Original Contribution

A simple and useful coma scale for patients with neurologic emergencies: the Emergency Coma Scale^?,??

Chiaki Takahashi MD^a,?, Hiroshi Okudera MD, PhD^a, Hideki Origasa PhD^b, Eiichi Takeuchi MD, PhD^c, Kazuhito Nakamura MD^d, Osamu Fukuda MD, PhD^e, Isao Date MD, PhD^f, Takashi Tokutomi MD, PhD^g, Tohru Aruga MD, DMSc^h, Tetsuya Sakamoto MD, PhDⁱ, Hitoshi Kobata MD, PhD ^j, Tomio Ohta MD, PhD^k

^aDepartment of Crisis Medicine, Graduate School of Medicine, University of Toyama, Toyama 930-0194, Japan ^bDivision of Biostatistics and Clinical Epidemiology, University of Toyama Graduate School, Toyama 930-0194, Japan ^cKyoto Fushimishimizu Hospital, Kyoto 612-8321, Japan

^dDepartment of neurosurgery, Osaka City General Hospital, Osaka 534-0021, Japan ^eDepartment of Neurosurgery, Saito Memorial Hospital, Niigata 949-6602, Japan ^fDepartment of Neurosurgery, Okayama University Hospital, Okayama 700-8558, Japan ^gDepartment of Neurosurgery, Kurume University, Fukuoka 830-0011, Japan

^hDepartment of Emergency and Critical Care Medicine, Showa University, Tokyo 142-8666, Japan ⁱTrauma and Critical Care Center, Teikyo University School of Medicine, Tokyo 173-0003, Japan ^jDivision of Neurosurgery, Osaka Mishima Emergency and Critical Care Center, Osaka 569-1124, Japan ^kOsaka Neurological Institute, Osaka 561-0836, Japan

Received 15 August 2009; revised 11 September 2009; accepted 15 September 2009

Abstract

Objectives: The Emergency Coma Scale was developed in Japan in 2003. We planned a multicenter study to evaluate the utility of the ECS by comparison of the ECS and the Glasgow Coma Scale (GCS).

Methods: Ten medical facilities, including 4 university hospitals in Japan, participated in this study. We evaluated and recorded the level of consciousness, using the ECS and GCS, of all patients transported to these medical facilities by ambulance. We then performed a statistical analysis of the level of rater agreement of each scale using the average weighted ? coefficient according to the types of diagnosis at time of discharge and the occupations of the raters. We then evaluated the relationship between outcome of patients and their scores on the ECS and GCS by logistic regression analysis.

Results: The ECS showed the greater agreement among raters in patient scoring (0.802). In patients with traumatic brain injury and cerebrovascular disease, the ECS also yielded the higher agreement (0.846 and 0.779, respectively). The ECS score appears to be more strongly related than the GCS to patient outcome as measured by the Glasgow Outcome Scale .

^? This trial was supported by the ECS Society, which is a joint committee of the Japanese Congress on NeuroSurgical emergencies and the Japan Neurological Emergency Society.

^?? Prior presentations: The Emergency Coma Scale for patients in the ED: concept, validity and simplicity. Am J Emerg Med. 2009 Feb;27(2):240-3.

* Corresponding author. Tel.: +81 76 434 7786; fax: +81 76 434 5110.

E-mail address: [email protected] (C. Takahashi).

0735-6757/$ - see front matter (C) 2011 doi:10.1016/j.ajem.2009.09.018

Conclusions: Our results showed that the ECS is a simple and readily understandable coma scale for a wide range of professionals in the field of neurologic emergencies. Furthermore, ECS appears to be suitable for evaluating patients in neUrologic emergency settings.

(C) 2011

Introduction

Patients with disturbances of consciousness of varying severity are often encountered in both outpatient clinics and emergency departments (EDs). On such occasions, the rapid and accurate evaluation of the patient’s level of conscious- ness by coma scales is essential. The Glasgow Coma Scale (GCS) [1-3], proposed by Teasdale and Jennett in 1974, is widely used and accepted. We occasionally discuss the advantages and disadvantages of the GCS in the ED and on the ward. The GCS has problems attributable to its complexity [4,5]. The total score of the GCS represents multiple means due to the numerous possible combinations of the 3 gauges which comprise it [5].

