|Home | About | Journals | Submit | Contact Us | Français|
Head injury in young adults is often associated with motor vehicle accidents, violence, and sports injuries. In older adults it is often associated with falls. Severe head injury can lead to secondary brain damage from cerebral ischaemia resulting from hypotension, hypercapnia, and raised intracranial pressure. Severity of brain injury is assessed using the GCS. While about a quarter of people with severe brain injury (GCS score less than 8) will make a good recovery, about a third will die, and a fifth will have severe disability or be in a vegetative state.
We conducted a systematic review and aimed to answer the following clinical question: What are the effects of interventions to reduce complications of moderate to severe head injury as defined by Glasgow Coma Scale? We searched: Medline, Embase, The Cochrane Library and other important databases up to April 2007 (BMJ Clinical Evidence reviews are updated periodically, please check our website for the most up-to-date version of this review). We included harms alerts from relevant organisations such as the US Food and Drug Administration (FDA) and the UK Medicines and Healthcare products Regulatory Agency (MHRA).
We found 17 systematic reviews, RCTs, or observational studies that met our inclusion criteria.
In this systematic review we present information relating to the effectiveness and safety of the following interventions: antibiotics, anticonvulsants, corticosteroids, hyperventilation, hypothermia, and mannitol.
Head injury in young adults is often associated with motor vehicle accidents, violence, and sports injuries. In older adults it is often associated with falls. This review covers moderate to severe head injury only.
There is no strong evidence of benefit from any treatment in reducing the complications of moderate to severe head injury. Despite this, most clinicians implement various combinations of treatments discussed here.
Hyperventilation and mannitol are frequently used to lower intracranial pressure. Anticonvulsants, barbiturates, antibiotics, and hypothermia are less commonly implemented.
CAUTION: Corticosteroids have been shown to increase mortality when used acutely in people with head injury.
The basic operational components of a head injury are a history of blunt or penetrating trauma to the head, which may be followed by a period of altered consciousness, and the presence of physical evidence of trauma. The specific elements of a head injury are related to its severity. Some guidelines define head injury more broadly as any trauma to the head other than superficial injuries to the face. Head injuries are classified in a variety of ways: severity of injury as assessed by the Glasgow Coma Scale (GCS; mild, moderate, severe); mechanism (blunt or penetrating); or morphology (skull fractures or intracranial lesions). Since its introduction in 1974, the GCS has been widely used as an initial measure of the severity of brain injury. The scale incorporates neurological findings such as voluntary movements, speech, and eye movements, into a 3-15 point scale. GCS allows measurement of neurological findings, and has been used to predict immediate and long-term outcome after head injury. A GCS of 8 or lower is considered representative of a severe brain injury, 9-13 of a moderate head injury, and 14-15 a mild head injury. The GCS is complicated by difficulties of communication and cooperation in the younger child. In children over the age of 5 years, the adult GCS can be used. In younger children the verbal response is modified, and in very young children the motor response is also modified because these children are unable to obey commands. In this review we cover only moderate to severe head injury as classified by GCS. Diagnosis and monitoring: The Advanced Trauma Life Support (ATLS) and Advanced Paediatric Life Support (APLS) guidelines contain standardised protocols for the initial assessment of traumatic head-injured adults and children, respectively. Most moderate to severe head injuries will require investigations after standard history and physical examination. Computed tomography (CT) scan is the investigation of choice in people with traumatic head injuries. Numerous organisations, including the National Institute for Health and Clinical Excellence, the Scottish Intercollegiate Guidelines Network, and the Royal College of Paediatrics and Child Health, have developed evidence-based pathways to provide physicians with guidance regarding whether a CT scan is required, and how urgently it should be performed. Monitoring of people with head injury may range from monitoring of intracranial pressure (ICP) with ventricular drains in people with severe head injuries to regular clinical neurological observations in people with less severe head injuries.
Head injury remains the leading cause of death in trauma cases in Europe and the USA, and accounts for a disproportionate amount of morbidity in trauma survivors. Worldwide, several million people, mostly children and young adults, are treated each year for severe head injury. In the UK, 1.4 million people, 50% of whom are children, present to emergency departments every year after a head injury. This represents 11% of all new emergency department presentations. About 80% of people presenting to emergency departments can be categorised as having mild head injury, 10% as moderate, and 10% as severe.
