Introduction

Head injury is associated with tremendous morbidity and mortality. The CRASH Trial which included data from 46 countries showed that head-injury is one of the leading causes of death and there was a significant difference in the outcome of head injury between the high income and low-middle income countries (1). In the UK, 1% of all deaths were attributed to head injury and up to 85% of all severely head injured patients have remained disabled after 1 year (2).

In Trinidad & Tobago, 1% of admissions to Government hospitals and 2 out of 1000 deaths were due to head injuries (3). Trauma accounts for the majority of head injured patients admitted to tertiary care institutes and when severe, requires mechanical ventilation in an ICU environment. The overall mortality in severe traumatic brain injury, defined as a post resuscitation Glasgow Coma Scale score of ? 8, is 23 % (4). Traumatic brain injury has a devastating financial, emotional, and social impact on survivors left with lifelong disability as well as their families (5).

Primary injury to the brain is irreversible; thus management of the head injured patient involves minimizing secondary brain injury. Hyperventilation has been a controversial, but fundamental principle in the management of these patients. Hyperventilation has been employed in the head injured patient due to the physiological mechanism of promoting cerebral vasoconstriction and thus reducing cerebral blood volume in an attempt to decrease intracranial pressure (6). However in recent times, studies have shown that prophylactic hyperventilation may do more harm than good.

Some studies have shown that hyperventilation in the first 24 hours following head injury may cause reduction in cerebral blood flow at a time when flow is already decreased, compounding cerebral ischemia (7). While there is evidence to support the reduction in cerebral blood flow and increased areas of ischemic brain following hyperventilation, few studies report the effect of hyperventilation on clinical outcome.

Although hyperventilation has remained a component of the management of the head injured patient for decades, it is now recommended to reserve its use for the patient with signs of impending brain herniation and to not employ hyperventilation routinely (8).

Hyperventilation is still practiced routinely in the management of the head injured patient in Trinidad and Tobago. To our knowledge, there has been no published paper from the Caribbean evaluating the effect of hyperventilation on clinical outcome of head injury patients. With this background, this study was conducted to determine the effect of hyperventilation on the outcome of brain injury in an adult ICU in Trinidad.

Methods

A retrospective observational study was conducted to determine the pattern of hyperventilation, range of PaCO2, and their relationship to the outcome of patients with severe brain injury at the Intensive Care Unit, Port-of-Spain General Hospital, Trinidad. Approval was sought from the University Ethics Committee, the Institutional Review Board of Port-of-Spain General Hospital and the Clinical Director of the ICU. Informed consent from individual patient was waived due to the retrospective observational nature of the study. A sample size was calculated prior to the study using the incidence of head injury in Trinidad & Tobago, which demonstrated that at least 100 patients should be included in the study for a power of 0.8.
All adult patients with severe brain injury who were admitted to the ICU at Port-of-Spain General Hospital (POSGH) for mechanical ventilation during the ten year period from 1998 through 2008 were included. Patients who died within 6 hours of ICU admission were excluded.
Demographic data included age, sex, ethnicity and history of substance abuse. Glasgow Coma Scale score on admission, etiology of traumatic brain injury and radiographic diagnosis by CT scan were recorded. The proposed plan for hyperventilation with respect to the duration and PaCO2 target levels was also noted. The serial PaCO2 levels were recorded for a period of 72 hours following admission from the arterial blood gas (ABG) reports.
Serial PaO2 levels, intracranial pressure, and the use of mannitol or intravenous dextrose were recorded. The length of stay in ICU, Glasgow Coma Scale score on discharge and patient outcome were also measured.
The patients were categorized into two groups for purpose of analysis: the first group included patients with mean PaCO2 level ? 30 mmHg and the second group included patients with mean PaCO2 level < 30 mmHg.
Descriptive analyses were performed for data such as the patients

Results

197 patients were enrolled into the study, of which 164 (83.2%) were male and 33 (16.8%) were female. The age of patients ranged from 13 to 87 and the median age was 39 years (interquartile ranges 26 – 53.3). The length of stay (LOS) ranged from 1 to 58 days and the median LOS was 9 days (interquartile ranges 5 – 15). The mean ICP was 20.3 [14.5 standard deviation (SD)] and the mean admission GCS was 6.3 (3.3). Table 1 shows the overall distribution of the age, length of stay, ICP, admission and discharge GCS.

Table 1

Age, Length of stay, Intracranial pressure and Glasgow Coma Score (GCS)

51% of the patients belonged to the East Indian ethnic descent, 36% were African ethnicity, 8.8% mixed origin, 3.4% Caucasian and 0.8% Chinese. Substance abuse was present in 28.4% of the population and alcohol, tobacco, marijuana and cocaine were the agents being used. 15.2% of these patients were alcohol users.

The various causes of brain injury included motor vehicle accidents (MVA), fall, assault, gun shot wound (GSW) and others. CT scan diagnoses regarding the type of head injury are shown in Table 2. Mannitol was given in 18.3% of the patients and intravenous glucose containing solutions was given in 40%.

Table 2

Diagnosis on admission

The patterns and period of hyperventilation varied among patients; 35% were hyperventilated for 48 hours, 15.7% for 24 hours, 6.1% for 72 hours and 0.5% for 36 hours. No hyperventilation was carried out for 17.8% of patients and the duration of proposed hyperventilation was unspecified for 17.2%. Hyperventilation was performed when intracranial pressure exceeded 20 mmHg in 7.6% of patients. The PaCO2 values ranged from 13.8 to 54.4 mmHg with a mean value of 32.3 [5.7, SD, (standard deviation)]. The planned target levels of PaCO2 and the achieved mean levels of PaCO2 are shown in Table 3.

