Hypothermia Therapy after Traumatic Brain Injury in Children

T h e n e w e ng l a n d j o u r na l o f m e dic i n e
original article
Hypothermia Therapy after Traumatic Brain
Injury in Children
James S. Hutchison, M.D., Roxanne E. Ward, B.A., Jacques Lacroix, M.D.,
Paul C. Hébert, M.D., M.H.Sc., Marcia A. Barnes, Ph.D., Desmond J. Bohn, M.B.,
Peter B. Dirks, M.D., Steve Doucette, M.Sc., Dean Fergusson, Ph.D.,
Ronald Gottesman, M.D., Ari R. Joffe, M.D., Haresh M. Kirpalani, M.B., M.Sc.,
Philippe G. Meyer, M.D., Kevin P. Morris, M.D., David Moher, Ph.D.,
Ram N. Singh, M.D., and Peter W. Skippen, M.D., for the Hypothermia Pediatric
Head Injury Trial Investigators and the Canadian Critical Care Trials Group*
Abs tr act
Background
Hypothermia therapy improves survival and the neurologic outcome in animal mod- The affiliations of the authors are listed
in the Appendix. Address reprint requests
els of traumatic brain injury. However, the effect of hypothermia therapy on the neuro- to Dr. Hutchison at the Department of
logic outcome and mortality among children who have severe traumatic brain injury Critical Care Medicine, Hospital for Sick
is unknown.
Children, 555 University Ave., Toronto,
ON M5G 1X8, Canada, or at jamie.
hutchison@sickkids.ca.
Methods
In a multicenter, international trial, we randomly assigned children with severe trau- *The investigators who participated in the
study are listed in the Appendix.
matic brain injury to either hypothermia therapy (32.5°C for 24 hours) initiated within
8 hours after injury or to normothermia (37.0°C). The primary outcome was the pro- N Engl J Med 2008;358:2447-56.
portion of children who had an unfavorable outcome (i.e., severe disability, persistent Copyright © 2008 Massachusetts Medical Society.
vegetative state, or death), as assessed on the basis of the Pediatric Cerebral Perfor-
mance Category score at 6 months.
Results
A total of 225 children were randomly assigned to the hypothermia group or the
normothermia group; the mean temperatures achieved in the two groups were
33.1±1.2°C and 36.9±0.5°C, respectively. At 6 months, 31% of the patients in the hy-
pothermia group, as compared with 22% of the patients in the normothermia group,
had an unfavorable outcome (relative risk, 1.41; 95% confidence interval [CI], 0.89 to
2.22; P = 0.14). There were 23 deaths (21%) in the hypothermia group and 14 deaths
(12%) in the normothermia group (relative risk, 1.40; 95% CI, 0.90 to 2.27; P = 0.06).
There was more hypotension (P = 0.047) and more vasoactive agents were adminis-
tered (P<0.001) in the hypothermia group during the rewarming period than in the
normothermia group. Lengths of stay in the intensive care unit and in the hospital
and other adverse events were similar in the two groups.
Conclusions
In children with severe traumatic brain injury, hypothermia therapy that is initiated
within 8 hours after injury and continued for 24 hours does not improve the neu-
rologic outcome and may increase mortality. (Current Controlled Trials number,
ISRCTN77393684.)
n engl j med 358;23 www.nejm.org june 5, 2008
2447
Downloaded from www.nejm.org on August 11, 2009 . For personal use only. No other uses without permission.
Copyright © 2008 Massachusetts Medical Society. All rights reserved.

