Original article 29

Comparison of the Full Outline of Unresponsiveness

Score Coma Scale and the Glasgow Coma Scale in an
emergency setting population
Cenker Eken, Mutlu Kartal, Ayse Bacanli and Oktay Eray

Background The Glasgow Coma Scale (GCS) is the outcome (Modified Rankin Scale: 3–6) was 0.720 (P = 0.001
most widely used tool for the evaluation of the level of and 95% CI: 0.650–784) for GCS and 0.751 (P = 0.0001 and
consciousness. The Full Outline of Unresponsiveness 95% CI: 0.682–0.812) for FOUR Score.
(FOUR) Score is a new coma Scale that was developed
considering the limitations of the GCS, and has been found Conclusion The new coma Scale, FOUR Score, is not
to be useful in an intensive care setting. We aimed to superior to the GCS. However, the combination of the
compare FOUR Score and GCS in the emergency setting. eye and motor components of FOUR Score is a valuable
tool that can be used instead of either the FOUR Score
Methods All patients older than 17 years who presented or GCS. European Journal of Emergency Medicine
with an altered level of consciousness, after any trauma to 16:29–36 c 2009 Wolters Kluwer Health | Lippincott
the head or with neurological complaints were included in Williams & Wilkins.
this study. Three-month mortality, in-hospital mortality, and
poor outcome using a Modified Rankin Scale (MRS) of 3–6 European Journal of Emergency Medicine 2009, 16:29–36
points were used as the primary outcome measures.
Keywords: altered level consciousness, emergency population, Full Outline
of Unresponsiveness Score, Glasgow Coma Scale
Results A total of 185 patients were included in the study.
Area under the curve (AUC) values in predicting 3-month Department of Emergency Medicine, Akdeniz University Medical Faculty,
Antalya, Turkey
mortality for GCS was 0.726 [P = 0.0001 and 95%
confidence interval (CI): 0.656–0.789] and 0.776 (P = 0.0001 Correspondence to Dr Cenker Eken, MD, Akdeniz University, Akdeniz Universitesi
and 95% CI: 0.709–834) for FOUR Score. AUC in predicting Hastanesi, Acil Tip Anabilim Dali, 07059, Antalya, Turkey
Tel: + 90 242 249 6176; e-mail: [email protected]
hospital mortality for GCS was 0.735 (P = 0.0001 and
95% CI: 0.655–0.797) and 0.788 (P = 0.0001 and 95% Received 28 November 2007 Accepted 2 April 2008
CI: 0.722–0.844) for FOUR Score. AUC in predicting poor

Introduction validity of the GCS, many alternative scoring systems

Defining the level of conscious (LOC) is a core clinical evaluating consciousness by either using additional
skill, which can be a challenge even for experienced physiological parameters or only using subunits of GCS
physicians. The Glasgow Coma Scale (GCS), defined by or different simple approaches have been developed.
Teasdale and Jennett in 1974 [1], remains the most Acute Physiology and Chronic Health Evaluation Score
commonly used scoring system for LOC. Although the [7], Mainz Emergency Evaluation Score [8], Reaction
GCS was created only for patients with head trauma, it Level Scale [9], Alert, response to Verbal stimulus,
has become widely used for patients with other causes response to Pain, Unresponsiveness Scale for trauma
of decreased LOC. patients [10], and Revised Trauma Score [11] are some
alternative scoring systems. Most of them were found to
Although it has been commonly used and reported to be be better or as good as the GCS in evaluating comatose
a reliable tool in predicting the prognosis of patients with patients using the subunits of GCS. However, none of
altered mental status, there are some limitations to its them has been entirely accepted in the place of the GCS.
use. One of these limitations is the lack of consistent
interobserver agreement while using the GCS [2–4]. Wijdicks et al. [12] developed a new coma Scale, Full
Furthermore, experienced staff tends to apply the GCS Outline of Unresponsiveness (FOUR) Score, considering
better [5] and low interobserver agreement was reported the limitations of the GCS. Verbal response is not a
in evaluating patients with moderate Scores [6]. component of the FOUR Score. Brainstem reflexes and
respiratory pattern, which can be used in the evaluation
Another limitation for the GCS is that intubated patients of patients with decreased LOC and not used in GCS,
cannot manifest a verbal response. Withdrawal from pain, were integrated into the FOUR Score. Withdrawal from
which is easily misinterpreted as a flexion response to pain, which is a parameter for the motor component of
pain, is another confounder for GCS. To enhance the the GCS, was also removed in the FOUR Score. Widjicks
0969-9546 c 2009 Wolters Kluwer Health | Lippincott Williams & Wilkins DOI: 10.1097/MEJ.0b013e32830346ab

