Correlation between Glasgow coma score components and survival in patients with traumatic brain injury

Injury, Int. J. Care Injured 42 (2011) 940–944

Contents lists available at ScienceDirect

Injury journal homepage: www.elsevier.com/locate/injury

Correlation between Glasgow coma score components and survival in patients with traumatic brain injury Woon-Man Kung a,e,h, Shin-Han Tsai a,c, Wen-Ta Chiu a,c,d, Kuo-Sheng Hung a,c, Shin-Ping Wang c, Jia-Wei Lin a,d, Muh-Shi Lin a,b,d,f,g,* a

Department of Neurosurgery, Taipei Medical University-Wan Fang Hospital, Taipei, Taiwan Department of Neurosurgery, Taipei County Hospital, Taipei, Taiwan Graduate Institute of Injury Prevention and Control, Taipei Medical University, Taipei, Taiwan d Graduate Institute of Clinical Medicine, College of Medicine, Taipei Medical University, Taipei, Taiwan e Institute of Biomedical Engineering, College of Medicine and College of Engineering, National Taiwan University, Taipei, Taiwan f Department of Neurosurgery, Taipei City Hospital, Zhong, Xiao Branch, Taipei, Taiwan g Department of Surgery, School of Medicine, Taipei Medical, University, Taipei, Taiwan h Department of Neurosurgery, Chang Bing Show Chwan Memorial Hospital, Changhua, Taiwan b c

A R T I C L E I N F O

A B S T R A C T

Article history: Accepted 13 September 2010

Background: The Glasgow coma scale (GCS) score is used in the initial evaluation of patients with traumatic brain injury (TBI); however, the determination of an accurate score is not possible in all clinical situations. Our aim is to determine if the individual components of the GCS score, or combinations of them, are useful in predicting mortality in patients with TBI. Methods: The components of the GCS score and the receiver-operating characteristic (ROC) curves were analyzed from 27,625 cases of TBI in Taiwan. Results: The relationship between the survival rate and certain eye (E), motor (M) and verbal (V) score combinations for GCS scores of 6, 11, 12 and 13 were statistically significant. The areas under ROC curve of E + V, M + V and M alone were 0.904, 0.903 and 0.900, respectively, representing the 3 most precise combinations for predicting mortality. The area under the ROC curve for the complete GCS score (E + M + V) was 0.885. Patients with lower E, M and V score respectively, and lower complete GCS scores had higher hazard of death than those with the highest scores. Conclusion: The results of this study indicate that the 3 fundamental elements comprising the Glasgow coma scale, E, M, and V individually, and in certain combinations are predictive of the survival of TBI patients. This observation is clinically useful when evaluating TBI patients in whom a complete GCS score cannot be obtained. ß 2010 Elsevier Ltd. All rights reserved.

Keywords: Traumatic brain injury (TBI) Glasgow coma scale (GCS) Eye Motor Verbal combinations Survival prediction

Introduction The Glasgow coma scale (GCS) has been universally established as a common diagnostic tool to evaluate the consciousness, clinical status, as well as the prognosis of traumatic brain injury (TBI) patients.1,9,16 The GCS score is based on the simple addition of the 3 components evaluated, eye (E), motor (M) and verbal (V). Multiple studies have reported the predictive value of the GCS, alone or in combination with other clinical factors, in determining the outcomes of patients who have sustained brain injury.7,8,10,13 Though the score is useful in the evaluation of TBI patients, an

* Corresponding author at: Department of Neurosurgery, Taipei Medical University-Wan Fang Hospital, Number 111, Section 3, Hsing-Long Road, Taipei, Taiwan. Tel.: +886 2 29307930x6942; fax: +886 2 29347054. E-mail address: [email protected] (M.-S. Lin). 0020–1383/$ – see front matter ß 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.injury.2010.09.019

accurate and complete GCS score is difficult to obtain in many situations, e.g., if the patient is intubated or they have excessive swelling of the eyelids, thus decreasing the ability of physicians to provide appropriate care. A simplified assessment system using fewer parameters for predicting survival of TBI patients may make training of medical professionals easier, lead to less variability in reporting, and allow more accurate assessment across a broader range of clinical scenarios. Thus, many authors have recognised the limitations of the GCS and have sought to improve its functionality and simplify its use.3,4,11 A report by Healey et al.5 suggested that the motor component of the GCS contains all of the information of the complete score, and the authors recommended using the motor score alone in outcome prediction models. TBI accounts for 12.5% of all the traumatic injuries in Taiwan, and the mortality rate of the patients with moderate to severe TBI is as high as 35%. Data of patients who died from traumas indicate that 55% of deaths were caused by TBI. The Head Injury Registry is