Ohta et al developed the Emergency Coma Scale (ECS) in 2003, which is a hybrid scale combining the advantages of the GCS and the Japan Coma Scale, which has a single gauge and a 3 major category structure, and is accepted in Japan (Table 1) [6,7]. The most notable difference between the ECS and GCS is the structure of grading. The structure of the GCS includes 3 gauges, and the total score is the sum of each component score. Meanwhile, the ECS has a single-gauge structure [8,9]. We anticipated that ECS scores would show a higher agreement among multiple raters, as its structure is simple enough to be applied by a variety of professionals in this field. And the ECS scores might correlate with the outcome of the patients strongly, if it could describe the severity of Consciousness disturbance of patients accurately. To test this hypothesis, we designed a multicenter study entitled “ECS Co-Operative Multi-center Evaluation Trial”

(E-COMET) [10,11].

Table 1 The ECS

Category 1: The patients open their eyes, speak and/or behave spontaneously (awake) and

1 Can say correct date, place and person

2 Cannot say correct date, place and person

Category 2 : The patients can open their eyes, speak and/or behave (aroused) by

10 Speech

20 Painful stimuli

Category 3 : The patients can neither open their eyes nor speak by painful stimuli (not aroused) but respond with

100L Localization

100W Withdraw forearm with opened armpits 200F Flex forearm with closed armpits

200E Extend forearm with closed armpits 300 None

L indicates localization; W, withdrawal; F, flexion; E, extension.

Methods

Study design

E-COMET was a multicenter, prospective, statistical comparative study. It consisted 2 series of trials (E-COMET STEP I and STEP II). In STEP I, the agreement among multiple raters and the relationships between outcome and coma scale scores were evaluated [12]. In STEP II, the accuracy of each scale was evaluated. We report herein the final results of E-COMET STEP I. The institutional review board of all designated facilities approved this study.

Study setting and population

Ten medical facilities, including 4 university hospitals in Japan, participated in this study. Eight facilities belong to the ECS Society, which is a joint committee of the Japanese Congress on Neurosurgical Emergencies and the Japan Neurological Emergency Society [8,11]. The central site of this trial was the Department of Crisis (Emergency and Disaster) Medicine at the University of Toyama.

Between April 2007 and April 2008, we prospectively collected data for these patients who ranged in age from 5 to 99 years. We excluded patients younger than 5 years because we determined that because of immaturity, they may not be able to completely understand verbal orders and respond to orientation questions as is required for evaluating the level of consciousness. Reilly et al [12] have previously described that children older than 2 years age can obey commands and that accurate answers to orientation questions are expected by 5 years of age. In addition, we stopped the evaluation process and excluded patients for whom the evaluation posed a risk of aggravating their condition due to time loss. The timely provision of medical care to patients took priority over evaluation by coma scales in the clinical study.

Study protocol

When the patients arrived at the ED, designated members of the medical team immediately evaluated the level of consciousness using the ECS and GCS. At this time, all raters were inhibited to tell other raters the scores they were recorded. We postulated that the raters had already been instructed in how to score a patient’s consciousness level using the ECS and GCS, with reference to the manual of guidebook from the Immediate Stroke Life Support course

[13] (Table 2). In each facility, before the study began, the designated person in charge gave lectures on the evaluation

Table 2 Manual for evaluation by the ECS (2004) Manual for evaluation by the ECS (2004)

[Category 1] Discerning whether the patients are wakeful or not. STEP1 Check the spontaneous eye opening, speaking,

movements of patients (category 1 or not?)

If raters could observe at least one of the following 3 actions: eye opening, speaking, movements of the patients, they are wakeful.

Be wakeful?

Yes -> go to STEP 2 No -> go to STEP 3

STEP2 Check the orientation of the patients (category 1 or category 2?)

Ask the patients whether they can say time, person, place correctly.

