If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Glasgow coma scale compared to other trauma scores in discriminating in-hospital mortality of traumatic brain injury patients admitted to urban Indian hospitals: A multicentre prospective cohort study
Department of Global Public Health, Karolinska Institutet, Stockholm, SwedenFunction Perioperative Medicine and Intensive Care, Karolinska University Hospital, Solna, Sweden
WHO Collaborating Centre for Research in Surgical Care Delivery in LMIC, Mumbai, IndiaDepartment of Global Public Health, Karolinska Institutet, Stockholm, SwedenInjury Division, The George Institute, New Delhi, India
GCS is one of the most commonly used trauma scores and is a good predictor of outcome in TBI patients.
•
There are other more complex scores like RTS, MGAP, GAP, KTS with additional physiological parameters to GCS.
•
This study compares discrimination of GCS to that of the above complex trauma scores for 24-hour and 30-day in-hospital mortality in adult TBI patients, in a resource limited LMIC setting.
•
This study indicates that discrimination of GCS is comparable to that of more complex trauma scores in predicting 24-hour and 30-day mortality of adult TBI patients in resource limited LMIC settings.
Abstract
Background
Glasgow Coma Scale (GCS) is one of the most commonly used trauma scores and is a good predictor of outcome in traumatic brain injury (TBI) patients. There are other more complex scores with additional physiological parameters. Whether they discriminate better than GCS in predicting mortality in TBI patients is debatable. The aim of this study was to compare the discrimination of GCS with that of MGAP, GAP, RTS and KTS for 24-hour and 30-day in-hospital mortality in adult TBI patients, in a resource limited LMIC setting.
Method
We analysed data from the multicentre, observational trauma cohort Towards Improved Trauma Care Outcome (TITCO) in India. We included all patients 18 years or older, admitted from the emergency department with TBI. The Area Under the Receiver Operating Characteristic (AUROC) curve was used to quantify and compare the discrimination of all scores: GCS; Revised Trauma Score (RTS); mechanism, GCS, age, systolic blood pressure (MGAP); GCS, age, systolic blood pressure (GAP) and Kampala Trauma Score (KTS) in the prediction of 24-hour and 30-day in-hospital mortality.
Results
A total of 3306 TBI patients were included in this study. The majority were within the GCS range 3-8. The commonest mechanism of injury was road traffic injuries [1907(58.0%)]. In-hospital mortality was 27.2% (899). There was no significant difference in discrimination in 24-hour in-hospital mortality when comparing GCS with MGAP and GAP. While GCS performed better than KTS, RTS performed better than GCS. For 30-day in-hospital mortality, GCS discriminated significantly better compared with KTS, but there was no significant difference when compared to MGAP and RTS. GAP discriminated significantly better when compared with GCS.
Conclusion
This study shows that the discrimination of GCS is comparable to that of more complex trauma scores in predicting 24-hour and 30-day in-hospital mortality in adult TBI patients in a resource limited LMIC setting.
GBD 2016 Traumatic Brain Injury and Spinal Cord Injury Collaborators. Global, regional, and national burden of traumatic brain injury and spinal cord injury, 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016.
]. Increasing population density, urbanization and a growing number of motor vehicles have led to an increase in the incidence of road traffic injuries (RTI) worldwide which in-turn has led to an escalation in the rate of TBI among the adult population, as RTI are the commonest mechanism of injury among these patients [
GBD 2016 Traumatic Brain Injury and Spinal Cord Injury Collaborators. Global, regional, and national burden of traumatic brain injury and spinal cord injury, 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016.
A Comparison of the Predictive Value of the Glasgow Coma Scale and the Kampala Trauma Score for Mortality and Length of Hospital Stay in Head Injury Patients at a Tertiary Hospital in Uganda: A Diagnostic Prospective Study.