Table 3

Planned ventilation pattern and achieved PaCO2

The mortality in this study was 38.6%. Multivariate tests were performed on variables including age, length of stay, GCS upon admission and discharge, mean PaCO2, mean ICP and PaO2 to determine their influence on the outcome. None of these variables were found to significantly affect outcome except the discharge GCS. This is shown in Table 4.

Table 4

Multivariate analysis for factors influencing outcome

Although there was a higher mortality in patients with a mean PaCO2 < 30 mmHg (46.8%), compared to those with a PaCO2 ≥30 mmHg (33.6%), this difference was not statistically significant by Chi squared analysis (Chi square=3.4, df:1, p=0.06).

Figure 1 shows the Receiver Operating Characteristic (ROC) Curves for the two variables: mean CO2 and GCS admission score. Area under the curve for Mean CO2 was 0.43, (95% Confidence Interval (CI): (0.34, 0.52)) and that for the GCS on admission was 0.36, (95% CI: 0.28, 0.46). The areas under the curves for both variables were < 0.5 implying that neither variable had a good discriminant ability to be used as predictors of outcome.

Figure 1

Receiver Operating Characteristic (ROC) curve for PaCO2 and admission Glasgow Coma Score (GCS)

Figure 2 shows the Kaplan-Meier survival curves for a duration of 30-days following ICU admission. Although the group with CO2 > 30 mmHg had a higher survival compared to the group with CO2 < 30 mmHg, the difference between groups was not statistically significant (Log-rank, p=0.08).

Figure 2

Kaplan Meier survival curves for two levels of PaCO2

Discussion

The major finding of the study is that deliberate hyperventilation did not influence the ICU outcome of brain injury patients.

The mean age of patients in this study was 41.3, a little higher compared to a study by Muizelaar et al., but very similar to a study published from Australia and New Zealand (7, 9). Patients with traumatic brain injury are often male young adults (9, 10). There was male preponderance in the present study also.

The mean ICU length of stay was 10.5 days in the present study, which was longer than the 3 days reported in a previous study (11). Patients in the previous study were discharged from the ICU to a high dependency unit (HDU). The study hospital does not have a HDU; hence these patients were retained in the ICU for longer time.

The mean GCS on admission was 6.3 in the present study, which is similar to a report by Schreiber et al. (12). This mean GCS score would be classified as severe head injury according to Thornhill et al. (13). These authors reported that severe head injury is associated with death or vegetative state in at least 38% of patients, which is comparable to the results of our study.

The mean PaO₂ in our study was 136.4 mmHg. In the study by Marion et al., the mean PaO₂ was 149 mmHg (14). The mean PaO₂ levels were calculated for the entire ICU stay in the present study, and not only during the hyperventilated period, which may explain the lower PaO₂ level.In a trial by Thiagaran et al., hyperoxygenation was shown to improve SjO₂ during hyperventilation, however such improvements were seen only when PaO₂ were in the range of 200-250 mmHg (15).

The mean PaCO₂ in the present study was 32.3 mmHg. Studies by Muizelaar et al., Marion et al., and Coles et al., reported much lower levels of mean PaCO₂ (26.1-29 mmHg) (7, 14, 16). These studies were prospective in nature and therefore the targets for hypocapnia were lower than that of the present study. No target levels for hypocapnia were dictated by the study.

Hyperventilation remains part of the recommended management for patients with raised intracranial pressure; however, it should not be performed prophylactically and should only be used with intracranial pressure monitoring (8). In the study ICU, hyperventilation was used in most patients with TBI.

The study by Muizelaar et al., compared groups of patients who were hypocapnic and normocapnic and found that in the normocapnia group, patient outcome was better after 3 and 6 months following head injury (7). Our study did not evaluate long-term outcomes. Nevertheless, findings of both the studies may imply that normocapnia should suffice while treating brain injury patients without the need for hypocapnia by deliberate hyperventilation. Furthermore, the effect of hypocapnia on ICP is not sustained and may only last for ~ 6-18 hours with subsequent hyperemia (17).Several studies have shown that hyperventilation induces deleterious effects on the brain (14, 16, 18). Although the present study did not clearly establish the adverse impact of hyperventilation, it showed that hyperventilation may not necessary for better outcome.

Our study found that age, length of ICU stay, GCS admission score, mean PaCO₂, mean ICP and PaO₂ did not affect patient outcome. Additionally, the present study also showed that the admission GCS score was a poor discriminator of the ICU outcome as shown by the ROC analysis (Figure 1). However, other studies have shown that a low GCS in association with advanced age were predictors of poor prognosis (12, 19).

There are some limitations to this study. Due to the retrospective nature of the study, not all data points were retrievable, which limited the number of patients. Infrequent sampling for blood gas analysis, inadequate calibration and machine errors might have also affected the stated levels of PaCO₂. Additionally, only a limited number of patients had ICP monitoring.

In summary, the study demonstrated that hyperventilation may not improve the ICU outcome of the head injured patients, and confirmed that deliberate hyperventilation need not be performed routinely in traumatic brain injury patients in the ICU.