---

T h e n e w e ng l a n d j o u r na l o f m e dic i n e
Hypothermia therapy significantly ticipating institution. Written informed consent
improves survival and the neurologic out- was obtained from the parents or guardians; de-
come in rodent models of traumatic brain ferred consent was obtained when parents or
injury.1,2 An early case series involving 18 children guardians were not available within 8 hours after
suggested that hypothermia therapy could improve injury.7
survival and the neurologic outcome among chil-
After each patient had been assessed and the
dren with traumatic brain injury.3 These observa- patient’s condition had been stabilized, a study
tions led to two randomized trials involving chil- physician randomly assigned the patient to a treat-
dren.4,5
ment group with the use of a central telephone-
In these two trials, investigators analyzed a based system that was available 24 hours a day.
total of 96 children with severe traumatic brain The randomization, prepared by an independent
injury and reported that hypothermia therapy ap- statistician, was blocked in groups of four (par-
peared to be safe and caused no significant in- ticipating centers were unaware of the block size)
crease in serious adverse events; however, these and included two stratification variables: center
trials were not powered to detect significant im- and age (less than 7 years of age and 7 years of age
provements in survival or neurologic recovery.4,5 or more). The rationale for stratification by age
In a single-center, randomized trial reported in at the time of injury was based on several studies
1997, Marion et al.6 found that 24 hours of hy- that showed less recovery in IQ scores, attention,
pothermia therapy decreased the risk of a poor and executive functions among children who sus-
outcome, defined as death, a persistent vegetative tained severe injuries earlier in childhood, as com-
state, or severe disability, in a subgroup of adults pared with those who were injured later in child-
with a score of 5 to 7 on the Glasgow Coma Scale hood.8-10
on admission after traumatic brain injury. These
data provided a rationale for a trial of 24 hours Treatment Guidelines
of hypothermia therapy in children, and we began Guidelines for cooling, rewarming, and manage-
our study shortly after publication of these find- ment of intracranial pressure and cerebral perfu-
ings.6 We hypothesized that, as compared with sion pressure were established by consensus of the
normothermia (36.5 to 37.5°C), treatment with participating investigators, after a review of all
hypothermia (32 to 33°C) for 24 hours, started relevant evidence.11 Patients were cooled with the
within 8 hours after severe traumatic brain injury, use of surface cooling techniques. Esophageal
would reduce the risk of an unfavorable outcome temperature was maintained at a mean (±SD) of
at 6 months.
32.5±0.5°C for 24 hours.11 For rewarming, the
temperature was increased at a rate of 0.5°C every
Methods
2 hours. After rewarming in the hypothermia
group, and beginning immediately in the normo-
Patients and Sites
thermia group, temperature was maintained at
We conducted this study at 17 centers in three 37±0.5°C until intracranial hypertension resolved.
countries (see the Appendix). Patients were eligi- We documented baseline characteristics, includ-
ble if they were 1 to 17 years of age and had trau- ing demographic and injury data, the score on
matic brain injury, a score on the Glasgow Coma the Glasgow Coma Scale, and Pediatric Trauma
Scale of 8 or less at the scene of the accident or in Score.12
the emergency room, a computed tomographic
(CT) scan that showed an acute brain injury, and Study Outcomes
a need for mechanical ventilation. We excluded pa- The primary outcome for the study was the pro-
tients who were screened more than 8 hours after portion of patients who had an unfavorable out-
injury, as well as patients with refractory shock, come — defined as severe disability, a persistent
suspected brain death, nonaccidental injury, pro- vegetative state, or death — at 6 months, which
longed cardiac arrest at the scene of the accident, was assessed without knowledge of the treatment
high cervical spinal cord injury, severe neurode- assignments. With the use of a scripted telephone
velopmental disability before the injury, brain in- interview, a trained site psychologist assessed each
jury due to a gunshot wound, acute isolated epi- patient according to the six-point Pediatric Cere-
dural hematoma, or pregnancy. The study was bral Performance Category scale (with a score of
approved by the research ethics board at each par- 1 representing normal performance, 2 mild dis-
2448
n engl j med 358;23 www.nejm.org june 5, 2008
Downloaded from www.nejm.org on August 11, 2009 . For personal use only. No other uses without permission.
Copyright © 2008 Massachusetts Medical Society. All rights reserved.

