n engl j med 364;16 nejm.org april 21, 2011
1493
The new england
journal
of medicine
established in 1812
april 21, 2011
vol. 364 no. 16
Decompressive Craniectomy in Diffuse Traumatic Brain Injury
D. James Cooper, M.D., Jeffrey V. Rosenfeld, M.D., Lynnette Murray, B.App.Sci., Yaseen M. Arabi, M.D.,
Andrew R. Davies, M.B., B.S., Paul D’Urso, Ph.D., Thomas Kossmann, M.D., Jennie Ponsford, Ph.D.,
Ian Seppelt, M.B., B.S., Peter Reilly, M.D., and Rory Wolfe, Ph.D., for the DECRA Trial Investigators
and the Australian and New Zealand Intensive Care Society Clinical Trials Group*
Abstr act
From the Departments of Intensive Care
(D.J.C., L.M., A.R.D.) and Neurosurgery
(J.V.R.), Alfred Hospital; the Depart-
ments of Epidemiology and Preventive
Medicine (D.J.C., L.M., A.R.D., J.P., R.W.)
and Surgery (J.V.R.), Monash University;
the Neurosciences Clinical Institute
(P.D.) and the Monash–Epworth Reha-
bilitation Research Centre (J.P.), Epworth
Healthcare; and the Epworth Hospital
(T.K.) — all in Melbourne, VIC; the De-
partment of Intensive Care Medicine,
Nepean Hospital, University of Sydney,
Sydney, NSW (I.S.); and the Department
of Neurosurgery, Royal Adelaide Hospi-
tal, Adelaide, SA (P.R.) — all in Australia;
and the Intensive Care Department, King
Saud Bin Abdulaziz University for Health
Sciences, Riyadh, Saudi Arabia (Y.M.A.).
Address reprint requests to Dr. Cooper at
the Department of Intensive Care, Alfred
Hospital, Commercial Road, Melbourne,
VIC 3004, Australia, or at jamie.cooper@
monash.edu.
*Investigators in the Decompressive
Craniectomy (DECRA) trial are listed in
the Supplementary Appendix, available
at NEJM.org.
This article (10.1056/NEJMoa1102077) was
published on March 25, 2011, and updated
on November 23, 2011, at NEJM.org.
N Engl J Med 2011;364:1493-502.
Copyright © 2011 Massachusetts Medical Society.
Background
It is unclear whether decompressive craniectomy improves the functional outcome
in patients with severe traumatic brain injury and refractory raised intracranial
pressure.
Methods
From December 2002 through April 2010, we randomly assigned 155 adults with
severe diffuse traumatic brain injury and intracranial hypertension that was refrac-
tory to first-tier therapies to undergo either bifrontotemporoparietal decompressive
craniectomy or standard care. The original primary outcome was an unfavorable
outcome (a composite of death, vegetative state, or severe disability), as evaluated
on the Extended Glasgow Outcome Scale 6 months after the injury. The final pri-
mary outcome was the score on the Extended Glasgow Outcome Scale at 6 months.
Results
Patients in the craniectomy group, as compared with those in the standard-care
group, had less time with intracranial pressures above the treatment threshold
(P<0.001), fewer interventions for increased intracranial pressure (P<0.02 for all
comparisons), and fewer days in the intensive care unit (ICU) (P<0.001). However,
patients undergoing craniectomy had worse scores on the Extended Glasgow Out-
come Scale than those receiving standard care (odds ratio for a worse score in the
craniectomy group, 1.84; 95% confidence interval [CI], 1.05 to 3.24; P = 0.03) and a
greater risk of an unfavorable outcome (odds ratio, 2.21; 95% CI, 1.14 to 4.26;
P = 0.02). Rates of death at 6 months were similar in the craniectomy group (19%)
and the standard-care group (18%).
Conclusions
In adults with severe diffuse traumatic brain injury and refractory intracranial hy-
pertension, early bifrontotemporoparietal decompressive craniectomy decreased
intracranial pressure and the length of stay in the ICU but was associated with more
unfavorable outcomes. (Funded by the National Health and Medical Research
Council of Australia and others; DECRA Australian Clinical Trials Registry number,
ACTRN012605000009617.)
