Therapeutic hypothermia for head injury

  • Review
  • Intervention

Authors


Abstract

Background

Induced hypothermia has been used in the treatment of head injury for many years. Encouraging results from small trials and laboratory studies led to renewed interest in the area and some larger trials.

Objectives

To estimate the effects of mild induced hypothermia in moderate and severe head injury on mortality, long-term functional outcome, complications, and short-term control of intracranial pressure (ICP).

Search strategy

We searched the Injuries Group Specialised register, CENTRAL, MEDLINE and EMBASE. We handsearched conference proceedings and checked reference lists of relevant articles, including a systematic review published in 2003.

Selection criteria

Randomised controlled trials of mild hypothermia to 34 to 35ºC for at least 12 hours versus control (open or normothermia) in patients with any closed head injury requiring hospitalisation. Two authors independently assessed all trials.

Data collection and analysis

Data on death, Glasgow outcome scale, complications and ICP were sought and extracted, either from published material or by contacting the investigators. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated for each trial on an intention-to-treat basis. Quantitative synthesis of data on complications other than pneumonia or ICP was not attempted. Trials of immediate and deferred hypothermia were analysed separately.

Main results

We found 14 trials with 1094 participants. Active immediate hypothermia was associated with an OR for death of 0.80, (1061 patients, OR 0.80, 95% CI 0.61 to 1.04), and 0.75 for odds of being dead or severely disabled, (746 patients, OR 0.75, 95% CI 0.56 to 1.00). Hypothermia treatment was associated with a statistically significant increase in odds of pneumonia (281 patients, OR 1.95, 95% CI 1.18 to 3.23).

The trial of deferred hypothermia (33 patients) reported a huge but not statistically significant reduction in the odds of death at six months, (OR 0.21, 95% CI 0.04 to 1.05). For death or severe disability, deferred hypothermia was associated with an OR of 0.10 (95% CI 0.01 to 1.00).

Authors' conclusions

There is no evidence that hypothermia is beneficial in the treatment of head injury. The earlier, encouraging, trial results have not been repeated in larger trials. The reasons for this are unclear. Hypothermia increases the risk of pneumonia and has other potentially harmful side-effects. Therefore, it would seem inappropriate to use this intervention outside of controlled trials in subgroups of patients for whom there is good reason to think the treatment would be beneficial.

Plain language summary

Hypothermia therapy after traumatic brain injury has not been shown to reduce death or disability, and it increases the risk of pneumonia

Severe head injury can damage the brain, and damage continues occurring in the brain for some time after the initial injury. Mild hypothermia treatment (cooling) has been a common strategy to slow down some of the brain changes that cause continuing damage. However, while hypothermia might slow down some harmful actions in the brain, potential major adverse effects include pneumonia (chest infection), blood flow problems or heart complications. The review of trials found that hypothermia therapy has not been shown to reduce death or disability after traumatic brain injury. However, hypothermia increases the risk of pneumonia and other life-threatening complications.

Background

Severe traumatic head injury is a major cause of death and disability amongst a predominantly young population, with an estimated ten million people experiencing such a head injury worldwide every year (Alexander 1992). There is, however, a significant lack of coherent evidence about effective therapies in the acute care of these patients. A long-term effort to review the literature and produce management guidelines by the American Association of Neurological Surgeons (Bullock 1996; Kirkpatrick 1997) could only make four definitive statements about treatment effectiveness that were supported by strong evidence from randomised studies.

Mild to moderate induced hypothermia has been used in the treatment of head injury for over 50 years (Fay 1945). Although there were several promising experimental studies (Laskowski 1960; Clasen 1968) and case series (Sedzimir 1959; Shapiro 1974), no controlled clinical studies were performed and the therapy fell from favour. In the last decade, however, several investigators have reported encouraging results of Phase II and III randomised clinical trials (Clifton 1995; Marion 1997; Shiozaki 1993), corroborated by consistent findings of high levels of cerebral protection associated with systemic cooling in well validated laboratory models of global ischaemia (Busto 1987). The early trials were small, single-centre investigations, which were sufficiently promising to lead on to larger, multi-centre trials.

Whilst the mechanism of action of such temperature control therapy was originally thought to be primarily a reduction in cerebral metabolic rate (Bering 1961), there is now evidence that mild hypothermia can also influence the excessive post-traumatic release of excitatory neurotransmitters (Busto 1989), and attenuate the opening of the blood-brain barrier (Smith 1996). The main risks associated with induced systemic hypothermia are an increased risk of sepsis and pneumonia, coagulation abnormalities, and possible myocardial ischaemia and atrial fibrillation (Schubert 1995).

Objectives

To determine whether the use of mild therapeutic hypothermia in the treatment of moderate and severe head injury:

  • reduces the risk of death (either during the treatment period or at the end of follow-up);

  • reduces the proportion of patients who at final follow-up are either dead or severely disabled;

  • reduces the mean intracranial pressure (ICP) recorded during the treatment period;

  • increases the risk of pneumonia and other complications.

Methods

Criteria for considering studies for this review

Types of studies

A search was conducted for all randomised controlled trials of mild hypothermia versus control (open or normothermia).

Types of participants

Patients with any closed head injury requiring hospitalisation.