---

Hypothermia Therapy after Traumatic Brain Injury in Children
ability, 3 moderate disability, 4 severe disability, compared with children 7 years of age or older.
5 a persistent vegetative state, and 6 death).13,14 Exploratory analyses were performed with the use
A score on this scale was also assessed by means of logistic-regression models to adjust for the ef-
of an interview of the parents or guardians 1 week fects of clinical factors that may be associated with
after the head injury, in which they were asked to the outcome in children with traumatic brain
estimate the child’s level of function before the injury — intervention group (hypothermia vs. nor-
injury, and 1, 3, and 12 months after the injury. mothermia), an age of less than 7 years as com-
In addition, measures of intelligence,15-17 memory pared with an age of 7 years or more, score on
functioning,18,19 and speed of information pro- the Glasgow Coma Scale (scores range from 3 to
cessing were assessed in all of the children who 15, with higher scores indicating better function)
were able to participate in testing 3 and 12 months on admission to the hospital (3 or 4 vs. 5 to 8),
after the injury; at these time points, parents were temperature on admission of less than 35°C ver-
also interviewed with the use of an instrument sus 35°C or more, intracranial pressure of more
that assesses a child’s executive functions.20 Blood than 20 mm Hg versus 20 mm Hg or less, hypo-
pressure, intracranial pressure, cointerventions, tension or hypoxia present or absent on admis-
lengths of stay in the intensive care unit (ICU) and sion, number of therapies used to control intra-
in the hospital, and the rates of adverse events, in- cranial pressure (0 to 3 vs. 4 or 5), hypertonic
cluding hypotension, infection, bleeding, arrhyth- saline used or not used to control intracranial
mias, and electrolyte abnormalities, were also re- pressure, and three variables that were noted on
corded.
CT scans (presence or absence of extradural he-
matoma, cerebral edema, and midline shift). Mor-
Statistical Analysis
tality was analyzed by means of chi-square tests.
We estimated that enrolling 202 children would Further exploratory analyses of mortality were
allow us to detect a reduction of 20 percentage performed with the use of Cox proportional-
points in the absolute risk21 of an unfavorable out- hazards models, with unadjusted and adjusted
come, from 50%22 in the control (normothermia) analyses of the time to death in the two groups.
group to 30% in the hypothermia group, with a Scores on the Pediatric Cerebral Performance
two-sided alpha level of 0.05 and a statistical pow- Category scale were also compared over time,
er of 80%. Assuming a 10% rate of loss to follow- with the use of an analysis of variance with re-
up, our estimated sample size was 222 children. peated measures.
Two planned interim analyses of the safety and
All secondary outcomes were analyzed accord-
efficacy of the study treatment were reviewed by ing to the intention-to-treat principle. Continuous
a blinded, independent data and safety monitoring variables were analyzed first with independent
committee after 33% and 66% of the patients had Student’s t-tests and then with generalized linear
been enrolled and followed for 6 months after in- models. Categorical variables, including rates of
jury.23 The rates of an unfavorable outcome, death, adverse events, were analyzed with the use of the
and adverse events were compared between groups, chi-square test. Additional analyses of variables
with P<0.001 designated as the threshold for stop- related to the process of care, including lengths
ping the trial if there was compelling evidence of of stay in the ICU and hospital, were performed
significant benefit or harm in either one of the by means of nonparametric procedures (the Wil-
study groups. There were no plans to stop the trial coxon rank-sum test).
early if there appeared to be no evidence of un-
equal benefit or harm. At each interim analysis,
R esults
the data and safety monitoring committee recom-
mended the continuation of the trial.
Patients and Treatment Assignment
The statistical analysis of the primary outcome From February 1999 to October 2004, a total of
was conducted with the use of the chi-square test 1441 consecutive patients with traumatic brain in-
according to the intention-to-treat principle and jury were admitted to the pediatric ICUs that par-
then according to the treatment received. Sensitiv- ticipated in the study (see the figure in the Sup-
ity analyses were performed to account for pa- plementary Appendix, available with the full text
tients with missing data for primary outcomes. of this article at www.nejm.org). Three hundred
We planned eight a priori subgroup analyses, in- twenty-seven of the 1441 patients (23%) met the
cluding one for children less than 7 years of age as eligibility criteria. Of the 327 eligible patients, 69
n engl j med 358;23 www.nejm.org june 5, 2008
2449
Downloaded from www.nejm.org on August 11, 2009 . For personal use only. No other uses without permission.
Copyright © 2008 Massachusetts Medical Society. All rights reserved.