The main causes of head injury include injuries incurred from motor vehicle accidents, falls, acts of violence, and sports injuries. Motor vehicle crashes account for most fatal and severe head injuries. Young adults (15-35 years old) are the most commonly affected group, reflecting increased risk-taking behaviour. A second peak occurs in the elderly (more than 70 years old), related to an increased frequency of falls. For most age groups, with the exception of extremes of age, there is a 2:1 male predominance. Severe head injury marks the beginning of a continuing encephalopathic process — secondary brain damage from ongoing cerebral ischaemia closely linked to factors such as hypotension, hypercapnia, and elevated ICP is a potential cause of morbidity and mortality.
Head injury can result in death or a lifelong impairment in physical, cognitive, and psychosocial functioning. Several factors have been shown to correlate with poor outcome — including low post-resuscitation GCS score, older age, eye pupil abnormalities, hypoxia or hypotension before definitive treatment, traumatic subarachnoid haemorrhage, and inability to control ICP. Data from the Traumatic Coma Data Bank found that people with an initial GCS score of 3 had 78% mortality, whereas those with a GCS score of 8 had 11% mortality. Overall, prognoses for people with severe head injury (GCS score 3-8) were: good recovery 27%, moderate disability 16%, severe disability 16%, vegetative 5%, and mortality 36%. Despite such data, the role of GCS in determining prognosis in head injury remains controversial. The impacts of head injury range from mild cognitive and psychosocial changes to severe physical disability and cognitive and sensory losses.
To reduce mortality and disability (neurological and other) from head injury; to reduce secondary physiological complications of head injury such as hypercapnia and intracranial hypertension; to reduce secondary clinical complications such as seizures and central-nervous-system infections; to reduce length of hospital stay time; to maximise chances of full recovery (moderate to good recovery according to GCS score), with a minimum of adverse effects of treatment.
All-cause mortality, death or severe neurological disability (according to GCS or other standardised functional scale, including psychological sequelae), seizures, mean ICP, mean arterial pressure, infection, adverse effects of treatment.
BMJ Clinical Evidence search and appraisal April 2007. For this review, the following sources were used for the identification of studies: Medline 1966 to April 2007, Embase 1980 to April 2007 and The Cochrane Library Issue 1, 2007. Additional searches were carried out using these websites: NHS Centre for Reviews and Dissemination (CRD) — for Database of Abstracts of Reviews of Effects (DARE), Health Technology Assessment (HTA), Turning Research into Practice (TRIP), and the National Institute for Health and Clinical Excellence (NICE). We also searched for retractions of studies included in the Review. Abstracts of studies retrieved in the search were assessed by an information specialist. Selected studies were then sent to the author for additional assessment, using pre-determined criteria to evaluate relevant studies. Study design criteria for inclusion in this review were: published systematic reviews and RCTs in any language, at least single blinded and containing more than 20 individuals of whom more than 80% were followed up. There was no minimum length of follow-up required to include studies. We excluded all studies described as "open", "open label", or not blinded unless blinding was impossible. In addition, we also use a regular surveillance protocol to capture harms alerts from organisations such as the US Food and Drug Administration (FDA) and the UK Medicines and Healthcare products Regulatory Agency (MHRA), which are added to the review as required. We have performed a GRADE evaluation of the quality of evidence for interventions included in this review (see table ).
The information contained in this publication is intended for medical professionals. Categories presented in Clinical Evidence indicate a judgement about the strength of the evidence available to our contributors prior to publication and the relevant importance of benefit and harms. We rely on our contributors to confirm the accuracy of the information presented and to adhere to describe accepted practices. Readers should be aware that professionals in the field may have different opinions. Because of this and regular advances in medical research we strongly recommend that readers' independently verify specified treatments and drugs including manufacturers' guidance. Also, the categories do not indicate whether a particular treatment is generally appropriate or whether it is suitable for a particular individual. Ultimately it is the readers' responsibility to make their own professional judgements, so to appropriately advise and treat their patients.To the fullest extent permitted by law, BMJ Publishing Group Limited and its editors are not responsible for any losses, injury or damage caused to any person or property (including under contract, by negligence, products liability or otherwise) whether they be direct or indirect, special, incidental or consequential, resulting from the application of the information in this publication.