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
 30 European Journal of Emergency Medicine 2009, Vol 16 No 1

et al. studied the FOUR Score in patients admitted in the plaints, GCS and FOUR Scores, 3-month mortality,
intensive care unit for nontraumatic medical conditions. hospital mortality, and MRS. The final diagnoses of
They suggested the FOUR Score to be superior to GCS patients were also recorded in the enrollment form.
owing to the availability of brainstem reflexes, breathing However, final diagnoses of patients were displayed in the
patterns, and the ability to recognize different stages of manuscript after classifying them as:
herniation. They also found a higher probability of in- 1. Traumatic head injury: subarachnoid hemorrhage,
hospital mortality using the lowest FOUR Score than the subdural hemorrhage, intraventricular hemorrhage,
lowest GCS. skull fracture, and cerebral edema.
2. Stroke: ischemic or hemorrhagic.
We evaluated the hypotheses that the area under the 3. Nontraumatic intracranial hemorrhage: aneurismal
curve (AUC) for the FOUR Score would significantly subarachnoid hemorrhage, subdural hemorrhage, and
exceed the AUC for GCS Scores for the unfavorable parenchymal or ventricular hemorrhage.
outcomes, in-hospital mortality, 3-month mortality, and a 4. Central nervous system infections: meningitis and
low (3–6) Modified Rankin Scale (MRS). encephalitis.
5. Metabolic encephalopathy: electrolyte disturbances,
hepatic or hypoxic encephalopathy, and hypoglycemia.
Materials and methods 6. And the other diagnoses like brain tumor.
Study setting and design
This prospective observational study was performed in a Description of the new coma Scale, FOUR Score,
tertiary care hospital emergency department (ED), which and GCS
is also a level IV trauma center, with an annual census The FOUR Score differs from the GCS. The FOUR
of 55 000 from May to April in 2006. This study was Score consists of components that Score eye, motor,
approved by the local ethics committee. brainstem reflex, and respiration patterns of the patients.
The FOUR Score and the GCS are detailed in Table 1.
Selection of participants
All patients older than 17 years who presented with an Outcome measures
altered level of consciousness, after any trauma to the We selected three primary outcome measures; 3-month
head (whether or not LOC was altered), neurological mortality, hospital mortality, and 3-month morbidity using
complaints of lateralizing motor, and/or sensory deficits, an MRS [13]. The MRS can be defined briefly as; 0 = no
dysarthria, dysphasia, or facial asymmetry were eligible
for inclusion in this study. A different study population
Table 1 Definition of FOUR score and Glasgow Coma Scale
from the former studies comparing the coma Scales was
selected. The study population of the former studies only FOUR Score Glasgow Coma Scale
consisted of patients with altered mental status. How- Eye response Eye response
ever, to calculate sensitivity, specificity, and a cut-off 4 = eyelids open or opened, tracking, 4 = open spontaneously
or blinking to command
value for two Scales, we preferred to enroll alert patients 3 = eyelids open, but not tracking 3 = eye opening to verbal command
who had the possibility of developing future adverse 2 = eyelids closed, but open to loud 2 = eye opening to pain
events such as neurological deficit or death. Therefore, all voice
1 = eyelids closed, but open to pain 1 = no eye opening
alert patients with head trauma and neurological deficit 0 = eyelids remain closed with pain
were included in the study. Patients were excluded from Motor response Motor response
4 = thumbs-up, fist, or peace sign 6 = obeys commands
the study if they were intubated or if they had been 3 = localizing to pain 5 = localizing to pain
administered sedative or paralytic agents before their 2 = flexion response to pain 4 = withdrawal from pain
presentation to the ED. 1 = extension response to pain 3 = flexion response to pain
0 = no response to pain or 2 = extension response to pain
generalized myoclonus status 1 = no motor response
Data collection and processing Brainstem reflexes Verbal response
4 = pupil and corneal reflexes present 5 = oriented
Before starting the study, all attending physicians and 3 = one pupil wide and fixed 4 = confused
residents in a 4-year residency program were trained on 2 = pupil or corneal reflexes absent 3 = inappropriate words
the new coma Scale, FOUR Score, and also on the GCS 1 = pupil and corneal reflexes absent 2 = incomprehensible sounds
0 = absent pupil, corneal, and cough 1 = no verbal response
to standardize physicians’ definitions. Written definitions reflex
of two Scales were included in the study form and Respiration
4 = not intubated, regular breathing
illustrative definitions of the FOUR Score were located in pattern
various areas of the department. Patients were deemed 3 = not intubated, Cheyne–Stokes
eligible for the study by attendees and residents who breathing pattern
2 = not intubated, irregular breathing
filled out an enrollment form. The FOUR Score and GCS 1 = breathes above ventilator rate
of the patients were calculated before any invasive or 0 = breathes at ventilator rate or
medical procedure. The enrollment form included apnea