---

T h e n e w e ng l a n d j o u r na l o f m e dic i n e
were not identified and their parents or guardians 108 patients (3%) in the hypothermia group and
were not approached for consent within 8 hours 4 of 117 (3%) in the normothermia group. Base-
after injury, 33 had parents or guardians who de- line characteristics of the patients who were en-
clined consent, and 225 (69% of eligible patients) rolled in the study are presented in Table 1.
were enrolled. One hundred eight patients were
randomly assigned to hypothermia therapy, and Intervention and Monitoring
117 patients to normothermia. A total of 7 patients One hundred two of the 108 patients (94%) who
(3% of enrolled patients) did not have a monitor were assigned to hypothermia therapy received the
inserted to measure intracranial pressure — 3 of intervention (mean temperature, 33.1±1.2°C for 24
Table 1. Baseline Characteristics of the Patients.*
Hypothermia
Normothermia
Group
Group
Characteristic
(N = 108)
(N = 117)
Age — yr
9.8±4.9
10.2±4.8
Male sex — no. (%)
70 (65)
71 (61)
Weight — kg
39.4±21.2
40.3±21.1
GCS on admission — median (IQR)
5 (4?6)
5 (3?6)
Pediatric Trauma Score — median (IQR)
3 (2?5)
3 (2?5)
Cause of injury — no. (%)
Motor vehicle
70 (65)
64 (55)
Passenger
30 (43)
32 (51)
Pedestrian
40 (57)
31 (49)
Bicycle
12 (11)
18 (15)
Fall
17 (16)
21 (18)
Other
9 (8)
14 (12)
Initial presentation — no. (%)
Hypotension on admission
8 (7)
3 (3)
Hypoxia on admission
3 (3)
0
Transfer from another institution
62 (57)
79 (68)
CT findings — no. (%)
Extradural hematoma
10 (9)
22 (19)
Intracerebral hematoma
62 (57)
62 (53)
Cerebral edema
85 (79)
83 (71)
Midline shift
33 (31)
27 (23)
Skull fracture
59 (55)
59 (50)
Other injuries — no. (%)†
Spinal cord injury
1 (1)
1 (1)
Thoracic injury
38 (35)
36 (31)
Cardiovascular injury
4 (4)
1 (1)
Abdominal injury
10 (9)
13 (11)
Genitourinary injury
11 (10)
10 (9)
Major fracture or dislocation
25 (23)
22 (19)
* Plus?minus values are means ±SD. GCS denotes Glasgow Coma Scale, and IQR interquartile range.
† Thoracic injuries included major airway trauma, hemothorax, pulmonary contusion, pulmonary laceration, and ruptured
diaphragm. Cardiovascular injuries included major-vessel injuries, cardiac lacerations, and myocardial contusion. Ab-
dominal injuries included injuries to the liver, spleen, bowel, and pancreas. Major fractures and dislocations included
facial, spinal, pelvic, and long-bone fractures and dislocation of major joints.
2450
n engl j med 358;23 www.nejm.org june 5, 2008
Downloaded from www.nejm.org on August 11, 2009 . For personal use only. No other uses without permission.
Copyright © 2008 Massachusetts Medical Society. All rights reserved.

---

Hypothermia Therapy after Traumatic Brain Injury in Children
hours). The mean time to initiation of cooling was
6.3±2.3 hours (range, 1.6 to 19.7) after injury, the
Normothermia
mean time to attainment of the target tempera-
37
ture range was 3.9±2.6 hours (range, 0.0 to 11.8),
and the mean time to completion of rewarming
36
after the 24-hour period at the target temperature
was 18.8±14.9 hours (range, 2.5 to 148.0) (Fig. 1).
In 114 of the 117 patients (97%) in the normother-
35
mia group, a normal temperature (36.9±0.5°C) was
maintained for 24 hours. No patient who was as-
Temperature (°C)
34
signed to the normothermia group was treated
with hypothermia. Any failures to follow the tem-
Hypothermia
perature protocol and treatment guidelines were
33
reviewed by a clinical care committee, and rapid
feedback was given to the principal and site inves-
0Initial Before 0 4 8 12 16 20 24 28 32 36 40 44 48
tigators to improve compliance with the protocol.
Randomi-
zation
Hours after Randomization
Cointerventions
Figure 1. Temperature of Patients in the Hypothermia and Normothermia
A significantly higher proportion of patients in the
Groups.
normothermia group than in the hypothermia
Data are shown as AUTHOR:
means an Hutchison
d 95% confidence inte RETAKE
rvals.
1st
ICM
group received hypertonic saline to control intra-
FIGURE:
2nd
REG F
1 of 2
cranial pressure during the first 24 hours (Table 2).
3rd
CASE
Revised
A significantly higher proportion of patients in the logistic-regression
EMail models adjusted for
Line clinica
4-C l fac- SIZE
ARTIST: ts
hypothermia group than in the normothermia tors that may be
Enon associated with the
H/T o
Combo utcom
H/T e in 22p3
group received vasoactive drugs for hypotension children with traumatic brain injury, the adj
AUTHOR, PLEASE NOTE: u
sted
during the rewarming period (Table 2). Otherwise, odds ratio for an unfavorable outcome with hypo-
Figure has been redrawn and type has been reset.
Please check carefully.
there were no significant imbalances in the rate thermia therapy was 2.33 (95% CI, 0.92 to 5.93;
at which therapies were used to treat intracranial P = 0.08). JOB: 35823
ISSUE: 06-05-08
hypertension or in the fluid balance between the
We also performed an analysis of the primary
groups.
outcome according to the treatment received, but
we noted no major differences from the intention-
Study Outcomes
to-treat analysis. In a subgroup analysis of patients
Data on primary outcomes were available for 205 7 years of age or older, the risk of an unfavorable
patients (91%). Overall, 20 of the 225 patients (9%) outcome was higher with hypothermia therapy
were lost to follow-up at 6 months — 6 of 108 than with normothermia (relative risk, 1.71; 95%
patients (6%) in the hypothermia group and 14 of CI, 0.96 to 3.06; P = 0.06). The relative risk of an
117 (12%) in the normothermia group. Thirty-two unfavorable outcome was also higher with hypo-
of 102 patients (31%) in the hypothermia group thermia therapy in the subgroup that included
and 23 of 103 (22%) in the normothermia group patients whose recorded measurements of intra-
had an unfavorable outcome at 6 months (relative cranial pressure were all less than 20 mm Hg
risk of an unfavorable outcome with hypothermia (relative risk, 2.12; 95% CI, 1.07 to 4.19; P = 0.03).
therapy, 1.41; 95% confidence interval [CI], 0.89 There were no significant differences in the other
to 2.22; P = 0.14) (Table 3). In a sensitivity analysis subgroups that were analyzed. The Pediatric Cere-
that accounted for the 20 patients who were lost bral Performance Category scores improved with
to follow-up at 6 months and assuming the worst time after the injury in both groups; the improve-
case in the hypothermia group and the best case ment was greater in the normothermia group than
in the normothermia group, hypothermia thera- in the hypothermia group 1, 3, 6, and 12 months
py was associated with an unfavorable outcome after the injury, although the difference was not
(P = 0.001); with the opposite scenario (best case significant (P = 0.07).
and worst case in the two groups, respectively),
There were 23 deaths (21%) in the hypothermia
there was no increased risk of an unfavorable out- group, as compared with 14 deaths (12%) in the
come with hypothermia therapy (P = 0.82). With normothermia group (relative risk of death with
n engl j med 358;23 www.nejm.org june 5, 2008
2451
Downloaded from www.nejm.org on August 11, 2009 . For personal use only. No other uses without permission.
Copyright © 2008 Massachusetts Medical Society. All rights reserved.

