Gujjar 2013

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Journal of Critical Care (2013) 28, 316.e1–316.

e8

Full Outline of UnResponsiveness score and Glasgow Coma


Scale in medical patients with altered sensorium: Interrater
reliability and relation to outcome
Arunodaya R. Gujjar MBBS, DM a,⁎, Poovathru C. Jacob MD a ,
R. Nandhagopal MBBS, DM a , S.S. Ganguly PhD b , Ammar Obaidy MD a ,
Abdullah R. Al-Asmi MRCP(C) a
a
Department of Medicine, College of Medicine and Health Sciences, Sultan Qaboos University, Muscat, Oman
b
Department of Family Medicine and Public Health, College of Medicine and Health Sciences, Sultan Qaboos University,
Muscat, Oman

Keywords:
Abstract
Coma;
Purpose: Full Outline of UnResponsiveness, or FOUR score (FS), is a recently described scoring
Coma score;
system for evaluation of altered sensorium. This study examined interrater reliability for FS and
Full Outline of
Glasgow Coma Scale (GCS) among medical patients with altered mental status and compared outcome
Unresponsiveness;
predictability of GCS, FS, and Sequential Organ Failure Assessment score.
FOUR score;
Patients and Methods: Adult patients with altered mental status due to medical causes were rated by
Glasgow Coma Scale;
neurology consultants and internal medicine residents on FS and GCS. Interobserver reliability for GCS
SOFA score
and FS was assessed using κ score. Relation with outcomes was explored using univariate and
multivariate analyses.
Main Results: Of the 100 patients (age, 62 ± 17 years), 60 had neurologic conditions; 26, metabolic
encephalopathy; 9, infections; and 7, others. Thirty-nine patients died at 3 months. κ Scores ranged
from 0.71 to 0.85 for GCS and from 0.71 to 0.95 for FS. On multivariate analysis, GCS was predictive
of outcome at 3 months; FS was predictive of mortality. Area under the receiver operating characteristic
curves suggested equivalent performance of both scoring systems.
Conclusions: Interrater reliability and outcome predictability for FS were comparable with those for
GCS. This study supports the use of FS for evaluation of altered mental status in the medical wards.
© 2013 Elsevier Inc. All rights reserved.

1. Introduction

⁎ Corresponding author. Department of Medicine (Neurology), College


Assessment of level of consciousness is an important part
of Medicine, Sultan Qaboos University, PO Box 35, PC 123, Muscat, of clinical neurologic assessment. Glasgow Coma Scale
Oman. Fax: +968 24121198. (GCS) has been the most widely used scale for assessment of
E-mail address: grarunoday@hotmail.com (A.R. Gujjar). sensorium [1]. Earlier studies have shown reasonable

0883-9441/$ – see front matter © 2013 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcrc.2012.06.009
316.e2 A.R. Gujjar et al.