]. The GCS is easy to use by members of the health care team and provides a mechanism to quickly evaluate the severity, see changes in the level of consciousness on repeated evaluation and predict the outcome of the patient [
Despite having several advantages, over the years limitations of GCS was observed by some researchers. It becomes unreliable in patients under the influence of drugs and alcohol, those who are intubated and those with injuries to the eye. Some studies also showed that GCS does not accurately recognize the severity of TBI with increasing age. There is also an inter- observer variability, which may lead to inaccurate outcome prediction, particularly in patients with severe TBI [
Other scores developed to assess the severity of trauma patients and these scores often include GCS or some other level of consciousness measure. Some of the most well-known of these scores are Revised Trauma Score (RTS), mechanism, GCS, age, systolic blood pressure (MGAP), GCS, age, systolic blood pressure (GAP) and Kampala Trauma Score (KTS). Studies conducted in HICs have suggested that these trauma scores perform better in predicting outcome in TBI patients compared to GCS alone [
It is not known if these more complex scores perform better than GCS alone for predicting TBI patient outcome in low resource settings. Particularly in these settings, the potential added value of increasing complexity should be weighed against increased effort, as the calculation of these complex scores require more trained man power. The aim of this study was to compare the discrimination of GCS with that of MGAP, GAP, RTS and KTS for 24-hour and 30-day in-hospital mortality in adult TBI patients, in a low resource LMIC setting.
Method
Study design
The study was an analysis of the ‘Towards Improvement of Trauma Care Outcomes’ (TITCO) in India cohort (www.titco.org) [
Revised trauma scoring system to predict in-hospital mortality in the emergency department: Glasgow Coma Scale, Age, and Systolic Blood Pressure score.
]. The TITCO cohort is a multicentre, observational cohort collected between September 2013 and December 2015 at four urban tertiary care referral centres spread across India.
Setting
The participating hospitals were Jai Prakash Narayan Apex Trauma Centre of All India Institute of Medical Sciences (AIIMS), New Delhi; Lokmanya Tilak Municipal General Hospital (LTMGH) and King Edward Memorial (KEM) Hospital, Mumbai and Institute of Postgraduate Medical Education and Research-Seth Sukhlal Karnani Memorial Hospital (SSKM), Kolkata. All four hospitals had dedicated neurosurgery facilities catering to neuro-trauma patients.
Inclusion criteria
Patients of age ≥18years, with a history of TBI admitted to the participating hospitals.
Exclusion criteria
Patients with incomplete data and those discharged against medical advice.
Variables
In this study, five trauma scores were compared - GCS, MGAP, GAP, RTS and KTS. Heart rate (HR), systolic blood pressure (SBP), respiratory rate (RR), and GCS were recorded on arrival at the hospitals where the study was conducted. The GCS was measured on arrival at the participating hospital regardless of their level of sedation or intubation. We used these vital signs to calculate the above-mentioned scores according to published equations [
]. (Details of the scores are described in Supplementary Digital Content)
The variables age, gender, mechanism of injury, and type of injury was collected from the patients’ records. Information regarding whether a patient came directly to the participating hospital, or was transferred from another facility was recorded. Delay in arrival, i.e. time between injury and presentation at the participating hospital was also recorded. To calculate KTS, we used GCS to estimate the AVPU (Alert, Voice, Pain, Unresponsive)Scale and used GCS 14–15 as ‘alert’, GCS 10–13 as ‘responds to voice’, GCS 5–9 as ‘responds to pain’, and GCS 3–4 as ‘unresponsive’) [
Value of the Glasgow coma scale, age, and arterial blood pressure score for predicting the mortality of major trauma patients presenting to the emergency department.
The outcome of this study was 24-hour in-hospital mortality (death occurring within 24 h of index hospital admission) and 30-day in-hospital mortality (death occurring within 30 days of index hospital admission).
Data source
Data on demographics, outcomes, vital signs, and injury data was collected by direct observation or from the patients’ records, patients and their relatives.
Bias
To minimise bias and maintain consistency in data collection in TITCO, two steps were followed. Firstly, weekly telephonic meetings were held between the four centers to discuss potential data collection issues. Secondly, to make sure there was no major deviation in data collection, yearly retrospective quality control was done.
Quantitative variables
Age was categorised as 18-24, 25-44, 45-64, ≥65. We used the International Classification of Diseases version 10 codes S02.0,S02.1,S02.3,S02.7- 02.9,S06.0- S06.9, S07.0- S07.9, and S09.7- S09.9 used to identify patients with TBI [
]. Median (inter-quartile range - IQR) was used to describe quantitative variables, while count and percentage were used for qualitative variables. We used the area under the receiver operating characteristics (AUROC) curve to quantify and compare the discrimination of the different trauma scores. The AUROC range is from 0 to 1, where 1 represents perfect discrimination. A p-value of less than 0.05 was considered statistically significant using Delong's test [
World Health Organization. (2004). ICD-10: international statistical classification of diseases and related health problems: Tenth revision, 2nd ed. World Health Organization.