---

T h e n e w e ng l a n d j o u r na l o f m e dic i n e
Table 2. Cointerventions to Manage Intracranial Pressure and Support Blood Pressure.*
Hypothermia
Normothermia
Group
Group
Therapy
(N = 108)
(N = 117)
P Value
0?24 Hr
Cerebrospinal fluid drainage — no. (%)
45 (42)
55 (47)
0.42
Mannitol — no. (%)
54 (50)
70 (60)
0.14
Hypertonic saline (3%) — no. (%)
34 (31)
54 (46)
0.02
Hyperventilation — no. (%)
0.99
PaCO2, 30?35 mm Hg
11 (10)
12 (10)
PaCO2, <30 mm Hg
47 (44)
50 (43)
Barbiturates — no. (%)
15 (14)
22 (19)
0.32
Dopamine, epinephrine, or norepinephrine — no. (%)
59 (55)
56 (48)
0.31
Transfusions of packed cells — no. (%)
38 (35)
30 (26)
0.12
Fluid balance — ml/24 hr
0.78
Median
700
750
Interquartile range
250?1700
300?1500
25?72 Hr
Cerebrospinal fluid drainage — no. (%)
52 (48)
56 (48)
0.97
Mannitol — no. (%)
61 (56)
67 (57)
0.91
Hypertonic saline (3%) — no. (%)
54 (50)
64 (55)
0.48
Hyperventilation — no. (%)
0.09
PaCO2, 30?35 mm Hg
6 (6)
9 (8)
PaCO2, <30 mm Hg
48 (44)
35 (30)
Barbiturates — no. (%)
23 (21)
26 (22)
0.87
Dopamine, epinephrine, or norepinephrine — no. (%)
92 (85)
66 (56)
<0.001
Transfusions of packed cells — no. (%)
33 (31)
36 (31)
0.97
Fluid balance — ml/24 hr
0.8
Median
828
617
Interquartile range
380?1550
500?1300
* The guidelines that were used for the management of intracranial pressure and cerebral perfusion pressure are outlined
in Hutchison et al.11 If the intracranial pressure was greater than 20 mm Hg and if reversible causes had been ruled
out or treated, therapies were given in the following order: cerebrospinal fluid drainage if an external ventricular drain
was present, mannitol, hypertonic saline, hyperventilation, and barbiturates; patients were treated with the next listed
therapy only if the previous therapy was ineffective. Therapies for cerebral perfusion pressure were given if cerebral per-
fusion was less than 60 mm Hg for patients 10 years of age or older or less than 50 mm Hg for those less than 10
years of age. PaCO2 denotes partial pressure of arterial carbon dioxide.
hypothermia therapy, 1.40; 95% CI, 0.90 to 2.27;
There were no significant differences in the
P = 0.06) (Table 3 and Fig. 2). In the unadjusted durations of intracranial pressure monitoring, me-
Cox proportional-hazards model, the hazard ratio chanical ventilation, or stays in the pediatric ICU
for death with hypothermia therapy was 1.84 (95% or the hospital between the two groups (Table 3).
CI, 0.95 to 3.58; P = 0.07), whereas in the model Intracranial pressures were lower during the cool-
adjusted for clinical factors that may be associated ing period and higher during the rewarming pe-
with the outcome in children with traumatic brain riod in the hypothermia group, as compared with
injury, the hazard ratio for death was 2.36 (95% the normothermia group; the difference was sig-
CI, 1.04 to 5.37; P = 0.04).
nificant at 16 hours (P = 0.02), 24 hours (P = 0.01),
2452
n engl j med 358;23 www.nejm.org june 5, 2008
Downloaded from www.nejm.org on August 11, 2009 . For personal use only. No other uses without permission.
Copyright © 2008 Massachusetts Medical Society. All rights reserved.

