The ability of GCS, FOUR, and APACHE II in predicting the outcome of patients with traumatic brain injury: A comparative study

1Doctor of Medicine (MD), Fellow of Critical Care Medicine (FCCM), Department of Anesthesiology and Critical Care, Khatam-Al-Anbiya Hospital, Zahedan University of Medical Sciences, Zahedan, IR, Iran 2Department of Anesthesiology and Critical Care, Zabol University of Medical Sciences, Zabol, IR, Iran 3Imam Reza Hospital, Mashhad University of Medical Sciences, Mashhad, IR, Iran 4Department of anesthesiology and critical care, Iran university of medical sciences, Tehran, IR, Iran


INTRODUCTION
Traumatic brain injury (TBI) is one of the world's most important causes of mortality and severe disability and imposes a devastating financial burden on communities 1 . This issue is exacerbated by the fact that most people suffering from TBI are young 2 . Mortality from TBI can be reduced by improving patient assessment and management systems 3 . In modern evidence-based medicine, access to reliable assessment scales is essential for predicting mortality risk and the ultimate outcome of intensive care so as to allow caregivers to determine the severity and outcome of patient condition and to make important health care decisions accordingly 4 . One of the scales used for such evaluations is the Glasgow Coma Scale (GCS). This scale was first designed in 1974 to assess the level of consciousness of patients with head injury 5 . GCS can predict the primary outcome of TBI (mortality and morbidity) and help healthcare professionals devise a model for care delivery [6][7][8] . However, some researchers have questioned the validity of this scale since it does not evaluate brainstem reflexes and cannot assess verbal responses in patients with intubation and tracheostomy [9][10][11] . To avoid these shortcomings, researchers have developed other scales for measuring consciousness in TBI patients. One of these scales is the FOUR (Full Outline of UnResponsiveness) score, developed by researchers at Mayo Clinic (United States), which consists of four components: eye responses, motor responses, brainstem reflexes, and respiration pattern 12,13 . Several studies have suggested that the FOUR scales can provide a better assessment of patients' final outcome than GCS 7,8,12 . However, these studies have reached this conclusion by comparing these two scales with each other and have not assessed the consistency of these two scales with the results of APACHE II or other major scales. Therefore, this study aimed to compare the ability of GCS and FOUR, versus other scales (Karnofsky, APACHE-II, etc.), to evaluate the level of consciousness and dysfunction in TBI patients.

METHODS
This descriptive cross-sectional study was performed with the authorization of the Department of Research and Technology of Zahedan University of Medical Sciences (Iran) and the approval of the Ethics Committee of the university (with Code IR.ZAUMS.REC.1396.33). Using the reports of previous studies and the sample size formula, the sample size was calculated as 102 14 . Sampling was performed using the convenience sampling method according to the defined inclusion criteria. Inclusion criteria were as follows: age of 18-65 years, hospitalization for head trauma, mechanical ventilation, and at least 24 hours of stay in the intensive care unit. Exclusion criteria were as follows: lack of consent of family members, the discovery of malignant and fatal diseases, cervical spinal cord injury, musculo-skeletal paralysis, deafness, blindness, and history of eye surgery resulting in a change in pupil size and shape. All eligible patients were monitored using GCS, FOUR, APACHE II, and Karnofsky scales until discharge or death. For sedated patients, the infusion was stopped one hour before measurements. The scores obtained from the scales were analyzed in SPSS using descriptive statistics, one-way ANOVA, and Pearson's correlation test. The FOUR scale consists of four components: eye responses, motor responses, brainstem reflexes, and respiration patterns (Figure 1). The validity and reliability of GCS and FOUR for Iranian patients were assessed in a 2011 study conducted in Tehran on 155 patients with 4 nurses and anesthesiology residents (including the author) in two groups of trained and untrained staff, which observed excellent inter-group agreement. In this study, the coefficient of correlation of FOUR scores in the trained and untrained groups were 0.998 and 0.993, respectively; the kappa coefficient of intra-group agreement in these groups were 0.981 and 0.986, respectively; as well, the coefficient of correlation between FOUR and GCS scores in these groups were 0.980 and 0.925, respectively 15 . APACHE II (Table 1) consists of 3 main dimensions: a physiological dimension, which includes GCS, temperature, mean arterial blood pressure, heart rate, respiratory rate, Pao2, arterial blood pH, serum sodium, potassium, creatinine, hematocrit, and white blood cell count. The score of each item in the physiological dimension ranges from 0 to 4, giving this dimension a total score of 0-59. The second part of this scale is a score between 0 and 6, which is awarded based on the patient's age group. The third part involves the evaluation of chronic diseases and the failures of one or more organs. The sum of scores obtained from these three parts is the APACHE II score, which ranges from 0 to 71. The higher the APACHE II score, the higher the patient's risk of mortality 16 . Karnofsky scale is used for assessing the patient's degree of dysfunction. This scale takes values between 0 (death) and 100 (no dysfunction), and reflect the patient's activity level; the lower the Karnofsky score, the more intensive care the patient requires 17,18 .