demographic features of the patients, presenting com- FOUR, Full Outline of Unresponsiveness.

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
 FOUR Score and Glasgow Coma Scale Eken et al. 31

symptoms at all; 1 = no significant disability despite Table 2 Patient characteristics

symptoms, able to carry out all usual duties and activities; Variable No
2 = slight disability, unable to carry out some previous
Total number of patients 185
activities, but able to look after own affairs without Median age (range) 59 (18–97)
assistance; 3 = moderate disability, requiring some help, Sex
Male (%) 119 (64.3)
but is able to walk without assistance; 4 = moderately Female (%) 66 (35.7)
severe disability, unable to walk without assistance and Diagnosis (%)
unable to attend to own bodily needs without assistance; Stroke 66 (35.7)
Minor head trauma 43 (23.2)
5 = severe disability, bedridden, incontinent, and requir- Traumatic head injury 23 (12.4)
ing constant nursing and attention; and 6 = dead. Nontraumatic intracranial hemorrhage 16 (8.6)
Metabolic encephalopathy 11 (5.9)
Brain tumor 9 (4.9)
Primary data analysis Seizure 4 (2.2)
CNS infection 3 (1.6)
Data were analyzed by SPSS 15.0 for Windows (SPSS Nontraumatic brain edema 2 (1.1)
Inc., Chicago, Illinois, USA)and receiving operating Sepsis 2 (1.1)
characteristic (ROC) curve analyses were performed by Pneumonia 2 (1.1)
Traumatic intra-abdominal hemorrhage 1 (0.5)
Med Calc Version 9.2.0.1 (Frank Schoonjans, Mariakerke, Gastrointestinal system hemorrhage 1 (0.5)
Belgium). Continuous variables were expressed by Hypoxic encephalopathy 1 (0.5)
Altered mental status with unknown etiology 1 (0.5)
mean ± standard deviation, ordinal variables as median
and minimum–maximum, and frequent variables as rates. CNS, central nervous system.
Correlation between MRS and coma Scales was deter-
mined by Spearman’s correlation coefficient. Predictive
value of GCS and FOUR Score in predicting primary
outcome measures was established by ROC curve by
ROC curve was performed to predict the 3-month
calculating AUC values by 95% confidence intervals (CI).
mortality, in-hospital mortality, and poor outcome for
A binary logistic regression analysis was performed to
MRS (3–6). AUC values in predicting 3-month mortality
reveal the adjusted odds ratios of GCS and FOUR Scores
for GCS was 0.726 (P = 0.0001 and 95% CI: 0.656–0.789)
in predicting the primary outcome measures. This study
and 0.776 (P = 0.0001 and 95% CI: 0.709–834) for FOUR
had a 95% power of predicting the 3-month mortality,
Score. The cut-off values for GCS in predicting 3-month
which failed for a 3-month low (3–6) MRS; 288 patients
mortality was 5 (positive likelihood ratio = 11.7) and 9 for
were needed to detect the difference between FOUR
the FOUR Score (positive likelihood ratio = 11.7). AUC
Score and GCS in predicting the 3-month low (3–6) MRS
in predicting in-hospital mortality for GCS was 0.735
with an 80% power. All hypotheses were constructed two-
(P = 0.0001 and 95% CI: 0.655–0.797) and 0.788
tailed and P r 0.05 was considered significant.
(P = 0.0001 and 95% CI: 0.722–0.844) for FOUR Score.
AUC in predicting poor outcome (MRS: 3–6) was 0.720
(P = 0.001 and 95% CI: 0.650–784) for GCS and 0.751
Results (P = 0.0001 and 95% CI: 0.682–0.812) for FOUR Score.
A total of 188 patients were included in the study. Three However, the comparison of AUC values of total FOUR
of the eligible patients were excluded from the study Score and GCS lack statistical significance as the
because of not being able to be reached for telephone overlapping 95% CIs demonstrate (Table 3). ROC curves
follow-up at the end of the 3 months. The remaining 185 depicting the comparisons of FOUR Score and GCS in
patients composed the study population. The median age predicting primary outcome measures are displayed in
of the study patients was 59 years (range: 18–97 years) Figs 1–3. Figures 4 and 5 also display the frequency of
and 64.3% (n = 119) were men. Characteristics of the FOUR Score and GCS according to the patients’ assigned
study patients and the ultimate diagnosis are displayed Scores.
in Table 2. Five patients had blood alcohol levels over
10 mg/dl, the upper limit of the blood alcohol level, and The subunit analyses of FOUR Score and GCS revealed
three patients had hypoglycemia. that the eye and motor components of FOUR Score
had AUC values higher than GCS and lower than the
Forty-seven (25.4%) patients died at the end of the complete FOUR Score, but these differences lack
3 months and 55% (n = 26) of them died in the hospital. statistical significance. Furthermore, brainstem reflexes
Seventy-two patients (38.8%) had poor outcomes accord- and respiratory pattern had significantly lower AUC
ing to the MRS (MRS = 3–6). values than eye and motor components of the FOUR
Score (Table 3).
A moderate correlation was observed between the MRS
at the end of the 3 months both for GCS (r = – 0.430; To determine the predictive value of coma Scales in
P = 0.000) and FOUR Scores (r = – 0.483; P = 0.000). traumatic and nontraumatic patients, ROC was reperformed