W.-M. Kung et al. / Injury, Int. J. Care Injured 42 (2011) 940–944

an electronic database of TBI cases in Taiwan, archived by the Injury Prevention Center at Taipei Medical University, and supported by the Department of Health. This registry provides a valuable resource for research into the causes and outcomes of TBI. In this retrospective study we sought to determine the association of individual GCS component scores, and combinations of the scores, as predictors of mortality in TBI patients by reviewing the records of TBI cases archived in the Head Injury Registry. Methods TBI subject inclusion and exclusion criteria A TBI patient was defined as a patient who, after having received direct or indirect trauma to the head, exhibited brain concussion, contusion, skull fracture, or any of their clinical manifestations such as loss of consciousness, amnesia, neurological deficits, and seizures. Clinical evidence of skull fractures and intracranial hemorrhage, e.g., positive findings on radiographs or computed tomography (CT) were used to define the extent of TBI. In general, a patient with one or more of these symptoms or diagnoses was registered as a TBI patient.

941

moderate, score of 9–12, patient received brain surgery or had abnormal CT scans; mild, score of 13–15, or conditions not meeting any of the above criteria. Statistical analysis The basic parameters associated with TBI were statistically tested by the Wilcoxon rank sum test or chi-square test. Differences in survival under respective combinations of E, M and V of GCS scores of 3–15 were examined by Gehan’s test. A receiver-operating characteristic (ROC) curve was generated to compare the accuracy in predicting mortality among combinations of E, M and V scores and complete GCS score. Cox’s proportional hazard model was utilised to estimate the hazard ratio of death. A difference was considered statistically significant for p < 0.05. All statistics were two-sided and performed on SPSS statistical software (version 15.0, SPSS Inc., Chicago, IL, USA). Results Patient characteristics and E, M, V, and GCS scores are presented in Table 1. The mean  standard deviation (SD) of age and E, M, V and GCS score were 41.6  22.8, 3.6  0.8, 5.6  1.0, 4.4  1.2, and

Data collection In this retrospective, case-series study, we had collected TBI data of injuries that occurred during the period from July 2001 to June 2006 from 55 major teaching hospitals in Taiwan. Approximately 85% of Taiwan’s population is covered by these hospitals, making the data representative of the whole population. Patients with direct or indirect trauma outside the head, those dead on arrival to the hospital, those who died after hospital discharge, and nonhospitalised patients were excluded from this analysis. Data were obtained from the Head Injury Registry, an electronic database of TBI cases in Taiwan, archived by the Injury Prevention Center at Taipei Medical University, and supported by the Department of Health. Review of the Head Injury Registry database allowed the inclusion of 27,625 patients with TBI for analysis in this study. Data on head injuries were recorded by experienced neurosurgeons from each hospital, and were extracted by the same research assistant from the Injury Prevention Center to maximise reliability and consistency. International classification of diseases Data were obtained based on the Ninth Revision ICD-9 codes. Medical records containing the following codes were reviewed: 800804 (fracture of skull or face bones), 850.0 (concussion without loss of consciousness), 850.1–850.2 (concussion with loss of consciousness), 851–851.1 (cerebral/cerebellar/brain stem contusion or laceration), 852.0–853 (extracerebral hematoma, intracerebral hematoma, subarachnoid hemorrhage, subdural hemorrhage, epidural hemorrhage), 854.0 (unspecified intracranial injury), 873.4– 879.8 (gunshot), 900.0 (injury to blood vessels of head and neck), and 950.0–951.5 (injury to cranial nerves). E coding, which was not the standard procedure in Taiwan, was not available in this study. A thorough review of inpatient medical records and related examinations for all head injury patients was carried out. We also recorded data pertaining to the identification of possible TBI causative factors, which included age, gender, GCS score, presence of absence of multiple systemic injuries, dead on arrival status and etiology, and those who died after hospital discharge. TBI severity classification The severity of TBI was classified by the patient’s GCS9,16 score on admission. Injury was classified as: severe, score of 8 or below;

Table 1 Baseline characteristics of 27,625 TBI patients.