Ian Maconochie, Department of Paediatric Accident and Emergency, St Mary's Hospital, London, UK.
Mark Ross, Department of Paediatric Accident and Emergency, St Mary's Hospital, London, UK.
SYMPTOM SEVERITY Compared with placebo: Antibiotics do not reduce the risk of meningitis in people with head injury compared with placebo ( moderate-quality evidence ). MORTALITY Compared with placebo: Antibiotics do not reduce the rate of mortality in people with head injury compared with placebo (moderate-quality evidence).
We found one systematic review (search date 2005, 4 RCTs, 17 observational studies, 2376 people with base of skull fractures) comparing antibiotics versus placebo. All studies assessed meningitis as their primary outcome. Meta-analysis of the RCTs identified by the review found no significant difference in all-cause mortality, meningitis-related mortality, or meningitis between antibiotics and placebo (4 RCTs, 208 people; all-cause mortality: 5/109 [5%] with antibiotics v 3/99 [3%] with placebo; OR 1.68, CI 0.41 to 6.95; meningitis-related mortality: 1/109 [1%] with antibiotics v 1/99 [1%] with placebo; OR 1.03, CI 0.14 to 7.40; meningitis: 10/109 [9%] with antibiotics v 14/99 [14%] with placebo; OR 0.69, CI 0.29 to 1.21). Trial duration and time to assessment of outcome were not reported by the review. Similar reductions in rates of meningitis were found in the subgroup of people with cerebrospinal fluid leakage. The review also meta-analysed results for the retrospective controlled studies it identified (2168 people with basal skull fracture) and found similar results.
No new evidence
MORTALITY Compared with control: We don't know whether hyperventilation alone or in combination with a buffer may reduce rates of mortality or improve neurological recovery in adults and children with intracranial lesions compared with normal ventilation ( very low-quality evidence ). NOTE Hyperventilation may worsen cerebral ischaemia by increasing cerebral tissue concentrations of toxic metabolites.
We found one systematic review (search date 2005), which identified one RCT of sufficient quality (113 people aged 3 years or more with intracranial lesions, Glasgow Coma Score [GCS] score lower than 8) comparing three interventions: hyperventilation alone; hyperventilation plus buffer (tris-hydroxy-methyl-amino methane [THAM]); or normal ventilation. All participants also received intracranial pressure-lowering agents such as mannitol, and barbiturates. The RCT found that, compared with normal ventilation, hyperventilation alone or combined with a buffer significantly reduced mortality at 1 year (9/36 [25%] with hyperventilation alone v 14/41 [34%] with normal ventilation; RR 0.73, CI 0.36 to 1.49; 11/36 [31%] with hyperventilation plus buffer v 14/41 [34%] with normal ventilation; RR 0.89, CI 0.47 to 1.72). However, this reduction in mortality did not correlate with an improvement in neurological recovery: there was no significant difference between groups in the combined outcome of death or disability (25/36 [69%] with hyperventilation alone v 25/41 [61%] with normal ventilation; RR 1.14, 95% CI 0.82 to 1.58; 19/36 [53%] with hyperventilation plus buffer v 25/41 [61%] with normal ventilation; RR 0.87, 95% CI 0.58 to 1.28). People receiving hyperventilation with higher GCS scores (4–5), suggesting a better initial prognosis, did significantly worse (P = 0.05) at 3 and 6 month follow-up than did other subgroups. The RCT had some weaknesses in its methods; it was not double blind, and randomisation was compromised early in the RCT, because people in whom informed consent could not be obtained were automatically assigned to control. This practice was stopped as soon as the authors became aware of it. One prospective cohort study (20 adults with GCS score lower than 8, mean age 38.8 years) was identified in the “awaiting assessment” section of the systematic review. The study only assessed harms (see below).