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
 32 European Journal of Emergency Medicine 2009, Vol 16 No 1

Table 3 Receiving operating characteristic curve analyses in predicting 3-month mortality, hospital mortality, and poor outcome for GCS,
FOUR score, and subunits of two scales
Variable Three-month mortality AUC (95% CI) Hospital mortality AUC (95% CI) Rankin Score 3–6 AUC (95% CI)

Total FOUR score 0.776 (0.709–0.834) 0.788 (0.722–0.844) 0.751 (0.682–0.812)

FOUR score
Eye 0.751 (0.682–0.812) 0.743 (0.674–0.805) 0.711 (0.640–0.775)
Motor 0.737 (0.667–0.799) 0.770 (0.703–0.829) 0.707 (0.635–0.771)
Brainstem reflexes 0.586 (0.512–0.658) 0.598 (0.523–0.669) 0.562 (0.487–0.634)
Respiration 0.572 (0.497–0.644) 0.585 (0.510–0.657) 0.567 (0.492–0.639)
Eye and motor responsesa 0.773 (0.706–0.831) 0.786 (0.720–0.843) 0.733 (0.633–0.795)
Total GCS 0.726 (0.656–0.789) 0.735 (0.655–0.797) 0.720 (0.650–0.784)
GCS
Eye 0.646 (0.573–0.715) 0.631 (0.557–0.701) 0.620 (0.546–0.690)
Motor 0.679 (0.606–0.745) 0.662 (0.589–0.730) 0.651 (0.578–0.720)
Verbal 0.701 (0.629–0.766) 0.705 (0.634–0.770) 0.720 (0.649–0.783)

AUC, area under the curve; CI, confidence interval; FOUR, Full Outline of Unresponsiveness; GCS, Glasgow Coma Scale.
a
Eye and motor responses of FOUR score combined because of having individually high AUC values whether a more simple approach could be established.

