Factors influencing intracranial pressure monitoring guideline
compliance and outcome after severe traumatic brain injury*
Heleen A.R. Biersteker, MD; Teuntje M.J.C. Andriessen, MSc; Janneke Horn, MD, PhD; Gaby Franschman, MD;
Joukje van der Naalt, MD, PhD; Cornelia W.E. Hoedemaekers, MD, PhD; Hester F. Lingsma, PhD; Iain Haitsma, MD;
Pieter E. Vos, MD, PhD
Objective: To determine adherence to Brain Trauma Foundation
guidelines for intracranial pressure monitoring after severe traumatic brain injury, to investigate if characteristics of patients
treated according to guidelines (ICP+) differ from those who were
not (ICP-), and whether guideline compliance is related to 6-month
outcome.
Design: Observational multicenter study.
Patients: Consecutive severe traumatic brain injury patients
(16 yrs, n = 265) meeting criteria for intracranial pressure
monitoring.
Measurements and Main Results: Data on demographics, injury
severity, computed tomography findings, and patient management were registered. The Glasgow Outcome Scale Extended was
dichotomized into death (Glasgow Outcome Scale Extended = 1)
and unfavorable outcome (Glasgow Outcome Scale Extended 1–4).
Guideline compliance was 46%. Differences between the monitored and nonmonitored patients included a younger age (median
44 vs. 53 yrs), more abnormal pupillary reactions (52% vs. 32%),
and more intracranial pathology (subarachnoid hemorrhage 62%
vs. 44%; intraparenchymal lesions 65% vs. 46%) in the ICP+ group.
Patients with a total intracranial lesion volume of ~150 mL and a
midline shift of ~12 mm were most likely to receive an intracranial
T
pressure monitor and probabilities decreased with smaller and
larger lesions and shifts. Furthermore, compliance was low in patients with no (Traumatic Coma Databank score I −10%) visible
intracranial pathology. Differences in case-mix resulted in higher
a priori probabilities of dying (median 0.51 vs. 0.35, p < .001)
and unfavorable outcome (median 0.79 vs. 0.63, p < .001) in the
ICP+ group. After correction for baseline and clinical characteristics with a propensity score, intracranial pressure monitoring
guideline compliance was not associated with mortality (odds
ratio 0.93, 95% confidence interval 0.47–1.85, p = .83) nor with
unfavorable outcome (odds ratio 1.81, 95% confidence interval
0.88–3.73, p = .11).
Conclusions: Guideline noncompliance was most prominent
in patients with minor or very large computed tomography abnormalities. Intracranial pressure monitoring was not associated
with 6-month outcome, but multiple baseline differences between
monitored and nonmonitored patients underline the complex nature of examining the effect of intracranial pressure monitoring in
observational studies. (Crit Care Med 2012; 40: 1914–1922)
Key Words: computed tomography; Glasgow Outcome Scale;
guideline adherence; intracranial pressure; multivariate analysis;
traumatic brain injury
raumatic space occupying lesions and cerebral edema may
result in a reduction of the intracranial volume reserve followed by a rise in intracranial pressure
(ICP). Subsequently, elevated ICP may
lead to herniation of brain tissue, inadequate cerebral perfusion, ischemia, and
death (1, 2). Monitoring and treatment
of raised ICP are therefore considered key
elements of clinical management of severe traumatic brain injury (TBI) (3).
ICP monitoring allows early detection of pressure changes and can guide
treatment of elevated ICP (4, 5). Based
on observational and case studies (6–8),
international guidelines recommend routine ICP monitoring in severe TBI (9–11).
However, the efficacy of ICP monitoring
has never been verified in randomized
*See also p. 1993.
From the Departments of Neurology (HARB, TMJCA,
PEV) and Intensive Care Medicine (CWEH), Radboud
University Nijmegen Medical Center, Nijmegen,
The Netherlands; Department of Intensive Care
Medicine (JH), Academic Medical Center, University of
Amsterdam, Amsterdam, The Netherlands; Department
of Anesthesiology (GF), VU University Medical Center,
Amsterdam, The Netherlands; Department of Neurology
(JvdN), University Medical Center Groningen, Groningen,
The Netherlands; Department of Public Health (HFL),
Erasmus Medical Center, Center for Medical Decision
Making, Rotterdam, The Netherlands; and Department
of Neurosurgery (IH), Erasmus Medical Center,
Rotterdam, The Netherlands.
The POCON study is funded by the Dutch Brain
Foundation (Hersenstichting - HSN-07-01).
The authors have not disclosed any potential conflicts of interest.
For information regarding this article, E-mail:
[email protected]
Copyright © 2012 by the Society of Critical Care
Medicine and Lippincott Williams & Wilkins
1914
DOI: 10.1097/CCM.0b013e3182474bde
controlled trials and recent studies have
questioned the benefits of ICP monitoring.
In one retrospective cohort study, a center
using ICP monitoring was compared to a
center not using ICP monitoring. Patients
in the ICP monitoring center received
longer mechanical ventilation and more
intense therapy but did not have better
outcome (12). Other reports concluded
that routine ICP monitoring is associated
with worse outcome (13) and higher risk
of extracranial complications (14). These
counterintuitive findings generated many
responses from the field (15–22), pointing
out that there is confounding by indication. Patients undergoing ICP monitoring
probably sustained more severe injuries
than those not undergoing ICP monitoring, and therefore have a worse outcome.
These comments underscore that in studies evaluating the effect of guidelines on
outcome, identification of and controlling
for confounding factors are important.
Crit Care Med 2012 Vol. 40, No. 6
The Brain Trauma Foundation (BTF)
guidelines are thoroughly constructed
and widely accepted guidelines for the
management of TBI. The BTF advises ICP
monitoring in all salvageable severe TBI
patients with a computed tomography
(CT) scan revealing intracranial pathology
(level II recommendation) or in severe TBI
patients with a normal CT scan but with
two or more of the following risk factors:
age over 40 yrs, unilateral or bilateral motor posturing, or systolic blood pressure
<90 mm Hg (level III recommendation)
(9). It has been suggested that implementation and strict adherence to the international guidelines results in a reduction
of mortality (6, 7, 23, 24). However, considerable variation in the appliance of ICP
monitors has been reported across studies
(3, 6, 25–27).