interobserver reliability of GCS and its predictive value, emergency physicians, and medical intensive care physicians
particularly in head injury and other neurologic disorders [2]. and observed good interrater reliability among these groups
Despite its widespread use, certain limitations of GCS are on nonneurology physicians and nurses [10-12]. They
known. The verbal component of GCS cannot be reliably conclude that FS can easily be used in the general medical
assessed in patients with aphasia and those with tracheal ICU and the emergency department with good reliability.
intubation. Glasgow Coma Scale has been observed to be FOUR score has been hitherto validated only in few centers
difficult to use by untrained staff such as nurses and other than Mayo hospitals, in the settings of ICU [13,14],
paramedical staff [3]. One study reported difficulty in emergency department [15], or neurosurgical patients [16],
interpretation of intermediate scores by emergency physi- with similar favorable observations. Recent studies have also
cians [4]. Attempts to modify GCS and compare it with other demonstrated the correlation of FS with outcome in patients
similar scores have been reported [5]. A study by Diringer with traumatic brain injury and cardiac arrest [17,18].
and Edwards [6] compared GCS with Innsbruck Coma Scale We aimed to compare FS with 2 other scoring systems in
(ICS) and observed that, among 84 patients with various use: GCS and SOFA score. The study compared (a) interrater
acute neurologic disorders, both scales were best at reliability between neurology consultants and internal
predicting complete improvement (GCS and ICS, 71% medicine residents in rating FS and GCS among patients
correct) and mortality (GCS, 60% correct; ICS, 56% correct). with altered mental status due to medical conditions and (b)
After deleting verbal score (which had poor discrimination) the 3 scoring systems in their efficacy in predicting outcomes
from GCS, only a modest improvement was noted in its at hospital discharge and 3 months.
prediction of outcome. They opined that combining clinical
scales with other demographic, physiological, functional,
and radiographic data would be needed to achieve useful
predictions of functional outcome. Certain scoring systems 2. Patients and methods
such as Acute Physiology and Chronic Health Evaluation
(APACHE), Multiple Organ Dysfunction Score, Sequential This prospective, observational study, approved by the
Organ Failure Assessment (SOFA) score, and others use ethical and scientific review boards, was conducted for 10
several clinical and laboratory parameters; however, these months (January to October 2010). Adult patients 15 years or
are rather unwieldy for bed-side use and are limited mostly to older admitted to the medical wards or medical high-
the intensive care units (ICUs) [7]. dependency unit with altered mental status due to any
On the above background, Wijdicks et al [8,9] developed medical cause were recruited into the study, after consent
the Full Outline of UnResponsiveness (FOUR) score (FS) from a family caregiver. Patients with head injury,
and compared its performance with GCS. The FS includes 4 postsurgery, or current hemodynamic instability (blood
components of neurologic evaluation—eye response, motor pressure [BP] b 80 mm Hg systolic) were excluded. Also
response, brainstem reflexes, and respiration—each with a excluded were patients in preterminal state and those who
scale of 0 to 4. The maximum score possible is 16 and were sedated/paralyzed for intubation. Demographic data,
minimum, 0. This scoring system does not include verbal cause and duration of altered sensorium, vital signs, and
response. The authors observed that FS is easily assessed diagnosis were documented for each patient. The following
even in patients who are sedated, intubated, and ventilated; it were documented nearest to the initial assessment of FS on
is able to recognize patients in locked-in syndrome (only day 1 of admission: plasma creatinine, bilirubin, BP (mean
vertical eye movements present), possible vegetative state arterial pressure), use of vaspopressors in case of hypoten-
(eyes open but not tracking), and possible brain death. sion, platelet count, oxygen saturation (by pulse oximetry),
In a prospective study of 120 patients in the neurologic arterial blood gas (among patients with desaturation), and
ICU, interrater reliability was excellent for FS (overall κ fractional inspired oxygen necessary to maintain normal
statistic = 0.82; 95% confidence interval [CI], 0.77-0.88) and oxygen saturation. These clinical and laboratory parameters
for GCS (κ statistic = 0.82; 95% CI, 0.76-0.87). Patients with were used to compute SOFA score [7]. Outcome status at
the lowest GCS score could be further distinguished using hospital discharge and 3 months were documented using
the FS. FOUR score also provided greater neurologic detail modified Rankin Score (scale of 0-6; 0, no symptoms; 6,
than did the GCS, recognizing locked-in syndrome and died). The outcomes at 3 months were obtained either by
possible vegetative state. Wijdicks et al [8] also compared direct patient evaluation or by a telephonic interview with the
the 2 scores for their ability to predict outcome. For FS, every patient's caregiver/family member using a questionnaire.
1-point increase in total score was associated with improved All participants in the study (4 neurology consultants and 4
in-hospital mortality by 20% (odds ratio [OR], 0.80; 95% CI, medicine residents/Senior House Officer (SHOs) posted to
0.72-0.88), as compared with 26% for GCS (OR, 0.74; 95% neurology unit) were briefed at the beginning of the study in
CI, 0.65-0.85). The probability of in-hospital mortality was the use of FS and GCS, with the help of printed tables and
higher for the lowest FS when compared with the lowest total PowerPoint slides. Opportunities were provided to practice the
GCS score. Other studies have subsequently reported on the evaluation in patients before beginning of the study. Both FS
validity of their scoring system by the intensive care nurses, and GCS were assessed for each patient by paired investigators
Four score and GCS in patients with altered sensorium 316.e3