A total of 16,000 trauma patients were included in the TITCO cohort. Of these, 9268 patients were admitted with TBI and 7244 patients were ≥18 years of age and hence eligible for our study. 3938 patients were excluded as they had one or more missing data for age, SBP, RR, GCS or mechanism of injury. The parameter most commonly missed was RR (48.5%), followed by SBP (19.85%). GCS was the least commonly missed data (7.4%). Finally, a total of 3306 patients were included in the study (Fig. 1). We found no important differences in demographic patterns or mortality comparing patients with complete and incomplete data. (Supplementary Digital Content).
The median age was 35 (IQR : 26, 50) years and the majority of patients were males [2737 (82.8%)]. Most of the patients were admitted following RTI [1907 (58.0%)] followed by fall [885 (26.9%)]. A total of 2294 (69.4%) patients were transferred from other hospitals, with a median delay in arrival to the hospital of 6.5 h (2.3, 22.1) (Table 1), of whom 165 patients were already intubated prior to arrival at the hospital where the study was conducted. Most of the patients had blunt trauma, with only about 1% patient (n = 35) suffering from penetrating injuries. In view of the small number of cases with penetrating injury, a subgroup analysis was not useful in this cohort.
Of all the TBI patients, 31.7% (n = 1049) underwent surgical intervention. While the mortality of operated patients was 30.4%, patients undergoing conservative management (n = 2257) had a mortality of 25.7%. In-hospital 24-hour and 30-day mortality rates were 6.0% (200) and 26.0% (860) respectively. Overall mortality was 27.2% (899) (Table 1). The proportion of mortality in direct and transferred patients were 25% and 28% respectively (p = 0.12) (Table 2).
Table 2Comparison of transferred and directly admitted patients.
The median, IQR, 95% Confidence Interval (CI) and AUROC of GCS, MGAP, GAP, RTS and KTS are shown in Table 3. For 24-hour in-hospital mortality, GCS discriminated significantly better compared with KTS but there was no significant difference in discrimination of GCS when compared with MGAP and GAP. When GCS was compared with RTS, the latter discriminated better than the former. For 30-day in-hospital mortality, GCS discriminated significantly better compared with KTS but there were no significant differences in the discrimination of GCS when compared with MGAP and RTS. GAP discriminated marginally better when compared with GCS (Table 3, Fig. 2a, b).
Table 3Median (IQR), AUROC and 95% CI for different trauma scoring systems for 24-hour and 30-day in-hospital mortality in TBI patients.
24-hour In Hospital Mortality
30-day In Hospital Mortality
Median IQR)
AUROC
95% CI
p-value
AUROC
95% CI
p-value
GCS
10 (7-15)
0.86
0.83-0.88
0.82
0.80-0.83
MGAP
23 (20-28)
0.86
0.83-0.88
0.83
0.82
0.81-0.84
0.12
GAP
18 (14-22)
0.86
0.84-0.88
0.77
0.83
0.81-0.84
0.002
RTS
6.90 (5.96-7.84)
0.88
0.86-0.90
0.008
0.81
0.80-0.83
0.67
KTS
13 (11-14)
0.80
0.78-0.83
<0.001
0.78
0.77-0.80
<0.001
(p-value = pairwise comparison of AUC between each score and GCS)
We conducted a subgroup analysis for patients with isolated TBI and poly-trauma with TBI. In isolated TBI patients, GCS discriminated significantly better than KTS but on comparing with MGAP, GAP and RTS there was no significant difference. Among poly-trauma patients with TBI, there was no significant difference in discrimination between GCS and MGAP, GAP and KTS. RTS discriminated better than GCS (Table 4, Fig. 3).
Table 4AUROC and 95% CI for different trauma scoring systems for Isolated TBI, Poly-trauma (with TBI), Directly admitted patients and transferred patients for 24-hour in-hospital mortality.
Isolated TBI (n = 2239)
Poly-trauma with TBI (n = 1067)
Directly admitted Patients
Transferred Patients
AUROC
95% CI
p-value
AUROC
95% CI
p-value
AUROC
95% CI
p-value
AUROC
95% CI
p-value
GCS
0.86
0.83-0.89
0.87
0.83-0.91
0.88
0.85-0.91
0.85
0.82-0.88
MGAP
0.87
0.84-0.89
0.72
0.85
0.80-0.89
0.50
0.89
0.86-0.90
0.50
0.85
0.82-0.88
0.31
GAP
0.87
0.84-0.90
0.77
0.86
0.81-0.90
0.75
0.89
0.86-0.92
0.50
0.86
0.83-0.89
0.99
RTS
0.87
0.85-0.90
0.09
0.90
0.87-0.93
0.03
0.90
0.87-0.93
0.49
0.87
0.84-0.90
0.25
KTS
0.81
0.78-0.84
0.01
0.81
0.76-0.85
0.07
0.82
0.78-0.87
0.78
0.79
0.76-0.82
0.02
(p-value = pairwise comparison of AUC between each score and GCS).