Do they have orientation? Yes -> ECS 1

No -> ECS 2

[Category 2] Observation of the response for stimulation STEP3 Observe the response for speech.

Call out the patients with loud voice. Raters also can touch and shake the patients.

Can they awake by voice stimuli? Yes -> ECS 10

No -> go to STEP 4

STEP4 Observe the response for painful stimuli

If there is any attendance for patients, raters should get agreement for painful stimuli to them.

If there is no wound in the trunk of patients, raters push harder the sternum with a fist.

Raters also can push harder nails of the patients to evaluate the response.

Raters must stimulate the patients with calling out them. Can they awake by painful stimuli?

Yes -> ECS 20

No -> go to STEP5

[Category 3] The state the patients cannot awake with all stimulations

STEP5 Observe the response for painful stimuli

Evaluate the consciousness level rapidly with painful stimuli as same as STEP4

localize the parts with painful stimuli -> 100L withdraw forearm with opened armpits -> 100W flex forearm with closed armpits -> 200F

extend forearm with closed armpits -> 200E no movement ->300

bExample for usingN

A patient, complaining of Severe headache and closing his eyes is classified into “yes” in STEP 1. If he could not answer the place and time correctly, he is classified into ECS 2 in STEP 2.

methods for these scales to all raters. The evaluation records included the following: the age and sex of each patient, the coma scales in daily use, the time and date of the evaluation, the patient’s score on each scale applied, and the patient’s diagnosis. Each patient’s progress and outcome were also noted later. We documented whether or not the patients were admitted to the hospital, were intubated and/or underwent

surgery, the lengths of hospital and Intensive care unit stays, the cause of death if applicable, and the outcome. We applied the Glasgow Outcome Scale to evaluate patient outcomes [14,15]. Outcomes were evaluated at the time of discharge. The raters then input these data to the ECS Web site (www.ecs.gr.jp), and the data were then directly transferred to our database in the central study site.

Data analysis

We first performed a statistical analysis of the weighted agreement among multiple raters according to patient diagnosis and the occupations of the raters. The agreements among multiple raters were compared by calculating the average weighted ? coefficient and applying the 95% confidence interval (CI) calculated from the standard error for the test. First, the weighted ? coefficients on each patient were calculated, and then these coefficients were integrated as an average weighted ? coefficient. When there is perfect agreement, the ? coefficient is equal to 1.0. ? coefficients less than 0.4 indicate fair agreement, values

>=0.41 and b0.6 have moderate agreement, values >=0.61

and <=0.8 have substantial agreement, and values >=0.81 have almost perfect agreement [16]. In this study, we collected data from 33 raters, including 15 physicians, 8 nurses, 4 residents, 3 paramedics, and 3 medical students. All analyses were performed using SAS 9.1 J (SAS Institute Inc, Cary, NC).

We also performed analysis to examine the relationships between outcome and various independent variables. We reassigned the 5 grades of the GOS into 2 grades to use the GOS as a dependent variable. The severe grade included death (D), vegetative state (VS), and severe disability (SD), and the mild grade included the remaining GOS grades. We adopted forced entry logistic regression analysis and calculated the odds ratios (ORs) and 95% CIs according to the independent variables. After the first analysis, we excluded the independent variables that showed a signif- icantly increased OR (N100). We then repeated the forced entry logistic regression analysis. Next, we evaluated the relationships between outcome and the 3 major categories of each coma scale, and the motor component of the GCS. We also reorganized the 13 grades of the GCS into 3 major categories for descriptive purposes: Category 1 corre- sponded to GCS 12-15, Category II to GCS 8-11, and Category III to GCS 3-7. Data were analyzed using the statistical software program Dr SPSS II for Windows (SPSS, Inc., Chicago, IL, USA).

