, and specific gravity in severe head trauma. Intensive Care Med 2005;31:1042–50
16. Lescot T, Degos V, Zouaoui A, Preteux F, Coriat P, Puybasset L. Opposed effects of hypertonic saline on contusions and noncontused brain tissue in patients with severe traumatic brain injury. Crit Care Med 2006;34:3029–33
17. Phelps ME, Gado MH, Hoffman EJ. Correlation of effective atomic number and electron density with attenuation coefficients measured with polychromatic x rays. Radiology 1975;117:585–8
18. Mull RT. Mass estimates by computed tomography: physical density from CT numbers. Am J Radiol 1984;143:1101–4
19. Degos V, Lescot T, Zouaoui A, Hermann H, Preteux F, Coriat P, Puybasset L. Computed tomography-estimated specific gravity of noncontused brain areas as a marker of severity in human traumatic brain injury. Anesth Analg 2006;103:1229–36
20. Lescot T, Abdennour L, Boch AL, Puybasset L. Treatment of intracranial hypertension. Curr Opin Crit Care 2008;14:129–34
21. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988;44:837–45
22. Inaba K, Teixeira PG, David JS, Brown C, Salim A, Rhee P, Browder T, Dubose J, Demetriades D. Computed tomographic brain density measurement as a predictor of elevated intracranial pressure in blunt head trauma. Am Surg 2007;73:1023–6
23. Beaumont A, Marmarou A, Hayasaki K, Barzo P, Fatouros P, Corwin F, Marmarou C, Dunbar J. The permissive nature of blood brain barrier (BBB) opening in edema formation following traumatic brain injury. Acta Neurochir Suppl 2000;76:125–9
24. Marmarou A, Fatouros PP, Barzo P, Portella G, Yoshihara M, Tsuji O, Yamamoto T, Laine F, Signoretti S, Ward JD, Bullock MR, Young HF. Contribution of edema and cerebral blood volume to traumatic brain swelling in head-injured patients. J Neurosurg 2000;93:183–93
25. Baldwin SA, Fugaccia I, Brown DR, Brown LV, Scheff SW. Blood-brain barrier breach following cortical contusion in the rat. J Neurosurg 1996;85:476–81
26. Barzo P, Marmarou A, Fatouros P, Corwin F, Dunbar J. Magnetic resonance imaging-monitored acute blood-brain barrier changes in experimental traumatic brain injury. J Neurosurg 1996;85:1113–21
27. Chen Y, Constantini S, Trembovler V, Weinstock M, Shohami E. An experimental model of closed head injury in mice: pathophysiology, histopathology, and cognitive deficits. J Neurotrauma 1996;13:557–68
28. Habgood MD, Bye N, Dziegielewska KM, Ek CJ, Lane MA, Potter A, Morganti-Kossmann C, Saunders NR. Changes in blood-brain barrier permeability to large and small molecules following traumatic brain injury in mice. Eur J Neurosci 2007;25:231–8
29. Bullock R, Smith R, Favier J, du Trevou M, Blake G. Brain specific gravity and CT scan density measurements after human head injury. J Neurosurg 1985;63:64–8
30. Israelsson C, Wang Y, Kylberg A, Pick CG, Hoffer BJ, Ebendal T. Closed head injury in a mouse model results in molecular changes indicating inflammatory responses. J Neurotrauma 2009;26:1307–14
31. Degos V, Pereira AR, Lescot T, Sanchez-Pena P, Daoudi M, Zouaoui A, Coriat P, Puybasset L. Does brain swelling increase estimated specific gravity? Neurocrit Care 2008;9:338–43
32. Andrew Kofke W, Sharma D. Predicting survival in NeuroICU patients. J Neurosurg Anesthesiol 2011;23:177–8
© 2012 International Anesthesia Research Society