(1 neurology consultant and a resident/SHO) within 1 hour of Table 1 Demographic profile, causes of altered mental status,
each other (ensuring that sensorium had not changed between and outcome among patients admitted with medical conditions
assessments). Serial assessment with FS and GCS were (n = 100)
performed for each patient only by a consultant on 2 further Mean age 62 ± 17 y
consecutive days, at least 24 hours apart. After the 3 M/F 58:42
assessments, patients were managed either in the same wards Median hospital stay 13 d
or in the ICU in case of worsening, as dictated by their needs. Mean (± SD) stay 26 ± 17 d
Cause of altered sensorium ⁎ n
2.1. Statistical analysis Neurologic conditions 58
Ischemic stroke 40
Intracerebral hemorrhage 4
Descriptive analyses for continuous variables were used to Seizures/Epilepsy 4
calculate mean values and SDs, whereas frequencies were Hypoxic ischemic brain injury 2
expressed as percentages. Interobserver reliability between Other encephalopathy 8
consultant scores vs those by the resident were assessed using Metabolic disorders 32
correlation coefficient estimates and κ statistics. Cronbach α Diabetes related 3
statistic was determined to assess internal consistency of the Renal failure 8
scoring systems. For outcome analysis, GCS and FS Hepatic failure 6
generated by the consultants and residents were evaluated Respiratory failure 4
separately. Univariate analysis for parameters predictive of Cardiac failure 3
Electrolyte disorders 6
outcome was performed using unpaired t test. For outcome
Adrenal dysfunction 2
evaluation at discharge and 3 months, a multiple logistic Infections 21
regression model (backward conditional method) was fitted Sepsis 17
with various patient attributes, which were found significant Pneumonia 3
on univariate analysis (at P b .1), and ORs with 95% CIs were Urinary infection 2
estimated. Receiver operating characteristic (ROC) curves Meningitis/HIV 1
were examined to compare the performance of GCS, FS, and Others 7
SOFA scores in predicting outcomes at discharge and 3 Cancer related 5
months. Reported P values are 2 tailed, and P ≤ .05 was Drug overdose 2
considered statistically significant. Statistical analysis was Outcome At discharge At 3 mo
Good outcome (mRS, 0-3) 31 32
performed using SPSS (version 13; SPSS Inc, Chicago, Ill).
Poor outcome (mRS, 4-6) 69 68
Died (mRS, 6) 24 39
⁎ A total of more than 100, as some patients had more than 1 factor
3. Results responsible for altered mental status.

One hundred patients (male-to-female ratio, 58:42; mean


age, 62 ± 17 years) with altered mental status due to a On comparing the overall trends of FS and GCS in
medical condition were studied. Table 1 summarizes their individual patients, there was good correlation (Pearson
demographic, clinical, and outcome data. Fifty-eight correlation coefficient, 0.881; P b .001; Fig. 1B). Internal
patients had altered mental status due to neurologic consistency for the evaluation of GCS and FS was noted to
conditions; 32, due to metabolic encephalopathy; 21, due be good to excellent on Cronbach α test, with scores of 0.815
to infections; and 7, due to other causes. Range of for GCS and 0.902 for FS.
sensorium and scatter of GCS and FS scores are depicted
in Fig 1. Duration of altered mental status ranged from few
hours to 3 days before recruitment into the study. The
median hospital stay was 13 days.
4. Outcome evaluation—using GCS and FS
The mean GCS at admission as assessed by a resident was scores from consultants
11.2 ± 3.3 and, by a neurology consultant, 12 ± 8; the mean
FS was 13.3 ± 3.5 for resident and 13 ± 3 for consultant 4.1. At discharge from hospital
(Table 2). κ Scores for GCS ranged from 0.7 to 0.85,
whereas those for FS ranged from 0.0.71 to 0.95—both At discharge, 69 patients had poor functional outcome
suggesting good to excellent interobserver agreement. (modified Rankin Score [mRS], 4-6), including 24 who died.
Among the component scores in the 2 scales, the best On univariate analysis (independent-samples t test), the
agreement was noted with respiration assessment (κ = 0.95), following were significantly related to mortality at discharge:
whereas the poorest was for brainstem response assessment systolic BP (P = .02); GCS—total and all component scores on
(κ = 0.71), both in FS. days 1, 2, and 3 (P b .006); FS—total and all component scores
316.e4 A.R. Gujjar et al.
45 20

40
B
35
30 Died

Four Score total


10
25
Survived
20
15
0
10
5
Pearson Corr Coeff: 0.881, P < .001
0
-10
Alert Drowsy Coma* 2 4 6 8 10 12 14 16
A
GCS total