Fig. 3ROC curve for GCS, MGAP, GAP, RTS, KTS scores for 24 h in hospital mortality comparing isolated TBI (a) with poly-trauma associated with TBI (b) and direct admission (c) with transferred admission (d).
We also conducted a subgroup analysis of directly admitted patients and transferred patients. GCS discriminated significantly better than KTS in transferred patients, but GCS did not discriminate significantly better than MGAP, GAP, or RTS (Table 4, Fig. 3). Finally, the discrimination of the scores was comparable in both operative and conservative management arms.
Discussion
Our key finding is that the discrimination of GCS alone is comparable, and in some instances superior, to the discrimination of more complex trauma scores, for both 24-hour and 30-day in-hospital mortality. This finding held true in patients with isolated TBI, as well as in poly-trauma patients with TBI, and in both directly admitted and transferred TBI patients. In patients with isolated TBI and in transferred TBI patients, GCS discriminated significantly better than KTS, whereas in poly-trauma patients with TBI, RTS discriminated significantly better than GCS.
This key findingis in contrast to many studies from HICs, where scores with additional physiological parameters to GCS were found to discriminate better in predicting outcome when compared to GCS alone [
]. Salottolo et al added age adjustment to GCS and showed that it discriminated better in predicting-in hospital mortality as compared to standard GCS [
]. In another study, the FOUR score, which in addition to GCS includes respiratory rate and pattern, had better inter-rater agreement than GCS, but not better discrimination [
Few studies from other low resource settings also compared GCS with that of other trauma scores and found that discrimination of GCS was comparable to other scores. Ariaka H et al. reported in their study conducted at Uganda that GCS was superior to KTS in predicting both mortality and length of hospital stay [
A Comparison of the Predictive Value of the Glasgow Coma Scale and the Kampala Trauma Score for Mortality and Length of Hospital Stay in Head Injury Patients at a Tertiary Hospital in Uganda: A Diagnostic Prospective Study.
]. Ramazeni et al compared the prognostic value of GCS and GAP in older patients (>65 years) with TBI admitted in the intensive care unit and found both scores to be reliable predictors of outcome, with GCS scoring slightly better than GAP [
Though our initial cohort was over 7244, over half had to be excluded in view of incomplete data. Interestingly, GCS was the least missed data. Respiratory rate being the maximally missed data, calculation of RTS and KTS was not feasible in 48.5% patients. However, analysis was also done for scores where maximally missed vitals like RR and SBP were not required. AUC for GCS for the cohort of patients (n = 6707) excluding data with only GCS missing was 0.83. AUC for GAP and MGAP for patients with complete GCS and SBP, but missing RR data (n = 5269) were also 0.83 in both cases. This analysis thus revealed a comparable discrimination between GCS, GAP and MGAP which was consistent with the primary result.
Nearly 70% of the patients included in the study were transferred from others hospitals. Ideally, the vitals used for computation of the scores should have been those measured on arrival at the first health facility. However, in view of absence of any pre-hospital care or transfer protocols between hospitals, information of the patient on arrival at the initial hospital was often not available to the hospital where the patient was transferred. Knowing that vital signs change with time and resuscitation, this was a limitation which was unavoidable, given the context under which the data was collected. In view of these shortcomings, vital signs measured on arrival at hospitals where data is being collected are routinely used for injury scoring in most LMIC datasets without adjusting for time since injury [
Feasibility is an important consideration when choosing which trauma score to use, especially in low resource settings. Both RTS and GAP had slightly better discrimination than GCS for 24 h and 30 days mortality respectively, but the calculations of both these scores are more complicated compared to GCS alone, because these scores require other variables in addition to GCS. In a low resource LMIC setting, where the number of trauma patients arriving at the emergency department (ED) is very high, the calculation of these scores will take time, and may delay the triaging process and thereby patient care [
]. In addition, GCS had the least missing data out of the vital signs needed to calculate the scores assessed in our study, indicating that it is feasible to record for most patients.