Types of interventions

Therapeutic cooling, either locally or systemically, by means of a fluid-filled cooling blanket, a 'bear-hugger' air-cooling device, ice water lavage, any combination of the above, or other methods, to a target temperature of at most 34 to 35ºC for a period of at least 12 hours. Cooling could begin immediately upon admission to the intensive therapy unit or be deferred until ICP has become uncontrollable by conventional management.

Types of outcome measures

Primary outcomes
  • All-cause mortality and death at the end of the scheduled follow-up period.

  • Severe disability at the end of the scheduled follow-up period.

Severe disability was defined as a Glasgow outcome scale score of 'severe disability' or 'persistent vegetative state', or an equivalent measure if this was unavailable.

Secondary outcomes
  • The frequency of complications such as pneumonia, coagulopathy and other serious adverse events during treatment and follow-up period.

  • The mean intracranial pressure (ICP) achieved during the acute treatment phase.

Search methods for identification of studies

Electronic searches

For the initial version of the review the following search was done. The Specialist Trials Register for the Injuries Group was searched in May 1998 for any relevant randomised trials relating to temperature control using the search terms: hypotherm* OR normotherm* OR cool* OR cold* OR temperature.

The search strategy for the register is primarily an electronic search of both MEDLINE and CENTRAL, supplemented by various hand-searching activities listed in the Group details. This was supplemented by a comprehensive EMBASE search, also performed in May 1998, to identify all potential RCTs involving human head injury and temperature control from 1980 onwards (for full strategy see Appendix 1).

Titles and abstracts for the results of this search were reviewed by DFS for possibly relevant trial reports, and the appropriate articles retrieved.

Searching other resources

These two searches were supplemented by further handsearching of recent conference proceedings and abstracts as follows:

  • International Conference on Recent Advances in Neurotraumatology, Italy 1996

  • 2nd International Neurotrauma Symposium, Glasgow 1993

  • 3rd International Neurotrauma Symposium, Toronto 1995

  • 4th International Neurotrauma Symposium, Seoul 1997

  • 27th Meeting of the Society for Critical Care Medicine, USA 1998

  • 10th International Symposium on Intracranial Pressure, USA 1997

In addition, reference lists of all relevant trials and review articles were checked, and leading investigators in the field were contacted for information about any other published or unpublished trials which may have been overlooked.

Searches for updates
2001

A search, of MEDLINE, EMBASE and the Cochrane Controlled Trials Register, for controlled trials in head injury was conducted at the editorial base of the Injuries Group. The results of this search where then screened by two members of the Injuries Group and any trials on hypothermia therapy were identified. Full details of this search strategy can be obtained from the editorial base.

2004

The reference list of a review article in a paper journal (McIntyre 2003) was cross checked with the studies in this review.

Data collection and analysis

Selection of studies

One author (DFS) examined titles and abstracts of the references extracted from both the Specialist Register and EMBASE for reports of possibly relevant trials and these reports were then retrieved in full.

Data extraction and management

All potential trial reports were read and assessed by both authors. The selection criteria for this review were applied independently, resolving disagreements by discussion. Both then extracted data on the method of randomisation, the inclusion and exclusion criteria, the exact hypothermia protocol used, the number of patients lost to follow-up or excluded, the method of outcome measurement and the method of assessment of serious complications. Also assessed were the functional and neurological outcome, if possible using the Glasgow Outcome Scale (GOS), and the number of serious adverse events observed in each treatment group.

For the 2001 update of this review, the Injuries Group searched its register to identify new studies, and then PA and CG examined those identified for eligibility. PA and CG extracted data independently. CG helped in updating the analysis and the manuscript and DFS checked the final manuscript of the update.

For the 2004 update of this review, the included studies were checked against those included in a systematic review of the topic published in a paper journal in 2003 (McIntyre 2003). PA then included additional studies in the review, reran the analyses and updated the text. Data entry was checked as accurate by a member of the Injuries Group editorial team.

Assessment of risk of bias in included studies

Rather than use one of the many quality scales available, the assessment of susceptibility to bias was limited to the assessment of the adequacy of allocation concealment and the degree of blinding in the assessment of outcome. Both of these factors have been shown empirically to be associated with possible bias in the estimation of the true treatment effect (Schulz 1995).

Data synthesis

Mantel-Haenzel odds ratios (ORs) and 95% confidence intervals (CIs) were calculated for death, death and disability for each trial on an intention-to-treat basis. The odds ratio was chosen because of the large variation in baseline event rates between the trials (mortality in the control groups ranges from 20% to 58%), implying that the relative risk would not be a good summary measure. Also, the Mantel-Haenzel approach was used because of the inaccuracy of Peto's approximation when the estimated treatment effect is large, as it was in several of the trials considered. Heterogeneity of treatment effect between trials was assessed using a standard chi-square test, and, if appropriate, a weighted estimate of the typical treatment effect across all studies was calculated.

Studies which initiated cooling therapy immediately upon arrival in the intensive therapy unit (ITU) were analysed separately from those which initiated cooling in patients in whom ICP was uncontrollable by other means.

Subgroup analysis and investigation of heterogeneity

Subgroup analyses were performed to determine whether the treatment effect varies with: a) trial quality (quality of allocation concealment, blinded outcome vs non-blinded outcome assessment), b) duration of hypothermia and c) length of follow-up.

In the 2001 update, quantitative synthesis of the most commonly reported complication that is, pneumonia was performed. Mantel-Haenzel ORs and 95% confidence intervals were calculated.

Due to the poor quality of reporting and differences in methods of summarising ICP values and other complications, no quantitative analysis of these outcomes were performed.