As part of a multicenter study we determined compliance to the BTF guidelines for ICP monitoring. Our primary
goal was to identify demographic and
injury characteristics associated with
guideline compliance. We hypothesized
that a low a priori probability as well as
a high a priori probability for death or
unfavorable outcome would decrease
the likelihood of guideline compliance.
Second, we assessed in a multivariate
model, correcting for potential confounders by using a propensity score,
whether ICP monitoring guideline compliance is a predictor for long-term outcome after severe TBI.
MATERIALS AND METHODS
Design and Setting. The Prospective
Observational COhort Neurotrauma project is
a study of epidemiology, acute care, and outcome in the first year after moderate and severe TBI. Five of 11 specialized (level I) trauma
centers in The Netherlands participated. These
centers all use standardized hospital developed
protocols for treatment of severe TBI patients,
which are based on international guidelines
set up by the BTF (www.braintrauma.org). See
reference (28) for detailed information about
the study.
Patient Selection. In Prospective Obser­
vational COhort Neurotrauma, all patients
with TBI admitted to the emergency department (ED) with a Glasgow Coma Scale (GCS)
score ≤13 between June 1, 2008 and May 31,
2009 were registered in a database. When intubated at the scene of injury, the GCS score
obtained before intubation (≤13) was used
as the qualifier to determine eligibility for
study inclusion. Exclusion criteria were age
<16 yrs, and hospital admission >72 hrs after
Figure 1. Flow chart study inclusion. BTF, Brain Trauma Foundation; CT, computed tomography;
CTabn+, CT scan which reveals hematomas, contusions, swelling, herniation, or compressed basal
cisterns; CTabn-, CT scan which does not reveal hematomas, contusions, swelling, herniation, or
compressed basal cisterns; ED, emergency department; GCS, Glasgow Coma Scale; GOSE, Glasgow
Outcome Scale Extended; ICP, intracranial pressure; ICP+, patients who received an ICP monitor; ICP-,
patients who did not receive an ICP monitor; ICU, intensive care unit; OR, operating room; POCON,
Prospective Observational COhort Neurotrauma.
Crit Care Med 2012 Vol. 40, No. 6
the injury was sustained. For this study, we
selected those patients meeting BTF criteria
for ICP monitoring (9). Two selection criteria were defined: 1) patients with severe TBI
(GCS ≤8 on ED admission) and an abnormal
CT scan (revealing hematomas, contusions,
swelling, herniation, or compressed basal
cisterns); 2) patients with severe TBI without CT abnormalities but with at least two of
the following criteria: age >40 yrs, unilateral
or bilateral motor posturing (ED GCS motor score ≤3), or an episode of systolic blood
pressure <90 mm Hg before hospital arrival
or at the ED.
Patients in whom a first CT scan was unavailable or who were not admitted to one of
the participating hospitals (i.e., transferred
from the ED to a nonparticipating hospital,
discharged home, or who died at the ED or
operating room) were excluded, as monitoring indication or guideline compliance could
not be determined. In addition, patients with
a gunshot injury were excluded.
Data Collection and Definitions. Data
were collected from medical records and
entered into a database by trained research
staff. Collected variables included age,
gender, injury mechanism, GCS scores,
pupillary reactions, hypotensive episode
(yes = systolic blood pressure <90 mm Hg,
suspected = on clinical grounds), hypoxic
episode (at injury scene or ED, yes = Sao2
<90% or Pao2 <8 kPa, suspected = on clinical grounds), glucose (mmol/L) and hemoglobin (g/dL) levels. Severity scores included
the Injury Severity Score and Abbreviated
Injury Scores. Major extracranial trauma
was defined as Abbreviated Injury Score ≥3
in one or more body regions other than the
head.
We registered the length of hospital and
intensive care unit (ICU) stay, sedation and
mechanical ventilation at the ICU, intra- and
extracranial surgery. Craniotomy was defined
as “acute” if scheduled after assessment of
the initial head CT scan. Brain-specific treatment included osmotherapy (mannitol or
hypertonic saline), vasopressor medication
to maintain cerebral perfusion pressure (not
registered if a patient received vasopressive
agent for systemic blood pressure support),
hyperventilation ([yes = Paco2 ≤4 kPa] and
only registered when documented as treatment strategy in medical records), cerebrospinal fluid drainage, hypothermia (body
temperature <35°C), and use of barbiturates.
Increased ICP was defined as any period with
a measured ICP >20 mm Hg.
The first acquired CT scan was evaluated.
If not available, the second scan was assessed
provided that it was made within 6 hrs after
the initial scan and prior to any neurosurgical intervention. Scans were scored using a standardized data sheet (29–31). The
Traumatic Coma Databank (TCDB) score
(32) and the presence and volume of subdural hematoma, epidural hematoma and intraparenchymal lesions were recorded. Lesion
1915
Table 1. Patient characteristics and injury severity parameters
ICP+
Patients
Age
Gender, male
ICP-
123
44 (26–54)
53 (37–69)
90 (73)
90 (63)
Mechanism of injury
65 (46)
Fall
48 (39)
57 (40)
5 (4)
16 (11)
Secondary referral
p
<.001a
0.96 (0.94–0.98)
<.001
.09b
1.87 (0.87–4.03)
.11
.97
.12
68 (55)
Other/unknown
OR (95% CI)
b
Traffic
Violence
p
142
2 (2)
4 (3)
17 (14)
30 (21)
GCS at injury scene
.12b
<.001b
GCS ≤8
107 (87)
95 (67)
0.98 (0.36–2.64)
GCS >8
12 (9.8)
24 (17)
Reference category
Unknown
4 (3.3)
23 (16)
0.20 (0.04–0.96)
.05
GCS at ED
3 (3–3)
0.72 (0.56–0.93)
.01
3 (3–6)
Pupillary reactions at ED
<.01a
<.01b
Both reacting
49 (40)
90 (63)
Reference category
One reacting
13 (11)
11 (7.7)
4.30 (1.29–14.3)
.02
Both nonreacting
50 (41)
34 (24)
3.64 (0.78–3.61)
<.01
Hypoxic episode
29 (25)
30 (23)
.65b
Hypotensive episode
27 (22)
37 (26)
.44b
Major extracranial injury
69 (57)
54 (39)
<.01b
1.68 (0.78–3.61)
.18
Complications
ICP, intracranial pressure; ICP+, ICP monitor guideline compliance group; ICP-, ICP monitor
guideline noncompliance group; OR, odds ratio; CI, confidence interval; GCS, Glasgow Coma Scale;
ED, emergency department.