The systematic review gave no information on adverse effects. The prospective cohort study (20 adults with severe head injury) evaluated potential adverse effects of hyperventilation. Assessments of the effects of 30 minutes of hyperventilation on extracellular metabolites associated with cerebral ischaemia and on local cerebral blood flow were done 24–36 hours and 3–4 days after injury. The study found that hyperventilation increased extracellular glutamate at 24–36 hours (at least 10% increase in 14/20 [70%] people; P less than 0.05) and lactate (at least 10% increase in 7/20 [35%] people; P less than 0.05), and decreased local cerebral blood flow (at least 10% decrease in 5/20 [25%] people; significance not reported). Results at 3–4 days were similar.
No new evidence
MORTALITY Compared with normothermia: We don't know whether hypothermia reduces the rate of mortality at 6 months to 2 years in people with head injury compared with normothermia ( moderate-quality evidence ). SYMPTOM SEVERITY Compared with normothermia: We don't know whether hypothermia improves neurological outcomes and cerebral perfusion, and reduces intracranial pressure in people with head injuries, compared with normothermia ( low-quality evidence ). NOTE Immediate hypothermia has been associated with increased rates of pneumonia, pulmonary complications, and thrombocytopenia.
We found one systematic review (search date 2003, 14 RCTs, 1094 people, primarily adults, with Glasgow Coma Scale [GCS] score 8 or lower, mechanism and morphology of injury unclear) and six subsequent RCTs. The review compared mild therapeutic hypothermia (34–35 °C) for at least 12 hours versus normothermia. In one of the RCTs, people were included only if intracranial pressure (ICP) was uncontrollable using fluid resuscitation, hyperventilation, and high-dose barbiturates. In the other RCTs, it is unclear whether drug treatment was also given to participants. Results were analysed separately for immediate hypothermia (13 RCTs) and delayed hypothermia (1 RCT). The review found no significant difference between immediate hypothermia and normothermia in mortality, or in the proportion of people who were dead or severely disabled 12 months after the end of treatment, although fewer people having immediate hypothermia were dead or disabled. The single RCT of delayed hypothermia identified by the review (33 people, age range not reported) found no significant difference in mortality between groups at 6 months, although fewer people having delayed hypothermia died. However, it found that, compared with normothermia, delayed hypothermia significantly reduced the proportion of people who had a poor outcome at 6 months. The first subsequent RCT (396 people) compared immediate (within 24 hours of injury) hypothermia (32–35 °C) versus normothermia for 1–7 days. It found that, compared with normothermia, immediate hypothermia significantly reduced mortality and improved the proportion of people with a good neurological outcome at the end of treatment. ICP was significantly lower in the hypothermia group at 24 hours, 3 days, and 1 week compared with the normothermia group. The second subsequent RCT (86 people) compared mild-moderate hypothermia (33–35 °C) immediately after hospital admission or 3–5 days after craniotomy versus normothermia. It found a significant reduction in mortality and a significant improvement in the proportion of people who had mild or no disability at 2 years with hypothermia compared with normothermia. The third subsequent RCT (30 adults) compared immediate hypothermia (34 °C, within 15 hours of head injury) versus normothermia. The RCT found no significant difference in good neurological outcome (GCS 4 or 5) at 6 months between hypothermia and normothermia. Hypothermia significantly decreased ICP and increased cerebral perfusion pressure. However, these observations are of questionable clinical value because they were not compared with parameters obtained from people receiving normothermia. The fourth subsequent RCT (48 children) compared immediate hypothermia (32–33 °C) versus normothermia for 48 hours. The RCT found no significant difference in mortality at 5 days between immediate hypothermia and normothermia, although fewer children receiving immediate hypothermia died. The fifth and sixth subsequent RCTs compared three interventions: immediate selective brain hypothermia, systemic hypothermia, and normothermia for 3 days. Both RCTs found that hypothermia of either type significantly increased the proportion of people with a good neurological outcome, and one found that hypothermia also reduced mortality. See table 1 for full results .