Results

Four hundred ninety-five patients were included in the study, with an age range of 5 to 99 years (mean, 58.6 years; SD, 22.4 years). Ninety-nine patients (20.0%) had

cerebrovascular disease (CVD), including 18 (3.6%) with subarachnoid hemorrhage , 27 (5.5%) with intracere- bral hemorrhage (ICH), and 53 (10.7%) with cerebral infarction (Table 3). We analyzed 1382 data points from our database and calculated the rating agreements using the average weighted ? coefficient of the 3 coma scale scores among multiple raters. One hundred seventy-two data points were excluded for 2 reasons: defect of data item or mismatching of evaluation time. We attempted to have raters evaluate the level of consciousness of patients at the same time whenever possible, as the level of consciousness of patients may change from moment to moment. For this reason, we excluded some data if the time of evaluation exceeded a 20-minute difference between raters.

Table 4 shows the agreements of each scale score among multiple raters and the 95% CI calculated for the various elements. The upper part of the table shows the agreements according to the occupations of the raters. The other portions of the table represent the results for residents, paramedics, and medical students (“other”). For the categories of whole raters, nurses, and other raters, the weighted agreements were almost perfect in ECS (0.80 [95% CI, 0.78-0.82], 0.88 [95%

CI, 0.83-0.93], and 0.81 [95% CI, 0.76-0.86], respectively);

however, comparison between the ECS and GCS showed no statistically significant difference (P N .05). For physicians, however, the weighted agreements showed almost the same results between the ECS and the GCS. The lower part of the table shows the weighted agreements of each scale score and the 95% CI according to the disease type. Cerebrovascular disease includes cerebral infarction, ICH, and SAH. In all patients with CVD, ICH, Traumatic brain injury , and epilepsy, the ECS showed significantly greater agreements (0.78 [95% CI, 0.78-0.80], 0.85 [95% CI, 0.82-0.87], and

0.59 [95% CI, 0.55-0.63]). In epilepsy and intoxication cases, the agreements were greater in ECS.

Table 3 Patient demographics

N =		495 (%)
Sex
Male	261	(52.7)
Female	224	(45.3)
Unknown	10	(2.0)
Age group (y)
7-49	156	(31.5)
50-	349	(71.0)
Type of the diseases
Cerebrovascular disease	99	(20.0)
CI	53	(10.7)
ICH	27	(5.5)
SAH	18	(3.6)
Trauma	135	(27.3)
TBI	66	(13.3)
Epilepsy	36	(7.2)
Toxicosis	35	(7.1)
Other	45	(9.1)

GCS		ECS
Whole rater	0.78 (0.77-0.79)	0.80 (0.78-0.82)
Medical physician	0.84 (0.84-0.86)	0.83 (0.81-0.86)
Nurse	0.80 (0.74-0.85)	0.88 (0.83-0.93)
Other	0.73 (0.68-0.78)	0.81 (0.76-0.86)
Diseases CVD	0.65 (0.63-0.68)	0.78 (0.76-0.80)
Cerebral infarction	0.53 (0.49-0.57)	0.55 (0.50-0.60)
ICH (including SAH)	0.75 (0.71-0.78)	0.84 (0.80-0.88)
TBI	0.74 (0.71-0.76)	0.85 (0.82-0.87)
Epilepsy	0.40 (0.36-0.45)	0.59 (0.55-0.63)
Intoxication 0.49 (0.44-0.55) 0.63 (0.59-0.67) The category “other” includes paramedics and medical students.