a
Mann-Whitney U test; bChi-square test.
Unless stated otherwise, n (%) or median (interquartile range) is reported.
volume was calculated using the ellipsoid
method (33, 34). Furthermore, we recorded
the presence of microbleeds (maximum diameter of 5 mm), subarachnoid hemorrhage,
and midline shift and the status of the ambient cisterns and fourth ventricle.
Six-month outcome was assessed with the
Glasgow Outcome Scale Extended (GOSE),
an eight-point scale ranging from 1 (dead) to
8 (good recovery). The GOSE was determined
through a postal questionnaire or a structured
telephone interview (35, 36).
Protocol Approval and Patient Consent.
The study protocol was approved by the local
ethics committee of the coordinating hospital (Radboud University Nijmegen Medical
Center). The other participating hospitals all
provided a feasibility statement. For follow-up
by telephone interview, verbal informed consent was obtained, and for outcome assessment
through postal questionnaires, we gained written informed consent.
Statistical Analysis. All statistical analyses were performed using SPSS version 16.0.
(SPSS, Inc., Chicago, IL). Unless stated otherwise, a p value <.05 was considered statistically
significant. Patients were grouped as guideline
compliant (ICP+) or noncompliant (ICP-).
1916
Characteristics of the ICP+ and ICP- groups
were compared using a Mann-Whitney U test in
case of ordinal variables or when assumptions
of normality were violated, and Pearson’s chisquare tests were used for nominal variables.
As a nonlinear relationship was expected, the
association between intracranial lesion volume
and midline shift with ICP monitor placement
was examined in univariate analyses with restricted cubic spline functions (37).
The propensity score corresponds to the
probability of a patient receiving an ICP monitor given a range of baseline demographic
and clinical variables (38). To calculate the
patient’s propensity score, a multivariate logistic regression model was performed with ICP
monitoring (yes/no) as outcome variable and
all baseline demographic and injury characteristics with p < .10 (in univariate analyses) as
explaining variables. Missing values were included as a separate “unknown” category. The
proportion explained variance (Nagelkerke R2)
and predictive accuracy (area under the receiver operating characteristic curve) were
assessed.
To investigate whether ICP monitoring
guideline compliance is a predictor for patient outcome, we performed two multivariate
logistic regression analyses with 6-month outcome dichotomized as dead (yes/no) and unfavorable (yes = GOSE 1–4 /no = GOSE 5–8).
Guideline compliance was entered into the
model as independent variable together with
the patient’s propensity score to adjust for potential confounding by indication.
Finally, the International Mission for
Prognosis and Analysis of Clinical Trials in TBI
prediction rule was used to calculate the probability of death (Pdeath6) and unfavorable outcome
(Punfav6) at 6 months after injury. Where possible,
the full laboratory model was used. In case of insufficient data, probability of death was calculated
using either the extended or the core model (39).
RESULTS
Patient Demographics and Early
Injury Characteristics. A total of 265 patients met BTF criteria for ICP monitoring and were included (Fig. 1). Patients
were predominantly male (68%) and
involved in road traffic accidents (50%)
(Table 1). An ICP monitor was inserted in
123 (46%) patients and guideline compliance ranged between 21% and 64% across
centers. Comparison of demographic and
injury characteristics between the ICP+
and ICP- groups revealed multiple differences. The ICP+ group was younger
(median 44 vs. 53 yrs, p < .001), had
lower GCS scores (GCS at ED 3 [3–3] vs.
3 [3–6], p < .01), more abnormal pupillary reactions (52% vs. 32%, p < .01), and
more major extracranial injuries (57%
vs. 39%, p < .01) than the ICP- group.
Evaluation of the CT scan was based on
252 (95%) first and on 9 (3.4%) second
acquired scans. In four cases (1.5%), a CT
scan was unavailable but patients were
included in this study because they had
undergone acute craniotomy. Intracranial
abnormalities were detected in 245 (93%)
patients of whom 121 (49%) received an
ICP monitor. Twenty (7.5%) patients met
BTF criteria but had a normal first CT
scan (TCDB I). Two (10%) of these patients received an ICP monitor. The TCDB
classification of the remaining patients
was as follows: 97 with diffuse injury II (32
[33%] ICP+), 26 with diffuse injury III (20
[77%] ICP+), three with diffuse injury IV
(2 [67%] ICP+), 78 with evacuated mass
lesions (51 [65%] ICP+), and 41 with nonevacuated mass lesions (16 [39%] ICP+).
Comparison of intracranial pathology between ICP+ and ICP- groups (Table
2) revealed that almost all injury types
were more common in the ICP+ group,
including the presence of subarachnoid
hemorrhage (62% vs. 44%, p < .01), subdural hematoma (59% vs. 40%, p < .01),
Crit Care Med 2012 Vol. 40, No. 6
Figure 2. Restricted cubic spline functions; relationship between intracranial pressure (ICP) monitor placement and lesion volume (A) and midline
shift (B).