Results

Description of studies

See: Characteristics of included studies; Characteristics of excluded studies; Characteristics of ongoing studies.

Results of the search

The search strategy produced 14 full papers and 13 abstracts which were read in full by both authors. Although it was by no means clear, these 27 reports referred to 13 distinct studies. Three studies were excluded on the grounds of either confounding, not being properly randomised or investigating a different intervention. Two other studies were ongoing and hence excluded. The updated search in 2001 identified a further five relevant studies. One of the abstracts (Jiang 1996) included in the original analysis was excluded in the update because of the identification of a paper (Jiang 2000) that referred to the same study. The search conducted in late 2003 identified two further studies for inclusion (Yan 2001, Zhang 2000) and one possible relevant study (Shen 2000) that has not been fully assessed as we are not able to retrieve the full text of the study.

Included studies

The 14 studies finally included randomised a total of about 1100 patients, approximately four times the total in the original version of this review. One trial (Shiozaki 1993) randomised between hypothermia and control only in those patients in whom ICP was uncontrollable using fluid resuscitation, hyperventilation and high-dose barbiturate therapy. The other 13 trials were of immediate hypothermia versus either active normothermia or open control. All trials except one (Ishikura 1998) reported deaths, and nine trials reported GOS scores at either three, six or 12 months.

Five trials (Clifton 1993; Clifton 2001; Hirayama 1994; Jiang 2000; Marion 1997) reported mean ICP values with standard errors for both hypothermia and control patients. The time periods over which these were summarised however did not correspond to the treatment periods, nor were they similar across the trials. For this reason no quantitative synthesis of these results was attempted.

Pneumonia, the most frequently recorded complication was reported in seven trials and quantitative analysis of this data was performed in the 2001 update (Aibiki 2000; Clifton 1992; Clifton 1993; Jiang 2000; Meissner 1998; Shiozaki 1999; Shiozaki 2001). The largest trial (Clifton 2001) did not report data on pulmonary infections but did state that the difference was not statistically significant. The reporting of other complications was variable, with no common definition of either the type or severity of these adverse events. Therefore, quantitative analysis of other complications were not performed.

Risk of bias in included studies

Allocation

Three of the trials included were identified solely on the basis of abstracts, and so the judgement of trial quality was particularly difficult in these cases. Three trials had a reasonable standard of allocation concealment and blinded outcome assessment (Clifton 1993; Marion 1997; Clifton 2001). Two trials had good allocation concealment but non-blinded assessment of outcome (Clifton 1992; Meissner 1998). Three trials had unclear allocation concealment ('by lot') but blinded outcome assessment (Aibiki 2000; Jiang 2000; Shiozaki 1993), and six trials had unclear allocation concealment and did not state whether the outcome assessment was blinded to the treatment allocation (Hirayama 1994; Ishikura 1998; Shiozaki 1999; Shiozaki 2001; Yan 2001; Zhang 2000).

Effects of interventions

Immediate hypothermia

Twelve trials involving 1061 patients reported deaths. Hypothermic treatment was associated with a 20% non-significant reduction in the odds of death at the end of treatment or final follow-up, (OR 0.80, 95% CI 0.61 to 1.04).

For death or severe disability nine trials involving 746 patients reported GOS scores at either three, six or 12 months. In this case, hypothermia was associated with a 25% non-significant reduction in the odds of being dead or severely disabled at final follow-up, (OR 0.75, 95% CI 0.56 to 1.00). However, there was some evidence of between study heterogeneity for this result, with P = 0.06 and I squared of 47% for the test for heterogeneity. With nine studies, it is difficult to investigate this further. The planned subgroup analyses on length of follow-up, study quality and length of intervention do not show a clear pattern that might explain this heterogeneity. It could, of course, be due to a number of other differences in the patients, interventions and co-interventions in the studies, but we had not planned to investigate this before seeing the results. A random effects meta-analysis of these data produce an estimate for the odds ratio of being dead or severely disabled of 0.61 (0.37 to 1.02), which reinforces the need for cautious interpretation of this result.

Seven trials involving 281 patients reported pneumonia as a complication. Hypothermia treatment was associated with a statistically significant increase in the odds ratio for pulmonary infections (OR 1.95, 95% confidence interval 1.18 to 3.23).

Deferred hypothermia

The single trial of deferred hypothermia in 33 patients reported both deaths and GOS scores at six months. Hypothermic treatment was associated with a huge but marginally significant (P = 0.06) reduction in the odds of death at six months, (OR 0.21, 95% CI 0.04 to 1.05). For death or severe disability hypothermic treatment was associated with a 90% reduction (P = 0.05) in the odds of a poor outcome, (OR 0.10, 95% CI 0.01 to 1.00). Hypothermia therapy was not associated with an increased risk of pneumonia (OR 1.13, 95% CI 0.18 to 6.93).

Discussion

The overall results of this systematic review have failed to detect an effect of immediate hypothermia, but the relatively wide confidence intervals for death and death or severe disability do not exclude a clinically significant benefit or small harm. The apparent positive result for deferred hypothermia must be treated with caution as it comes from one small trial.

This review also highlights the harm that may be incurred on patients due to hypothermia. The quantitative analysis showed a statistically significant near doubling of the odds of pneumonia with hypothermia treatment. It should be noted however, that the largest trial (Clifton 2001) was not included in this pneumonia analysis.