Age, years E 1 2 3 4 M 1 2 3 4 5 6 V 1 2 3 4 5 GCS 3 4 5 6 7 8 9 10 11 12 13 14 15 Male gendera Cause of injurya Traffic accident Fall Insult Hit by falling object Other Skull bone fracturea Intracranial hemorrhagea Mortality

Mean/number

Standard deviation/percentage

41.61 3.63 1935 707 2933 22,050 5.62 787 296 314 1014 2489 22,725 4.45 2261 988 644 1867 21,865 13.7 665 222 226 430 429 371 399 495 554 882 900 1681 20,363 17,401

22.8 0.84 7.0% 2.6% 10.6% 79.8% 1.04 2.9% 1.1% 1.1% 3.7% 9.0% 82.3% 1.22 8.2% 3.6% 2.3% 6.8% 79.2% 2.88 2.4% 0.8% 0.8% 1.6% 1.6% 1.3% 1.4% 1.8% 2.0% 3.2% 3.3% 6.1% 73.7% 63.2%

14,997 8324 2469 322 1491 2973 8399 977

54.3% 30.2% 8.9% 1.2% 5.4% 10.9% 30.9% 3.5%

E, eye opening score; M, motor reaction score; V, verbal response score; GCS, Glasgow coma scale. a Data not obtained for some patients.

W.-M. Kung et al. / Injury, Int. J. Care Injured 42 (2011) 940–944

(58.8) (35.4) (2.4) (1.0) (2.4) (16.0) (63.9)

266 233 24 5 25 120 330

(48.1) (42.1) (4.2) (.9) (4.5) (21.9) (60.1)

535 258 49 7 32 134 375

(60.7) (29.3) (5.6) (.8) (3.6) (15.4) (42.9)

488 306 70 9 26 121 415

(54.3) (34.0) (7.8) (1.0) (2.8) (13.6) (46.5)

994 490 127 8 59 254 743

(59.2) (29.2) (7.6) (0.5) (3.5) (15.3) (44.7)

10718 6004 2133 270 1227 1488 3997

(52.7) (29.5) (10.5) (1.3) (6.0) (7.4) (20.0)

<0.001* <0.001*

14.0  2.9, respectively. Most of the subjects were males (63%) and had maximum E, M, V and GCS score (above 70%), and most injuries were sustained due to a traffic accident (54%) or fall (30%). Approximately 31% of subjects had an intracranial hemorrhage, and 11% had a skull bone fracture. The mortality rate was 3.5%. Univariate analysis indicated the GCS score varied by gender and cause of injury (both, p < 0.05, Table 2). The percentages of males and people who were injured in traffic accidents became smaller as the GCS score increased; that is, males hurt in traffic accidents had lower GCS scores than females who were hurt by other causes. Additionally, subjects who had a skull fracture or intracranial hemorrhage had lower GCS scores (both, p < 0.05; Table 2). To examine the difference in survival rates among combinations of E, M and V within the same GCS score, 120 combinations of E–M–V comprising GCS scores 3–15 were tested. For GCS scores 6, 11, 12 and 13, the survival rates of each E–M–V combination under the same score were significantly different (all, p < 0.05; Table 3). In Table 4, the survival rates of E–M–V combinations that have the highest number of patients are shown, though the survival rates among various combinations were similar (all, p > 0.05). The predictive model of survival status which only contained V and M, or any combination of two elements (V + M, E + M, E + V) (Table 5) exhibited a higher area under curve (AUC) than the

(64.6) (30.5) (1.3) (0.0) (3.5) (33.0) (88.3)

265 138 5 3 18 115 372

(61.8) (32.2) (1.2) (0.7) (4.2) (27.4) (87.3)

299 109 7 4 8 114 337

(70.0) (25.5) (1.6) (0.9) (1.9) (26.8) (79.7)

208 138 8 2 13 82 288

(56.4) (37.4) (2.2) (0.5) (3.5) (22.7) (78.7)

211 154 7 0 27 87 278

(52.9) (38.6) (1.8) (0.0) (6.8) (22.2) (70.4)

291 175 12 5 12 77 313

Table 3 40 significant combinations of E, M and V of GCS scores of 6, 11, 12, 13 predictive of survival.