We could follow up 189 patients (95 men and 94 women) at discharge. Glasgow Outcome Scale scores indicated that 68.7% of the patients had good outcomes (good recovery and moderate disability). Fourteen patients died: 8 of brain herniation, 1 of Liver dysfunction, 1 of septic shock, 1 of multiple organ failure, 2 of ventricular fibrillation due to myocardial infarction, and 1 of Cardiac rupture secondary to injury in a traffic accident. We analyzed our data and calculated the OR and 95% CIs of the coma scale scores, and independent variables. The independent variable represent- ing intubation was fully excluded and that representing hospital admission was partially excluded in the process of statistical analysis. Among all raters, of the 2 scales, ECS scores (OR, 1.58; 95% CI, 1.42-1.76) appeared to correlate more strongly with outcome based on GOS scores (Table 3). The Motor component of the GCS, which has 6 grades, had an OR of 1.71. Hospitalized and/or older patients were found to have significantly poorer outcomes. We also analyzed the data according to disease type, that is, CVD or TBI only. In patients with CVD, the ECS score (OR, 3.37; 95% CI, 1.91- 5.97) again apparently correlated most strongly with outcome based on the GOS score (Table 4). The OR (4.18; 95% CI, 1.95-8.92) of the Motor component of the GCS was higher than that of the ECS. In patients with TBI, a similar tendency was found. With each decade increase in age, the OR for patients with TBI was higher than that for patients with CVD. With respect to sex, the OR was higher for males than females among patients with CVD, whereas among patients with TBI, females had higher OR. We also analyzed the relationships between the 3 major categories of each scale and outcome (Table 5). The 3 major categories of the ECS (OR, 4.93; 95% CI, 3.21-7.58) showed the strongest statistically significant correlation with outcome based on the GOS score. The major categories of the ECS (OR, 30.1; 95% CI, 8.51-106.4) also showed the strongest correlation with outcome based on the GOS score in CVD. In patients

Table 4 The average weighted agreements and 95% CI of each scale score among multiple raters according to the occupations of the raters and the disease type

Occupations Agreement (95% CI)

Item		OR (95%CI), P
		ECS		GCS		GCS M
All	The score of scale (per 1-grade increase)	1.58 (1.42-1.76)	b.001	1.35 (1.26-1.44)	b.001
	Age (per 10-y increase)	1.32 (1.15-1.51)	b.001	1.33 (1.16-1.52)	b.001
	Sex (compared to male)	1.46 (0.85-2.51)	.174	1.18 (0.69-2.03)	.544
	Length of hospital stay (per 1-d increase)	1.02 (1.01-1.03)	.002	1.01 (1.00-1.03)	.011
CVD	The score of scale (per 1-grade increase)	3.37 (1.91-5.97)	b.001	1.71 (1.38-2.13)	b.001
	Age (per 10-y increase)	1.15 (0.76-1.74)	.502	1.06 (0.68-1.60)	.722
	Sex (compared to male)	0.69 (0.22-2.23)	.540	0.56 (0.18-1.78)	.326
	Length of hospital stay (per 1-d increase)	1.04 (1.01-1.08)	.018	1.03 (0.99-1.07)	.131
TBI	The score of scale (per 1-grade increase)	4.19 (1.75-10.01)	.001	2.15 (1.54-3.02)	b.001
	Age (per 10-y increase)	2.26 (1.37-3.67)	.001	2.39 (1.87-4.01)	.001
	Sex (compared to male)	10.94 (1.53-78.27)	.011	8.46 (1.24-57.69)	.029
	Length of hospital stay (per 1-d increase)	1.01 (1.00-1.03)	.552	1.02 (0.99-1.15)	.144
OR of 3 major components and Motor component of the GCS
All	The score of scale (per 1-grade increase)	4.93 (3.21-7.58)	b.001	3.01 (2.12-4.29)	b.001	1.71 (1.45-2.12)	b.001
	Age (per 10-y increase)	1.41 (1.22-1.63)	b.001	1.41 (1.22-1.63)	b.001	1.34 (1.20-1.58)	b.001
	Sex (compared to male)	1.67 (0.94-2.99)	.082	1.54 (0.87-2.75)	.139	1.64 (0.94-2.84)	.081
	Admission (compared to never)	5.64 (1.58-20.10)	.008	3.16 (0.90-11.08)	.072
	Length of hospital stay (per 1-d increase)	1.02 (1.01-1.03)	b.001	1.01 (1.00-1.03)	.010	1.01 (1.00-1.02)	.012
CVD	The score of scale (per 1-grade increase)	30.1 (8.51-106.4)	b.001	7.31 (2.97-18.32)	b.001	4.18 (1.95-8.92)	b.001
	Age (per 10-y increase)	1.07 (0.71-1.61)	.742	0.96 (0.67-1.38)	.841	0.90 (0.74-1.43)	.781
	Sex (compared to male)	0.77 (0.22-2.67)	.768	0.711 (0.25-1.99)	.517	0.92 (0.32-2.63)	.877
	Length of hospital stay (per 1-d increase)	1.03 (1.00-1.07)	.059	1.04 (1.00-1.07)	.052	1.05 (1.01-1.09)	.006
TBI	The score of scale (per 1-grade increase)	31.0 (8.51-106.4)	b.001	28.4 (5.46-147.2)	b.001	5.24 (2.41-11.42)	b.001
	Age (per 10-y increase)	1.89 (1.22-2.59)	.004	2.24 (1.28-3.87)	.004	2.30 (1.40-4.53)	.002
	Sex (compared to male)	7.9 (1.29-48.25)	.025	6.58 (1.17-36.96)	.032	7.38 (1.27-43.0)	.026
	Length of hospital stay (per 1-d increase)	1.0 (0.97-1.03)	.864	1.01 (0.99-1.04)	.416	1.01 (0.99-1.04)	.413