This result is far less optimistic than that in the original version of the review, where there appeared to be a benefit of immediate hypothermia on death or severe disability. There are several possible explanations for this. The earlier result relied upon a few small trials. Each of these trials was performed in a single centre, and the investigators involved were experienced both in inducing and maintaining low body temperatures in injured patients. It is possible that the intervention was applied differently in different centres. One other possible explanation is time lag bias - the phenomenon that studies with dramatic results are published more quickly than those with less dramatic findings. This seems unlikely to be important here as three of the four small, more positive studies had been completed and published before the largest study began recruiting.

The authors of the largest study (Clifton 2001) suggest that the variation in effect might be due to differences in the proportion of patients admitted with existing hypothermia. One explanation may be that in some studies such patients allocated to normothermia were actively rewarmed and that this might be harmful. Alternatively, it may be crucial to induce hypothermia very early on to achieve the neuroprotective effect. The Clifton group is currently running another multicenter study in order to characterise the utility of hypothermia therapy in specific subgroups of patients.

A recent systematic review of this topic (McIntyre 2003) concluded that there was a statistically significant effect of hypothermia on death and poor neurological outcome. McIntyre's review did not include one trial that is in this review (Meissner 1998), and its main analysis pooled trials both of immediate and deferred hypothermia, which are separated in this review. Pooling these two subgroups adds the Shiozaki 1993 trial into the analysis, causing the results to become statistically significant. As the statistical significance of the overall result depends crucially on the criteria for pooling studies, it seems sensible to be cautious in interpreting the results.

Authors' conclusions

Implications for practice

The evidence for an effect of hypothermia on death or severe disability is unclear. By altering the criteria for pooling studies it is possible to change the statistical significance of the overall result, suggesting that the results should be interpreted with caution. In addition, there is a statistically significant increase in pulmonary infections and other potential for harm, so at the moment it seems reasonable to continue to use the intervention within the context of controlled trials to clarify in which situations hypothermia might be beneficial.

Implications for research

The largest study (Clifton 2001) was stopped early because the monitoring committee felt the probability of detecting a benefit was too small to justify further exposure of patients to a potentially harmful intervention. Hence, future studies should be undertaken with caution and only in subgroups strongly believed to benefit from hypothermia.

The Clifton group is conducting another multi-centre study to characterize the utility of hypothermia therapy in specific subgroups such as younger participants (18 to 45 years) who on admission have low body temperatures.

Acknowledgements

Thanks to:

  • Brenda Thomas (Stroke Review Group) for help and advice with the EMBASE search strategy.

  • Ian Whittle, Kate Signorini, Elena Telaro, Yoichi Nagayama, Irene Kwan and Lisa Xue for help with manuscripts in languages other than English.

  • Reinhard Wentz and Irene Kwan of the Injuries Group for updating the search.

  • the editorial team of the Cochrane Injuries Group for updating the searches for the review.

Data and analyses

Download statistical data

Comparison 1. Immediate hypothermia versus normothermia
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Death at final follow-up131063Odds Ratio (M-H, Fixed, 95% CI)0.80 [0.61, 1.04]
2 Death or severe disability at final follow-up9746Odds Ratio (M-H, Fixed, 95% CI)0.75 [0.56, 1.00]
3 Death or severe disability stratified by quality  Odds Ratio (M-H, Fixed, 95% CI)Subtotals only
3.1 Concealed allocation, blinded outcome assessment3494Odds Ratio (M-H, Fixed, 95% CI)0.80 [0.56, 1.14]
3.2 Concealed allocation, non-blinded outcome assessment110Odds Ratio (M-H, Fixed, 95% CI)0.38 [0.02, 6.35]
3.3 Non-concealed allocation, blinded outcome assessment2113Odds Ratio (M-H, Fixed, 95% CI)0.34 [0.16, 0.76]
3.4 Non-concealed allocation, non-blinded outcome assessment3129Odds Ratio (M-H, Fixed, 95% CI)1.17 [0.58, 2.35]
4 Death or severe disability stratified by treatment duration  Odds Ratio (M-H, Fixed, 95% CI)Subtotals only
4.1 24 hours291Odds Ratio (M-H, Fixed, 95% CI)0.38 [0.16, 0.90]
4.2 48 hours5542Odds Ratio (M-H, Fixed, 95% CI)0.97 [0.69, 1.36]
5 Death or severe disability at various times during follow-up  Odds Ratio (M-H, Fixed, 95% CI)Subtotals only
5.1 3 months5250Odds Ratio (M-H, Fixed, 95% CI)0.72 [0.43, 1.21]
5.2 6 months4492Odds Ratio (M-H, Fixed, 95% CI)0.79 [0.55, 1.13]
5.3 12 months2168Odds Ratio (M-H, Fixed, 95% CI)0.41 [0.22, 0.77]
6 Pneumonia during the treatment period8283Odds Ratio (M-H, Fixed, 95% CI)1.95 [1.18, 3.23]
Analysis 1.1.

Comparison 1 Immediate hypothermia versus normothermia, Outcome 1 Death at final follow-up.

Analysis 1.2.

Comparison 1 Immediate hypothermia versus normothermia, Outcome 2 Death or severe disability at final follow-up.

Analysis 1.3.

Comparison 1 Immediate hypothermia versus normothermia, Outcome 3 Death or severe disability stratified by quality.

Analysis 1.3.1.