Table 2. Characteristics of first computed tomography scan
ICP+
ICP-
123
142
Subarachnoid hemorrhage
76 (62)
61 (44)
<.01a
1.28 (0.57–2.85)
.55
Subdural hemorrhage
72 (59)
55 (40)
<.01a
1.13 (0.40–3.25)
.82
Epidural hemorrhage
21 (17)
13 (9.4)
.70a
Intraparenchymal lesion
80 (65)
63 (46)
<.01a
2.36(0.90–3.25)
.08
Punctate hemmorhages
48 (39)
36 (26)
.03a
2.11 (0.99–4.51)
.06
Patients
OR (95% CI)
p
p
Intracranial pathology
Ambient cisterns
Compressed
Absent
.03a
34 (27)
22 (16)
13.2 (3.3–52.6)
<.001
4 (3)
8 (6)
3.25 (0.69–15.2)
.13
Fourth ventricle
<.001
a
Compressed
43 (35)
10 (7)
0.33 (0.09–1.22)
.10
Absent
27 (27)
28 (20)
0.07 (0.01–0.50)
<.01
Lesion volume
<.01a
No lesion
20 (16)
49 (35)
Reference category
<25 mL
42 (34)
45 (32)
1.18 (0.33–4.24)
.81
25 to <100 mL
39 (32)
18 (13)
6.22(1.17–33.0)
.03
100 to <200 mL
13 (11)
14 (10)
5.85 (0.63–54.7)
.12
12 (9)
4.10 (0.38–43.8)
.25
≥200 mL
9 (7)
Midline shift
.002a
No shift
60 (49)
93 (65)
Reference category
<5 mm
21 (17)
11 (34)
1.51 (0.45–5.10)
.50
5 to <15 mm
33 (27)
20 (14)
0.65 (0.16–2.76)
.56
9 (7)
14 (10)
0.27 (0.03–2.40)
.25
≥15 mm
ICP, intracranial pressure; ICP+, ICP monitor guideline compliance group; ICP-, ICP monitor
guideline noncompliance group; OR, odds ratio; CI, confidence interval.
a
Chi-square test.
Values are reported as n (%) or median (interquartile range).
and intraparenchymal lesions (65% vs.
46%, p < .01). The ICP+ group revealed
larger total lesion volumes compared to
Crit Care Med 2012 Vol. 40, No. 6
the ICP- group (median 22.7 mL vs. 4.1
mL, p < .01). Figure 2 shows the probability of ICP monitoring in relation
to total intracranial lesion volume and
midline shift, examined with a restricted
cubic spline model. Note that for lesion
volume, outliers (1.5%) were truncated to
the upper limit of 355 mL. Both figures
reveal an n-shaped curve with the highest
probabilities of ICP monitor placement
in lesion volumes at approximately 150
mL and a midline shift at around 12 mm.
Patients with smaller but also with larger
lesion volumes or midline shifts were less
likely to receive ICP monitoring.
Predicting ICP Monitoring Compliance.
Baseline demographics and injury characteristics that differed (at p < .10) between
the ICP+ and ICP- groups were entered
into a multivariable model to predict ICP
monitor placement. The corresponding
odds ratios (ORs), 95% confidence intervals (CIs), and p values are reported in
Tables 1 and 2. The following variables
were significant predictors of guideline
compliance: age, GCS at injury scene and
ED, pupillary reactions, status of the ambient cisterns and the fourth ventricle, and
total lesion volume. The model explained
52% of the variance in ICP monitor placement (Nagelkerke R2) and had an area under the curve of 0.86 (95% CI 0.81–0.90).
Patient Management. The median duration of ICP monitoring was 4.6 days, and
a raised ICP was measured in 71% (Table
3). Patients with ICP monitoring had a significantly longer ICU (median 10.8 vs. 2.7
days, p < .001) and hospital (median 22.0
vs. 7.5 days, p < .001) stay compared to
patients without ICP monitoring. Except
for 13 (9.2%) patients in the ICP- group,
all were mechanically ventilated and duration of ventilation was longer in the ICP+
group (median 7.7 days vs. 1.3 days, p <
1917
Table 3. Interventions and length of hospital stay
ICP+
Patients
123
ICP-
p
142
Length of stay
ICU (days)
10.8 (4.2–21)
2.65 (1.00–6.9)
<.001a
Hospital (days)
22.0 (8.3–44)
7.48 (1.9–20)
<.001a
4.58 (2.0–8.5)
Not applicable
ICP monitoring
Duration (days)
Elevated ICP
87 (71)
Not applicable
Mechanical ventilation
123 (100)
129 (91)
<.001b
Duration (days)
7.7 (3.3–14)
1.3 (0.5–5.0)
<.001a
Sedation
119 (97)
98 (69)
<.001b
Osmotherapy
74 (61)
8 (5.6)
<.001b
Vasopressors
90 (73)
14 (9.9)
<.001b
Hypothermia
15 (12)
0 (0)
<.001c
CSF drainage
19 (15)
2 (1.4)
<.001c
Hyperventilation
26 (21)
1 (0.7)
<.001c
ICP/CPP-targeted interventions
Barbiturates
3 (2.4)
Craniotomy
Acute
Delayed
Other surgery
0 (0)
.10c
50 (41)
26 (18)
<.001b
43 (35)
21 (15)
<.001b
7 (5.7)
5 (3.5)
.40b
35 (29)
30 (31)
.17b
ICP, intracranial pressure; ICP+, ICP monitor guideline compliance group; ICP-, ICP monitor
guideline noncompliance group; ICU, intensive care unit; CPP, cerebral perfusion pressure; CSF,
cerebrospinal fluid.
a
Mann-Whitney U test; bChi-square test; cFisher’s exact test.
Values are reported as n (%) or median (interquartile range).
.001). All brain-specific therapies, except
delayed craniotomy, were more common
in the ICP+ group. In two patients without ICP monitoring, cerebrospinal fluid
drainage was started because of a developing hydrocephalus.