with TBI, the OR of the ECS and GCS were nearly even, showing no statistically significant difference. The fitness of the model was good in all analyses, as confirmed by Nagelkerke’s R² value [17].

Table 5 Odds ratios and 95% CI in all patients, those with CVD, and those with TBI among the 2 coma scale scores, the Motor component of the GCS, and other items

Coma scale scores and items that show higher ORs correlate strongly with poor outcome as measured by GOS. The lower part of the table shows OR and 95% CI among the 3 major categories of each coma scale and other items.

Discussion and limitations

The GCS is recognized for its accuracy and strong correlation with patient outcome [18-21]. The result of this trial sufficiently proved it. However, the GCS users need to evaluate 3 independent gauges and calculate the total score if they want to accurately determine the consciousness level of a patient. The total score then consists of 3 gauges, that is, E, V, and M, representing multiple means due to the numerous possible combinations of the 3 components. Theoretically, there are 120 different combinations [5]. Riechers et al [4] stated that the GCS is therefore somewhat complex, and users generally need considerable time to master its use.

We speculated that an optimal coma scale, one that could be used by all medical staff members would result in rapid and simple score determinations for patients without

confusion. We hypothesized that the expression of new ECS would incorporate simplicity in expression and structure. The ECS has single-gauge structure and 3 major classes according to the level of arousal, and each class has detailed subclasses. The most severe class, category III, describes a coma state in which the patient cannot be awakened by any stimulation. The intermediate class, category II, is a state in which patients can be aroused by stimulation and the mild class, category I, describes patients who remain wakeful without any stimulation [22]. The definition of differentiation between purposeful movement in category I and random movement depend on whether patients can obey the movement order of raters [6]. This structure, based on 3 major classes such as a triage tag, is very useful and easy to understand. Even if medical staff members are only able to determine the major class of consciousness level, for example, if there is not enough time for detailed scoring using the ECS because of the emergent situation in the ED, other health care providers can obtain an approximate understanding of the patient’s condition. Thereafter, when the patient is stabilized, their level of consciousness can be reassessed in detail based on the 9 subclasses of the ECS.

We found that for all occupations, the ECS showed almost perfect agreement, but the difference between the ECS and GCS scores were not significant (P N .05), except in nurses. The GCS showed substantial agreement between the rater categories of all raters, nurses, and others. This tendency may be explained by physicians understanding the evaluation methods of the GCS. With respect to nurses and others, paramedics in particular, it is difficult to evaluate by the GCS accurately compared to physicians. The data of “other” include many data from medical students who have no experience, at this point in their training, with Patient evaluation by coma scales. They showed markedly greater rating agreement using the ECS. This result suggests that the raters could evaluate consciousness of the patients concurrently by ECS even for multi- professionals, as well as among staff members engaged in emergency medicine. Moreover, in using the ECS, it appears that raters can, based on consensus, evaluate a disturbance in the level of consciousness resulting from brain compression because the agreement of ECS is significantly greater in CVD, ICH, and TBI, which can cause an intracranial mass lesion in the super acute phase. The ECS may have sufficient capability to evaluate the level of consciousness in CVD and TBI, which are 2 common major neurologic emergencies. The ECS was made aiming to detect consciousness disturbance due to brain herniation quickly [22]. In all CVD cases, the weighted agreements were quite high, but there may be specific diseases for which the ECS is either very well or rather poorly adapted, and further analysis of this issue is required. In epilepsy and intoxication cases, the ECS showed the greater agreement among the 2 scales.