*

a

b

Data were not obtained for some patients. Chi-square test was used. Significantly different among categories, p < 0.05.

146 69 3 0 8 72 197 (61.3) (30.6) (1.8) (0.0) (6.3) (29.9) (84.9) 136 68 4 0 14 64 186 (65.3) (27.1) (3.0) (1.4) (3.3) (37.5) (87.3) 434 180 20 9 22 245 562

<0.001* <0.001* 12388 (61.0)

15 14

1113 (66.4) 560 (62.4)

13

404 (73.1)

12 11 10

353 (71.3)

9

276 (69.5)

8

301 (70.0)

289 (78.5)

7

319 (74.4)

6 5

470 (70.9)

171 (75.7)

4

153 (68.9)

3

Male gender Cause of injury Traffic accident Fall Insult Hit by falling object Others Skull bone fracture Intracranial hemorrhage

Glasgow coma scale

Table 2 Univariate analysis of GCS score for epidemiological characteristics in 27,625 TBI patients.a,b

598 (68.2)

p

942

GCS

E

M

V

Numbers

Percentage (%)

Survival Rate

p*

13 13 13 13 13 13 12 12 12 12 12 12 12 12 12 12 11 11 11 11 11 11 11 11 11 11 11 11 11 11 6 6 6 6 6 6 6 6 6 6

2 3 3 4 4 4 1 2 2 3 3 3 4 4 4 4 1 1 2 2 2 3 3 3 3 4 4 4 4 4 1 1 1 1 2 2 2 3 3 4

6 5 6 4 5 6 6 5 6 4 5 6 3 4 5 6 5 6 4 5 6 3 4 5 6 2 3 4 5 6 1 2 3 4 1 2 3 1 2 1

5 5 4 5 4 3 5 5 4 5 4 3 5 4 3 2 5 4 5 4 3 5 4 3 2 5 4 3 2 1 4 3 2 1 3 2 1 2 1 1

22 101 386 33 269 99 11 11 6 24 545 66 6 31 117 65 2 4 1 49 13 10 60 143 43 5 4 26 106 91 13 2 27 334 0 6 31 2 1 13

2.4 11.1 42.4 3.6 29.6 10.9 1.3 1.3 0.7 2.7 61.8 7.5 0.7 3.5 13.3 7.4 0.4 0.7 0.2 8.8 2.3 1.8 10.8 25.7 7.7 0.9 0.7 4.7 19.0 16.3 3.0 0.5 6.3 77.9 0.0 1.4 7.2 0.5 0.2 3.0

0.89 1.00 0.97 1.00 0.97 0.92 1.00 0.82 1.00 0.67 0.95 0.95 1.00 1.00 0.86 0.77 1.00 0.75 1.00 0.89 1.00 1.00 1.00 0.98 0.97 0.80 1.00 1.00 0.84 0.97 0.70 1.00 0.87 0.80 NA 0.67 0.54 0.50 1.00 0.61

0.002

0.019

0.048

0.018

E, eye opening score; M, motor reaction score; V, verbal response score; GCS, Glasgow coma scale. p-values determined with Gehan’s test. * Significantly different among various combinations, p < 0.05.