Predicted and Actual 6-Month
Outcome. Table 4 shows the actual
6-month outcome and the probabilities
for death and unfavorable outcome for the
ICP+ and ICP- groups. The follow-up rate
for 6 month survival was 94%, and in 89%
a GOSE score could be obtained (Fig. 1).
No difference in observed 6-month mortality was found between the ICP+ and
ICP- groups, but the ICP+ group did show
a higher rate of unfavorable outcome
(74% vs. 53%, p = .01) and unfavorable
survival (44% vs. 20%, p < .01).
To calculate Pdeath6 and Punfav6, the full
IMPACT model was used in 219 (83%), the
extended model in 10 (4.0%), and the core
model in 18 (6.8%) patients. In 18 (6.8%)
patients, probabilities could not be calculated due to missing data. Monitored patients showed a higher a priori probability
of death (median 0.51 vs. 0.35, p < .001)
and unfavorable outcome (median 0.79
vs. 0.63, p < .001). Guideline compliance
1918
was highest in patients with an intermediate probability of death (61% compliance
in Pdeath6 .4–.6), whereas compliance was
lowest in patients with low probability
of dying (21% compliance in Pdeath6 0–.2)
(bottom row in Fig. 3A). When probability for unfavorable outcome was considered (bottom row in Fig. 3B), guideline
compliance was highest in patients with
a Punfav6 between .6 and .8 (56%) and declined with lower probability of unfavorable outcome.
Relationship Between ICP Monitoring
and Outcome. Multivariate logistic regression analyses for death and unfavorable outcome were performed with ICP
monitoring and the patient’s propensity
score as explaining variables. Placement
of an ICP monitor was not associated with
death (OR 0.93 [95% CI 0.47–1.85]) nor
with unfavorable outcome (OR 1.81 [95%
CI 0.88–3.73]). As univariate analyses
showed significantly more unfavorable
survival (GOSE 2–4) but not more mortality in the ICP+ group, we performed post
hoc analyses to investigate the hypothesis
that ICP monitoring causes a shift from
death to vegetative state or severe disability. Two alternative cutoff values (GOSE
1–2 vs. 3–8 and GOSE 1–3 vs. 4–8) were
used to dichotomize outcome into unfavorable and favorable. In neither of these
post hoc analysis, ICP monitoring was associated with outcome (p > .50).
To explore whether ICP monitoring
was related to outcome in specific subgroups, we stratified patients by predicted
outcome probability. Within each predicted outcome category, we compared
actual outcome between the ICP+ and
ICP- group. (Fig. 3A and 3B) No significant differences in survival were detected
in any of the subgroups. However, patients with a predicted unfavorable outcome probability between .6 and .8 who
received an ICP monitor had a significantly higher rate of observed unfavorable outcome (78%) than patients who
did not receive an ICP monitor (54%,
p = .04). Nevertheless, after correction for
case-mix by using the propensity score
this effect became nonsignificant, (OR
2.39 [95% CI 0.53–10.8]).
DISCUSSION
In this prospective observational multicenter study, adherence to BTF guidelines
for ICP monitoring was <50%. Compared
to patients who were not monitored, the
ICP monitored group was younger and
had lower GCS scores, more abnormal
pupillary reactions, more severe systemic
injuries, and more lesions on the initial
CT scan. This variability in baseline characteristics between the ICP-compliant and
noncompliant groups resulted in a different a priori different probability of dying
and unfavorable outcome as calculated
with the IMPACT prediction model. The
proportion of patients receiving an ICP
monitor was highest in patients with a
40%–60% chance of death while guideline
compliance sharply declined in patients
with a low probability of dying. In addition, guideline compliance was highest
in patients with intermediate total lesion
volumes and midline shifts and decreased
in patients with smaller but also with very
large lesions or shifts. These findings corroborate with a previous study (40) illustrating that both the “best” and “worst”
patients are least likely to undergo ICP
monitoring. Furthermore, our results underline the need for adequate correction
of confounding variables when assessing
the effect of interventions in observational cohort studies.
In crude analyses, guideline compliance was associated with poorer outcome,
most notably in patients with a 60%–80%
Crit Care Med 2012 Vol. 40, No. 6
Figure 3. Actual 6-month outcome stratified by predicted 6-month outcome for death (A) and unfavorable outcome (B). ICP, intracranial pressure; ICP+, ICP monitor guideline compliant group; ICP-, ICP
monitor guideline noncompliant group. Probability of death and unfavorable outcome was calculated
using the International Mission for Prognosis and Analysis of Clinical Trials in TBI model. n, number of
patients with and without ICP monitor in each probability group.
chance of unfavorable outcome. However,
after adjustment for variability in baseline
characteristics by including a propensity score in a logistic regression model,
no relationship was found between ICP
monitoring and 6-month outcome. Our
findings are in agreement with studies
reporting no effect of ICP monitoring on
outcome (12, 14, 40) but contrast with
two studies reporting ICP monitoring to
be associated with either increased (8) or
decreased survival (13). Although all previous studies applied multivariate correction
methods, only few CT characteristics were
assessed (i.e., the TCDB or Abbreviated
Injury Score head score). Individual CT
characteristics such as lesion type, lesion
volume, midline shift, the status of the
ambient cisterns and the fourth ventricle
Crit Care Med 2012 Vol. 40, No. 6
are predictors of outcome (29, 31, 41). As
this study demonstrates, detailed knowledge on intracranial pathology probably
plays an important role in clinical decision making. Many individual signs of severe injury on CT were more prominent in
the guideline compliant than in the noncompliant group and had high ORs for ICP
monitoring. Not incorporating specific information on CT findings, like volume or
shift, may explain the contrasting findings
in previous studies. We advocate that individual CT characteristics deserve more attention as confounding factors in analyses
of the effect of ICP monitoring.
The overall guideline compliance rate
in our study (46%) is in line with compliance rates (43%–67%) reported elsewhere (12, 13, 25, 40, 42, 43). Compliance
ranged between 21% and 64% across
centers, reflecting variability in approach
and treatment of TBI patients. However,
also other between-center differences in
baseline characteristics, like age, GCS,
and CT characteristics (data not shown in
this study), may explain variation in ICP
monitoring rates.