---

Hypothermia Therapy after Traumatic Brain Injury in Children
Table 3. Primary and Secondary Outcomes.*
Relative Risk
Hypothermia
Normothermia
or Absolute
Group
Group
Difference
Outcome
(N = 108)
(N = 117)
(95% CI)
P Value
Primary
PCPC score 4?6 — no./total no. (%)
32/102 (31)
23/103 (22)
1.41 (0.89 to 2.22)
0.14
Secondary
Overall mortality — no. (%)
23 (21)
14 (12)
1.40 (0.90 to 2.27)
0.06
Duration of care — days
ICP monitoring
6.4±4.0
6.0±3.0
0.4 (?0.6 to 1.3)
0.45
Mechanical ventilation
9.5±6.1
8.9±5.7
0.7 (?0.9 to 2.2)
0.41
Intensive care unit
11.5±7.1
11.3±7.2
0.2 (?1.7 to 2.1)
0.85
Hospital
30.2±31.7
28.3±24.2
1.9 (?5.8 to 9.5)
0.63
Physiological variables at 0 to 24 hr
ICP — mm Hg (95% CI)
14.7±10.7
17.1±11.1
0.12
(12.7 to 16.8)
(15 to 19.1)
CPP — mm Hg (95% CI)
66.4±12
64.3±11.5
0.19
(64.1 to 68.8)
(62.2 to 66.5)
Mean blood pressure — mm Hg (95% CI)
80.6±9.8
81.4±10.1
0.56
(78.7 to 82.5)
(79.5 to 83.2)
Heart rate — beats/min (95% CI)
81.5±16.7
108.1±19.1
<0.001
(78.3 to 84.7)
(104.6 to 111.6)
Physiological variables at 25 to 72 hr
ICP — mm Hg (95% CI)
17.1±7.1
17.4±10.7
0.77
(15.6 to 18.5)
(15.4 to 19.4)
CPP — mm Hg (95% CI)
60.8±7.8
66±10.8
<0.001
(59.2 to 62.4)
(64 to 68.1)
Mean blood pressure — mm Hg (95% CI)
77.7±7.6
83.4±8
<0.001
(76.2 to 79.2)
(81.9 to 84.9)
Heart rate — beats/min (95% CI)
100.7±18.3
105.3±18.2
0.07
(97.1 to 104.3)
(101.9 to 108.6)
Adverse events — no. (%)
Hypotension
0–24 hr
27 (25)
18 (15)
0.07
25–72 hr
49 (45)
38 (32)
0.047
Ventricular tachycardia
1 (1)
0
0.48
Ventricular fibrillation
0
1 (1)
0.52
ARDS
8 (8)
6 (5)
0.47
Pneumonia
39 (36)
51 (44)
0.25
Septic shock
2 (2)
2 (2)
0.38
Other infections
16 (15)
19 (16)
0.81
Bleeding
Late intracranial
3 (3)
5 (4)
0.72
Extracranial
1 (1)
1 (1)
0.96
* Plus?minus values are means ±SD. Relative risks are given for the Pediatric Cerebral Performance Category (PCPC) score and for overall
mortality. Absolute differences between the values for the hypothermia and normothermia groups are given for the duration of ICP monitor-
ing and of mechanical ventilation and for the length of the intensive care unit and hospital stays. The primary outcome was the proportion of
patients with an unfavorable outcome, defined as severe disability, a persistent vegetative state, or death (PCPC score of 4 to 6) at 6 months.
ARDS denotes acute respiratory distress syndrome, CPP cerebral perfusion pressure, and ICP intracranial pressure.
n engl j med 358;23 www.nejm.org june 5, 2008
2453
Downloaded from www.nejm.org on August 11, 2009 . For personal use only. No other uses without permission.
Copyright © 2008 Massachusetts Medical Society. All rights reserved.