Analysis 1.3.1.

Comparison 1 Immediate hypothermia versus normothermia, Outcome 3 Death or severe disability stratified by quality, Subgroup 1 Concealed allocation, blinded outcome assessment.

Analysis 1.3.2.

Analysis 1.3.2.

Comparison 1 Immediate hypothermia versus normothermia, Outcome 3 Death or severe disability stratified by quality, Subgroup 2 Concealed allocation, non-blinded outcome assessment.

Analysis 1.3.3.

Analysis 1.3.3.

Comparison 1 Immediate hypothermia versus normothermia, Outcome 3 Death or severe disability stratified by quality, Subgroup 3 Non-concealed allocation, blinded outcome assessment.

Analysis 1.3.4.

Analysis 1.3.4.

Comparison 1 Immediate hypothermia versus normothermia, Outcome 3 Death or severe disability stratified by quality, Subgroup 4 Non-concealed allocation, non-blinded outcome assessment.

Analysis 1.4.

Comparison 1 Immediate hypothermia versus normothermia, Outcome 4 Death or severe disability stratified by treatment duration.

Analysis 1.4.1.

Analysis 1.4.1.

Comparison 1 Immediate hypothermia versus normothermia, Outcome 4 Death or severe disability stratified by treatment duration, Subgroup 1 24 hours.

Analysis 1.4.2.

Analysis 1.4.2.

Comparison 1 Immediate hypothermia versus normothermia, Outcome 4 Death or severe disability stratified by treatment duration, Subgroup 2 48 hours.

Analysis 1.5.

Comparison 1 Immediate hypothermia versus normothermia, Outcome 5 Death or severe disability at various times during follow-up.

Analysis 1.5.1.

Analysis 1.5.1.

Comparison 1 Immediate hypothermia versus normothermia, Outcome 5 Death or severe disability at various times during follow-up, Subgroup 1 3 months.

Analysis 1.5.2.

Analysis 1.5.2.

Comparison 1 Immediate hypothermia versus normothermia, Outcome 5 Death or severe disability at various times during follow-up, Subgroup 2 6 months.

Analysis 1.5.3.

Analysis 1.5.3.

Comparison 1 Immediate hypothermia versus normothermia, Outcome 5 Death or severe disability at various times during follow-up, Subgroup 3 12 months.

Analysis 1.6.

Comparison 1 Immediate hypothermia versus normothermia, Outcome 6 Pneumonia during the treatment period.

Comparison 2. Deferred hypothermia versus normothermia
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Death at final follow-up133Odds Ratio (M-H, Fixed, 95% CI)0.21 [0.04, 1.05]
2 Death or severe disability at final follow-up133Odds Ratio (M-H, Fixed, 95% CI)0.10 [0.01, 1.00]
3 Death or severe disability at various times during follow-up  Odds Ratio (M-H, Fixed, 95% CI)Subtotals only
3.2 6 months133Odds Ratio (M-H, Fixed, 95% CI)0.10 [0.01, 1.00]
4 Death or severe disability stratified by quality  Odds Ratio (M-H, Fixed, 95% CI)Subtotals only
4.3 Non-concealed allocation, blinded outcome assessment133Odds Ratio (M-H, Fixed, 95% CI)0.10 [0.01, 1.00]
5 Death or severe disability stratified by treatment duration  Odds Ratio (M-H, Fixed, 95% CI)Subtotals only
5.2 48 hours133Odds Ratio (M-H, Fixed, 95% CI)0.10 [0.01, 1.00]
6 Pneumonia during the treatment period122Odds Ratio (M-H, Fixed, 95% CI)1.13 [0.18, 6.93]
Analysis 2.1.

Comparison 2 Deferred hypothermia versus normothermia, Outcome 1 Death at final follow-up.

Analysis 2.2.

Comparison 2 Deferred hypothermia versus normothermia, Outcome 2 Death or severe disability at final follow-up.

Analysis 2.3.

Comparison 2 Deferred hypothermia versus normothermia, Outcome 3 Death or severe disability at various times during follow-up.

Analysis 2.3.2.

Analysis 2.3.2.

Comparison 2 Deferred hypothermia versus normothermia, Outcome 3 Death or severe disability at various times during follow-up, Subgroup 2 6 months.

Analysis 2.4.

Comparison 2 Deferred hypothermia versus normothermia, Outcome 4 Death or severe disability stratified by quality.

Analysis 2.4.3.

Analysis 2.4.3.

Comparison 2 Deferred hypothermia versus normothermia, Outcome 4 Death or severe disability stratified by quality, Subgroup 3 Non-concealed allocation, blinded outcome assessment.

Analysis 2.5.

Comparison 2 Deferred hypothermia versus normothermia, Outcome 5 Death or severe disability stratified by treatment duration.

Analysis 2.5.2.

Analysis 2.5.2.

Comparison 2 Deferred hypothermia versus normothermia, Outcome 5 Death or severe disability stratified by treatment duration, Subgroup 2 48 hours.

Analysis 2.6.

Comparison 2 Deferred hypothermia versus normothermia, Outcome 6 Pneumonia during the treatment period.