Fig. 1 Fig. 2

100 100

80 80

60 60
Sensitivity
Sensitivity

40 40

20 20 Total FOUR Score

Total FOUR Score
Total GCS Total GCS

0 0
0 20 40 60 80 100 0 20 40 60 80 100

100-Specificity 100-Specificity

Receiving operating characteristic curves comparing Full Outline of Receiving operating characteristic curves comparing Full Outline of
Unresponsiveness (FOUR) score and Glasgow Coma Scale (GCS) Unresponsiveness (FOUR) score and Glasgow Coma Scale (GCS)
in predicting 3-month mortality. in predicting hospital mortality.

using MRS as the dependent variable. We did not A logistic regression analysis including age, sex, systolic
use mortality as the dependent variable because of and diastolic blood pressures, respiration rate, alcohol,
insufficient mortality in the trauma group. hypoglycemia, and trauma revealed both the FOUR Score
and GCS as independent variables in predicting 3-month
AUC for GCS in trauma patients was 0.776 (P = 0.0045 mortality. The adjusted odds ratios for FOUR Score and
and 95% CI: 0.657–0.869) and 0.786 (P = 0.0024 and 95% GCS were 0.638 (95% CI: 0.511–0.796; P = 0.000) and
CI: 0.668–0.876) for the FOUR Score. In nontraumatic 0.796 (95% CI: 0.695–0.911; P = 0.001), respectively.
patients, AUC for the GCS was 0.655 (P = 0.0026 and
95% CI: 0.562–0.740) and 0.713 (P = 0.0001 and 95% CI:
0.623–0.793) for the FOUR Score. No statistical differ- Discussion
ence was observed between the AUC values of total The FOUR Score has been developed to overcome
FOUR Score and GCS. The AUC values with 95% CI are certain limitations of the GCS, imposed by the need to
displayed in Table 4. characterize a verbal response. Furthermore, the authors

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
 FOUR Score and Glasgow Coma Scale Eken et al. 33

Fig. 3 Fig. 5

100 100

80
80

Frequency
60
60
Sensitivity

40

40
20

20 Total FOUR Score 0

Total GCS 2.50 5.00 7.50 10.00 12.50 15.00
GCS Score
0
Frequency of Glasgow Coma Scale (GCS) (range: 3–15) among study
0 20 40 60 80 100 population.
100-Specificity

Receiving operating characteristic curves comparing Full Outline of

Unresponsiveness (FOUR) score and Glasgow Coma Scale (GCS) in Table 4 Receiving operating characteristic curve in predicting poor
predicting poor outcome (Modified Rankin scale: 3–6). outcome in patients with and without trauma
Patients without trauma Patients with trauma
Variable MRS from 3 to 6 MRS from 3 to 6

Total FOUR score 0.713 (0.623–0.793) 0.786 (0.668–0.876)

FOUR score
Fig. 4 Eye 0.674 (0.582–0.758) 0.776 (0.657–0.869)
Motor 0.652 (0.559–0.738) 0.810 (0.695–0.895)
Brainstem reflexes 0.565 (0.471–0.656) 0.508 (0.383–0.632)
100 Respiration 0.561 (0.467–0.653) 0.611 (0.484–0.728)
Total GCS 0.655 (0.562–0.740) 0.776 (0.657–0.869)
GCS
Eye 0.594 (0.500–0.684) 0.679 (0.553–0.787)
80
Motor 0.597 (0.503–0.686) 0.706 (0.582–0.811)
Verbal 0.646 (0.553–0.732) 0.782 (0.664–0.873)
Frequency

60 FOUR, Full Outline of Unresponsiveness; GCS, Glasgow Coma Scale; MRS,

modified Rankin Scale.

40
indicates that brainstem and respiratory components of
the FOUR Score had much lower AUC numerical values
20
than did the eye and motor components of the FOUR
Score. This demonstrates that brainstem reflexes and
0 respiratory pattern do not provide the expected benefit
0.00 5.00 10.00 15.00 to using the eye and motor components; contrary to
FOUR Score what the authors of the FOUR Score anticipated. The
developers of the FOUR Score also did not state the AUC
Frequency of Full Outline of Unresponsiveness (FOUR) score (range: values of brainstem reflexes and respiratory pattern. They
0–16) among study population.
only found a significant odds ratio for the brainstem
component, but not for respiratory pattern in predicting
in-hospital mortality when adjusted for confounding
who developed the FOUR Score anticipated that it would variables in the logistic regression analysis.
be more easily remembered than the GCS, because the
FOUR Score assigns all Scores within the same 0–4 range, Although the AUC values of the FOUR Score were
unlike the GCS. Also, the FOUR Score adds the slightly better than the GCS in predicting all primary
evaluation of brainstem reflexes and respiratory patterns outcome measures not only for the whole study popula-
rather than a verbal response Score. However, this study tion, but also for nontraumatic and traumatic patient