Our large study based on multicentre data adds to the body of literature comparing GCS with other trauma scores. Collectively the evidence indicates a potential for simplifying trauma system benchmarking and research on TBI patients in these settings. Both trauma system benchmarking and research are key components of Trauma Quality Improvement Programs that rely on high quality and readily available data for risk adjustment of outcomes [
]. We suggest that it may be enough to adjust for GCS to account for severity in these analyses, especially in analyses of patients with isolated TBI, when resources are scarce. This would increase the feasibility of both data collection and analysis, and thus the feasibility of quality improvement programs for TBI patients in low resource settings.
Methodological considerations and limitations
In the large number of patients who were transferred from other hospitals, HR, SBP, RR, and GCS were recorded on arrival at the participating centres. This approach differs from the approach used in the original MGAP study [
], in which they had access to both prehospital values as well as values on initial presentation. The original RTS study does not report what values they used and in the original KTS study they used values on presentation to the participating hospitals, but do report the number of transferred patients [
]. The discrimination of the assessed scores may have been different if we had used the values recorded on initial presentation to the primary receiving hospitals, but these were not available in the TITCO cohort and would not have reflected the clinical reality of the participating centres.
Though the initial cohort was high (7244), over 50% (3938) patients could not be included due to missing data. RR was the most missing data (48.5%), followed by SBP (19.8%) and contributed to a large number of patients being eliminated from the analysis. GCS was the least missing data. So, we also conducted an available case analysis in which we used all non-missing data available for each pair-wise comparison. The results of this analysis were consistent with the main analysis.
The comparison with KTS is limited because we had to indirectly estimate the AVPU and number of serious injuries components included in this score, using GCS and AIS respectively. The direction of the bias introduced by these estimations is hard to guess, and it could have affected the discrimination either way. To ensure a more faithful representation of the original score future research comparing KTS with other scores should make an effort to collect data on AVPU and number of serious injuries directly. Finally, our findings are likely to be generalisable to other urban referral hospitals in India and other resource limited setting that receive large numbers of transferred patients. Our findings may not be representative of rural hospitals in India or otherwise.
Conclusion
This study indicates that the discrimination of GCS is comparable to that of more complex trauma scores in predicting 24-hour and 30-day mortality of adult TBI patients in low resource settings. In terms of discrimination, there is very little to gain from adding more variables. Thus, it may be enough to adjust for GCS when accounting for severity in risk adjusted analysis as part of trauma quality improvement programs in these settings. In addition, GCS had the least missing data, indicating that the ease of assessment of GCS by health care workers makes it a very practical score to predict in hospital mortality in low resource settings.
Acknowledgement
We thank the Towards Improved Trauma Care Outcomes (TITCO), India, research team for their support.
Funding
The TITCO dataset by the research consortium of Indian Universities was funded by grants from the Swedish National Board of Health and Welfare and the Laerdal Foundation for Acute Care Medicine, Norway. The funding agencies had no influence on the conduct of the study, the contents of the manuscript, or the decision to send the manuscript for publication.
Ethical considerations
The institutional ethics committee of all participating hospitals LTMGH (IEC/11/13 dated 26 Jul 2013), KEM (IEC(I)/out/222/14 dated 4 Mar 2014), SSKM (IEC/279 dated 21 Mar 2013) and Apex Delhi (IEC/NP-327/2013 RP-24/2013 dated 25 Sep 2013) individually approved the collation of the database and analysis.
Traumatic Brain Injury and Spinal Cord Injury Collaborators. Global, regional, and national burden of traumatic brain injury and spinal cord injury, 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016.
A Comparison of the Predictive Value of the Glasgow Coma Scale and the Kampala Trauma Score for Mortality and Length of Hospital Stay in Head Injury Patients at a Tertiary Hospital in Uganda: A Diagnostic Prospective Study.
Revised trauma scoring system to predict in-hospital mortality in the emergency department: Glasgow Coma Scale, Age, and Systolic Blood Pressure score.
Value of the Glasgow coma scale, age, and arterial blood pressure score for predicting the mortality of major trauma patients presenting to the emergency department.
World Health Organization. (2004). ICD-10: international statistical classification of diseases and related health problems: Tenth revision, 2nd ed. World Health Organization.
00694-5&id=gr1.jpg){kind=link}
00694-5&id=gr2.jpg){kind=link}
00694-5&id=gr3.jpg){kind=link}