W.-M. Kung et al. / Injury, Int. J. Care Injured 42 (2011) 940–944 Table 4 Predominant combinations of E, M and V under the same GCS score: percentage and survival rate influencing survival rate. GCS

E

M

V

Number

Percentage (%)

Survival rate

15 14 13 12 11 10 9 8 7 6 5 4 3

4 3 3 3 3 3 2 1 1 1 1 1 1

6 6 6 5 5 5 5 5 5 4 3 2 1

5 5 4 4 3 2 2 2 1 1 1 1 1

20,355 1089 386 545 143 141 135 100 194 334 160 177 665

100.0 65.0 42.4 61.8 25.7 28.4 34.2 26.4 44.9 77.9 70.8 79.0 100

0.98 0.98 0.98 0.95 0.98 0.95 0.89 0.80 0.80 0.80 0.71 0.46 0.31

E, eye opening score; M, motor reaction score; V, verbal response score; GCS, Glasgow coma scale.

Table 5 Receiver-operating characteristic analysis of mortality prediction.

E M V E+M V+E V+M GCS

Area under ROC curve

95% confidence interval

0.863 0.900 0.890 0.877 0.904 0.903 0.885

0.847–0.878 0.875–0.903 0.877–0.903 0.865–0.890 0.892–0.916 0.891–0.916 0.873–0.897

ROC, receiver-operating characteristic; E, eye opening score; M, motor reaction score; V, verbal response score; GCS, Glasgow coma scale.

E + M + V model (Table 4). In other words, even a model containing only a single element of the GCS score may be as effective, or even more effective, than the GCS model at predicting survival after TBI. Table 6 presents factors related to survival status. The hazard of death increased as age increased [hazard ratio (HR) = 1.02, p < 0.001]. Compared with patients with the highest score, patients with a lower E score had a higher hazard of death (HR = 30.0–2.6, p < 0.001). Similarly, patients with lower M (HR = 76.5–7.1, p < 0.001), V (HR = 38.6–2.6, p < 0.001), and GCS scores (HR = 103.3–3.5, p < 0.001) had higher hazard of death than those with the highest scores, and the hazards were reduced as the scores increased. Male gender (HR = 1.3, p = 0.002), skull bone fracture (HR = 19.1, p < 0.001), and intracranial hemorrhage (HR = 3.1, p < 0.001) were also risk factors of death. When compared with subjects injured in a traffic accident, those with an injury due to a fall (HR = 1.2, p = 0.007) had a 1.23 times higher hazard of death. Discussion Simplifying predictive models of survival in TBI has significant clinical utility. Accurate and complete GCS scores are difficult to obtain in many situations and a simplified assessment system using fewer parameters may make training of medical professionals easier, lead to less variability in reporting and more accurate patient assessments. In this large, retrospective study of TBI patients, we have found that combinations of the individual components of the GCS score, and even a single element in some instances, are predictive of mortality in patients with TBI, and this predictive value is greater in some cases than the complete GCS score. Since its introduction to the medical community, the GCS has become one of the primary tools in the assessment of TBI, and alone or in combination with other clinical factors has been shown to be useful in predicting outcomes after traumatic and non-

943

Table 6 Hazard ratios predicting mortality of the parameters. Hazard ratio

95% confidence interval

p

Age E 1 2 3 4

1.1

(1.1–1.2)

<0.001*

30.1 10.9 2.62 Ref.

(25.2–35.9) (8.3–14.1) (2.0–3.5) –

<0.001* <0.001* <0.001* –

M 1 2 3 4 5 6

76.5 43.7 22.8 14.2 7.1 Ref.

(62.3–93.8) (33.6–56.9) (16.7–31.0) (11.0–18.2) (5.6–9.1) –

<0.001* <0.001* <0.001* <0.001* <0.001* –

38.6 17.1 5.8 2.6 Ref.

(31.4–47.5) (13.2–22.1) (3.8–8.8) (1.7–3.9) –

<0.001* <0.001* <0.001* <0.001* –

GCS score 3 4 5 6–8 9 10–12 13 14–15

103.3 66.8 34.7 21.8 12.3 6.1 3.5 Ref.