Two coma scales included in this trial were originally developed to evaluate consciousness levels of patients in the acute phase of a coma, but it was suggested that they might also be useful for predicting patient outcomes [23]. Under all conditions examined, the ECS showed the highest OR between score and outcome, possibly because the ECS is highly accurate in evaluating consciousness level. Glasgow Coma Scale scores sometimes vary among multiple raters because the definition of the E and V components of the GCS leads to overestimation [4,5]. In patients with CVD and TBI, the ECS also had the highest OR among the 3 scale scores and is suitable for predicting the outcomes of neurologic disorders. When we compared OR by sex, only females with CVD showed a lower OR, possibly because the CVD group included many female patients with SAH but who were in comparatively good condition. In addition, there were more male patients with arteriosclerotic or cardioembolic infarction, conditions that frequently worsen. This tendency is consistent with general statistical results for CVD in Japan [24].

The OR between the 3 major categories of each scale and patient outcome revealed the ECS to have the strongest correlation with outcome, suggesting that its simple 3-category structure might be useful in predicting

patient outcome. Our results suggest that the 3-major- category structure of the ECS, as a tool for predicting the outcomes of patients, might be suitable for clinical application, even if the 9 detailed subcategories of ECS are not used. Although we divided the GCS scores into 3 categories for this analysis, this approach can be problematic. The GCS cannot be rearranged to single gauge simply and is divided into 3 categories such that it is the sum of 3 independent gauges and it differs in concept from the ECS. Therefore, we also performed a logistic regression analysis using the Motor component of the GCS as an independent variable and compared its OR with that of each coma scale. Teoh et al [5] reported that the Motor component of the GCS predict the outcomes of severely ill patients well. The Motor score component of the GCS showed greater OR than the total score of the GCS, for all diseases. From these results, we consider it feasible to evaluate the consciousness levels of patients using only the

3 categories of the ECS, with the ultimate aim of accurately predicting outcome.

Although the same training manuals were used, it was not possible to standardize the teaching methods of the individual lecturers who taught the raters about the coma scales.

Bias may have resulted from the fact that some raters in Japan were already familiar with the GCS but learned about the ECS for the first time at the start of this study.

The GOS used in the present study was developed to evaluate the outcomes of patients with TBI [11,12]. However, for the purposes of this study, we adapted it for use with other disease types such as stroke, epilepsy, and infectious diseases.

Conclusions

The ECS score could yield greater agreements among multiple classes of medical professionals and could predict outcome of the patients well. The ECS may be widely accepted among all medical professionals, even public, because of Its simplicity and usability.

Acknowledgments

We thank our colleagues, Dr Masahiro Wakasugi, Dr Takashi Asahi, Dr Daisuke Tange, and Dr Joji Hamada, as well as the nurses and young physicians in training in the ED of Toyama University Hospital. We gratefully acknowledge the support of the following individuals and institutions: Division of Biostatistics and Clinical Epide- miology, University of Toyama Graduate School; Ms Kumagai, Department of Neurosurgery, Saito Memorial Hospital; Dr Koyama and Dr Ohara, ED staff members at Kyoto Renaiss Hospital, Itoigawa General Hospital, and Toyama City Hospital.