Appendices

Appendix 1. EMBASE search strategy

1. exp head injury/
2. pneumocephalus/.
3. cerebrospinal fluid/
4. otorrhea/
5. exp skull fracture/
6. exp spine fracture/...
7. cerebrospinal fluid rhinorrhea/
8. exp asphyxia/
9. exp spine injury/
10. helmet/
11. brain protection/
12. brain edema/
13. exp brain hemorrhage/
14. brain hypoxia/
15. coma/
16. persistent vegetative state/
17. traumatic epilepsy/
18. or/1-17
19. (head or brain or cerebr$ or skull or crani$ or spin$).tw.
20. (wound$ or injur$ or trauma$ or oedema$ or edema$).tw.
21. damage$.tw.
22. 20 or 21
23. 19 and 22
24. 18 or 23
25. human/
26. "888".tg.
27. 25 or 26
28. nonhuman/
29. "777".tg.
30. 28 or 29
31. 27 and 30
32. 30 not 31
33. clinical trial/
34. multicenter study/
35. phase 2 clinical trial/
36. phase 3 clinical trial/...
37. phase 4 clinical trial/
38. randomized controlled trial/
39. controlled study/
40. meta analysis/
41. crossover procedure/
42. double blind procedure/...
43. single blind procedure/...
44. randomization/...
45. major clinical study/...
46. placebo/....
47. drug comparison/...
48. clinical study/...
49. "0197".tg....
50. "0150".tg....
51. "03738".dc....
52. (clin$ adj25 trial$).tw....
53. ((singl$ or doubl$ or tripl$ or trebl$) adj25 (blind$ or mask$)).tw.
54. placebo$.tw....
55. random$.tw....
56. control$.tw....
57. or/33-56....
58. 24 not 32....
59. induced hypothermia/...
60. profound induced hypothermia/...
61. exp temperature/...
62. exp low temperature procedures/..
63. cold/....
64. cold air/....
65. cold exposure/...
66. (hypotherm$ or normotherm$ or cool$ or cold$ or temperature$).tw.
67. or/59-66....
68. 58 and 67...
69. 68 and 57.

What's new

DateEventDescription
14 May 2008AmendedConverted to new review format.

History

Protocol first published: Issue 2, 1999
Review first published: Issue 2, 1999

DateEventDescription
28 July 2004New citation required and conclusions have changed

Substantive amendment. New studies found and included or excluded.

Two more trials have been included, and one added to the awaiting assesment list as we have not been able to obtain a copy of the trial.
Conclusions have been reviewed and compared to those found in a review of the topic in JAMA.

12 November 2001New search has been performedFour studies were included, which were published since the original version of this review.
Quantitative synthesis of the incidence of pulmonary infections was conducted in the 2001 update.

Contributions of authors

DFS wrote the protocol, performed the searches and reviewed the titles and abstracts, extracted the data, performed the analyses and wrote the draft of the review. PA reviewed the manuscripts of potential trials, extracted the data and edited the draft review.

For the 2001 update, the Injuries Group performed the search and screened studies. PA and CG assessed eligibility, extracted data, performed the analysis and redrafted the text.

For the 2004 update, the Injuries Group performed the search and screened studies. PA and the Injuries Group extracted data, and PA performed the analysis and rewrote the text.

Declarations of interest

None known.

Sources of support

Internal sources

  • DFS was supported by MRC project grant G9604637, UK.

  • NHS R&D Programme, UK.

External sources

  • CG was supported by the Doris Duke Research Fellowship, USA.

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Aibiki 2000

MethodsRandomised controlled trial. Allocation concealment unclear.
Four patients excluded from the normothermic group after randomisation because of abdominal or chest injuries.
ParticipantsPatients aged 4 to 76 with traumatic brain injury and Glasgow coma scale from 3 to 8.
InterventionsHypothermia patients: Cooling to 32-33C within 4 hours on injury for 3-4 days. Rewarming at 1C per day.
Normothermia patients: maintained at 36-37C.
OutcomesDeath and GOS at 6 months.
Thromboxane A2 and prostaglandin I2 levels during study. Complications during treatment.
NotesGOS assessed by "independent neurosurgeon who were not aware of the study".
Risk of bias
ItemAuthors' judgementDescription
Allocation concealment?UnclearB - Unclear

Biswas 2002

MethodsRandomised controlled trial.
Allocation concealment
ParticipantsTwenty-one children up to 18 years old, with closed traumatic brain injury and a GCS of 8 or less.
InterventionsHypothermia patients (n=10?): cooled to 32 to 34 degress celsius for 48 hours. Rewarming over a period of 12 hours.
Control patients (n=11): rectal temperature was maintained between 36.5 and 37.5 degrees celsius.
OutcomesDeath.
GOS at three, six and 12 months.
ICP and CPP.
NotesGOS assessed blind to allocation.
Analysis on an intention-to-treat basis.
Risk of bias
ItemAuthors' judgementDescription
Allocation concealment?UnclearD - Not used

Clifton 1992

MethodsRandomised controlled trial. Allocation by 'sealed envelopes'.
ParticipantsPatients with GCS 4-8 with closed head injury but no major systemic injuries, in whom cooling could begin within 6 hours of injury.
InterventionsHypothermia patients: cooling to 30-32C for 24 hours using cooling blankets and iced saline stomach lavage. Rewarming over a period of 24 hours.
Control patients: No active temperature management.
OutcomesDeath and GOS at 3 months.
Complications during treatment phase.
NotesGOS not assessed blind to treatment allocation.
Risk of bias
ItemAuthors' judgementDescription
Allocation concealment?YesA - Adequate