The review found that, compared with normothermia, immediate hypothermia significantly increased rates of pneumonia. It should be noted that the largest RCT of immediate hypothermia in the review did not report data on pulmonary infections, and so was not included in the meta-analysis. The reporting of other complications in the review was variable, so meta-analysis was not possible. Four subsequent RCTs found that, compared with normothermia, hypothermia significantly increased complications (pulmonary infection, thrombocytopenia, cerebral perfusion pressure). In one of the RCTs, all platelet counts during both hypothermia treatments normalised within 3 days of hypothermia cessation. The fifth subsequent RCT gave no information on adverse effects. The sixth subsequent RCT, conducted in children, found no significant difference in temperature deviation, infection, arrhythmia, or coagulopathy at 5 days between immediate hypothermia and normothermia. See table 1 for full results .
MORTALITY Compared with placebo or hypertonic saline: Mannitol does not reduce mortality rates at 3 months in people with moderate to severe intracranial lesions compared with placebo or hypertonic saline ( moderate-quality evidence ). Compared with barbiturates: Mannitol may not reduce mortality rates at 3 months in people with severe head injuries and raised intracranial pressure compared with phenobarbitone ( low-quality evidence ).
We found one systematic review (search date 2006, 4 RCTs, 120 adults with moderate to severe intracranial lesions, Glasgow Coma Score [GCS] score 8 or lower in most RCTs) comparing mannitol versus placebo, hypertonic saline, or phenobarbitone.
The review identified two RCTs. The first RCT (41 adults with moderate to severe head injury) compared mannitol given before hospital admission versus placebo. It found no significant difference in mortality at 3 months between mannitol and placebo (5/20 [25%] with mannitol v 3/21 [14%] with placebo; RR 1.75, CI 0.48 to 6.38). The width of the confidence interval suggests that the RCT is likely to have been underpowered to detect a clinically important difference between groups. The second RCT (20 people with head injury and persistent coma requiring ongoing treatment for refractory intracranial hypertension) identified by the review compared mannitol versus hypertonic saline. It also found no significant difference between mannitol and hypertonic saline in mortality at 3 months (5/10 [50%] with mannitol v 4/10 [40%] with placebo; RR 1.25, 95% CI 0.47 to 3.33). The RCT was too small to draw reliable conclusions.
The review identified one RCT (59 adults with severe head injury, GCS score less than 8, and raised intracranial pressure [ICP]) comparing mannitol versus phenobarbitone. It found no significant difference in mortality at 3 months between mannitol and phenobarbitone (15/31 [48%] with mannitol v 16/28 [57%] with phenobarbitone; RR 0.85, 95% CI 0.52 to 1.38). The RCT is likely to have been underpowered to detect a clinically important difference between groups. In the RCT, some participants later received the alternative treatment if the allocated treatment did not control ICP. It is unclear from the review whether the analysis was by intention to treat.
The review gave no information on adverse effects. Physiological and animal studies have found an increased risk of acute renal failure with large doses of mannitol. Multiple doses of mannitol may accumulate in the brain, causing a reverse osmotic shift and raising brain osmolarity, thus theoretically increasing ICP.
There were very few adequate RCTs of mannitol in people with moderate to severe head injury. The review excluded three RCTs that compared high- versus low-dose mannitol, because of ongoing concerns about the validity of the original data they report. Current clinical use of mannitol is based on its pharmacological mode of action, and on observational human and animal studies that show a potential beneficial effect.
No new evidence
MORTALITY Antiepileptic drugs compared with placebo: Antiepileptic drugs may not decrease mortality rates at 3 months to 2 years in people with head injuries compared with placebo ( very low-quality evidence ). Barbiturates compared with placebo: Barbiturates may not reduce mortality rates in people with head injuries compared with placebo ( low-quality evidence ). Barbiturates compared with mannitol: Barbiturates may not reduce mortality rates at 3 months in people with severe head injuries and raised intracranial pressure compared with mannitol (low-quality evidence). SYMPTOM SEVERITY Antiepileptic drugs compared with placebo: Antiepileptic drugs may reduce reducing early seizures, but may not reduce post-traumatic seizures at 48 hours, or improve neurological outcomes at 30 days to 2 years compared with placebo (very low-quality evidence). Barbiturates compared with placebo: Barbiturates may not reduce adverse neurological outcomes and intracranial pressure in people with head injuries compared with placebo (low-quality evidence).