(82.5–129.3) (50.2–88.9) (25.0–48.1) (17.1–28.0) (8.3–18.2) (4.5–8.3) (2.1–5.9) –

<0.001* <0.001* <0.001* <0.001* <0.001* <0.001* <0.001* –

Gender Male Female

1.3 Ref.

(1.1–1.5) –

Skull bone fracture Yes No

19.1 Ref.

(14.7–24.9) –

<0.001* –

Intracranial hemorrhage Yes No

3.1 Ref.

(2.7–3.5) –

<0.001* –

Cause of injury Traffic accident Fall Insult Hit by falling object Others

Ref. 1.2 0.3 0.4 0.9

– (1.0–1.4) (0.2–0.5) (0.2–1.2) (0.6–1.2)

–

V 1 2 3 4 5

0.002* –

0.007* <0.001* 0.108 0.473

E, eye opening score; M, motor reaction score; V, verbal response score; GCS, Glasgow coma scale. * Significantly correlated with the outcome, p < 0.05.

traumatic brain injury.7–10,13,16 A report by Levin et al.8 looking at the clinical course of the vegetative state after closed head injury by analyzing data from the Traumatic Coma Data Bank found GCS and pupillary findings correlated with persistent vegetative state. Another report by Murray et al.10 found GCS to be one of the most powerful independent prognostic variables in predicting outcome after TBI. Despite the usefulness of the GCS, many authors have sought to improve its functionality and simplify its use.5,11,16 Foreman et al.3 compared GCS score and other assessments of injury severity and found that the GCS score, abbreviated injury score (AIS), and injury severity score (ISS) were only weakly correlated with 12-month outcomes; however, the combination of GCS and AIS or ISS correlated with outcome better than any of the 3 methods alone. The male to female ratio in this study was approximately 2–1, a finding close to results reported by the Center for Disease Control (CDC) in the United States (US)6 and consistent with the result of

944

W.-M. Kung et al. / Injury, Int. J. Care Injured 42 (2011) 940–944

many other studies.2,14 We found traffic accident to be the main cause of TBI, accounting for 54% of injuries, similar to other reports.2,14 Injuries related to motorcycle use comprised the majority of injuries caused by traffic accidents (75%). In our study, 19% of injuries were associated with alcohol use, a finding lower than the 35–50% reported in other series.15 This result is likely because testing for alcohol (breath or blood tests) in the emergency room is not performed for every TBI patient in Taiwan. The mortality rate of 3.5% in our study is lower than that reported by the CDC in the US.6 Underestimation of the mortality rate in our study was most likely because patients who were dead on arrival in the emergency department and those who died after hospital discharge were excluded from our analysis. The distribution of TBI patients according to severity in our study is similar to the result of other series, where the rates of minor, moderate and severe TBI were reported to be 83.0%, 8.4% and 8.5%, respectively.6,12 Our data demonstrated that the survival rate of each E–M–V combination under the same GCS score (GCS score 13, for example) were significantly different. Analyzing those predominant combinations, we found that the E–M–V combinations in TBI patients with GCS scores of 13 and 14 were similar (E3M6V4 and E3M6V5, respectively), and that with GCS scores 10, 11 and 12 were also similar (E3M5V2, E3M5V3 and E3M5V4, respectively). This observation suggests that for the predominant number of TBI patients with GCS scores higher than 10, the level of consciousness and clinical status of TBI patients was highly determined by verbal (V) score. Whereas, for the TBI patients with GCS scores of 3–7, the predominant E–M–V combinations were E1M1V1, E1M2V1, E1M3V1, E1M4V1 and E1M5V1, respectively, suggesting that for the predominant number of TBI patients with GCS scores lower than 7, the level of consciousness and clinical status was highly associated with motor (M) score instead. In the analysis of individual E, M and V scores, as well as various combinations of E, M and V, the area under the ROC curve (E = 0.863, M = 0.900, V = 0.889, E + V = 0.904, M + V = 0.903, E + M = 0.877) was similar to the results of E + M + V (0.885). This suggests that evaluation of TBI patients using the score of various E, M, V combinations or E, M or V alone is as accurate a predictor of mortality as the total GCS score. There are only a few published reports exploring the relationship between E, M or V, and score combinations and survival.4,5 Gill et al.4 in an analysis of 8412 TBI cases concluded that the 3 individual GCS components alone, as well as 2 simplified 3-point scores, were as effective in predicting outcomes as the total GCS score. Healey et al.5 in an analysis of over 200,000 patients in the US National Trauma Databank (NTDB), determined that the motor component of the GCS alone is as effective, and perhaps more useful than the total GCS score in predictive models. A minor limitation of the study which should be considered is that separate tests have been conducted at GCS scores of 4, 5, 6, etc. Thus, multiple tests of significance have been performed and the potential exists that a significant result may really be a false positive. Though this does not invalidate the tests that have been performed, caution should be exercised in the interpretation of the results.