The New England Journal of Medicine
Downloaded from nejm.org at UQ Library on August 2, 2018. For personal use only. No other uses without permission.
Copyright © 2011 Massachusetts Medical Society. All rights reserved.
The
new engla nd journal
o f
medicine
n engl j med 364;16 nejm.org april 21, 2011
1494
A
mong patients who are hospital-
ized with severe traumatic brain injury,
60% either die or survive with severe dis-
ability.
1-3
Of Australia’s population of 22 million,
4
approximately 1000 patients annually sustain a
severe traumatic brain injury, with associated life-
time costs estimated at $1 billion.
5
In the United
States, the annual burden of traumatic brain in-
jury is more than $60 billion.
6
After severe traumatic brain injury, medical
and surgical therapies are performed to minimize
secondary brain injury.
7-9
Increased intracranial
pressure, which is typically caused by cerebral
edema, is an important secondary insult.
7,9,10
Al-
though few data regarding the monitoring of in-
tracranial pressure are available from randomized,
controlled trials, such monitoring is recommend-
ed by international clinical practice guidelines,
and first-tier therapies are used to control intra-
cranial pressure.
11
However, many patients with
severe traumatic brain injury have raised intracra-
nial pressure that is refractory to first-tier thera-
pies.
11,12
In such cases, surgical decompressive
craniectomy is performed with increasing fre-
quency to control intracranial pressure.
10
We de-
signed the multicenter, randomized, controlled
Decompressive Craniectomy (DECRA) trial
13,14
to
test the efficacy of bifrontotemporoparietal de-
compressive craniectomy in adults under the age
of 60 years with traumatic brain injury in whom
first-tier intensive care and neurosurgical thera-
pies had not maintained intracranial pressure
below accepted targets.
Methods
Trial Design
From December 2002 through April 2010, we re-
cruited adults with severe traumatic brain injury
in the intensive care units (ICUs) of 15 tertiary
care hospitals in Australia, New Zealand, and Sau-
di Arabia. The trial protocol (available with the
full text of this article at NEJM.org) was designed
by the study’s executive committee and approved
by the ethics committee at each study center.
Patients
Patients were eligible for participation in the trial
if they were between the ages of 15 and 59 years
and had a severe, nonpenetrating traumatic brain
injury. Among patients who were evaluated either
after resuscitation or before intubation, this injury
was defined as a score of 3 to 8 on the Glasgow
Coma Scale (on which scores range from 3 to 15,
with lower scores indicating reduced levels of con-
sciousness) or Marshall class III (moderate diffuse
injury on computed tomography [CT]).
15
Patients
were excluded if they were not deemed suitable for
full active treatment by the clinical staff caring
for the patient or if they had dilated, unreactive
pupils, mass lesions (unless too small to require
surgery), spinal cord injury, or cardiac arrest at
the scene of the injury. In all cases, the patients’
next of kin provided written informed consent.
Study Procedures
All patients in the study were treated in ICUs with
advanced neurosurgical management capabilities
and equipment, including the availability of intra-
cranial-pressure monitoring with the use of either
an external ventricular drain or a parenchymal
catheter. Patients received treatment for intracra-
nial hypertension whenever the intracranial pres-
sure was greater than 20 mm Hg.
8,9,11,12,16
We
defined an early refractory elevation in intracra-
nial pressure as a spontaneous (not stimulated)
increase in intracranial pressure for more than
15 minutes (continuously or intermittently) with-
in a 1-hour period, despite optimized first-tier
interventions. Such interventions included opti-
mized sedation, the normalization of arterial car-
bon dioxide pressure, and the use of mannitol,
hypertonic saline, neuromuscular blockade, and
external ventricular drainage.