---

T h e n e w e ng l a n d j o u r na l o f m e dic i n e
differences in neuropsychological outcomes be-
100
tween the groups.
Normothermia
In the hypothermia group, as compared with
the normothermia group, the mean serum glucose
level was significantly higher in the first 24 hours
75
Hypothermia
(171.2±91.9 mg per deciliter [9.5±5.1 mmol per li-
ter] vs. 138.7±46.8 mg per deciliter [7.7±2.6 mmol
per liter], P = 0.002), the platelet count was signifi-
50
cantly lower (174,900±61,000 per cubic millimeter
vs. 192,000±67,200 per cubic millimeter, P = 0.05),
and the prothrombin time and serum lactate level
Proportion Surviving (%)
were significantly higher between 25 and 72 hours
25
(prothrombin time, 15.3±2.6 seconds vs. 14.3±2.5
Censored observations
seconds; P = 0.03; lactate level, 11.7±7.2 mg per
deciliter vs. 9.0±5.4 mg per deciliter; P = 0.03).
0 0
50
100
150
200
Discussion
Days
In this multicenter, randomized, controlled trial,
Figure 2. Kaplan–Meier Estimates of Survival.
AUTHOR: Hutchison
RETAKE
1st
we found that among children with major head in-
The causes of ICM
death in the hypothermia group were brain death (4 patients),
FIGURE:
2nd
2 of 2
jury, moderate hypothermia therapy (32 to 33°C),
brain trauma REG
(12 F
), brain ischemia (1), hypoxia resulting from 3rd
severe lung
injury (3), an CASE
d septic shock (1), with unknown causes Revised
in 2 patients. The
initiated within 8 hours after the injury and main-
4-C
causes of dea EMail
Line
th in the normothermia group were brain d SIZE
eath (2 patients),
tained for 24 hours, did not improve the functional
ARTIST: ts
H/T
H/T
brain trauma (9), and brain ischemia (2), with an unknow 22p3
Enon
n cause in 1 patient.
Combo
outcome at 6 months. We observed a trend toward
AUTHOR, PLEASE NOTE:
increased mortality in the hypothermia group and
Figure has been redrawn and type has been reset.
found no evidence of a benefit with respect to any
48 hours (P = 0.01), and 7
Please check carefully. 2 hours (P = 0.03). The secondary outcomes, including functional and neu-
JOB:
hea
35823 rt rate was significantly lowe
ISSUE: r in pa
06-05-08 tients who ropsychological outcomes at 3 and 12 months,
were undergoing hypothermia therapy than in length of stay in the ICU or hospital, and adverse
those in the normothermia group (P<0.001) (Table events.
3). During rewarming after hypothermia therapy,
While we were conducting this trial, the results
we noted significantly more episodes of hypoten- of a large study of hypothermia therapy in 392
sion (P = 0.047) and lower mean blood pressures adults with severe traumatic brain injury were pub-
and cerebral perfusion pressures (P<0.001 for both lished.24 This study, conducted by Clifton et al.,
comparisons) (Table 3). Hypotension was treated did not show meaningful benefits in the rate of
with boluses of intravenous fluids and vasopres- survival or in functional outcomes and documented
sors according to the study treatment guidelines. more complications, such as critical hypotension,
No other serious adverse events were significantly in adults who were treated with hypothermia for
associated with the use of hypothermia therapy 48 hours than in those who were treated with
(Table 3).
normothermia.24 Three of four systematic reviews
We performed neuropsychological follow-up for that pooled trial data on hypothermia, including
59% of the survivors at 3 months and for 63% of 1130 adults in the four systematic reviews, also
the survivors at 12 months. Patients were not as- noted the absence of a benefit from hypothermia
sessed if they were too young to participate in therapy after traumatic brain injury.25-28 Hypo-
testing (generally, younger than 5 years of age) or thermia therapy appears to be of benefit in some
had severe functional or physical impairment that adults and newborns with a hypoxic?ischemic
made assessment impossible, or if their parents or brain injury29-33 but not in adults with traumatic
guardians could not be contacted or refused fol- brain injury.24
low-up. Scores on assessments of long-term visual
A potential limitation of our trial is that the
memory were significantly worse in the hypother- mean time to the initiation of hypothermia was
mia group than in the normothermia group 12 6.3 hours. It is plausible that hypothermia therapy
months after injury (P = 0.05) (see the table in the might be more effective if it were initiated earlier,
Supplementary Appendix). There were no other as was reported in an animal model of traumatic
2454
n engl j med 358;23 www.nejm.org june 5, 2008
Downloaded from www.nejm.org on August 11, 2009 . For personal use only. No other uses without permission.
Copyright © 2008 Massachusetts Medical Society. All rights reserved.