Interestingly, compliance was almost
five times lower in patients with a normal CT scan and two or more risk factors
(10%) compared to patients with visible
intracranial pathology (49%). The BTF
recommendation for ICP monitoring in
patients without visible intracranial pathology stems from a prospective study
executed in the 1970’s reporting raised
ICP in eight (15%) of 61 severe TBI patients with a normal CT (44). In our study,
the two patients without initial intracranial pathology who received an ICP monitor both experienced a period of raised
ICP. Our study was not designed to assess
the risk of increased ICP in patients with a
normal CT scan but confirms that severe
TBI patients without initial intracranial
pathology may develop raised ICP. Also
in patients with a TCDB II classification,
compliance was relatively low (33%). Yet,
in a recent series of ICP-monitored patients, 50% with a TCDB II classification
developed raised ICP (45). The noncompliance in patients without or with minor
visible pathology suggests that clinicians
are often not convinced that ICP monitoring in this subgroup of severe TBI patients
is necessary; therefore, the extent of CT
abnormalities that require ICP monitoring deserves further investigation.
When assessing the effect of ICP monitoring on patient outcome, it is not the
insertion of an ICP device in itself but the
subsequent treatment of raised ICP (or decreased cerebral perfusion pressure) that
is thought to mediate outcome. Patients
who received an ICP monitor had longer
ICU and hospital stays and received up to
nine times more (in case of osmotherapy)
ICP- and/or cerebral perfusion pressure–
directed therapy. More treatment in the
guideline compliant group may be a direct result of more intensive monitoring
but may also be related to variability in
injury severity. It has been suggested that
interventions to reduce ICP may sometimes be harmful or inappropriately applied, possibly explaining the absence of
an association between ICP monitoring
and patient outcome (13). As we did not
examine treatment in terms of dosage,
promptness, and length of a certain intervention, we were unable to assess the
1919
Table 4. Actual and predicted 6-month outcome
ICP+
n
ICP-
123
142
59 (48)
52 (37)
7 (5.7)
10 (7.0)
Trauma to the head
50 (85)
40 (77)
Extracranial injury
2 (3.4)
4 (7.7)
Head and extracranial injury
2 (3.4)
2 (3.8)
Not trauma related/unknown
5 (8.5)
6 (12)
2 (3.6)
0 (0)
Dead
Survival unknown
Cause of death
p
.07a
.70a
GOSE in survivors
2: Vegetative state
3: Lower severe disability
14 (25)
4: Upper severe disability
6 (11)
9 (11)
6 (7.5)
5: Lower moderate disability
12 (21)
11 (14)
6: Upper moderate disability
10 (18)
13 (16)
7: Lower good recovery
4 (7.0)
18 (23)
8: Upper good recovery
2 (3.6)
18 (23)
Unknown
7 (12)
5 (6.3)
Unfavorable outcome (1–4)
81 (74)
67 (53)
Unfavorable survival (2–4)
22 (44)
15 (20)
112
135
Pdeath6
.51 (.35–.67)
.35 (.22–.62)
<.001b
PUnfav6
.79 (.63–.89)
.63 (.44–.84)
<.001b
.001a
<.01a
Predicted outcome
n
ICP, intracranial pressure; ICP+, ICP monitor guideline compliance group; ICP-, ICP monitor
guideline noncompliance group; GOSE, Glasgow Outcome Scale Extended; Pdeath6, IMPACT calculated
probability of mortality 6 months after injury; PUnfav6, IMPACT calculated probability of unfavorable
outcome 6 months after injury.
a
Chi-square test; bMann-Whitney U test.
Values are reported as n (%) or median (interquartile range).
adequacy of patient management following ICP monitoring in detail.
Our multivariate model including demographic, injury severity, and CT characteristics explained 52% of the variance
in ICP monitor placement, suggesting
that other factors are involved in the clinical decision-making process. For instance
the patient’s premorbid level of functioning but also personal expectations of
the treating physician may play a role: A
Canadian survey among neurosurgeons
reported that only 20% had a high level of
confidence that ICP monitoring improves
outcome (42). Given that patient management is influenced by many (measurable
and nonmeasurable) factors, full correction for potential confounders in an observational cohort study is very difficult and
requires large patient numbers. Instead, a
randomized controlled trial would be the
study design of choice to prove an effect
of ICP monitoring on outcome. However,
it is unlikely such a trial will be performed
1920
as many think it is unethical to withhold
ICP monitoring after TBI (46).
Some study limitations need to be addressed. Early clinical parameters such
as pupillary reactions and complications
at the ED were unknown in <10% of the
patients. Six-month GOSE scores were
missing in 11% but are in line with follow-up rates reported in other prospective
observational studies (29, 47, 48). We defined severe TBI based on the ED admission GCS score, which may have been
influenced by prehospital interventions
such as endotracheal intubation and use
of sedatives. Furthermore, deterioration
of the GCS may occur at a later stage and
patients may at that point fulfill the criteria for ICP monitoring. These secondary
complications were not included in our
analyses. Similarly, we only assessed the
initial CT scan but intracranial abnormalities may evolve over time (49–51)
and may later on influence the course
of patient management. Some studies
have suggested that ICP monitoring is
associated with increased chance of developing complications such as infections
and hemorrhage (52–54), but these variables were not registered in our study.
Finally, as we did not collect data on the
frequency of neurological examinations
or the number of CTs made during ICU
admission, we were unable to assess the
adequacy of alternative monitoring methods across the two groups.
In conclusion, this multicenter cohort
study showed multiple baseline differences between severe TBI patients with
and without ICP monitoring in age, injury severity, and intracranial pathology.
These differences in case-mix resulted in
higher a priori probabilities of death and
unfavorable outcome in patients with ICP
monitoring and emphasize the complexity of studying the effect of ICP monitoring in an observational study. After
adjustment for baseline characteristics,
our study showed no significant relationship between ICP monitoring and death,
or unfavorable outcome. However, conclusions about the (absence of an) effect
of ICP monitoring should be drawn with
caution as correction for all potential confounders is difficult.