Within the first 72 hours after injury, we ran-
domly assigned patients either to undergo decom-
pressive craniectomy plus standard care or to re-
ceive standard care alone, using an automated
telephone system. Randomization was stratified
according to center and the technique that was
used to measure intracranial pressure (external
ventricular drain or parenchymal catheter) in
blocks of two or four patients. A standardized
surgical approach, modeled on the Polin tech-
nique,
17
was used. This approach included a large
bifrontotemporoparietal craniectomy with bilat-
eral dural opening to maximize the reduction in
intracranial pressure
13,14
(for details, see the
Supplementary Appendix, available at NEJM.org).
The sagittal sinus and falx cerebri were not di-
vided. After craniectomy, the excised bone was
stored at −70°C or in a subcutaneous abdominal
pouch, according to the standard practice of the
operating surgeon. After all swelling and infection
The New England Journal of Medicine
Downloaded from nejm.org at UQ Library on August 2, 2018. For personal use only. No other uses without permission.
Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Decompressive Craniectomy in Traumatic Brain Injury
n engl j med 364;16 nejm.org april 21, 2011
1495
had resolved, 2 to 3 months after craniectomy,
the bone was replaced.
Standard care from the time of enrollment
followed clinical practice guidelines
13
that were
based on those recommended by the Brain Trauma
Foundation.
8
In the two study groups, second-tier
options for refractory elevation of intracranial pres-
sure included mild hypothermia (to 35°C), the
optimized use of barbiturates, or both. For patients
receiving standard care, the trial protocol per-
mitted the use of lifesaving decompressive cra-
niectomy after a period of 72 hours had elapsed
since admission.
Assessments and Data Collection
Research coordinators at each institution col-
lected the trial data. All source data were verified
in every patient by monitors. At baseline, demo-
graphic and clinical characteristics were record-
ed from medical files. These data included the
initial CT findings, which were scored with the
use of the Marshall criteria, and the Injury Sever-
ity Score (on a scale ranging from 0 to 75, with
higher scores indicating greater injury severity).
The Trauma Score–Injury Severity Score
18
(on a
scale ranging from 0 to 1, with lower scores rep-
resenting a lower probability of survival) was
also calculated.
Hourly intracranial pressure and mean arterial
pressure measurements were recorded for 12 hours
before randomization and 36 hours after random-
ization. Also recorded were first- and second-tier
therapeutic interventions and surgical complica-
tions of craniectomy and of subsequent cranio-
plasty (surgical reversal of the craniectomy).
Outcome Measures
Outcome measures were evaluated by telephone
by three trained assessors who were unaware of
study-group assignments. The original primary
outcome was the proportion of patients with an
unfavorable outcome, a composite of death, a
vegetative state, or severe disability (a score of 1
to 4 on the Extended Glasgow Outcome Scale), as
assessed with the use of a structured, validated
telephone questionnaire
19-22
at 6 months after
injury.
21
(The Extended Glasgow Outcome Scale
ranges from 1 to 8, with lower scores indicating
a poorer functional outcome.) After the interim
analysis in January 2007, the primary outcome
was revised to be the functional outcome at
6 months after injury on the basis of proportion-
al odds analysis of the Extended Glasgow Out-
come Scale.
19
Secondary outcomes were intracra-
nial pressure measured hourly, the intracranial
hypertension index
23
(defined as the number of
end-hourly measures of intracranial pressure of
more than 20 mm Hg divided by the total num-
ber of measurements, multiplied by 100), the
proportion of survivors with a score of 2 to 4 on
the Extended Glasgow Outcome Scale (defined as
severe disability and requiring assistance in daily
living activities), the numbers of days in the ICU
and in the hospital, and mortality in the hospital
and at 6 months.
Study Oversight
Funding was provided by the National Health
and Medical Research Council of Australia; the
Transport Accident Commission of Victoria, Aus-
tralia; the Intensive Care Foundation of the Aus-
tralian and New Zealand Intensive Care Society;
and the Western Australian Institute for Medical
Research. The funders had no role in the design
of the trial protocol; in the collection, analysis,
or interpretation of the trial data; or in the writ-
ing of the manuscript. The members of the ex-
ecutive committee attest that the trial was per-
formed in accordance with the protocol, including
revision of the primary outcome measure as de-
scribed above, and vouch for the accuracy and
completeness of the reported data.