We found one systematic review and one subsequent RCT. The review (search date 2002, 6 RCTs, 1218 adults and children; severity, mechanism, and morphology of head injury unclear) that compared antiepileptic drugs or barbiturates versus placebo. The RCTs identified by the review assessed phenytoin (4 RCTs, 903 people), carbamazepine (1 RCT, 151 people), and pentobarbital (1 RCT, 164 people) commenced within 8 weeks of acute head injury. The RCTs had several weaknesses in methods: in the largest trial of phenytoin (586 people), 50% of people withdrew, and therapeutic drug levels were often not attained or even reported in some RCTs. The RCT of the barbiturate phenobarbital was not included in most of the meta-analyses. The RCTs found no significant difference in mortality over 2 years between antiepileptic drugs and placebo (5 RCTs: 95/540 [18%] with antiepileptic drugs v 78/514 [15%] with placebo; RR 1.15, CI 0.89 to 1.51). One RCT of phenytoin and one of carbamazepine assessed the combined outcome of death or neurological disability, and found no significant difference between groups at 2 years (RR 0.96, CI 0.72 to 1.26 for phenytoin v placebo; RR 1.49, CI 1.06 to 2.08 for carbamazepine v placebo). The review found that carbamazepine or phenytoin significantly reduced early seizure (within 1 week) compared with placebo (4 RCTs: 22/456 [5%] with antiepileptics v 65/434 [15%] with placebo; RR 0.34, 95% CI 0.21 to 0.54). The subsequent RCT (102 children aged 16 years or less with Glasgow Coma Score [GCS] score 10 or lower) compared phenytoin versus placebo. The primary outcome assessed was incidence of post-traumatic seizures within 48 hours. It found no significant difference between phenytoin and placebo in the proportion of children with seizures (3/46 [7%] with phenytoin v 3/56 [5%] with placebo; mean difference –0.015, 95% CI –0.127 to +0.091; mean increase +1.5%, 95% CI –9.1% to +12.7%). The RCT also found no significant difference at 30 days between groups in the secondary outcomes of mortality or neurological outcome. However, the RCT was underpowered to detect clinically important differences between groups — the study authors had legal difficulties with waiving consent, and so were unable to enrol the required number of participants.
We found one systematic review (search date 2005, 3 RCTs, 208 adults and children, GCS score 7 or lower, mechanism and morphology of head injury unclear), which compared barbiturates (phenobarbitol or pentobarbital) or barbiturates plus an anaesthetic (pentobarbital plus etomidate) versus no barbiturates. The review found no significant difference in mortality between barbiturates and no barbiturates (48/105 [46%] with barbiturates v 43/103 [42%] with placebo; RR 1.09, 95% CI 0.81 to 1.47). It also found no significant difference in the proportion of people with an adverse neurological outcome (death, vegetative state, or severe disability) between phenobarbitone or pentobarbital and placebo (34/68 [50%] with barbiturates v 29/67 [43%] with placebo; RR 1.15, 95% CI 0.81 to 1.64). Two RCTs (126 adults with severe head injury) examined the effect of barbiturates on intracranial pressure (ICP) and found no significant difference between barbiturates and placebo, although results were better in people taking barbiturates (first RCT, people with uncontrolled ICP: 25/37 [68%] with barbiturates v 30/36 [83%] with placebo; RR for uncontrolled ICP 0.81, 95% CI 0.62 to 1.06; second RCT: WMD in ICP –1.00, 95% CI –7.77 to +5.77). Trial duration and time to assessment of outcome were not reported by the review.
See benefits of mannitol.
The review gave little information on adverse effects. One RCT (586 people) identified by the review found that antiepileptic drugs were associated with negative cognitive effects (no further data reported). Two RCTs identified by the review found that phenytoin significantly increased the incidence of skin rashes compared with placebo (30/292 [10%] with phenytoin v 18/276 [7%] with placebo; RR 1.57, 95% CI 0.90 to 2.75). One additional RCT (80 people) compared discontinuation versus continuation of phenytoin or carbamazepine, in people recovering from head injury who had received either drug for the previous 6 months to 3.5 years. The RCT assessed neurological adverse effects, and found that people had significantly improved performance in motor and speed tasks and reduced anxiety when either drug was discontinued (absolute numbers tabulated; P less than 0.05 for all outcomes). However, practice and acquired learning could not be excluded as causative factors of these apparent improvements in performance upon drug withdrawal.