Clifton 1993

MethodsRandomised controlled trial. Allocation by 'sealed envelopes'.
ParticipantsPatients age 16 to 60, GCS 4-7 with closed head injury but no major systemic injuries, in whom cooling could begin within 6 hours of injury.
InterventionsHypothermia patients: cooling to 32-33C for 48 hours using cooling blankets. Rewarming over a period of 48 hours.
Control patients: Cooling blankets were used to maintain body temperature at 37C for 80 hours.
OutcomesDeath and GOS at 3 months.
Complications during treatment period.
ICP during treatment period.
NotesGOS assessed blind to treatment allocation.
Risk of bias
ItemAuthors' judgementDescription
Allocation concealment?YesA - Adequate

Clifton 2001

MethodsRandomised controlled trial.
Allocation concealment unclear, but report states that "only the study biostatistician was aware of each patient's treatment group assignment".
Outcome data missing for 7 patients, and not presented for 17 patients whose entry details were incomplete.
ParticipantsPatients aged 16 to 65 with a non-penetrating head injury and a Glasgow coma scale of 3 to 8 after resuscitation.
InterventionsHypothermia patients: cooling to 32.5-34C for 48 hours using ice, cold gastric lavage, unwarmed ventilator gases, and then temperature control pads. Rewarming at rate of up to 0.5C in 2 hours.
Control: Body temperature maintained at 37C.
OutcomesDeath and GOS at 6 months.
ICP monitored during treatment.
Nine neurobehavioural and neuropsychological scales at 6 months..
NotesGOS assessed blind to treatment allocation.
Risk of bias
ItemAuthors' judgementDescription
Allocation concealment?YesA - Adequate

Hirayama 1994

MethodsRandomised controlled trial. Allocation method not stated.
ParticipantsPatients age 18 to 81, GCS 3-7 with closed head injury.
InterventionsHypothermia patients: cooling to 32-33C for 48 hours using cooling blankets. Rewarming over a period of 48 hours.
Control patients: Not stated.
OutcomesDeath and GOS at 3 months.
ICP during treatment period.
NotesBlinding of outcome assessment not stated.
Risk of bias
ItemAuthors' judgementDescription
Allocation concealment?UnclearB - Unclear

Ishikura 1998

MethodsRandomised controlled trial. Allocation by 'random sampling'.
ParticipantsPatients with GCS 3-8 with closed head injury.
Interventions'Moderate hypothermia' without any details.
OutcomesThrombopoetin levels during treatment.
Deaths in hypothermia arm only.
NotesAbstract only.
Risk of bias
ItemAuthors' judgementDescription
Allocation concealment?UnclearB - Unclear

Jiang 2000

MethodsRandomised controlled trial. Allocation method not clear.
stated.
ParticipantsPatients with mean age of 41 years, GCS 3-8.
InterventionsHypothermia patients: 'Mild hypothermia' induced using cooling blankets until ICP within 'normal range' for 24 hours.
Control patients: Temperature maintained between 37-38C for 14 days.
OutcomesDeath and GOS at 12 months.
Complications.
NotesAssesment by MD blinded to treatment allocation.
Risk of bias
ItemAuthors' judgementDescription
Allocation concealment?YesA - Adequate

Marion 1997

MethodsRandomised controlled trial. Allocation by 'sealed envelopes'.
ParticipantsPatients age 16 to 75, GCS 3-7 with closed head injury, in whom cooling could begin within 6 hours of injury.
InterventionsHypothermia patients: Cooling to 32-33C for 24 hours using cooling blankets and nasogastric lavage. Rewarming over a period of 12 hours.
Control patients: Active management of temperature to 37-38.5C during five day treatment period.
OutcomesDeath and GOS at 3, 6 and 12 months.
ICP and CPP values during treatment phase.
Complications for subset.
NotesGOS assessment by psychiatrist blinded to treatment allocation.
Risk of bias
ItemAuthors' judgementDescription
Allocation concealment?YesA - Adequate

Meissner 1998

MethodsRandomised controlled trial. Allocation by 'sealed envelopes'.
ParticipantsPatients with severe blunt head injury, in whom cooling could begin within 8 hours of injury.
InterventionsHypothermia patients: Cooling to 32-33C for 48 hours.
Control patients: Temperature maintained at 36-37C.
OutcomesDeath.
Infections.
NotesGOS assessed at 6 months by non-blinded assessor, but not yet available.
Risk of bias
ItemAuthors' judgementDescription
Allocation concealment?YesA - Adequate

Meissner 2003a

Methods 
Participants 
Interventions 
Outcomes 
Notes 
Risk of bias
ItemAuthors' judgementDescription
Allocation concealment?UnclearD - Not used

Meissner 2003b

Methods 
Participants 
Interventions 
Outcomes 
Notes 
Risk of bias
ItemAuthors' judgementDescription
Allocation concealment?UnclearD - Not used

Shiozaki 1993

MethodsRandomised controlled trial. Allocation 'by lot'.
ParticipantsPatients age 10 or over, GCS 8 or less with head injury, in whom ICP could not be controlled by high-dose barbituate therapy.
InterventionsHypothermia patients: cooling to 33.5-34.5C using water-circulating cooling blankets for a minimum of 48 hours and until ICP was below 20 mmHg for 24 hours. Rewarming over a period of 24 hours.
Control patients: No active temperature management.
OutcomesDeath and GOS at 6 months.
Complications during treatment.
ICP and CPP values during treatment period for hypothermic arm only.
NotesGOS assessed blind to treatment allocation.
Risk of bias
ItemAuthors' judgementDescription
Allocation concealment?UnclearB - Unclear