RESULTS
Of the 102 patients, 80 (78.4%) were male, and 22 (21.6%) were female. The mean age of patients was 32.2 ± 14.8 (Figure 2). The mean GCS score at admission was 8.6 ± 3.5, and at the end of the study was 12.9 ± 3.5. The mean GCS score during the hospitalization period was 10.8 ± 2.9. The mean FOUR score at admission was 9.3 ± 4.5 and at the end of the study was 14.8 ± 4.8. Also, the mean FOUR score during the hospitalization was 11.7 ± 3.9. The mean Karnofsky core was 23.9 ± 5.5, with the lowest score being 10 and the highest being 40. The mean APACHE II score at admission was 11.5 ± 6.7, with the lowest score being 1 and the highest being 29 (Figure 3). There was a significant correlation between the FOUR and GCS scores for the level of consciousness (r=0.925), which was statistically significant (p=0.0001). The mean scores of FOUR, GCS and Karnofsky scales were significantly higher in survivors and patients with neurological deficits than in patients who died. Patients who died also had a significantly higher APACHE II score than those who did not ( Table 2). A high degree of positive correlation was observed between assessments by FOUR and GCS for patient outcome (healthy, neurological deficits, or death). Also, both of these scales showed very high negative correlation with APACHE II. The correlation between Karnofsky and APACHE II scores in healthy patients was r = -0.457; in patients with neurological deficits r = -0.481, and in deceased patients, was r = -0.796. These correlations were statistically significant in all cases except for patients with neurological deficits ( Table 3). Both GCS and FOUR were as capable as APACHE II in predicting the outcomes. The highest agreement between the scales was in the assessment of mortality rates, which was slightly higher between GCS and APACHE II.

DISCUSSION
Access to reliable and uniform scales for assessing a patient's level of consciousness can help caregivers provide a better assessment of patient condition 8 . The results of this study showed that GCS and FOUR are perfectly consistent in the assessment of patients' level of consciousness. Most of the previous studies have reported the better ability of the FOUR scale in assessing consciousness and have suggested that GCS is unable to measure brainstem reflexes, is inconsistent in measuring verbal reflexes in confused patients without consciousness disorder, and is unable to give reliable results for patients with intubation and tracheostomy 1,7,8,19 . However, the results of this study showed no significant difference between the initial score obtained from these two scales. There was also no significant difference between the two scales in the assessment of consciousness and outcome. In other words, GCS was as capable as FOUR in predicting patient outcomes. Considering the ease of use of GCS and the fact that it can be used by all healthcare team members and does not need any special measurement equipment, GCS can still serve as a good scale for measuring consciousness and outcome.
Contrary to this finding, Keykha et al. reported that the FOUR scale is a better measure of consciousness level and should be used instead of GCS when measuring consciousness in ICU patients 8 . This difference in the results can be attributed to the fact that the above study was performed on patients receiving continuous sedation infusion, whereas sedation of the patients in this study was stopped one hour before measurements. Stopping the sedation infusion before consciousness measurements have the benefit of ensuring that the patient's level of consciousness is not physically reduced. To avoid high levels of reduced consciousness, it is best to use on-off infusions, because sedative drugs may accumulate in the fat tissue, and their sedative effects can remain for as long as 24 hours after stopping infusion 20 . In cases where continuous sedation is absolutely needed for pain control or tolerance of mechanical ventilation, although the scores of both scales would drop, the FOUR scale appears to be more suitable for consciousness assessment. In a study by Jalali, it was reported that the FOUR scale was better in assessing consciousness and predicting mortality than GCS, but the low score of neither scale was associated with mortality because deceased patients had a higher mean score in both scales 1

CONCLUSIONS
The results of this study showed that GCS and FOUR are both very good scales for measuring the consciousness level and outcome of TBI patients and that neither scale can be considered preferable over the other in measuring the consciousness level. Since many healthcare professionals with varying degrees of expertise participate in the treatment process, the instruments used to evaluate patients' condition should be able to provide a uniform assessment of consciousness level and outcomes when applied by different people. Therefore, it is best to choose the right instrument according to the type of department and the skill of healthcare professionals who are going to employ it.

LIMITATIONS
The limitations of this study included the absence of any measurement during sedation and the small size of the sample. The authors recommend conducting a similar study with a greater number of subjects to confirm the findings.