ACKNOWLEDGMENTS
The authors express their gratitude
to Dick Drost, Annemiek Coers, Annelou
van der Veen, Joshua Field, and Vivian
de Ruijter for their help with data collection. We thank Amon Heijne for his
help with development and maintenance of the POCON database and Carel
Gosling (Trauma Unit AMC) for providing us with emergency department admission data.
REFERENCES
1. Padayachy LC, Figaji AA, Bullock MR:
Intracranial pressure monitoring for
traumatic brain injury in the modern era.
Childs Nerv Syst 2010; 26:441–452
2. Miller JD, Becker DP, Ward JD, et al:
Significance of intracranial hypertension
in severe head injury. J Neurosurg 1977;
47:503–516
3. Enblad P, Nilsson P, Chambers I, et al; Brain
IT Group: R3-survey of traumatic brain
injury management in European Brain IT
centres year 2001. Intensive Care Med 2004;
30:1058–1065
4. Lobato RD, Sarabia R, Rivas JJ, et al:
Normal computerized tomography scans in
severe head injury. Prognostic and clinical
management implications. J Neurosurg 1986;
65:784–789
Crit Care Med 2012 Vol. 40, No. 6
5. Narayan RK, Greenberg RP, Miller JD, et al:
Improved confidence of outcome prediction
in severe head injury. A comparative analysis
of the clinical examination, multimodality
evoked potentials, CT scanning, and
intracranial pressure. J Neurosurg 1981;
54:751–762
6. Patel HC, Menon DK, Tebbs S, et al: Specialist
neurocritical care and outcome from head
injury. Intensive Care Med 2002; 28:547–553
7. Palmer S, Bader MK, Qureshi A, et al: The
impact on outcomes in a community hospital
setting of using the AANS traumatic brain
injury guidelines. Americans Associations
for Neurologic Surgeons. J Trauma 2001;
50:657–664
8. Lane PL, Skoretz TG, Doig G, et al: Intracranial
pressure monitoring and outcomes after
traumatic brain injury. Can J Surg 2000;
43:442–448
9. Bratton SL, Chestnut RM, Ghajar J, et
al; Brain Trauma Foundation; American
Association of Neurological Surgeons;
Congress of Neurological Surgeons; Joint
Section on Neurotrauma and Critical Care,
AANS/CNS: Guidelines for the management of
severe traumatic brain injury. VI. Indications
for intracranial pressure monitoring. J
Neurotrauma 2007; 24 Suppl 1:S37–S44
10. Maas AI, Dearden M, Teasdale GM, et al:
EBIC-guidelines for management of severe
head injury in adults. European Brain Injury
Consortium. Acta Neurochir (Wien) 1997;
139:286–294
11. Andrews PJ, Citerio G, Longhi L, et al; NeuroIntensive Care and Emergency Medicine
(NICEM) Section of the European Society of
Intensive Care Medicine: NICEM consensus
on neurological monitoring in acute
neurological disease. Intensive Care Med
2008; 34:1362–1370
12. Cremer OL, van Dijk GW, van Wensen E, et
al: Effect of intracranial pressure monitoring
and targeted intensive care on functional
outcome after severe head injury. Crit Care
Med 2005; 33:2207–2213
13. Shafi S, Diaz-Arrastia R, Madden C, et al:
Intracranial pressure monitoring in braininjured patients is associated with worsening
of survival. J Trauma 2008; 64:335–340
14. Salim A, Hannon M, Brown C, et al:
Intracranial pressure monitoring in severe
isolated pediatric blunt head trauma. Am
Surg 2008; 74:1088–1093
15. Stevens RD, Bhardwaj A: Evolving paradigms
in the management of severe traumatic brain
injury. Crit Care Med 2005; 33:2415–2417
16. Stover JF, Steiger P, Stocker R: Need for
intracranial pressure monitoring following
severe traumatic brain injury. Crit Care Med
2006; 34:1582–1583; author reply 1583
17. Chesnut RM: The evolving management
of traumatic brain injury: Don’t shoot the
messenger. Crit Care Med 2006; 34:2262;
author reply 2262–2262; author reply 2263
18. Rincon F, Mayer SA: Does intracranial
pressure monitoring improve outcome after
Crit Care Med 2012 Vol. 40, No. 6
severe traumatic brain injury? Curr Neurol
Neurosci Rep 2006; 6:507–508
19. Maas AI, Stocchetti N, Bullock R: Moderate
and severe traumatic brain injury in adults.
Lancet Neurol 2008; 7:728–741
20. Chesnut RM: Intracranial pressure monitoring
in brain-injured patients is associated with
worsening of survival. J Trauma 2008;
65:500–501
21. Boret H, Montcriol A, Meaudre E: The
worsening of survival among patients with
brain injuries monitored with ICP versus
those not monitored with ICP. J Trauma
2008; 65:499
22. Grände PO, Reinstrup P: Do conventional
treatments of severe traumatic head
injury worsen outcome? J Trauma 2009;
67:415–416
23. Fakhry SM, Trask AL, Waller MA, et al; IRTC
Neurotrauma Task Force: Management of
brain-injured patients by an evidence-based
medicine protocol improves outcomes and
decreases hospital charges. J Trauma 2004;
56:492–499; discussion 499
24. Becker DP, Miller JD, Ward JD, et al: The
outcome from severe head injury with early
diagnosis and intensive management. J
Neurosurg 1977; 47:491–502
25. Bulger EM, Nathens AB, Rivara FP, et al;
Brain Trauma Foundation: Management of
severe head injury: Institutional variations
in care and effect on outcome. Crit Care Med
2002; 30:1870–1876
26. Heskestad B, Baardsen R, Helseth E, et al:
Guideline compliance in management of
minimal, mild, and moderate head injury:
High frequency of noncompliance among
individual physicians despite strong guideline
support from clinical leaders. J Trauma 2008;
65:1309–1313
27. Huizenga JE, Zink BJ, Maio RF, et al:
Guidelines for the management of severe head
injury: Are emergency physicians following
them? Acad Emerg Med 2002; 9:806–812
28. Andriessen TM, Horn J, Franschman G, et
al: Epidemiology, severity classification, and
outcome of moderate and severe traumatic
brain injury: A prospective multicenter study.