30
50

C D
40

20
30
n

n 20
10

10

0 0
3 4 5 6 7 8 9 10 11 12 13 14 15 0 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Total GCS Day 1 Total FOUR score Day 1

Fig. 1 A, Distribution of sensorium and mortality among 100 patients with altered mental status due to a medical condition. *Stuporose or
comatose. B, Correlation of GCS and FS on day 1. C and D, Number of patients on day 1 with different scores on GCS and FS,
respectively (y-axis in A, C, D: number of patients.

on 3 consecutive days (P b .024); serum creatinine (P = days 1, 2, and 3 (P b .04); FS—total and component scores
.05); serum albumin (P b .001); hemoglobin (P = .003); on 3 days ((P b .036) other than brainstem responses on
and SOFA score (P = .001; Table 3). Factors significantly days 1 and 2; and serum albumin (P = .05).
related to good outcome (vs poor outcome) were as follows: Based on multivariate analysis with inclusion of 9
age (P = .003); GCS—total and all component scores on parameters (age, systolic BP, diastolic BP, total GCS on
day 1, total FS on day 1, serum albumin, creatinine,
hemoglobin, and SOFA score), factors independently
Table 2 Mean scores and interrater reliability for GCS and FS predicting good outcome at discharge included age and
on day 1 of admission for 100 patients, rated simultaneously systolic BP. Factors predicting mortality at discharge
by neurology consultant and resident included systolic BP, diastolic BP, FS (total on day 1),
Neurology consultant Resident κ Score hemoglobin percentage (Hb%), and SOFA score (Table 4).
(mean score) (mean score) (P b .001)
GCS n = 100 n = 100 4.2. At 3-month follow-up
Eye 3.24 ± 1.1 3.25 ± 1.1 .84
opening At 3 months, 68 patients had poor outcome (mRS, 4-6),
Motor 4.95 ± 1.4 4.97 ± 1.3 .85 including 39 who died. The parameters related to mortality
Verbal 3.1 ± 1 3 ± 1.6 .77 at 3 months on univariate analysis were as follows: systolic
Total GCS 11.5 ± 3.4 11.2 ± 3.3 .71
(P = .013) and diastolic BP (P = .045); GCS—total and
FS
component scores on 3 days (P b .008), except verbal scores
Eye open 2.96 ± 1.4 2.93 ± 1.5 .89
Motor 3.13 ± 1.1 3.18 ±1 .86 on days 1 and 2; FS—total and all component scores on 3
Brainstem 3.59 ± 1 3.62 ± 0.9 .71 consecutive days (P b .015); serum albumin (P = .05);
Respiration 3.65 ± 0.97 3.68 ± 0.9 .95 hemoglobin (P b .001); and SOFA score (P b .001)
(Table 3). Parameters related to good outcome at 3 months
Four score and GCS in patients with altered sensorium 316.e5

