RESEARCH—HUMAN—CLINICAL STUDIES RESEARCH—HUMAN—CLINICAL STUDIES Qiang Yuan, MD*‡ Xing Wu, MD*‡ Hongwei Cheng, MD§ Chaohua Yang, MD¶ Yuhai Wang, MDk Ersong Wang, MD# Binghui Qiu, MD** Zhimin Fei, MD‡‡ Qing Lan, MD§§ Sirong Wu, MD¶¶ Yunzhao Jiang, MDkk Hua Feng, MD## Jingfang Liu, MD*** Ke Liu, MD‡‡‡ Fayun Zhang, MD§§§ Rongcai Jiang, MD¶¶¶ Jianmin Zhang, MDkkk Yue Tu, MD### Xuehai Wu, MD‡ Liangfu Zhou, MD‡ Jin Hu, MD‡ ‡Department of Neurosurgery, Huashan Hospital, Fudan University, Shanghai, China; §Department of Neurosurgery, The First Affiliated Hospital of Anhui Medical University, Hefei, China; ¶Department of Neurosurgery, West China Hospital of Sichuan University, Chengdu, China; kDepartment of Neurosurgery, 101 Hospital of People’s Liberation Army, Wuxi, China; #Department of Neurosurgery, Jinshan Hospital, Fudan University, Shanghai, China; **Department of Neurosurgery, The South Hospital of Southern Medical University, Guangzhou, China; ‡‡Department of Neurosurgery, Shanghai Shuguang Hospital, Shanghai, China; §§Department of Neurosurgery, The Second Affiliated Hospital of Soochow University, Suzhou, China; ¶¶Department of Emergency Medicine, The First Affiliated Hospital of Soochow University, Suzhou, China; kkDepartment of Neurosurgery, Wuxi Third People’s Hospital, Wuxi, China; ##Department of Neurosurgery, Southwest Hospital, Chongqing, China; ***Department of Neurosurgery, Xiangya Hospital Central South University, Changsha, China; ‡‡‡Department of Neurosurgery, Chongqing Emergency Medical Center, Chongqing, China; §§§Department of Neurosurgery, Enze Medical Center Luqiao Hospital, Taizhou, China; ¶¶¶Department of Neurosurgery, General Hospital of Tianjing Medical University, Tianjin, China; kkkDepartment of Neurosurgery, The Second Affiliated Hospital of Zhejiang University School of Medicine, Zhejiang, China; ###Department of Neurosurgery, Affiliated Hospital of Logistics University of People’s Armed Police Force, Tianjin, China *These authors have contributed equally to this article. Correspondence: Jin Hu, MD, Department of Neurosurgery, Huashan Hospital, Fudan University, 12 Wulumuqi Zhong Rd, Shanghai 200040, China. E-mail: [email protected]

Is Intracranial Pressure Monitoring of Patients With Diffuse Traumatic Brain Injury Valuable? An Observational Multicenter Study BACKGROUND: Although intracranial pressure (ICP) monitoring of patients with severe traumatic brain injury (TBI) is recommended by the Brain Trauma Foundation, any benefits remain controversial. OBJECTIVE: To evaluate the effects of ICP monitoring on the mortality of and functional outcomes in patients with severe diffuse TBI. METHODS: Data were collected on patients with severe diffuse TBI (Glasgow Coma Scale [GCS] score on admission ,9 and Marshall Class II-IV) treated from January 2012 to December 2013 in 24 hospitals (17 level I trauma centers and 7 level II trauma centers) in 9 Chinese provinces. We evaluated the impact of ICP monitoring on 6-month mortality and favorable outcome using propensity score–matched analysis after controlling for independent predictors of these outcomes. RESULTS: ICP monitors were inserted into 287 patients (59.5%). After propensity score matching, ICP monitoring significantly decreased 6-month mortality. ICP monitoring also had a greater impact on the most severely injured patients on the basis of head computed tomography data (Marshall computed tomography classification IV) and on patients with the lowest level of consciousness (GCS scores 3-5). After propensity score matching, monitoring remained nonassociated with a 6-month favorable outcome for the overall sample. However, monitoring had a significant impact on the 6-month favorable outcomes of patients with the lowest level of consciousness (GCS scores 3-5). CONCLUSION: ICP monitor placement was associated with a significant decrease in 6-month mortality after adjustment for the baseline risk profile and the monitoring propensity of patients with diffuse severe TBI, especially those with GCS scores of 3 to 5 or of Marshall computed tomography classification IV. KEY WORDS: Diffuse injury, Intracranial pressure monitoring, Outcome, Propensity score, Traumatic brain injury Neurosurgery 78:361–369, 2016

DOI: 10.1227/NEU.0000000000001050

T

raumatic brain injury (TBI) is a common and potentially devastating problem worldwide.1-3 Treatment of severe TBI principally seeks to reduce the elevated intracranial pressure (ICP) and to maintain adequate cerebral

Received, April 30, 2015. Accepted, August 20, 2015. Published Online, October 8, 2015. Copyright © 2015 by the Congress of Neurological Surgeons.

ABBREVIATIONS: AOR, adjusted odds ratio; CI, confidence interval; GCS, Glasgow Coma Scale; ICP, intracranial pressure; GOSE, Glasgow Outcome Scale Extended; TBI, traumatic brain injury Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (www.neurosurgery-online.com).

NEUROSURGERY

www.neurosurgery-online.com

blood flow and oxygenation.4,5 However, the pathophysiology of TBI is heterogeneous, featuring both focal and diffuse injuries. Thus, optimal therapeutic strategies differ.6-8 On computed tomography (CT), diffuse TBI usually lacks visible focal lesions but features compression of the basal cisterns or third ventricle, along with a midline shift.9 Since the Marshall CT classification was introduced in 1991,10 the system has become widely used to describe the features of patients with diffuse injuries and is a valuable predictor of patient outcomes.8 Patients with nonfocal diffuse lesions are at high risk of intracranial hypertension and mortality, although their mortality is lower than

VOLUME 78 | NUMBER 3 | MARCH 2016 | 361

Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited

YUAN ET AL

that of patients with mass lesions.11,12 Therefore, ICP monitoring is widely used to manage diffuse TBI; the ICP is measured both before and after surgery to explore whether decompressive craniectomy, either bifrontal (in cases of diffuse brain swelling) or unilateral (in cases of hemispheric swelling), effectively lowered the ICP.13-15 Although ICP monitoring of patients with severe TBI is recommended by the Brain Trauma Foundation,16 any benefits remain controversial. In addition, Chesnut et al17 conducted the first randomized controlled trial of such monitoring in 2012. This study showed no evidence of independent mortality benefit of monitoring, but the trial design was controversial. Clinical methods for interpretation of ICP data from individual patients are required; monitoring may be effective for some but not all TBI patients.18,19 A new technique called propensity score matching better controls for selection bias and more accurately estimates the true effects of observational data.20,21 Therefore, we used this technique to perform an observational multicenter study evaluating the effects of ICP monitoring on the mortality of and functional outcomes in patients with severe diffuse TBI.