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
 34 European Journal of Emergency Medicine 2009, Vol 16 No 1

subgroups, according to the results of this study, this expected to result with an unfavorable outcome. In this
difference was not statistically significant, which can instance, the level of altered consciousness may naturally
be seen by analyzing the overlapped 95% CIs in Table 3. be expected to be related to the poor outcome. Hence,
The critical question here is whether the FOUR Score including only the patients with altered LOC, as
replaces the GCS in clinical practice. As previously practiced in the previous studies, may cause referral bias.
mentioned, the coma Scales that were developed instead Studies investigating the value of simpler coma Scales
of GCS have not been accepted commonly in the clinical compared with the GCS that have reported close
practice of emergency medicine. Some suggest that the predictive values to the GCS may support the conclusion
Alert Verbal Painful Unresponsive Scale as an alternative, that complex coma Scores are not necessarily better. The
although this was not found to be as valuable as GCS. findings of this study on brainstem reflexes and
The findings of this study show the total FOUR Score respiratory pattern in predicting unfavorable outcomes
cannot replace the GCS Score although, interobserver also support the results stated in the previous literature.
reliability and the ease-of-use of the FOUR Score in EDs However, a dilemma has developed after these studies
are still unknown. were performed because of the simple methods, as
aforementioned, have not replaced the GCS. This
dilemma is still a challenge in the development of new
An interesting finding in this study was that the eye
coma Scales other than the GCS. As aforementioned, eye
and motor components of FOUR Score had higher AUC
and motor components of FOUR Scores either alone or
values than the total GCS, and these components were
in combination demonstrated high AUC values, which
also close to the total FOUR Score although, these
were not statistically different from total FOUR Score
differences lacked statistical significance. Studies com-
and GCS. The motor subunit of the FOUR Score does
paring the predictive values of components of the GCS
not include the component ‘withdrawal from pain’. The
with the total GCS showed insignificant differences
eye component classifies a patient with open eyes
according to the AUC values in the ROC analysis in
according to the ability to track. These findings may be
predicting unfavorable outcomes [14,15]. As this study
attributed to the changes in the eye and motor
demonstrated high predictive values of the eye and motor
components of FOUR Score.
components of FOUR Score, an additional ROC analysis
was performed by combining these two components to
determine whether the predictive value would be One of the aims of the authors in developing the FOUR
enhanced. This process slightly augmented the AUC Score was to eliminate the limitations caused by the
value, but did not constitute a statistically significant evaluation of verbal response in intubated patients.
difference than the total FOUR Score. Actually, these Verbal response does not help the evaluation of the eye
findings are in concordance with previous studies and and motor components. It can also be predicted from the
should be interpreted as a summary of the literature. eye and motor Scores of GCS as reported by Rutledege
Even though the predictive values of the subunits of et al. [20]. A recent study by Marmarou et al. [21], which
the GCS are close to the GCS, combining them does combined the results of multiple studies with a total
not increase their predictive values as expected. Never- number of 8721 patients, reported the entire eye, verbal,
theless, the close AUC value of the combined eye and and motor components of GCS and one or two unreactive
motor responses, which are easier to remember than both pupils individually were all strongly associated with
the entire GCS and FOUR Score, may be the most useful unfavorable outcomes. In this study, the advantage of
predictor of 3-month mortality, in-hospital mortality, and the FOUR Score over the GCS seems to arise from the
a final Rankin Score of 3–6. changes in eye and motor components more than
excluding the verbal response.
As the GCS is not only used for traumatic brain injuries,
but also for all patients with altered conscious, many The developers of the FOUR Score also aimed to
researchers have developed coma Scales by adding construct a more practical Scale than the GCS. They
physiological parameters to the GCS to take into assigned the components of FOUR Score from 0 to 4.
consideration their affects on the prognosis of comatose The interrater reliability of the FOUR Score and GCS
patients. Acute Physiology and chronic health evaluation among nurses, residents, and neurointensivists was also
and the Revised Trauma score are two of those that are investigated in their study. They reported an overall k
widely known. However, studies comparing these Scores value of 0.82 both for total GCS and total FOUR Score.
with the GCS have not shown the expected enhance- The interrater reliability among nurses was found to be
ment when compared with the predictive value of GCS lower than that among residents and neurointensivists.
both in patients with and without trauma [16–19]. These Although the previous studies stated that experienced
disappointing results regarding the physiological para- nurses have been found to have higher agreement rates
meters may be interpreted as physiological abnormalities than inexperienced ones in scoring GCS [5], a close
that are serious to be able to cause altered LOC and are agreement between emergency physicians and parame-