Table 3 Factors influencing mortality at 3 months among 100 predicting mortality at 3 months included the following: FS
patients (39 died) with acute medical conditions causing (total, day 1), systolic BP, and Hb% (Table 4).
altered sensorium: univariate analysis Thus, of the 3 scores, GCS was an independent predictor
Survived Died P of outcome at 3 months; FS was predictive of mortality at
(mean ± SD) (mean ± SD) both discharge and 3 months; and SOFA score was
predictive of mortality only at discharge. Including compo-
SYS BP 138.93 ± 21.23 126.18 ± 28.90 .013
DIAST BP 72.16 ± 12.90 65.38 ± 20.51 .045 nents of GCS and FS in logistic regression did not much
D1 GCS E 3.52 ± .87 2.79 ± 1.28 .001 change the predictive factors.
D1 GCS M 5.28 ± 1.19 4.44 ± 1.59 .003
D1 GCS V 3.33 ± 1.61 2.69 ± 1.61 .057 4.3. Outcome evaluation—using GCS and FS scores
D1 GCSTOT 12.13 ± 2.93 10.08 ± 3.90 .003 from residents
D2 GCS E 3.54 ± .91 2.65 ± 1.36 .000
D2 GCSM 5.41 ± 1.11 4.62 ± 1.64 .007
Resident-generated GCS and FS scores showed good
D2 GCS V 3.34 ± 1.67 2.77 ± 1.37 .095
D2 GCSTOT 12.29 ± 3.15 9.95 ± 3.76 .002 correlation (P b .05) with morality and outcome at 3 months
D3 GCS E 3.56 ± .88 2.83 ± 1.27 .002 (except breathing component on FS), on univariate analysis.
D3 GCS M 5.74 ± .56 4.74 ± 1.56 .000 On multivariate analysis, GCS total score was predictive of
D3 GCS V 3.56 ± 1.59 2.69 ± 1.68 .017 outcome, whereas total FS score was predictive of mortality
D3 GCSTOT 12.84 ± 2.60 10.26 ± 3.94 b.001 at 3 months (Table 4). These trends were similar to those
D1 FS-E 3.25 ± 1.21 2.51 ± 1.64 .012 from consultants.
D1 FS-M 3.39 ± .99 2.72 ± 1.23 .003
D1 FS-BS 3.82 ± .72 3.23 ± 1.31 .005
D1 FS-RS 3.87 ± .59 3.31 ± 1.30 .004
D1 FS-TOT 14.33 ± 2.50 11.77 ± 4.15 b.001 Table 4 Multivariate analysis: factors independently predicting
D2 FS-E 3.25 ± 1.30 2.32 ± 1.68 .004 good outcome or mortality at discharge and 3 months' follow-up
D2 FS-M 3.54 ± .87 2.86 ± 1.29 .004 among patients with altered mental status due to medical
D2 FS-BS 3.80 ± .75 3.03 ± 1.48 .001 conditions (n = 100)
D2 FS-RS 3.84 ± .68 3.22 ± 1.44 .006
D2 FS-TOT 14.43 ± 2.98 11.46 ± 4.92 b.001 OR 95% CI P
D3 FS-E 3.36 ± 1.22 2.37 ± 1.66 .002 for OR
D3 FS-M 3.72 ± .57 2.89 ± 1.30 .000 Lower Upper
D3 FS-BS 3.84 ± .68 3.26 ± 1.29 .008
D3 FS-RS 4.00 ± .00 3.51 ± 1.22 .006 Factors predicting good outcome
D3 FS-TOT 14.92 ± 2.07 12.03 ± 4.38 b.001 at discharge
Creatinine 96.13 ± 135.95 136.95 ± 101.56 .111 Age 1.038 1.01 1.067 .007
Bilirubin 14.37 ± 19.72 31.71 ± 58.93 .050 GCS—day 1 (Cons) a 0.835 0.715 0.974 .022
Albumin 32.57 ± 7.22 25.31 ± 7.50 b.001 Factors predicting mortality
Hemoglobin 11.818 ± 2.242 10.379 ± 2.467 .003 at discharge
SOFA score 2.36 ± 1.79 4.54 ± 3.26 b.001 Systolic BP 0.925 0.882 0.970 .001
Diastolic BP 1.083 1.015 1.156 .016
Days 2 and 3 have lesser patients for evaluation because some died or FS—day 1 (Cons) a 0.759 0.593 0.970 .028
were sedated/paralyzed for intubation. All GCS and FS scores were
Hb% 0.577 0.401 0.831 .003
generated by consultants. D1, D2, D3 indicate days of assessment; E,
SOFA score 1.501 1.045 2.158 .028
eye opening; M, motor response; V, verbal response; TOT, total GCS;
BS, brainstem function; RS, respiratory pattern. Factors predicting good
outcome at 3 mo
Age 1.043 1.011 1.076 .007
were as follows: age (P = .03); GCS—total and all component Systolic BP 0.973 0.950 0.996 .024
scores on 3 days (P b .024); FS—total and all component GCS—day 1 (Cons) a 0.760 0.635 0.911 .003
GCS (Res) a 0.769 0.64 0.92 .004
scores on 3 days (P b .03), except brainstem and respiratory
Factors predicting mortality at 3
scores on days 1 and 2 and brainstem score on day 3; serum mo
albumin (P = .001); hemoglobin (P = .021); and SOFA score Systolic BP 0.966 0.944 0.988 .003
(P = .01). At 3 months, 9 of 10 patients with a GCS of 3 to 6 Hb% 0.717 0.561 0.917 .008
died, whereas 11 (84.6%) of 13 with an FS of 0 to 9 died. FS—day 1 (Cons) a 0.694 0.574 0.839 b.001
On multivariate analysis with inclusion of 10 parameters Total FS (Res) b 0.726 0.61 0.87 b.001
(age, systolic BP, diastolic BP, total GCS on day 1, total FS Variables entered into analysis: age, systolic BP, diastolic BP, GCS
on day 1, serum albumin, bilirubin, creatinine, hemoglobin, score total day 1, FS—total day 1, creatinine, albumin, Hb%, and
and SOFA score), factors independently predicting good SOFA score.
a
outcome at 3 months included the following: GCS (total Scores by consultant.
b
Scores by resident on day 1.
score on day 1), age, and Hb%. Factors independently
316.e6 A.R. Gujjar et al.