METHODS Design and Setting A database was created by the neurosurgery department of Huashan Hospital of Fudan University to investigate the efficacy of TBI management guided by ICP monitoring in China. A standardized, structured questionnaire was completed by physicians and checked by local investigators; we collected data on all patients with TBI treated from January 2012 to December 2013 in 24 hospitals (22 established and 2 new hospitals; 17 level I trauma centers and 7 level II trauma centers) in 9 Chinese provinces. Study subjects were selected from the database if they met the following criteria: an adult patient (age $14 years) with a severe TBI (Glasgow Coma Scale [GCS] score on admission ,9) and a Marshall class Ⅱ to IV (diffuse TBI) as revealed by primary CT on admission. Exclusion criteria were a penetrating brain injury and admission with a diagnosis of brain death. Subjects were divided into 2 groups. The no-ICP group was not monitored; the ICP group was. The study protocol was approved by the local ethics committee of the coordinating hospital (Huashan Hospital), and all other participating hospitals provided feasibility statements. The next of kin, caretakers, or guardians consented on the behalf of participants whose capacity to consent was compromised. Verbal informed consent conducted by telephone or written informed consent conducted by mail was obtained before all follow-up interviews.

Patient Management Protocol Each center uses a standardized protocol to treat patients with severe TBI. The protocols are based on the international guidelines of the Brain Trauma Foundation.4 Patients in the ICP group had an ICP monitor placed as soon as possible and were treated to maintain an ICP of ,20 mm Hg. The care for patients in the no-ICP group was provided in accordance with imaging-clinical examination based on the treating physician’s experience. Clinical neurological status (GCS score, pupil size, and reactivity) was monitored every hour in both groups. A head CT was obtained at admission; before operation; immediately after

362 | VOLUME 78 | NUMBER 3 | MARCH 2016

operation; at 1, 2, 5, and 7 days after admission; and at any other time clinically indicated in both groups. All patients were positioned in a 30° head-up position and initially sedated with benzodiazepine and an opioid. If the patients had an elevated ICP, mannitol or hypertonic saline was administered to an osmolality of 310 to 320 mOsmL21. Cerebrospinal fluid (CSF) was drained via an intraventricular catheter if the ventricular pressure exceeded 20 mm Hg. Drainage of intraventricular fluid was intermittent; efforts were made to remove the smallest volume of fluid necessary to control ICP over the shortest possible period of time. Mild hyperventilation (PCO2, 30-33 mm Hg) was applied as necessary. Refractory intracranial hypertension was considered present if the ICP increased to .30 mm Hg or the cerebral perfusion pressure falls to ,60 mm Hg for .15 minutes with failure to respond to the maximal medical treatment described above. If refractory intracranial hypertension occurred, decompressive craniectomy or consecutive decompressive craniectomy on the other side was performed as soon as possible.

Data Collection and Definitions All data were retrospectively collected by physicians from medical records and included variables such as age, sex, mechanism of injury, GCS score on admission, pupillary reaction on admission, hypotensive episodes (defined as episodes of systolic blood pressure ,90 mm Hg during the first day after admission), and hypoxic episodes (defined as arterial oxygen saturation ,90% or arterial oxygen tension ,60 mm Hg during the first day after admission). Major extracranial trauma was considered present when the Abbreviated Injury Score was $3 in $1 body regions excluding the head. The type and duration of ICP monitoring, ICP value, and duration of intracranial hypertension during admission were also recorded. The first CT scan (acquired on admission) was evaluated if possible. If the first scan was not available, the second scan was assessed, provided that it had been taken within 6 hours of the initial scan and before any neurosurgical intervention. The Marshall CT score10 and the presence of any subdural hematoma, epidural hematoma, intraparenchymal lesion, or subarachnoid hemorrhage were recorded. Furthermore, the midline shift and the status of the ambient cisterns and fourth ventricle were recorded. The Marshall CT classification stratifies diffuse TBI into 4 categories of varying severity as follows: type I, no visible pathological component; type II, high- or mixed-density contusions ,25 mL in volume, visible basal cisterns, and brain shift #5 mm; type III, high- or mixed-density contusions ,25 mL in volume, compressed or obliterated basal cisterns, and brain shift #5 mm; and type IV, high- or mixeddensity contusions ,25 mL in volume, compressed or obliterated basal cisterns, and brain shift .5 mm. Because the hospital environment might influence TBI care and outcomes, we classified all centers as follows: trauma center level, teaching center level, and extent of routine use of ICP monitoring. The primary outcomes were 6-month mortality and a 6-month favorable outcome, assessed with the Glasgow Outcome Scale Extended (GOSE), which is an 8-point scale ranging from 1 (dead) to 8 (good recovery). GOSE scores were determined from either a (mailed) questionnaire or a structured telephone interview by both physicians and local investigators. GOSE scores of 1 to 4 were considered unfavorable; GOSE scores of 5 to 8 were deemed favorable.