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
 FOUR Score and Glasgow Coma Scale Eken et al. 35

dics was also stated for GCS by Menegazzi et al. [6], GCS Score is 15, which could be a limitation of this study.
k values were 0.66 for emergency physicians and 0.63 Another limitation to this study was the low mortality
for paramedics. And a recent study by Wolf et al. [22] rate in trauma patients and that only MRS was used as
reported an excellent agreement both for FOUR Score the outcome measure for the subgroup analysis. Studies
(weighted k: eye, 0.84; motor, 0.73; brainstem, 0.89; and with bigger sample sizes are needed for more accurate
respiration, 0.92) and GCS (weighted k: eye, 0.85; motor, subgroup analyses. The number of potentially eligible
0.74; and verbal, 0.89) among intensive care nurses, patients who were not enrolled in this study is not known.
unlike data from Widjicks et al. The high agreement rates
stated in the study by Widjicks et al. for the total GCS and The interrater reliability was also not investigated in
its verbal components (weighted k: total GCS, 0.82 and this study. Further studies are needed to determine the
verbal, 0.88) were not similar to the studies investigating interrater reliability of FOUR Score among emergency
the interrater agreement of GCS among emergency physicians.
physicians. Gill et al. [2] reported lower agreement rates
for total GCS and verbal response than motor and eye Conclusion
components among emergency physicians (k: eye, 0.57; The new coma Scale, FOUR Score, developed to
motor, 0.54; verbal, 0,44; and total GCS, 0.32). In another overcome the limitations of GCS, had slightly higher
study by Gill et al. [3], verbal components and total GCS AUC values than the GCS in predicting 3-month
were found to have lower weighted k values than motor mortality, in-hospital mortality, and low (3–6) MRS, but
and eye components, despite intubated patients which this lacked statistical and clinical significance. However,
were assigned with the same Scores. The interrater the combination of the eye and motor components of
agreement of the FOUR Score among emergency the FOUR Score should be a valuable tool instead of
physcians was not investigated in this study. This should either the FOUR Score or the GCS, which is also easy to
be investigated in future studies. remember. Brainstem reflexes and respiration pattern do
not provide additional benefits in predicting the unfavor-
The cut-off values in this study were 5 for the GCS and able outcomes as expected. The interobserver reliability
9 for the FOUR Score in predicting mortality, but if the of FOUR Score among emergency physicians and
hospital mortality was used as the outcome measure, both paramedics should also be a studied further.
the cut-off values for the GCS and FOUR Scores were
found to be 4. However, Wijdikcs et al. reported cut-off Acknowledgement
values of 7 for GCS and 9 for FOUR Score in predicting This study was supported by Akdeniz University
hospital mortality. These differences may be as a result of Foundation and great thanks to Dr Feras Khan for his
the statistical techniques used in determining the cut-off contributions to the syntax of the article.
values and the differences between study populations.
Wijdicks et al. used the maximum sum of sensitivity and References
specificity, which is similar to Youden Index (sensitivi- 1 Teasdale G, Jennet B. Assessment of coma and impaired consciousness.
A practical scale. Lancet 1974; 2:81–84.
ty + specificity – 1). However, we used the positive 2 Gill M, Marterns K, Lynch EL, Salih A, Green SM. Interrater reliability of 3
likelihood ratio to determine the cut-off values. Cho simplified Neurologic Scales Applied to Adults presenting to the ED with
and Wang [17] stated a cut-off value of 5 for GCS in altered levels of conciousness. Ann Emerg Med 2007; 49:403–407.
3 Gill MR, Reiley DG, Green SM. Interrater reliability of Glasgow Coma Scale
predicting hospital mortality. scores in the emergency department. Ann Emerg Med 2004;43:215–223.
4 Liddy CM, Claudio MM, Teresa ML, Sibbald WJ. Interobserver variability in
data collection of the APACHE II score in teaching and community hospitals.
Limitations Crit Care Med 1999; 27:1999–2004.
The population of this study has a composition different 5 Rowley G, Fielding K. Reliability and accuracy of the Glasgow Coma Scale
with experienced and inexperienced users. Lancet 1991; 337:535–538.
from the former studies. We also enrolled the patients 6 Menegazzi JJ, Davis EA, Sucov AN, Paris PM. Reliability of the Glasgow
with neurological deficits who did not have an altered Coma Scale when used by emergency physicians and paramedics. J Trauma
level of consciousness. We thought that all of these 1993; 34:46–48.
7 Knaus WA, Zimmerman JE, Wagner DP, Draper EA, Lawrence DE.
patients would not have a GCS of 15 and FOUR Score APACHE-acute physiology and chronic health evaluation: a physiologically
of 16 and they all had high risks of mortality and based classification system. Crit Care Med 1981; 9:591–597.
neurological disability, which was evaluated with MRS. 8 Hennes HJ, Reinnhardt T, Dick W. The Mainz Emergency Evaluation Scoring
for assessment of emergency patients. Emerg Med 1992; 18:130–136.
We also included the patients with head trauma without 9 Starmark JE, Stalhammar D, Holmgren E. The reaction level scale (RLS85).
altered mental status. GCS has been used in the clinical Manual and guidelines. Acta Neurochir 1988; 91:12–20.
practice not only for patients with a Score of under 15, 10 American College of Surgeons Committee on Truama. Advanced life
support course for physicians. Chicago IL: American College of Surgeons;
but also for patients without any neurological deficits. 1993.
The level of mental status of these patients is usually 11 Champion HR, Sacco WJ, Copes WS, Gann DS, Gennarelli TA, Flanagan
defined as having a GCS of 15, particularly for patients ME. A revision of the trauma score. J Trauma 1989; 29:623–629.
12 Wijdicks EFM, Bamlet WR, Maramattom BV, Manno EM, McClelland RL.
presenting with trauma to the ED. These patients also Validaiton of new coma scale: the FOUR Score. Ann Neurol 2005;
have a risk of mortality and morbidity even though their 58:585–593.