5. Receiver operating characteristic analysis Table 5 Performance of GCS and FS generated by


consultants in predicting outcomes at 3 months among 100
patients with medical causes of altered sensorium
Areas under the ROC curve (AUCs) for total scores of
GCS and FS from consultants and SOFA scores were Outcome at Good outcome Mortality
comparable for predicting mortality at 3 months (Fig. 2; 3 mo AUC P AUC P
Table 5). On comparing the performance of component
GCS motor 0.707 .001 0.676 .003
scores, motor scores of both GCS and FS were observed to GCS total 0.683 .003 0.655 .009
perform as well as the total scores, whereas GCS verbal FS motor 0.715 .001 0.678 .003
scores and FS respiration were noted to have poor FS total 0.708 .001 0.697 .001
performance. Performance of SOFA score was comparable
This information is illustrated in Fig. 2.
with those of the coma scores.

fairly robust outcome predictors. Specifically, GCS was an


6. Discussion independent predictor of outcome at 3 months, FS was
predictive of mortality at both discharge and 3 months, and
This study compares the interrater reliability of the new SOFA score was predictive of mortality only at discharge.
FS among neurology consultants and residents with that of Similar trends in outcome predictability were observed for
GCS among unselected patients with altered mental status GCS and FS generated by both consultants and residents.
managed in general medical wards. Our results suggest that Most of the earlier studies examined outcome predictability
interrater reliability is comparable for FS and GCS, both at discharge or at 28 days [14], whereas one examined 3-
being in the range of good to excellent [19]. These results are month outcome among patients with head injury or
comparable with earlier studies from Mayo Clinics and neurologic disorders [15]. Similar to our study, these studies
elsewhere [8-16]. All the earlier studies have addressed reported the outcome predictability of FS to be comparable
patients in the ICUs or emergency department. To our with or superior to GCS. We also examined the value of
knowledge, this study is the first to evaluate FS in the context serial GCS and FS documented on the initial 3 days in
of general medical wards, comparing neurologists and outcome prediction. There was no significant difference
internal medicine residents. These results suggest that the between scores on days 2 and 3 as compared with day 1 in
FS can be administered fairly reliably in the medical wards this regard. Although the SOFA score was associated with
even by residents. outcomes both at discharge and 3 months on univariate
This study further examined the predictive value of GCS, analysis, it failed to sustain this relation on multivariate
FS, and SOFA scores on outcomes at discharge and at 3 analysis. Our results suggest that similar to GCS, FS scores
months after discharge. Whereas univariate analysis demon- are robust outcome predictors up to 3 months. Their value
strated a relationship between outcomes with various clinical may be enhanced by the fact that all patients who died in this
and laboratory parameters in addition to the above scores, study died of disease progression, with no instance of
multivariate analysis demonstrated the coma scores to be withdrawal of care.

ROC: Good outcome at 3 mo ROC: Died at 3 mo


1.00 1.00

.75 .75
Sensitivity

Sensitivity

Source of the Curve


.50 .50
Reference Line
Reference Line
D1FSTOT
D1FSTOT
.25 D1FSM .25
D1FSM
D1GCSTOT
D1GCSTOT
0.00 D1GCSM D1GCSM
0.00
0.00 .25 .50 .75 1.00 0.00 .25 .50 .75 1.00
1 - Specificity 1 - Specificity

Fig. 2 Receiver operating characteristic curves for GCS (total and motor scores) and FS (total and motor scores) among patients with
medical causes of altered mental status (n = 100). D1FSTOT indicates FS total on day 1; D1FSM, motor score of FS on day 1; D1GCSTOT,
total GCS score on day 1; D1GCSM, motor score on GCS on day 1.
Four score and GCS in patients with altered sensorium 316.e7