Statistical Analysis Continuous variables are expressed as mean 6 SD and categorical variables as percentages. Univariate analyses of categorical data were

www.neurosurgery-online.com

Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited

ICP MONITORING OF DIFFUSE TBI

conducted with the aid of the x2 test. Normally distributed variables were compared by use of the Student t test, and nonnormally distributed variables were compared with the aid of the Mann-Whitney U test. We first identified variables that differed significantly between the ICP and no-ICP groups. Next, we predicted the probability of ICP monitor placement (the propensity score) for each patient using these variables. Each patient who underwent ICP monitoring was matched to the patient in the nonmonitored group who had the closest propensity score with a simple 1:1 nearest-neighbor matching algorithm.22 To exclude bad matches (in the sense that the estimated propensity scores were very different), we imposed a caliper of 0.15 of the standard deviation of the logit of the propensity score. After matching, we examined the balance of all covariates (see Supplemental Digital Content, http://links.lww.com/NEU/A785, which illustrates the details of propensity score matching). To verify that matching produced a balanced sample, differences in all observed variables were explored via univariate analysis; we asked whether the ICP was monitored on the unmatched sample as well as on the matched pair of samples. If the matching balance was good, any significant difference between the 2 original groups should become nonsignificant. We developed 2 separate multivariate logistic regression models: 1 model predicting 6-month mortality and 1 model predicting a 6-month favorable outcome. We next evaluated the impact of ICP monitoring on the actual outcomes using the matched samples after controlling for independent predictors of the outcomes. The adjusted odds ratios (AORs; with 95% confidence intervals [CIs]) of ICP monitoring for both outcomes were computed for the entire sample and for 2 specific subgroups: those with GCS scores of 3 to 5 and those with Marshall CT classification Ⅳ. An AOR ,1.00 implied that the factor significantly decreased the odds of developing an outcome; an AOR .1.00 implied that the factor significantly increased the odds of developing an outcome; and a CI that spanned 1.00 implied that the factor did not predict any outcome. All P values were 2 sided, and a value of P , .05 was considered to indicate statistical significance. Statistical analyses were performed with SPSS Statistics (version 20.0.0, IBM Corp, Somers, New York) and the R software environment (version 2.12.1, The R Foundation for Statistical Computing, Vienna, Austria).

RESULTS Patient Characteristics A total of 482 patients were included in the final analyses. A flow diagram outlining patient selection is shown in the Figure. Patients were predominantly male (80.5%); the mean patient age was 46.0 years. Motor vehicle accident was the most common injury mechanism (65.8%), followed by a fall (19.3%). Hypotension during the first day after admission was evident in 11.4% of all patients and hypoxia in 17.8%. Of all patients, 62.7% had GCS scores of 6 to 8 on admission, and 37.3% had GCS scores of 3 to 5. More than two-thirds of all patients (75.7%) had at least 1 reactive pupil on arrival. The most common head CT finding was intraparenchymal hematoma or contusion (74.1%). Of the total sample, 68.5% of patients underwent surgery (craniotomy to treat an intracranial hematoma or craniectomy). ICP monitors were inserted into 287 patients (59.5%): intraventricular monitoring was conducted in 162 (56.4%), parenchymal monitoring in 81 (28.2%), subdural monitoring in 42 (14.6%), and epidural monitoring in 2 (0.7%).

NEUROSURGERY

FIGURE. Flow diagram outlining patient selection in this study.

Patient Management The median duration of ICP monitoring was 6 days, and ICP was elevated in 68.3% of patients. Patients undergoing ICP monitoring had a significantly longer intensive care unit stay (median, 11 vs 7 days; P = .001) and hospital stay (median, 31 vs 18 days; P = .02). However, neither the duration of mechanical ventilation nor the duration of osmotherapy differed significantly between the 2 groups. Differences Between Patients Who Did and Did Not Undergo ICP Monitoring Many significant differences were evident when the demographic and injury characteristics of the ICP and non-ICP groups were compared (Table 1). ICP patients were younger; patients with Marshall CT classification III were more likely to undergo ICP monitoring (46.3% vs 30.8%, respectively; P = .002); and patients with Marshall CT classification IV were less likely to

VOLUME 78 | NUMBER 3 | MARCH 2016 | 363

Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited

YUAN ET AL

TABLE 1. Differences Between the Intracranial Pressure Monitoring Group and the No–Intracranial Pressure Monitoring Group Before and After Propensity Score Matchinga Before Matching

n Age (mean 6 SD), y Sex, n (%) Male Female Mechanism of injury, n (%) Motor vehicle accident Fall Stumble Blow to head Others GCS score on admission, n (%) 6-8 3-5 Pupillary reactions on admission, n (%) Both reacting 1 Reacting None reacting Hypotension, n (%) Hypoxia, n (%) Marshall CT classification, n (%) II III IV Major extracranial injury, n (%) Intracranial lesion, n (%) Epidural hematoma Subdural hematoma Traumatic subarachnoid hemorrhage Intraparenchymal hematoma or contusion Surgical management, n (%) Craniotomy for mass lesion Craniectomy No Trauma center level, n (%) I II Teaching status, n (%) University Nonteaching ICP monitoring use, n (%)b $50% ,50% 6-mo GOSE, n (%) Dead Vegetative state Lower severe disability Upper severe disability Lower moderate disability Upper moderate disability Lower good recovery Upper good recovery

After Matching

No ICP Monitoring

ICP Monitoring

P Value

No ICP Monitoring

ICP Monitoring

P Value

195 47.93 6 18.05

287 44.62 6 15.91

— .04

129 43.96 6 17.54

129 43.85 6 16.45

— .96

163 (83.6) 32 (16.4)

225 (78.4) 62 (21.6)

.16

110 (85.3) 19 (14.7)

100 (77.5) 29 (22.5)

.11

132 (67.7) 42 (21.5) 10 (5.1) 8 (4.1) 3 (1.5) 5.97 6 1.55 127 (65.1) 68 (34.9)

185 (64.5) 51 (17.8) 28 (9.8) 13 (4.5) 10 (3.5) 5.89 6 1.51 175 (61.0) 112 (39.0)

.21

84 (65.1) 27 (20.9) 9 (7.0) 6 (4.7) 3 (2.3) 5.86 6 1.57 82 (63.6) 47 (36.4)

83 (64.3) 22 (17.1) 12 (9.3) 7 (5.4) 5 (3.9) 5.64 6 1.56 71 (55.0) 58 (45.0)

.82

120 (61.5) 20 (10.3) 55 (28.2) 18 (9.2) 36 (18.5)

192 (66.9) 33 (11.5) 62 (21.6) 37 (12.9) 50 (17.4)

.25

58 60 77 71

(29.7) (30.8) (39.5) (36.4)

75 (26.1) 133 (46.3) 79 (27.5) 134 (46.7)

.002

45 (23.1) 76 (39.0) 141 (72.3) 117 (60.0)

75 (26.1) 159 (55.4) 203 (70.7) 240 (83.6)

9 (4.6) 129 (66.2) 57 (29.2)

.55 .36

.27 .16

72 (55.8) 15 (11.6) 42 (32.6) 12 (9.3) 27 (20.9)