Shiozaki 1999

MethodsRandomised controlled trial. Allocation concealment not clear.
No loss to follow up.
ParticipantsPatients with traumatic brain injury, a Glasgow coma scale of 8 or less, and ICP under 20mmHg on treatment with hyperventilation, barbiturates and fluid restriction.
InterventionsHypothermia patients: cooling to 33.5-34.5C for 48 hours, using water circulating blankets. Rewarming at 1C per day.
Normothermia patients: maintained at 36.5-37.5C.
OutcomesDeath and GOS at 6 months.
Complications.
NotesBlinding of outcome assessment not stated.
Risk of bias
ItemAuthors' judgementDescription
Allocation concealment?UnclearB - Unclear

Shiozaki 2001

MethodsRandomised controlled trial.
Allocation concealment not clear.
No loss to follow up.
ParticipantsPatients with traumatic brain injury, a Glasgow coma scale of 8 or less, and ICP of under 25mmHg with other therapies.
InterventionsHypothermia patients: Cooling to 33.5-34.5C for 48 hours, using cooling blankets and gastric lavage. Rewarming at 1C per day.
Normothermia patients: maintained at 36.5-37.5C.
OutcomesDeath and GOS at 3 months.
Complications
NotesBlinding of outcome assessment not stated.
Risk of bias
ItemAuthors' judgementDescription
Allocation concealment?UnclearB - Unclear

Yan 2001

MethodsRandomised controlled trial.
Allocation concealment not described.
No loss to follow up described.
ParticipantsPatients with traumatic brain injury within 10 hours of injury and a Glasgow Coma Scale of 3 to 8 on initial assesment.
InterventionsHypothermia patients: Cooling to 32-34C for 3-5 days, using a cooling bed and, in some, ice blocks. 'Natural' rewarming.
Normothermia patients: cooling not used.
OutcomesDeath, follow up period unclear.
Neuroelectrophysiological measurements.
NotesBlinding of outcome assessment not stated.
Risk of bias
ItemAuthors' judgementDescription
Allocation concealment?UnclearB - Unclear

Zhang 2000

MethodsRandomised controlled trial.
Allocation concealment not described. No loss to follow up mentioned.
ParticipantsPatients aged under 65 with traumatic brain injury and a Glasgow Coma Scale of 3-8 on admission to hospital.
InterventionsHypothermia patients: Cooling to 32-33C for 3-8 days.
Normothermia patients: temperature not stated.
OutcomesDeath, follow up period unclear.
NotesBlinding of outcome assessment not stated.
Risk of bias
ItemAuthors' judgementDescription
Allocation concealment?UnclearB - Unclear

Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion
Gentilello 1997Randomized trial of rewarming therapy after accidental hypothermia in trauma.
Nara 1997Unable to find sufficient information on study design.
Nordby 1984Not a randomised comparison, and hypothermia confounded with barbituate therapy
Yamagami 1997Not a randomised study; hypothermia group GCS 4-6, normothermia group GCS 8-10.

Characteristics of ongoing studies [ordered by study ID]

Clifton 2002

Trial name or titleNational Acute Brain Injury Study: Hypothermia II (NABISH II)
Methods 
ParticipantsPatients aged 16 to 45 years inclusive who have a closed head injury, present to the Emergency Dept. with a Glasgow Coma Score between 3-8, have a body temperature (bladder or rectal) of 35 degrees Celsius or less, and with an Abbreviated Injury Score (AIS) of 4 or less for the rest of the body.
InterventionsThe patients will be randomly allocated to either the hypothermia group or the normothermia group. A cooling suit will be used to cool the hypothermia patients down to a body temperature of 33 degrees celsius. This temperature of 33 degrees will be maintained in the hypothermia patients for 48 hours. After 48 hours, the study nurses will gradually re-warm the hypothermia patients no faster than one degree every four hours. This takes at least 16 hours sometimes longer depending upon the stability of the patient's vital signes. The control group - normothermia will be allowed to re-warm gradually upon arrival to the hospital with no medical intervention to raise or lower the body temperature.
OutcomesMortality and GOS. ICP and complications.
Outcomes will be measured 6 months post injury by Harvey Levin, MD at Baylor College of Medicine. The personnel conducting outcome measurements will be blinded to the patient's assigned treatment protocol (whether hypothermia or normothermia).
Starting dateRun-in period 21st October 2002 to 21st April 2003.
Randomisation begins after run-in period.
Contact information

Guy.L.Clifton, MD
Chairman Neurosurgery Dept., University of Texas Medical School
6431 Fannin St., Suite 7.148
Houston, TX 77030
713-500-6135
guy.l.clifton@uth.tmc.edu

Emmy R. Miller, RN, PhD, Co-investigator NABISH II
Associate Professor of Neurosurgery
University of Texas Medical School, Houston, TX
6431 Fannin St., Suite 7.148
Houston, TX 77030
713-500-6145
Emmy.R.Miller@uth.tmc.edu

Notes

There are other study sites participating in NABISH II.

They are:
University of Pittsburgh, Duke University, University of California at Los Angeles, University of California at Sacramento, University of California at San Francisco, University of Virginia at Fairfax, University of Cincinnatti, University of Mississippi at Jackson.

Ancillary