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
 36 European Journal of Emergency Medicine 2009, Vol 16 No 1

13 van Swieten JC, Koudstaal PJ, Visser MC, Schouten HJ, van Gijn J. 18 Cho DY, Wang YC, Lee MJ. Comparison of APACHE III, II and Glasgow
Interobserver agreement for the assessment of handicap in stroke patients. coma scale for prediction of mortality in a neurosurgical intensive care unit.
Stroke 1988; 19:604–607. Clin Intensive Care 1995; 6:9–14.
14 Michelle Gill, Ryan Windemuth, Robert Steele, Steven M. Green. 19 Hartley C, Cozens A, Mendelow AD, Stevenson JC. The Apache II scoring
A comparison of the Glasgow Coma Scale Score to simplified alternative system in neurosurgical patients: a comparison with simple Glasgow coma
scores for the prediction of traumatic brain injury outcomes. Ann EmergMed scoring. Br J Neurosurg 1995; 9:179–187.
2005; 45:37–42. 20 Rutledege R, Lentz C, Fakhry S, Hunt J. Appropriate use of the Glasgow
15 Davis DP, Serrano JA, Vilke GM, Sise MJ, Kennedy F, Eastman AB, et al. The Coma Scale in Intubated Patients: a linear regression prediction of the
predictive value of field versus arrival Glasgow coma scale score and TRISS Glasgow Verbal Score from the Glasgow Eye and Motor Scores. J Trauma
calculations in moderate to severe traumatic brain injury. J Trauma 2006; Inj Infect Crit Care 1996; 41:514–522.
60:985–990. 21 Marmarou A, Lu J, Butcher I, McHugh GS, Murray GD, Steyerberg I, et al.
16 S̆tefek Grmec, Vladimir Gas̆parovic. Comparison of APACHE II, MEES and Prognostic value of the Glasgow Coma Scale and pupil reactivity in
Glasgow Coma Scale in patients with nontraumatic coma for prediction of traumatic brain injury assessed pre-hospital and on enrollment: an impact
mortality. Crit Care 2001; 5:19–23. analysis. J Neurotrauma 2007; 24:270–280.
17 Cho DY, Wang YC. Comparasion of the APACHE III, APACHE II and 22 Wolf CA, Wijdicks EFM, Bamlet WR, McClelland RL. Further validation of
Glasgow Coma Scale in acute head injury for prediction of mortality and the FOUR score coma scale by intensive care nurses. Mayo Clin Proc
functional outcome. Intensive Care Med 1997; 23:77–84. 2007; 82:435–438.

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.