81 15 33 16 20

(62.8) (11.6) (25.6) (12.4) (15.5)

.45

46 36 47 55

(35.7) (27.9) (36.4) (42.6)

36 55 38 64

(27.9) (42.6) (29.5) (49.6)

.05

.45 ,.001 .71 ,.001

30 (23.3) 66 (51.2) 86 (66.7) 107 (82.9)

32 75 83 98

(24.8) (58.1) (64.3) (76.0)

.77 .26 .69 .17

18 (6.3) 174 (60.6) 95 (33.1)

.35

6 (4.7) 81 (62.8) 42 (32.6)

6 (4.7) 83 (64.3) 40 (31.0)

.97

155 (79.5) 40 (20.5)

234 (81.5) 53 (18.5)

.58

98 (76.0) 31 (24.0)

91 (70.5) 38 (29.5)

.33

135 (69.2) 60 (30.8)

215 (74.9) 72 (25.1)

.17

78 (60.5) 51 (39.5)

80 (62.0) 49 (38.0)

.80

71 (36.4) 124 (63.6)

222 (77.4) 65 (22.6)

,.001

71 (55.0) 58 (45.0)

80 (62.0) 49 (38.0)

.26

62 (31.8) 23 (11.8) 21 (10.8) 18 (9.2) 10 (5.1) 18 (9.2) 13 (6.7) 30 (15.4)

51 (17.8) 53 (18.5) 33 (11.5) 25 (8.7) 24 (8.4) 20 (7.0) 35 (12.2) 46 (16.0)

.03

39 (30.2) 11 (8.5) 15 (11.6) 11 (8.5) 8 (6.2) 10 (7.8) 9 (7.0) 26 (20.2)

27 (20.9) 19 (14.7) 15 (11.6) 11 (8.5) 8 (6.2) 6 (4.7) 16 (12.4) 27 (20.9)

.33

.22 .77

.025

.42 .26

.26

a

GCS, Glasgow Coma Scale; GOSE, Glasgow Outcome Scale extended; ICP, intracranial pressure. The hospitals that used ICP monitoring frequently; $50% means these hospitals with ICP monitoring use $50%.

b

364 | VOLUME 78 | NUMBER 3 | MARCH 2016

www.neurosurgery-online.com

Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited

ICP MONITORING OF DIFFUSE TBI

undergo monitoring (27.5% vs 39.5%, respectively; P =.002). Patients who sustained major extracranial injuries were more likely to undergo monitoring (46.7% vs 36.4%, respectively; P = .03). Subdural hematomas (55.4% vs 39.0%; P , .001) and intraparenchymal hematoma or contusions (83.6% vs 60.0%; P , .001) were more common in the ICP group. In addition, more patients in the ICP than the non-ICP group were treated in trauma centers that often monitored ICP (77.4% vs 36.4%, respectively; P , .001). All of these variables were considered during propensity score matching to predict the probability (the propensity score) of ICP monitoring. After matching, no variable differed significantly between the 2 groups (Table 1). Multivariate Regression to Define Predictors of Mortality and a Favorable Outcome Only 5 variables were significantly associated with 6-month mortality, and 6 variables were significantly associated with a 6month favorable outcome after controlling for the full set of covariates (Table 2). Age, GCS on admission, Marshall CT classification, hypotension during the first day, and ICP monitoring use $50% were independent predictors of 6-month mortality. Age, GCS on admission, a stumble as the mechanism of injury, Marshall CT classification, epidural hematoma, and subdural hematoma were independent predictors of a 6-month favorable outcome. Relationship Between ICP Monitoring and Clinical Outcome When unadjusted data were used, patients who underwent ICP monitoring had a lower rate of 6-month mortality than those who

did not (17.8% vs 31.8%; P , .001). After propensity score matching, however, ICP monitoring was also associated with a decrease in 6-month mortality rates even after controlling for independent predictors of mortality (AOR, 0.46; 95% CI, 0.240.90; adjusted P = .02). In addition, ICP monitoring had a greater impact on the most severely injured patients based on head CT data (Marshall CT classification IV; AOR, 0.31; 95% CI, 0.11-0.90; adjusted P = .03) and on patients with the lowest level of consciousness (GCS scores 3-5; AOR, 0.32; 95% CI, 0.12-0.87; adjusted P = .03; Table 3). ICP monitoring was not associated with a 6-month favorable outcome on the basis of unadjusted data (45.3% vs 36.4%; P = .05). After propensity score matching, monitoring remained nonassociated with a 6-month favorable outcome for the overall sample (AOR, 1.57; 95% CI, 0.87-2.82; adjusted P = .13). However, monitoring had a significant impact on the 6-month favorable outcomes of patients with the lowest level of consciousness (GCS scores 3-5; AOR, 3.20; 95% CI, 1.07-9.53; adjusted P = .04; Table 3).

DISCUSSION This is the only multicenter study to evaluate the effects of ICP monitoring on TBI patients with a specific subtype of injury (diffuse TBI). We chose to investigate such patients for 2 main reasons. First, TBI patients with diffuse lesions have been shown to be at a high risk of intracranial hypertension, especially those exhibiting effacement of basal cisterns. Toutant et al23 reported that 74% of patients with diffuse TBI lacking cisterns developed ICPs .30 mm Hg. ICP values in patients with diffuse TBI

TABLE 2. Independent Predictors of 6-Month Mortality and 6-Month Favorable Outcome (Glasgow Outcome Scale Extended $4)a Step of Forward Logistic Regression Analysis 6-mo mortality 1 2 3 4 5 6-mo favorable outcome (GOSE $4) 1 2 3

4 5 6

Variable

AOR (95% CI)

P Value

Age GCS score on admission Marshall CT classification Hypotension present on day 1 ICP monitoring use $50%

1.03 0.66 1.49 2.52 0.40

(1.01-1.04) (0.57-0.78) (1.08-2.05) (1.30-4.91) (0.25-0.64)

,.001 ,.001 .01 .006 ,.001

Age GCS score on admission Mechanism of injury Motor vehicle accident Fall Stumble Blow to head Other Marshall CT classification Epidural hematoma Subdural hematoma

0.97 (0.95-0.98) 1.76 (1.50-2.06)

,.001 ,.001

1 (Reference) 1.21 (0.71-2.07) 3.91 (1.75-8.73) 1.83 (0.67-5.02) 1.05 (0.28-3.94) 0.58 (0.44-0.76) 1.76 (1.08-2.85) 1.57 (1.02-2.41)

.48 .001 .24 .95 ,.001 .02 .04

a

AOR, adjusted odds ratio; CI, confidence interval; CT, computed tomography; GCS, Glasgow Coma Scale; GOSE, Glasgow Outcome Scale Extended; ICP, intracranial pressure. “Reference” in the motor vehicle accident means the reference value of other mechanism of injury included fall, stumble, blow to head, and other.