The review found that barbiturates significantly increased hypotension compared with placebo (2 RCTs: 37/64 [60%] with barbiturates v 20/62 [32%] with placebo; RR 1.80, CI 1.19 to 2.70). One RCT (53 people) identified by the review found that mean body temperature was significantly lower in people taking barbiturates compared with placebo (WMD –3.20, CI –4.66 to –1.74).
See harms of mannitol.
The hypotensive effect of barbiturates is likely to offset the beneficial effect on cerebral perfusion pressure, or of any barbiturate-related reduction in ICP.
No new evidence
MORTALITY Compared with placebo: Corticosteroids increase mortality rates at 2 weeks to 6 months in adults with mild, moderate, or severe head injuries compared with placebo ( moderate-quality evidence ).
We found one systematic review (search date 2004, 20 RCTs, 12,303 adults and children with head injury) comparing corticosteroids versus placebo or no corticosteroids in people with acute traumatic brain injury treated within 7 days of injury. The review did not meta-analyse results for the outcome of death, because of significant heterogeneity among the trials — probably as a result of the inclusion of the largest trial (Corticosteroid Randomisation After Significant Head Injury [CRASH]). The CRASH trial was 27 times larger than any of the previous RCTs identified by the review, and accounted for about 80% of all people included in the entire review. The RCT (10,008 adults with head injury, Glasgow Coma Scale [GCS] score 14 or lower, about 70% with moderate to severe head injury, mechanism and morphology of injury unclear) compared intravenous methylprednisolone for 48 hours versus placebo. It found that the risk of death from all causes within 2 weeks was significantly higher in people taking corticosteroids than in people taking placebo (1052/4985 [21%] with corticosteroids v 893/4979 [18%] with placebo; RR 1.18, 95% CI 1.09 to 1.27; P = 0.0001). The relative increase in mortality in people receiving corticosteroids did not differ by injury severity (P = 0.22) or by time since injury (P = 0.05). Data collected at 6 months (9673 adults, 97% of participants) were consistent with earlier results. Mortality was higher in people taking corticosteroids than in those taking placebo (1248/4985 [26%] with corticosteroids v 1075/4979 [22%] with placebo; RR 1.15, 95% CI 1.07 to 1.24; P = 0.0001), as was death or severe disability (1828/4985 [38%] with corticosteroids v 1728/4979 [36%] with placebo; RR 1.05, 95% CI 0.99 to 1.10; P = 0.079). Nine RCTs (1628 adults and children, proportion of children not reported, most with severe head injury) identified by the review assessed the combined outcome of death or disability, and found no significant difference between corticosteroids and placebo (9 RCTs; 505/927 [54%] with corticosteroids v 340/701 [49%] with placebo or no corticosteroids; RR 1.01, 95% CI 0.91 to 1.11). This analysis did not include data from the CRASH trial.
Five RCTs (10,798 adults and children) identified by the review, including the CRASH trial, assessed infectious complications, and found no significant difference between corticosteroids and placebo or no corticosteroids (2586/5347 [48%] with corticosteroids v 2560/5451 [47%] with placebo or no corticosteroids; RR 1.03, 95% CI 0.99 to 1.11). There was also no significant difference in gastrointestinal bleeding between corticosteroids and placebo or no corticosteroids (10 RCTs, including CRASH; 95/5722 [1.6%] with corticosteroids v 72/5580 [1.3%] with placebo or no corticosteroids; RR 1.23, 95% CI 0.91 to 1.67).
In the CRASH trial, the main cause of death was not analysed. The mechanism remains uncertain, but is unlikely to be related to corticosteroid complications. Neither the CRASH trial nor the systematic review reported data on hypotension, which is known to be strongly associated with mortality after traumatic brain injury.
No new evidence