Statistical Analysis
The trial was originally designed to identify an
increase in the proportion of favorable outcomes
(defined as a score of 5 to 8 on the Extended
Glasgow Outcome Scale) from 30% among pa-
tients receiving standard care to 50% among pa-
tients undergoing craniectomy, with a two-sided
type I error of 0.05 and a power of 80%
14
with a
sample size of 210 patients. (This design is equiv-
alent to the identification of a reduction in the
rate of unfavorable outcomes from 70% to 50%.)
At the interim analysis (with the study-group as-
signments concealed), it was determined that if
the score on the 8-grade Extended Glasgow Out-
come Score were analyzed by ordinal logistic re-
gression, 150 patients would be required to de-
tect a between-group difference of 1.5 in the me-
dian score with a power of 80% and a two-sided
type I error of 0.05. An ordinal logistic-regres-
sion analysis of the score on the Extended Glasgow
Outcome Scale was then defined as the main pri-
The New England Journal of Medicine
Downloaded from nejm.org at UQ Library on August 2, 2018. For personal use only. No other uses without permission.
Copyright © 2011 Massachusetts Medical Society. All rights reserved.
The
new engla nd journal
o f
medicine
n engl j med 364;16 nejm.org april 21, 2011
1496
mary outcome. To allow the trial to be completed
within a reasonable time frame, the sample size
was decreased to 150, with an additional enroll-
ment of 15 patients permitted if necessary to re-
place patients lost to follow-up.
14
Both the original
primary and final primary outcomes are report-
ed. At the point at which enrollment reached 150
patients, no patients had been lost to follow-up,
and recruitment ceased at 155 patients.
All analyses were performed according to the
intention-to-treat principle. We used ordinal lo-
gistic regression for univariate between-group
comparisons of scores on the Extended Glasgow
Outcome Scale and logistic regression for com-
parisons of unfavorable outcomes. These analyses
were followed by adjusted comparisons with inclu-
sion in the regression models of the prespecified
covariates
17
: age, the last Glasgow Coma Scale
Table 1. Baseline Characteristics of the Patients.*
Characteristic
Decompressive
Craniectomy
(N = 73)
Standard
Care
(N = 82) P Value†
Age — yr 0.89
Median 23.7 24.6
Interquartile range 19.4–29.6 18.5–34.9
Male sex — no. (%) 59 (81) 61 (74) 0.44
Systolic blood pressure — mm Hg 135.4±32.0 135.7±27.6 0.95
Glasgow Coma Scale
Overall score‡ 0.31
Median 5 6
Interquartile range 3–7 4–7
Motor score§ 0.49
Median 3 3
Interquartile range 1–4 1–5
Maximum score for head injury on Abbreviated Injury Scale — no. (%)¶ 0.52
3 or 4 35 (48) 44 (54)
5 38 (52) 38 (46)
Injury Severity Score‖ 0.88
Median 33 32
Interquartile range 25–38 24–41
Trauma Score–Injury Severity Score ** 0.46
Median 0.74 0.72
Interquartile range 0.42–0.88 0.51–0.90
Reactivity of pupils — no./total no. (%) 0.04
Neither pupil 19/71 (27) 10/80 (12)
One or both pupils 52/71 (73) 70/80 (88)
Hypotension — no. (%) 24 (33) 25 (30) 0.93
Hypoxemia — no. (%) 18 (25) 24 (29) 0.55
Traumatic subarachnoid hemorrhage — no. (%) 42 (58) 48 (59) 0.90
Cause of injury — no./total no. (%) 0.72
Motor-vehicle or motorcycle accident 45/70 (64) 55/81 (68)
Bicycle accident 4/70 (6) 2/81 (2)
Pedestrian accident 5/70 (7) 4/81 (5)
Other 16/70 (23) 20/81 (25)
The New England Journal of Medicine
Downloaded from nejm.org at UQ Library on August 2, 2018. For personal use only. No other uses without permission.
Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Decompressive Craniectomy in Traumatic Brain Injury
n engl j med 364;16 nejm.org april 21, 2011
1497
score before intubation, the Glasgow Coma Scale
motor score after resuscitation, and the Marshall
class.
15
A post hoc adjusted comparison included
one variable (pupil reactivity) that differed sig-
nificantly between groups at baseline. Cox pro-
portional-hazards regression was used for the
comparison of the numbers of days in the ICU
and in the hospital. A P value of less than 0.05
was considered to indicate statistical significance.
All analyses were performed with the use of
Stata statistical software.
Results
Patients
Of 3478 patients who were assessed for trial eli-
gibility, 155 were enrolled (
Fig. 1
in the Supple-
mentary Appendix). The first 5 patients who were
enrolled in the trial participated in a pilot study,
13
and data from these patients were included in all
the analyses. The most common reasons for ex-
clusion from the trial were the presence of a ce-
rebral mass lesion and successful control of
intracranial pressure with the use of first-tier
therapies. A total of 136 patients (88%) were from
either Australia or New Zealand.
The patients were randomly assigned to one
of the two treatment groups: 73 to undergo early
decompressive craniectomy and 82 to receive
standard care. Baseline characteristics of the two
study groups were similar in most respects, ex-
cept that fewer patients in the craniectomy group
had reactive pupils (
Table 1
). The median age was
23.7 years in the craniectomy group and 24.6 in
the standard-care group. The median intracra-
nial pressure during the 12 hours before ran-
domization was 20 mm Hg (interquartile range,
18 to 22) in the two groups (
Fig. 1
). The median
Table 1. (Continued.)
Characteristic
Decompressive
Craniectomy
(N = 73)
Standard
Care
(N = 82) P Value†
Time from injury to hospital — hr 0.90
Median 1.0 1.2
Interquartile range 0.8–1.8 0.7–1.9
Time from injury to randomization — hr 0.60
Median 35.2 34.8
Interquartile range 23.3–52.8 25.8–45.4
Marshall class — no. (%)†† 0.39
Diffuse injury II 17 (23) 27 (33)
Diffuse injury III or IV 53 (73) 53 (65)
Nonevacuated mass lesion (VI) 3 (4) 2 (2)
* Plus–minus values are means ±SD.
† All P values were calculated with the use of the chi-square test to compare proportions and the Wilcoxon rank-sum
test to compare distributions.
‡ The overall score on the Glasgow Coma Scale ranges from 3 to 15, with lower scores indicating reduced levels of
consciousness.
§ The motor score on the Glasgow Coma Scale ranges from 1 to 6, with lower scores indicating more limited motor
response.
¶ The score for head injury on the Abbreviated Injury Scale ranges from 1 to 6, with higher scores indicating more
severe injury.
‖ The Injury Severity Score ranges from 0 to 75, with higher scores indicating greater injury severity.
** The Trauma Score–Injury Severity Score ranges from 0 to 1, with lower scores indicating a lower probability of
survival.
†† The Marshall classification is based on findings on computed tomography as follows: class I, diffuse injury with no
visible signs; class II, diffuse injury with basal cisterns intact, a midline shift of 0 to 5 mm, and a high- or mixed-den-
sity lesion of 25 ml or less with the possibility of bone fragments or foreign bodies; class III, diffuse injury with swell-
ing, including compressed or absent cisterns with a midline shift of 0 to 5 mm and a high- or mixed-density lesion
of 25 ml or less; class IV, diffuse injury with shift, including a midline shift of more than 5 mm and a high- or mixed-
density lesion of 25 ml or less; class V, surgical evacuation of a mass lesion; and class VI, a high- or mixed-density
lesion of more than 25 ml that has not been surgically evacuated.
The New England Journal of Medicine
Downloaded from nejm.org at UQ Library on August 2, 2018. For personal use only. No other uses without permission.
Copyright © 2011 Massachusetts Medical Society. All rights reserved.