NEUROSURGERY

VOLUME 78 | NUMBER 3 | MARCH 2016 | 365

Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited

YUAN ET AL

TABLE 3. Impact of Intracranial Pressure Monitoring on 6-Month Mortality or 6-Month Favorable Outcome in the Full Sample and Selection Subgroups After Propensity Score Matchinga ICP Monitor Adjusted Placement, % OR 6-mo mortality Full sample GCS score 3-5 on admission Marshall CT Ⅳ 6-mo favorable outcome (GOSE $4) Full sample GCS score 3-5 on admission Marshall CT IV

95% CI

P Value

59.5 62.2

0.46 0.32

0.24-0.90 0.12-0.87

.02 .03

50.6

0.31

0.11-0.90

.03

59.5 62.2

1.57 3.20

0.87-2.82 1.07-9.53

.13 .04

50.6

1.26

0.31-5.05

.75

a

AOR, adjusted odds ratio; CI, confidence interval; GOSE, Glasgow Outcome Scale Extended; ICP, intracranial pressure; GCS, Glasgow Coma Scale.

cannot be precisely predicted with the Marshall CT scan classification,24 but a relevant relationship may, in fact, exist.25 Second, any lesion producing a mass effect should always be urgently evacuated. In patients with diffuse traumatic brain injuries (Marshall CT class I-IV), however, the decision to perform a neurosurgical procedure such as craniotomy or craniectomy is generally based on ICP monitoring data and the level of intracranial hypertension. A secondary decompressive craniectomy is usually performed when the ICP increases to .30 mm Hg or the cerebral perfusion pressure falls to ,60 mm Hg for .15 minutes and the patient does not respond to the maximum medical treatment. Therefore, compared with patients with focal TBI, those with diffuse TBI may benefit more from ICP monitoring. Many of the demographic and injury characteristics of the ICP and non-ICP groups of diffuse TBI patients included in the present study differed significantly. Such differences in baseline characteristics may be associated with different probabilities of ICP monitor placement. Therefore, in the assessment of the effects of monitoring on patient outcomes, propensity score matching allows such differences to be better controlled, affording more accurate estimates of the effects of monitoring. Although the guidelines of the Brain Trauma Foundation and the American Association of Neurological Surgeons recommend monitoring of all patients in our sample, only 59.5% of patients actually underwent ICP monitor placement during the acute phase of injury. Previous analyses of ICP monitoring also reported compliance rates of about 50%,26-30 similar to ours. We found that younger patients, those with Marshall CT classification III, those who sustained major extracranial injury, and patients who sustained subdural hematomas, intraparenchymal hematoma, or contusions were more likely to undergo ICP monitoring. Our results are in good accord with earlier data, which suggested that much of the difference in ICP monitor placement, or no

366 | VOLUME 78 | NUMBER 3 | MARCH 2016

placement, reflects implicit risk assessments by neurosurgeons at the time when a monitor can be placed.26,28 Older patients may be perceived to benefit less from monitoring because of their decreased ability to recover from even minor intracranial injuries.31 In contrast, patients with significant intracranial injuries, including subdural hematomas, intraparenchymal hematoma or contusions, or radiographic signs of intracranial hypertension (Marshall CT classification III), stand to gain the most from invasive monitoring. Such patients often require aggressive medical therapy and cannot be effectively monitored noninvasively. However, it is clear that the clinical decision-making process guiding ICP monitor placement is complex, being centered around determining which patients will receive the most benefit and incur the least harm. Then, we found that a significant 6-month mortality benefit was associated with ICP monitoring after controlling for other risk factors for mortality and the probability of monitor placement. In addition, ICP monitoring was also significantly associated with 6-month functional outcomes in patients with GCS scores of 3 to 5 on admission. Several studies have suggested that monitoring is bimodal32; both the sickest and the least sick patients have low likelihoods of monitor placement. However, our results suggest that the most severely injured patients (GCS scores of 3-5 or Marshall CT classification IV) may be the most likely to benefit from monitoring. Lobato et al33 found that the relative risk of a delayed operation was associated with the Marshall CT classification of initial CT scans (diffuse injury IV, 30.7%; diffuse injury III, 30.5%; nonevacuated mass, 20.0%; evacuated mass, 20.2%; diffuse injury II, 12.1%; diffuse injury I, 8.6%) and that patients with Marshall CT classification IV are more likely to experience elevated ICP and may derive greater benefits from ICP-targeted management. Therefore, the central question may not, in fact, be whether invasive monitoring should be applied to all TBI patients but rather how to identify patients for whom monitoring is most appropriate and who will benefit most, in the sense that monitoring will aid in the definition of a suitable course of treatment. Although a recent randomized controlled study of ICP monitoring of TBI patients concluded that monitoring did not improve outcomes,17 the cited study did not specifically target patients with diffuse TBI. Moreover, randomized controlled trials are often not generalizable to real-world patterns of care because the patient populations are carefully selected and the treatment protocols are regimented. Thus, many researchers are turning to large observational data sets to determine how ICP monitoring is used in nonexperimental settings and to better understand which patients may benefit from such monitoring. Previous observational studies involved only a single institution,26 were unable to control for as complete a set of patient characteristics as the set that we evaluated,34-36 or failed to address nonrandom patient distribution patterns between the monitored and nonmonitored groups.27,37 Limitations The present study has several limitations. First, this was an observational study rather than a prospective randomized study; thus, the inclusion criteria (age, time from injury, method of

www.neurosurgery-online.com

Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited

ICP MONITORING OF DIFFUSE TBI

management, type of TBI, etc) varied among institutions, and therapeutic interventions were at the discretion of many independent physicians. Second, the nature of the data available from patient medical records imposes inherent limitations on the work. Third, the frequencies of neurological examination and the number of CT scans taken during admission were not collected, so it was not possible to assess the adequacy of alternative monitoring methods. Finally, although propensity scoring is useful when selection bias may be in play, the method controls primarily for differences in observed variables between groups. The validity of any result is strongly dependent on appropriate selection of variables.