The performance of the 3 scoring systems in outcome administration of FS should not be a major issue. FOUR
prediction at 3 months evaluated by ROC curve analysis score appears to provide more objective information
revealed them to be fairly comparable. Overall, the regarding the patient's neurologic status, given the targeting
performance for predicting good vs poor outcome was better of specific systems by its different components; combina-
than for predicting mortality. Eken et al [15] observed the tions of observations may point to situations such as
eye and motor response components of FS to be comparable vegetative state or brain death, which would definitely
with the total scores in performance. We observed the motor influence management decisions. As regards outcome
components of GCS and FS to perform almost as efficiently prediction, FS, overall, appears to perform similarly to
as the total scores. GCS; the emerging experience that motor response in FS
The methodological strengths of this study are as may be equivalent to the total score is similar to that
follows: the inclusion of unselected patients presenting to observed in GCS. There would always be a significant
a general medical hospital practice, who are likely the number of patients at the extremes on the scale of mental
largest segment of patients seeking medical assistance in status whose outcomes would be contrary to that predicted
any population; comparison of FS with GCS, which is the their initial assessment of sensorium; this would manifest as
most commonly used coma score; comparison with SOFA “poor performance” on, for example, the ROC curve method
score, which is a combination of clinical and selected of assessment. Hence, it would theoretically be impossible to
laboratory parameters reflecting multiple-organ system obtain a “perfect” scoring system. In fact, improving medical
functions (and which also includes GCS); study of serial care would be expected to further mitigate the “performance”
coma scores on 3 consecutive days; and outcome studies at of such scoring systems. There would be no “gold standard”
discharge and at 3 months, with no deaths being consequent for evaluating the use of a test for predicting long-term
to withdrawal of care. outcome. The emerging experience with limitations in
This study had several limitations: exclusion of patients administering the FS and its use in varied situations would
who were sedated or paralyzed for intubation, which likely shape its adaptability. In the opinion of the authors, given the
excluded more sick patients from the cohort; inability to limitations and ambiguities of earlier scoring systems, the FS
obtain paired κ scores for consultant-resident pairs due to appears to have addressed most concerns and may likely
variable groupings; and obtaining 3 months' outcome data evolve as a widely adopted method of documenting mental
by telephone interview rather than direct patient evaluation. status. These conclusions appear to concur with emerging
In addition to these, a limitation observed in brainstem opinions at other centers [20].
function evaluation as part of FS in individual patients was
the influence of past iridectomy (for cataract surgery) and
corneal opacities (which is common in our patients, often
7. Conclusion
due to past trachoma infection). About 20% of the patients
older than 60 years had at least one of these conditions,
limiting papillary response evaluation; in such patients, This study examined the interobserver reliability of FS
scoring was limited to other methods of brainstem and GCS among medically ill, adult patients presenting with
evaluation (eg, corneal response, cough, and gag response). altered mental status, comparing the performance of
This limitation likely resulted in the relatively lower neurology consultants with internal medicine residents. The
interrater reliability observed for the brainstem component interobserver reliability was comparable for both the coma
of FS. scores, being in the range of good to excellent. FOUR score
With the emerging experience, would it be practical to performs comparably with GCS but better than SOFA score
replace GCS with FS as a routine measure of sensorium in in predicting outcomes at 3 months. Both consultant- and
the wards? A distinction should be made among several resident-generated scores were similar in outcome predict-
issues in this regard: ease of administration of the test; its ability. Serial GCS and FS extending to days 2 and 3 of
reliability and utility, essentially in providing useful admission did not improve outcome predictability in
information that could guide management; and outcome comparison with day 1 scores. On ROC curve analysis,
prediction. Although the number of components appears to FS, GCS, and SOFA scores perform fairly comparably in
be more in FS, the authors opine that they are all essential predicting outcomes at 3 months. In the elderly patients, past
parts of formal neurologic evaluation of any patient with iridectomy for cataract surgery or corneal opacity may limit
altered sensorium. At the end of a conventional neurologic accurate FS assessment. We conclude that FS could reliably
evaluation for altered sensorium, it should be possible to be used in the medical wards for evaluation of mental status.
score the patient on FS with no additional specific tests
required. Ambiguities of generating a GCS verbal response
score raised by absent verbal output in an aphasic or Acknowledgments
comatose patient or eye response score in “coma vigil”—
where eyes are open but without conscious “focusing”—are This study was funded by the Sultan Qaboos University,
eliminated in FS. With some experience, developing ease of Muscat, Oman. The authors wish to acknowledge and thank
316.e8 A.R. Gujjar et al.