Conclusion The results of this study indicate that the 3 fundamental elements comprising the Glasgow coma scale, E, M and V, individually, and in certain combinations are predictive of the survival of TBI patients. This observation is clinically useful when evaluating TBI patients in whom a complete GCS score cannot be obtained. Conflict of interest statement The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper. Acknowledgements The author(s) disclosed receipt of the following financial support for the research and/or authorship of this article: National Health Research Institute, Taiwan (Grant No. NHRI-EX99-9707PI); Department of Health, Executive Yuan, Taiwan (Grant No. DOH99TD-B-111-003). References 1. Chung CY, Chen CL, Cheng PT, et al. Critical score of Glasgow coma scale for pediatric traumatic brain injury. Pediatr Neurol 2006;34(5):379–87. 2. Fife D, Faich G, Hollinshead W, Boynton W. Incidence and outcome of hospitaltreated head injury in Rhode Island. Am J Public Health 1986;76(7):773–8. 3. Foreman BP, Caesar RR, Parks J, et al. Usefulness of the abbreviated injury score and the injury severity score in comparison to the Glasgow Coma Scale in predicting outcome after traumatic brain injury. J Trauma 2007;62(4):946–50. 4. Gill M, Windemuth R, Steele R, Green SM. A comparison of the Glasgow Coma Scale score to simplified alternative scores for the prediction of traumatic brain injury outcomes. Ann Emerg Med 2005;45(1):37–42. 5. Healey C, Osler TM, Rogers FB, et al. Improving the Glasgow Coma Scale score: motor score alone is a better predictor. J Trauma 2003;54(4):671–8. 6. Langlois JA, Rutland-Brown W, Thomas KE. Traumatic brain injury in the United States: emergency department visits, hospitalizations, and deaths. Atlanta, GA: Centers for Disease Control and Prevention, National Center for Injury Prevention and Control; 2004. 7. Lenartova L, Janciak I, Wilbacher I, et al. Severe traumatic brain injury in Austria III: prehospital status and treatment. Wien Klin Wochenschr 2007;119(1– 2):35–45. 8. Levin HS, Saydjari C, Eisenberg HM, et al. Vegetative state after closed-head injury. A Traumatic Coma Data Bank Report. Arch Neurol 1991;48(6):580–5. 9. McNett M. A review of the predictive ability of Glasgow Coma Scale scores in head-injured patients. J Neurosci Nurs 2007;39(2):668–75. 10. Murray GD, Butcher I, McHugh GS, et al. Multivariable prognostic analysis in traumatic brain injury: results from the IMPACT study. J Neurotrauma 2007;24(2):329–37. 11. Mushkudiani NA, Hukkelhoven CW, Herna´ndez AV, et al. A systematic review finds methodological improvements necessary for prognostic models in determining traumatic brain injury outcomes. J Clin Epidemiol 2008;61(4):331–43. 12. Nell V, Brown DS. Epidemiology of traumatic brain injury in Johannesburg – II. Morbidity, mortality and etiology. Soc Sci Med 1991;33(3):289–96. 13. Nijboer JM, van der Naalt J, Ten Duis HJ. Patients beyond salvation? Various categories of trauma patients with a minimal Glasgow Coma Score. Injury 2010;41(1):52–7. 14. Parkison D, Stephensen S, Phillips S. Head injuries-a prospective, computerized study. Can J Surg 1985;28(1):79–83. 15. Sloan EP, Zalenski RJ, Smith RF, et al. Toxicology screening in urban trauma patients: drug prevalence and its relationship to trauma severity and management. J Trauma 1989;29(12):1647–53. 16. Wijdicks EF. Clinical scales for comatose patients: the Glasgow Coma Scale in historical context and the new FOUR Score. Rev Neurol Dis 2006;3(3):109–17.