CONCLUSION ICP monitor placement was associated with a significant decrease in 6-month mortality after adjustment for the baseline risk profile and the monitoring propensity of patients with diffuse severe TBI, especially those with GCS scores of 3 to 5 or with Marshall CT classification IV. Moreover, ICP monitor placement was also significantly associated with improved 6-month functional outcomes for diffuse TBI patients with GCS scores of 3 to 5. Therefore, our retrospective, observational, multicenter study supports ICP monitoring of patients with severe diffuse TBI. A well-balanced randomized trial is needed to definitively determine whether ICP monitoring improves the outcomes of diffuse TBI patients. Disclosures This work was supported by the National Natural Science Foundation of China (grants 81471241, 81271375, and 81171133). The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper.

REFERENCES 1. Coronado VG, Xu L, Basavaraju SV, et al. Surveillance for traumatic brain injuryrelated deaths—United States, 1997-2007. MMWR Surveill Summ. 2011;60(5): 1-32. 2. Maas AI, Stocchetti N, Bullock R. Moderate and severe traumatic brain injury in adults. Lancet Neurol. 2008;7(8):728-741. 3. Selassie AW, Zaloshnja E, Langlois JA, Miller T, Jones P, Steiner C. Incidence of long-term disability following traumatic brain injury hospitalization, united states, 2003. J Head Trauma Rehabil. 2008;23(2):123-131. 4. Brain Trauma Foundation, American Association of Neurological Surgeons, Congress of Neurological Surgeons, Bratton SL, Chestnut RM, Ghajar J, et al. Guidelines for the management of severe traumatic brain injury, VIII: intracranial pressure thresholds. J Neurotrauma. 2007;24(suppl 1):S55-S58. 5. Mendelson AA, Gillis C, Henderson WR, Ronco JJ, Dhingra V, Griesdale DE. Intracranial pressure monitors in traumatic brain injury: a systematic review. Can J Neurol Sci. 2012;39(5):571-576. 6. Rosenfeld JV, Maas AI, Bragge P, Morganti-Kossmann MC, Manley GT, Gruen RL. Early management of severe traumatic brain injury. Lancet. 2012;380 (9847):1088-1098. 7. Andriessen TM, Jacobs B, Vos PE. Clinical characteristics and pathophysiological mechanisms of focal and diffuse traumatic brain injury. J Cell Mol Med. 2010;14 (10):2381-2392. 8. Zhu GW, Wang F, Liu WG. Classification and prediction of outcome in traumatic brain injury based on computed tomographic imaging. J Int Med Res. 2009;37(4):983-995.

NEUROSURGERY

9. Maas AI, Hukkelhoven CW, Marshall LF, Steyerberg EW. Prediction of outcome in traumatic brain injury with computed tomographic characteristics: a comparison between the computed tomographic classification and combinations of computed tomographic predictors. Neurosurgery. 2005;57(6):1173-1182; discussion 1173-1182. 10. Marshall LF, Marshall SB, Klauber MR, et al. The diagnosis of head injury requires a classification based on computed axial tomography. J Neurotrauma. 1992;9 (suppl 1):S287-S292. 11. Becker DP, Miller JD, Ward JD, Greenberg RP, Young HF, Sakalas R. The outcome from severe head injury with early diagnosis and intensive management. J Neurosurg. 1977;47(4):491-502. 12. Gennarelli TA, Spielman GM, Langfitt TW, et al. Influence of the type of intracranial lesion on outcome from severe head injury. J Neurosurg. 1982;56(1): 26-32. 13. Aarabi B, Hesdorffer DC, Ahn ES, Aresco C, Scalea TM, Eisenberg HM. Outcome following decompressive craniectomy for malignant swelling due to severe head injury. J Neurosurg. 2006;104(4):469-479. 14. Timofeev I, Kirkpatrick PJ, Corteen E, et al. Decompressive craniectomy in traumatic brain injury: outcome following protocol-driven therapy. Acta Neurochir Suppl. 2006;96:11-16. 15. Cooper DJ, Rosenfeld JV, Murray L, et al. Decompressive craniectomy in diffuse traumatic brain injury. N Engl J Med. 2011;364(16):1493-1502. 16. Brain Trauma Foundation, American Association of Neurological Surgeons, Congress of Neurological Surgeons, Bratton SL, Chestnut RM, Ghajar J, et al. Guidelines for the management of severe traumatic brain injury, VI: indications for intracranial pressure monitoring. J Neurotrauma. 2007;24(suppl 1):S37-S44. 17. Chesnut RM, Temkin N, Carney N, et al. A trial of intracranial-pressure monitoring in traumatic brain injury. N Engl J Med. 2012;367(26):2471-2481. 18. Ropper AH. Brain in a box. N Engl J Med. 2012;367(26):2539-2541. 19. Figaji AA. Re: Intracranial pressure monitors in traumatic brain injury: a systematic review. Can J Neurol Sci. 2012;39:571-576. Can J Neurol Sci. 2013;40(3): 433-434. 20. Haviland A, Nagin DS, Rosenbaum PR. Combining propensity score matching and group-based trajectory analysis in an observational study. Psychol Methods. 2007;12(3):247-267. 21. Stuart EA. Matching methods for causal inference: a review and a look forward. Stat Sci. 2010;25(1):1-21. 22. Austin PC. A critical appraisal of propensity-score matching in the medical literature between 1996 and 2003. Stat Med. 2008;27(12):2037-2049. 23. Toutant SM, Klauber MR, Marshall LF, et al. Absent or compressed basal cisterns on first CT scan: ominous predictors of outcome in severe head injury. J Neurosurg. 1984;61(4):691-694. 24. Hiler M, Czosnyka M, Hutchinson P, et al. Predictive value of initial computerized tomography scan, intracranial pressure, and state of autoregulation in patients with traumatic brain injury. J Neurosurg. 2006;104(5):731-737. 25. Miller MT, Pasquale M, Kurek S, et al. Initial head computed tomographic scan characteristics have a linear relationship with initial intracranial pressure after trauma. J Trauma. 2004;56(5):967-972; discussion 972-973. 26. Talving P, Karamanos E, Teixeira PG, et al. Intracranial pressure monitoring in severe head injury: compliance with Brain Trauma Foundation guidelines and effect on outcomes: a prospective study. J Neurosurg. 2013;119(5):1248-1254. 27. Shafi S, Diaz-Arrastia R, Madden C, Gentilello L. Intracranial pressure monitoring in brain-injured patients is associated with worsening of survival. J Trauma. 2008; 64(2):335-340. 28. Biersteker HA, Andriessen TM, Horn J, et al. Factors influencing intracranial pressure monitoring guideline compliance and outcome after severe traumatic brain injury. Crit Care Med. 2012;40(6):1914-1922. 29. Shafi S, Barnes SA, Millar D, et al. Suboptimal compliance with evidence-based guidelines in patients with traumatic brain injuries. J Neurosurg. 2014;120(3): 773-777. 30. Dawes AJ, Sacks GD, Cryer HG, et al. Intracranial pressure monitoring and inpatient mortality in severe traumatic brain injury: a propensity score-matched analysis. J Trauma Acute Care Surg. 2015;78(3):492-502. 31. Mosenthal AC, Lavery RF, Addis M, et al. Isolated traumatic brain injury: age is an independent predictor of mortality and early outcome. J Trauma. 2002;52(5):907-911. 32. Mauritz W, Steltzer H, Bauer P, Dolanski-Aghamanoukjan L, Metnitz P. Monitoring of intracranial pressure in patients with severe traumatic brain injury: an Austrian prospective multicenter study. Intensive Care Med. 2008;34(7):1208-1215.