References

1. Teasdale G, Jennet B. Assessment of coma and impaired conscious- ness. Lancet 1974;13(2):81-2.
2. Teasdale G, Jennett B. Assessment and prognosis of coma after head injury. Acta Neurochir (Wien) 1976;34(1-4):45-55.
3. Jennett B, Teasdale G. Aspects of coma after severe head injury. Lancet 1977;23(1):878-81.
4. Riechers II RG, Ramage A, Brown W, et al. Physician knowledge of the Glasgow Coma Scale. J Neurotrauma 2005;22(11):1327-34.
5. Teoh LSG, Gowardman JR, Larsen PD, et al. Glasgow Coma Scale: variation in mortality among permutations of specific total scores. Intensive Care Med 2000;26(2):157-61.
6. Ohta T. Transition of judgment on depth of consciousness disturbance and its perspectives-from the Japan Coma Scale to the Emergency Coma Scale. J Jpn Congr Neurol Emerg 2003;16:1-4.
7. Okudera H, Ohta T, Aruga T, et al. Development of an Emergency Coma Scale by the ECS task force: 2003 report. J Jpn Congr Neurol Emerg 2004;17:66-8.
8. Wakasugi M, Okudera H, Ohta T, et al. Development of the new coma scale: Emergency Coma Scale (ECS). In: Kanno T, Kato Y, editors. Minimally Invasive Neurosurgery and Multidisciplinary Neurotrau- matology. Tokyo: Springer-Verlag; 2006. p. 400-3.
9. Takahashi C, Okudera H, Wakasugi M, et al. Emergency Coma Scale- a basic concept of new coma scale for consciousness disturbance. In: European Association of Neurosurgical Societies, editor. 13th Congress of the European Association of Neurosurgical Societies. Bologna, Italy, Medimond; 2007. p. 321-5.
10. Takahashi C, Okudera H, Wakasugi M, et al. What is an excellent coma scale? Comparative study among JCS and ECS-first report. Neurosurg Emerg 2007;12:129-35.
11. Takahashi C, Okudera H, Wakasugi M, et al. A multi-center comparative study of Glasgow Coma Scale and Emergency Coma Scale-intermediate Report. In: European Association of Neurosur-

gical Societies, editor. 13th Congress of the European Association of Neurosurgical Societies. Bologna, Italy: Medimond; 2007. p. 317-20.

1. Reilly PL, Simpson DA, Sprod R. Assessing the Conscious level in infants and young children: a paediatric version of Glasgow Coma Scale. Child’s Nerv Syst 1988;4(1):30-3.
2. ISLS Course Guidebook Editorial Committee, editor. ISLS course guide book. 2nd ed. Tokyo: Herusu Shuppan; 2007.
3. Jennett B, Bond M. Assessment of outcome after severe brain damage: a practical scale. Lancet 1975;1(7905):480-4.
4. Jennet B, Snoek J, Bond M, et al. Disability after severe head injury: observations on use of the Glasgow Outcome Scale. J Neurol Neurosurg Psychiatry 1981;44(4):285-93.
5. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33(1):159-74.
6. Nagelkerke NJD. A note on a general definition of the coefficient of determination. Biometrika 1991;78:691-2.
7. Gill M, Windemuth R, Steele R, et al. A comparison of the Glasgow Coma Scale score to simplified alternative scores for the prediction of traumatic brain injury outcomes. Trauma 2005;45(1):37-42.
8. Teasdale G, Jennett B, Murray L, et al. Glasgow Coma Scale: To sum or not to sum? Lancet 1983;2(8351):678.
9. Teasdale GM, Murray L. Revisiting the Glasgow Coma Scale and Coma Score. Intensive Care Med 2000;26:157-61.
10. Haukoos JS, Gill MR, Rabon RE, et al. Validation of the simplified motor score for the prediction of brain injury outcomes after trauma. Ann Emerg Med 2007;50:18-24.
11. Ohta T. Phenomenological aspects of consciousness-its distur- bance in acute and chronic stages. Acta Neurochir [Suppl] 2005;93: 191-3.
12. Gotoh O, Tamura A, Yasui N, et al. Japan Coma Scale in prediction of outcome after early surgery for Aneurysmal subarachnoid hemorrhage. No Shinkei 1995;47:49-55.
13. Kitamura A, Nakagawa Y, Sato M, et al. Proportion of stroke subtypes among men and women >= 40 years of age in an urban Japanese city in 1992, 1997, and 2002. Stroke 2006;37(6):1374-8.