J Neurotrauma 2011; 28:2019–2031
29. Jacobs B, Beems T, van der Vliet TM, et al: The
status of the fourth ventricle and ambient
cisterns predict outcome in moderate and
severe traumatic brain injury. J Neurotrauma
2010; 27:331–340
30. Jacobs B, Beems T, Stulemeijer M, et al:
Outcome prediction in mild traumatic
brain injury: Age and clinical variables are
stronger predictors than CT abnormalities.
J Neurotrauma 2010; 27:655–668
31. Jacobs B, Beems T, van der Vliet TM, et
al: Computed tomography and outcome
in moderate and severe traumatic brain
injury: Hematoma volume and midline shift
revisited. J Neurotrauma 2011; 28:203–215
32. Marshall LF, Eisenberg HM: A new
classification of head injury based on
computerized tomography. J Neurosurg
1991; 75:S14-S20
33. Vos PE, van Voskuilen AC, Beems T, et al:
Evaluation of the traumatic coma data bank
computed tomography classification for
severe head injury. J Neurotrauma 2001;
18:649–655
34. Pasqualin A, Barone G, Cioffi F, et al: The
relevance of anatomic and hemodynamic
factors to a classification of cerebral
arteriovenous malformations. Neurosurgery
1991; 28:370–379
35. Wilson JT, Edwards P, Fiddes H, et al:
Reliability of postal questionnaires for the
Glasgow Outcome Scale. J Neurotrauma
2002; 19:999–1005
36. Wilson JT, Pettigrew LE, Teasdale GM:
Structured interviews for the Glasgow
Outcome Scale and the extended Glasgow
Outcome Scale: Guidelines for their use. J
Neurotrauma 1998; 15:573–585
37. Durrleman S, Simon R: Flexible regression
models with cubic splines. Stat Med 1989;
8:551–561
38. D’Agostino RB Jr: Propensity score methods
for bias reduction in the comparison of a
treatment to a non-randomized control
group. Stat Med 1998; 17:2265–2281
39. Steyerberg EW, Mushkudiani N, Perel P,
et al: Predicting outcome after traumatic
brain injury: Development and international
validation of prognostic scores based on
admission characteristics. PLoS Med 2008;
5:e165; discussion e165
40. Mauritz W, Steltzer H, Bauer P, et al:
Monitoring of intracranial pressure in
patients with severe traumatic brain injury:
An Austrian prospective multicenter study.
Intensive Care Med 2008; 34:1208–1215
41. Maas AI, Steyerberg EW, Butcher I, et
al: Prognostic value of computerized
tomography scan characteristics in traumatic
brain injury: Results from the IMPACT study.
J Neurotrauma 2007; 24:303–314
42. Sahjpaul R, Girotti M: Intracranial pressure
monitoring in severe traumatic brain injury–
results of a Canadian survey. Can J Neurol Sci
2000; 27:143–147
43. Jeevaratnam DR, Menon DK: Survey of
intensive care of severely head injured
patients in the United Kingdom. BMJ 1996;
312:944–947
44. Narayan RK, Kishore PR, Becker DP, et al:
Intracranial pressure: To monitor or not
to monitor? A review of our experience
with severe head injury. J Neurosurg 1982;
56:650–659
45. Bremmer R, de Jong BM, Wagemakers M,
et al: The course of intracranial pressure
in traumatic brain injury: Relation with
outcome and CT-characteristics. Neurocrit
Care 2010; 12:362–368
46. Stocchetti N, Longhi L, Zanier ER:
Intracranial pressure monitoring for
traumatic brain injury: Available evidence
and clinical implications. Minerva Anestesiol
2008; 74:197–203
47. Murray GD, Teasdale GM, Braakman R, et
al: The European Brain Injury Consortium
1921
survey of head injuries. Acta Neurochir
(Wien) 1999; 141:223–236
48. Myburgh JA, Cooper DJ, Finfer SR, et al;
Australasian Traumatic Brain Injury Study
(ATBIS) Investigators for the Australian;
New Zealand Intensive Care Society Clinical
Trials Group: Epidemiology and 12-month
outcomes from traumatic brain injury in
australia and new zealand. J Trauma 2008;
64:854–862
49. Compagnone C, d’Avella D, Servadei F, et
al: Patients with moderate head injury: A
prospective multicenter study of 315 patients.
1922
Neurosurgery 2009; 64:690–696; discussion
696
50. Narayan RK, Maas AI, Servadei F, et al;
Traumatic Intracerebral Hemorrhage Study
Group: Progression of traumatic intracerebral
hemorrhage: A prospective observational
study. J Neurotrauma 2008; 25:629–639
51. Servadei F, Murray GD, Penny K, et al: The
value of the “worst” computed tomographic
scan in clinical studies of moderate and
severe head injury. European Brain Injury
Consortium. Neurosurgery 2000; 46:70–75;
discussion 75
52. Bekar A, Dogan S, Abas F, et al: Risk factors
and complications of intracranial pressure
monitoring with a fiberoptic device. J Clin
Neurosci 2009; 16:236–240
53. Anderson RC, Kan P, Klimo P, et al:
Complications of intracranial pressure
monitoring in children with head trauma.
J Neurosurg 2004; 101(1 Suppl):53–58
54. Münch E, Weigel R, Schmiedek P, et al: The
Camino intracranial pressure device in clinical
practice: Reliability, handling characteristics
and complications. Acta Neurochir (Wien)
1998; 140:1113–1119; discussion 1119
Crit Care Med 2012 Vol. 40, No. 6