all the patients and their families for their kind cooperation, [9] Wijdicks EF. Clinical scales for comatose patients: the Glasgow Coma
Dr Gousia Jikky for help in data collection and maintaining Scale in historical context and the new FOUR score. Rev Neurol Dis
2006;3(3):109-17 [Review. PubMed PMID: 17047576].
database, residents and SHOs in the Medicine Department [10] Wolf CA, Wijdicks EF, Bamlet WR, et al. Further validation of the
for their enthusiastic participation, and nursing staff for their FOUR score coma scale by intensive care nurses. Mayo Clin Proc
valuable support. 2007;82(4):435-8.
[11] Stead LG, Wijdicks EF, Bhagra A, et al. Validation of a new coma
scale, the FOUR score, in the emergency department. Neurocrit Care
2009;10(1):50-4 [Epub 2008 Sep 20. PubMed PMID: 18807215].
References [12] Iyer VN, Mandrekar JN, Danielson RD, et al. Validity of the FOUR
score coma scale in the medical intensive care unit. Mayo Clin Proc
[1] Teasdale G, Jennett B. Assessment of coma and impaired conscious- 2009;84(8):694-701 [PubMed PMID: 19648386; PubMed Central
ness. A practical scale. Lancet 1974;2:81-4. PMCID: PMC2719522].
[2] Teasdale G, Knill-Jones R, van der Sande J. Observer variability in [13] Weiss N, Mutlu G, Essardy F, et al. French version of the FOUR score:
assessing impaired consciousness and coma. J Neurol Neurosurg a new coma score. Rev Neurol (Paris) 2009;165(10):796-802.
Psychiatry 1978;41:603-10. [14] Fischer M, Rüegg S, Czaplinski A, et al. Inter-rater reliability of the
[3] Rowley G, Fielding K. Reliability and accuracy of the Glasgow Coma Full Outline of UnResponsiveness score and the Glasgow Coma Scale
Scale with experienced and inexperienced users. Lancet 1991;337: in critically ill patients: a prospective observational study. Crit Care
535-8. 2010;14(2):R64.
[4] Menegazzi JJ, Davis EA, Sucov AN, et al. Reliability of the Glasgow [15] Eken C, Kartal M, Bacanli A, et al. Comparison of the Full Outline of
Coma Scale when used by emergency physicians and paramedics. Unresponsiveness score coma scale and the Glasgow Coma Scale in an
J Trauma 1993;34:46-8. emergency setting population. Eur J Emerg Med 2009;16(1):29-36.
[5] Balestreri M, Czosnyka M, Chatfield DA, et al. Predictive value of [16] Akavipat P. Endorsement of the FOUR score for consciousness
Glasgow Coma Scale after brain trauma: change in trend over the past assessment in neurosurgical patients. Neurol Med Chir 2009;49(12):
ten years. J Neurol Neurosurg Psychiatry 2004;75:161-2. 565-71.
[6] Diringer MN, Edwards DF. Does modification of the Innsbruck and [17] Sadaka F, Patel D, Lakshmanan R. The FOUR score predicts outcome
the Glasgow Coma Scales improve their ability to predict functional in patients after traumatic brain injury. Neurocrit Care 2012;16:95-101.
outcome? Arch Neurol 1997;54(5):606-11. [18] Fugate JE, Rabinstein AA, Claassen DO, White RD, Wijdicks EFM.
[7] Zygun D, Berthiaume L, Laupland K, et al. SOFA is superior to MOD The FOUR score predicts outcome in patients after cardiac arrest.
score for the determination of non-neurologic organ dysfunction in Neurocrit Care 2010;13:205-10.
patients with severe traumatic brain injury: a cohort study. Crit Care [19] Landis JR, Koch GG. The measurement of observer agreement for
2006;10:R115, http://dx.doi.org/10.1186/cc5007. categorical data. Biometrics 1977;33:159-74.
[8] Wijdicks EF, Bamlet WR, Maramattom BV, et al. Validation of a new [20] Kornbluth K, Bhardwaj A. Evaluation of coma: a critical appraisal of
coma scale: the FOUR score. Ann Neurol 2005;58(4):585-93. popular scoring systems. Neurocrit Care 2011;14:134-43.

You might also like