---

Hypothermia Therapy after Traumatic Brain Injury in Children
brain injury, in which hypothermia was instituted uted to the failure of hypothermia therapy. The
within 15 minutes34; however, there would be rate of loss to follow-up at 6 months was less
great logistical challenges to conducting such a than 10%.
trial. We did not detect a benefit in our subgroup
On the basis of the results of this multicenter
of patients who were treated early (data not trial, we conclude that the use of this hypothermia
shown). Another limitation of our trial is that protocol is not warranted for the treatment of se-
more prolonged therapy might have resulted in vere head injury in children. Further research may
beneficial outcomes. We chose to treat patients for elucidate whether earlier implementation of hypo-
24 hours on the basis of evidence from studies of thermia therapy or more prolonged hypothermia
adults.6 However, in one systematic review, a sub- therapy would improve the outcome in children
group analysis suggested that hypothermia ther- with severe traumatic brain injury.
apy given for more than 48 hours reduced the risks
of death and a poor neurologic outcome (relative
Supported by grants from the Canadian Institutes of Health Re-
search (Canadian Neurotrauma Research Program) (MCT50398),
risk, as compared with normothermia, 0.70 [95% the Ontario Neurotrauma Foundation (ONBO-00009 and
CI, 0.56 to 0.87] and 0.65 [95% CI, 0.48 to 0.89], ONRO-41), the Rick Hansen Institute, the Hospital for Sick Chil-
respectively).28 Hypothermia, with adjustment of dren Foundation (XG 99-057), Physicians Services Incorporated
(98-62), Fonds de la Recherche en Santé du Québec (004095-
degree and depth according to intracranial pres- 104), the Children’s Hospital of Eastern Ontario Research Insti-
sure, may be of benefit as a therapy for refractory tute (98/16S[E]), and Direction de la Recherche Clinique, Assis-
intracranial hypertension in many children with tance Publique–Hôpitaux de Paris (P010208).
Dr. Lacroix reports receiving consulting fees from Johnson &
severe traumatic brain injury.4,5 Another potential Johnson. No other potential conflict of interest relevant to this
limitation of our study is the small sample. Future article was reported.
studies should be powered to detect smaller treat-
We thank the study patients and their families for joining the
trial and helping us address the potential benefits and risks of
ment effects.
hypothermia therapy; the research assistants, psychologists, in-
This study has several strengths. We used a tensivists, neurosurgeons, and pediatric ICU nurses at each of
similar approach to control intracranial hyperten- the 17 study centers who participated in the study; members of
the Chalmers Research Group and Ottawa Hospital Methods
sion and to manage fluid balance in the two Centre for expert coordination and data analysis; and Drs. Anne-
groups.11 We found no evidence that cointerven- Marie Guerguerian, Brian Kavanagh, Maureen Meade, and Deb-
tions such as the management of intracranial pres- orah Cook for their critical review of the manuscript. We fondly
remember Dr. Sid Watkins, Southampton, United Kingdom,
sures and fluids or other aspects of care contrib- who died suddenly during the conduct of this study.
APPENDIX
From the Departments of Critical Care Medicine and Pediatrics (J.S.H., D.J.B.) and Surgery (Neurosurgery) (P.B.D.), Hospital for Sick
Children; Interdepartmental Division of Critical Care, Faculty of Medicine, University of Toronto (J.S.H., D.J.B.); and Neuroscience and
Mental Health Research Program, Hospital for Sick Children Research Institute (J.S.H., M.A.B.) — all in Toronto; Chalmers Research
Group, Children’s Hospital of Eastern Ontario Research Institute, (R.E.W., D.M.); and Clinical Epidemiology Program, Ottawa Health
Research Institute, Ottawa Hospital, Department of Medicine (Critical Care) (P.C.H., S.D., D.F.) and Departments of Pediatrics and
Epidemiology & Community Medicine, Faculty of Medicine (D.M.), University of Ottawa — all in Ottawa; Division of Intensive Care,
Department of Pediatrics, Ste.-Justine Hospital (J.L.); and Department of Pediatrics&