VOLUME 78 | NUMBER 3 | MARCH 2016 | 367

Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited

YUAN ET AL

33. Lobato RD, Gomez PA, Alday R, et al. Sequential computerized tomography changes and related final outcome in severe head injury patients. Acta Neurochir (Wien). 1997;139(5):385-391. 34. Farahvar A, Gerber LM, Chiu YL, Carney N, Hartl R, Ghajar J. Increased mortality in patients with severe traumatic brain injury treated without intracranial pressure monitoring. J Neurosurg. 2012;117(4):729-734. 35. Alali AS, Fowler RA, Mainprize TG, et al. Intracranial pressure monitoring in severe traumatic brain injury: results from the American College of Surgeons Trauma Quality Improvement Program. J Neurotrauma. 2013;30(20):1737-1746. 36. Fakhry SM, Trask AL, Waller MA, Watts DD, Force INT. Management of braininjured patients by an evidence-based medicine protocol improves outcomes and decreases hospital charges. J Trauma. 2004;56(3):492-499; discussion 499-500. 37. Cremer OL, van Dijk GW, van Wensen E, et al. Effect of intracranial pressure monitoring and targeted intensive care on functional outcome after severe head injury. Crit Care Med. 2005;33(10):2207-2213.

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (www.neurosurgery-online.com).

Acknowledgments We gratefully acknowledge the wholehearted cooperation of the doctors, the patients, the Trauma Centre, and the departments of neurosurgery in 24 hospitals of China (Neurosurgery Department, West China Hospital of Sichuan University; Emergency Medicine Department, The first Affiliated Hospital of Soochow University; Neurosurgery Department, The Second Affiliated Hospital of Soochow University; Neurosurgery Department, The Second Affiliated Hospital of Zhejiang University School of Medicine; Neurosurgery Department, Xiangya Hospital Central South University; Neurosurgery Department, Southwest Hospital; Neurosurgery Department, Shanghai Shuguang Hospital; Neurosurgery Department, The First Affiliated Hospital of Anhui Medical University; Neurosurgery Department, General Hospital of Tianjing Medical University; Neurosurgery Department, Hospital of Logistics University of People’s Armed Police Force; Neurosurgery Department, Fuyang People’s Hospital; Neurosurgery Department, Wuxi Third People’s Hospital; Neurosurgery Department, Enze Medical Center Luqiao Hospital; Neurosurgery Department, Chongqing Emergency Medical Center; Neurosurgery Department, Shanghai Sixth People’ Hospital; Neurosurgery Department, Huashan Hospital, Fudan University; Neurosurgery Department, People’s Hospital of Dazhu County; Neurosurgery Department, 101 Hospital of People’s Liberation Army; Neurosurgery Department, The South Hospital of Southern Medical University; Neurosurgery Department, Yuyao People’s Hospital; Neurosurgery Department, Jiashan People’s Hospital; Neurosurgery Department, Jinshan Hospital, Fudan University; Neurosurgery Department, Nanhui Central Hospital; and Qilu Hospital of Shandong University).

COMMENTS

T

he authors clearly describe outcomes in 482 patients with severe traumatic brain injury treated at 24 level 1 and 2 trauma centers in China. Forty percent were managed without intracranial pressure (ICP) monitors, and nearly 65% underwent decompressive craniectomies. On the basis of variables considered predictive of brain swelling, each patient in the ICP monitored group was matched to a patient in the non–ICP monitored group (propensity matching) for outcomes analyses. At 6 months after injury, the mortality rate was lower in the ICP monitored group, and the functional outcome was better in the subgroup that had an initial Glasgow Coma Scale score of 3 to 5 and ICP monitoring. This clinical study contrasts with the study by Chesnut et al1 of 324 patients admitted to hospitals in Bolivia and Ecuador with severe traumatic brain injury who were also assigned to ICP monitoring vs no ICP monitoring groups. That study concluded, “For patients with severe

368 | VOLUME 78 | NUMBER 3 | MARCH 2016

traumatic brain injury, care focused on maintaining monitored intracranial pressure at 20 mm Hg or less was not shown to be superior to care based on imaging and clinical examination.” Many have raised concerns about the prehospital care in the Chesnut study and how delays in transport may have affected the outcomes. Only a small proportion of those who were monitored had an external ventricular drain and the benefit of cerebrospinal fluid drainage for ICP control. An important distinction may also be in the level of critical care. Most currently agree that those patients with severe traumatic brain injuries who are at risk for brain swelling should have intracranial pressure monitoring.2,3 The benefits of treatment of brain swelling, and especially withholding treatments when there is no increase in ICP, clearly outweigh the very low risk for serious complications from placement of invasive monitors. Moreover, it is emphasized that an ICP monitor is simply a that—a monitor of one of many aspects of the altered physiology and metabolism in the severely traumatized patient. Control of brain swelling and ultimately functional recovery depend on how well the mechanisms of secondary brain injury are defined and how effectively they are treated in each individual patient. Donald W. Marion Silver Spring, Maryland

1. Chesnut RM, Temkin N, Carney N, et al; Global Neurotrauma Research Group. A trial of intracranial-pressure monitoring in traumatic brain injury. N Engl J Med. 2012;367(26):2471-2481. 2. Chesnut R, Bleck T, Citerio G, et al. A consensus-based interpretation of the benchmark evidence from South American Trials: treatment of intracranial pressure trial [published online ahead of print August 31, 2015]. J Neurotrauma. Available at: http://online.liebertpub.com/doi/10.1089/neu.2015.3976. Accessed August 1, 2015. 3. Chesnut R, Videtta W, Vespa P, Le Roux P; Participants in the International Multidisciplinary Consensus Conference on Multimodality Monitoring. Intracranial pressure monitoring: fundamental considerations and rationale for monitoring. Neurocrit Care. 2014;21(suppl 2):S64-S84.

T

he publication of the Benchmark Evidence from South American Trials: Treatment of Intracranial Pressure (BEST TRIP) trial has caused some debate and confusion about the value of intracranial pressure (ICP) monitoring in patients with severe traumatic brain injury (TBI).1,2 The authors evaluated the effect of ICP monitoring on outcome in a comparative effectiveness design using propensity score matching to construct comparable groups of patients who received and did not receive ICP monitoring. The authors report that ICP monitoring significantly decreased 6-month mortality after controlling for independent predictors of mortality. The greatest effect was seen in patients with a Marshall computed tomography (CT) classification IV and in patients with a Glasgow Coma Scale score of 3 to 5 on admission. However, no significant effect on favorable outcome was found. This is an important study that further sheds some light on the enigma of benefits of ICP monitoring in patients with TBI. Strong aspects are the matching of patients in a comparative effectiveness design and a focus on subgroups who may benefit most. Larger effects on mortality are reported in patients with Marshall CT classification IV (diffuse injury with shift). Of note, however, is that in this study the Marshall classification was not applied as originally intended and implemented in most studies. In the original publication on the Marshall classification,3 any lesion surgically evacuated is considered an evacuated mass lesion. This implies that CT class V may also include patients with smaller lesions treated surgically

www.neurosurgery-online.com

Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited

ICP MONITORING OF DIFFUSE TBI

and patients in whom surgery is scheduled on the basis of the CT findings. We do not consider deviation from the original description a major problem because the authors are transparent in their reporting. We see the observational design of this study more as a strength than as a weakness, given the greater generalizability compared with more selected populations recruited into clinical trials with strict enrollment criteria. The disadvantage is that the ICP monitoring is not allocated at random, with risk of differences between treatment groups as also clearly occurred in this study. This problem was dealt with by the authors in 2 ways: first by developing a propensity score and matching patients on the basis of this score to create comparable groups and second by performing covariate adjustment for the major predictors in the analysis phase. Propensity scores are currently a very popular method, but a major shortcoming is that this method, as “normal” adjustment, adjusts only for observed confounders, not for unobserved confounders. Thus, interpretation of differences in outcome between matched patient groups should be done with caution. Nevertheless, the sample size is considerable, and the authors have done the best they can in examining the performance of the propensity score and the characteristics of the unmatched and matched cohorts very critically. Unfortunately, even if we assume that the adjustment was sufficient and results are valid, the essential question underpinning interpretation of the results of this study remains unanswered. This question is whether the effect of ICP monitoring is related to differences in therapeutic approaches. Because ICP monitoring is in fact a diagnostic intervention, a crucial question is whether therapies are initiated on the basis of information that would not have been available without the ICP monitor. Although not reported in detail, the authors have stated that there were no major treatment differences between the matched treatment group with regard to hyperosmolar therapy or hyperventilation. However, a major difference could be that in the ICP monitored group cerebrospinal fluid was additionally drained, which was not the case in patients treated without ICP monitoring. An intriguing hypothesis would be that the lower mortality noted in the ICP monitoring group might be due to the external CSF drainage from the external ventricular drain. In a comparison of mortality between the North American and international tirilazad trials conducted in the 1990s,

NEUROSURGERY

Hukkelhoven et al4 described a 5% lower mortality in the North American patients. Differences were shown in methodology for ICP monitoring: Ventricular monitoring being more common in North American than in the international setting. Ghajar et al5 also previously reported improvement of outcome with external cerebrospinal fluid drainage. More recently, Gerber et al6 reported a decrease in case fatality rates in severe TBI over the years 2001 through 2009, which was associated with increased guideline compliance and use of ICP monitoring (increase from 56% to 75%), mainly by means of a ventricular catheter. Future studies on ICP monitoring should address documentation of the extent to which monitoring of ICP influences the intensity and type of treatment. We suggest a particular focus on possible benefit of external ventricular drainage. Andrew Maas Edegem, Belgium Hester Lingsma Rotterdam, The Netherlands

1. Chesnut RM, Temkin N, Carney N, et al. A trial of intracranial pressure monitoring in traumatic brain injury. N Engl J Med. 2012;367(26):2471-2481. 2. Stocchetti N, Picetti E, Berardino M, et al. Clinical applications of intracranial pressure monitoring in traumatic brain injury: report of the Milan Consensus Conference. Acta Neurochir (Wien). 2014;156:1615-1622. 3. Marshall LF, Marshall SB, Klauber MR, et al. A new classification of head injury based on computerized tomography. J Neurosurg. 1991;75:S14-S20. 4. Hukkelhoven CW, Steyerberg EW, Farace E, Habbema JD, Marshall LF, Maas AI. Regional differences in patient characteristics, case management, and outcomes in traumatic brain injury: experience from the tirilazad trials. J Neurosurg. 2002;97(3): 549-557. 5. Ghajar JB, Hariri RJ, Schreiber K, et al. Improved outcome from traumatic coma using ventricular CSF drainage. In: Nakamura N, Hashimoto T, Yasue M, eds. Recent Advances in Neurotraumatology, Tokyo: Springer-Verlag; 1993:327-330. 6. Gerber LM, Chiu YL, Carney N, Hartl R, Ghajar J. Marked reduction in mortality in patients with severe traumatic brain injury. J Neurosurg. 2013;119 (6):1583-1590.

VOLUME 78 | NUMBER 3 | MARCH 2016 | 369

Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited