DOI 10.1007/s12028-012-9708-y
ORIGINAL ARTICLE
Abstract
Background Optimal resuscitation after traumatic brain
injury (TBI) remains uncertain. We hypothesize that
cerebral metabolic crisis is frequent despite adequate
resuscitation of the TBI patient and that metabolic crisis
negatively influences outcome.
Methods We assessed the effectiveness of a standardized
trauma resuscitation protocol in 89 patients with moderate
to severe TBI, and determined the frequency of adequate
resuscitation. Prospective hourly values of heart rate, blood
pressure, pulse oximetry, intracranial pressure (ICP),
respiratory rate, jugular venous oximetry, and brain extracellular values of glucose, lactate, pyruvate, glycerol, and
glutamate were obtained. The incidence during the initial
72 h after injury of low brain glucose <0.8 mmol/L, elevated lactate/pyruvate ratio (LPR) >25, and metabolic
crisis, defined as the simultaneous occurrence of both low
glucose and high LPR, were determined for the group.
Results 5 patients were inadequately resuscitated and
eight patients had intractable ICP. In patients with successful resuscitation and controlled ICP (n = 76), within
72 h of trauma, 76 % had low glucose, 93 % had elevated
Introduction
Traumatic Brain Injury (TBI) is a major cause of death and
disability. In the United States, there are 52,000 deaths
annually due to TBI, accounting for nearly 1/3 of all
trauma-related deaths [1]. The initial treatment of TBI has
improved through the formalization of trauma triage protocols and treatment guidelines [2], with a resultant
reduction in mortality from 39 % in 1987 to 27 % in 1996.
Resuscitation is particularly important in patients with TBI
since early physiological compromise such as hypoxemia,
hypotension, and elevated intracranial pressure can result
in secondary insults that markedly worsen neurological
outcome [3, 4]. Resuscitation after TBI is designed to avoid
profound secondary ischemic injury and is often focused on
hemodynamic endpoints that are readily measured such as
blood pressure, oxygenation, and ICP [5, 6]. However, the
effects of standard trauma resuscitation on brain metabolism and long term outcomes have not been well studied.
Many questions remain about what physiological endpoints
are ideal for the injured brain.
123
50
123
Resuscitation
In this study, a formalized standardized approach to hemodynamic resuscitation was used that begun in the emergency
room and consisted of intravenous crystalloid infusion of
24 L of normal saline, half-normal saline, or D5W over 2 h
to achieve mean arterial pressure >80 mmHg, heart rate
<100 bpm, and urine output >0.5 cc/kg/h. Fluids were
infused through large bore trauma venous catheters inserted
in the femoral vein and subclavian central venous catheters.
Thereafter, normal saline was infused at 100 cc/h as maintenance fluids, with additional boluses of normal saline
infused to achieve desired hemodynamic goals. Mechanical
ventilation was used in all patients, with oxygenation and
end-tidal carbon dioxide monitoring used to achieve goals of
PaO2 > 100 mmHg, and ETCO2 3035 mmHg. After
insertion of the subclavian central venous catheters and
radial arterial lines, additional hemodynamic resuscitation
was employed to achieve these goals: CVP > 6 mmHg,
MAP > 80 mmHg and CPP 5070 mmHg. In addition
to intravenous crystalloids, norepinephrine was used to
support blood pressure. Blood transfusions of PRBCs,
platelets, FFPs, and cryoprecipitate were used liberally
to keep hemoglobin >8 g/dl, INR <1.5 s, and platelet
count >90,000. Serum sodium values were checked every
6 h, and 3 % NaCl infusions at 2550 cc/h were used to
maintain serum sodium > 140 mEq/L. Written guidelines
for nurse monitoring of the hemodynamic endpoints were
employed, with physician notification triggers. The neurointensive care and trauma surgery attending physicians made
twice daily rounds to assess the adequacy of hemodynamic
resuscitation.
In order to classify which of the research patients were
adequately resuscitated, we first retrospectively appraised
the neurocritical care attending notes on the patient the first
few days to see if the overall clinical impression was that
the patient was or was not adequately resuscitated. All
patients were treated by the same single attending physician over the entire period.
Second, to more objectively classify the clinical
impression of resuscitation, we created a resuscitation
scoring system (RSS) based on six parameters: fluid balance (I/O), systolic blood pressure (SBP), heart rate (HR),
partial pressure of oxygen (pO2), hematocrit (HCT), and
hemoglobin (HGB). Using published literature on optimal
or ideal values for each of these parameters in the trauma
patient [2, 5, 6, 1518], we determined cut-off points to
characterize resuscitation success in the first 24 h post
injury. If the patient was below the desired value for all
24 h, 1223 h, or 011 h then they were assigned 2, 1, or 0
points, respectively. For example, if the patient had a
hematocrit <28 for all 24/24 h they were given 2 points, if
they had a hematocrit <28 for 15/24 h they were given 1
51
123
52
Data Analysis
Statistical analyses were performed by the program R
version 2.13.2 (R Foundation for Statistical Computing,
Vienna, Austria; http://www.R-project.org; R Development
Core Team, 2011). Microdialysis data were analyzed on a
per-patient basis as was done in the largest microdialysis
study to date [12] (i.e., one data-point per patient, the
average for the period and parameter being evaluated) for
calculating medians, means, ranges, differences between
outcome groups, and predictors of outcome (logistic
model). We did not pool hourly data from multiple
patients. Intergroup statistical analysis was performed with
a simple non-parametric test (MannWhitney U-test) and
logistic regression. We did not adjust for multiple comparisons. Regression analyses were performed to determine
if metabolic crisis independently predicted outcome in this
population. Favorable outcome (GOSe C 7) was coded as
0 and unfavorable (GOSe B 6) coded as 1 in these models,
with predictors being age, incoming GCS, number of
pupils reactive on admission, and metabolic crisis. Metabolic crisis was entered into the model as % time spent in
crisis in the first 72 h, and then calculated for each 12-h
increment (half the standard deviation) spent in crisis to
facilitate interpretation of the odds ratio. We performed
additional robust regression analyses because the data
were not normally distributed. By means of R code provided by Wilcox [20] we computed a (1) bounded M
regression using Huber Psi and Schweppe weights and an
(2) adjusted M-estimator using Huber Psi and Schweppe
weights with regression outliers getting a weight of zero
[21]. In addition, we fit linear models via weighted likelihood which is robust against the presence of bad
leverage points [22].
Results
123
% Patients (n)
76.3 (58/76)
38.0 21.5
LPR > 25
93.4 (71/76)
44.9 23.8
Metabolic crisis
73.7 (56/76)
33.5 22.8
53
Parameter
6 month outcome
Favorable (GOSe C 7)
P value
Unfavorable (GOSe B 6)
Admission characteristics
Age
32.5 (26.345.0)
50.5 (38.362.0)
0.008
GCS
9.0 (7.312.5)
6.0 (3.010.8)
0.131
Hypoxia (% of patients)
18.50
0.146
Hypotension (% of patients)
9.10
26.20
0.284
20
25
0.243
1.4 (0.014.2)
25.2 (6.550.4)
0.011
Lactate/pyruvate ratio
25.4 (18.930.7)
32.8 (24.741.6)
0.011
46.8 (8.654.5)
66.0 (33.597.0)
0.074
Glucose (mmol/L)
Glucose < 0.8 mmol/L (% time)
1.3 (0.93.1)
9.1 (0.040.7)
0.9 (0.61.5)
47.3 (18.381.6)
0.061
0.021
Lactate (mmol/L)
1.8 (1.53.0)
2.7 (1.54.3)
0.132
Pyruvate (lmol/L)
78.7 (51.3117.5)
77.9 (49.9121.8)
0.978
Glycerol (lmol/L)
65.5 (49.196.9)
63.7 (42.3-124.6)
0.998
Monitoring parameters
Significant differences
(P < 0.05) in bold type
(MannWhitney U test)
Glutamate (lmol/L)
3.8 (3.36.4)
8.8 (5.815.7)
0.027
75.7 (73.479.7)
74.1 (69.375.2)
0.237
14.5 (13.015.8)
14.1 (10.417.2)
0.703
Table 3 Multivariate analyses showing age, GCS, pupil reactivity, and metabolic crisis as predictors of poor 6 month outcome with P-value for
four different statistical iterations of the model
Predictor
B (standard
error)
OR
95 % CI
P value
1. Logistic
regression
2. M-regression
3. M-estimator
4. Weighted
likelihood
Age (years)
0.09 (0.03)
1.09
1.021.17
0.011
0.010
0.013
0.003
GCS (315)
-0.25 (0.11)
0.78
0.620.98
0.031
0.150
0.127
0.017
Pupils (# reactive)
0.28 (0.58)
1.32
0.434.08
0.629
0.965
0.938
0.678
0.77 (0.37)
2.16
1.054.45
0.036
<0.001
<0.001
0.012
-1.52 (1.66)
0.22
Constant
2
Model summary: x = 22.2, P < 0.0001, R = 0.481 (Nagelkerke), Brier Score = 0.094
Significant differences (P < 0.05) in bold type
123
54
Glucose
100
L/P Ratio
90
80
Lactate / Pyruvate
Glucose (mmol/L)
70
60
50
40
30
20
10
0
0
24
48
72
24
48
10
72
Lactate
300
Pyruvate
250
Pyruvate (mol/L)
Lactate (mmol/L)
200
150
100
50
0
24
48
72
24
48
300
Glycerol
50
Glutamate
40
Glutamate (mol/L)
Glycerol (mol/L)
250
200
150
100
30
20
10
50
0
24
48
72
123
72
24
48
72
known predictors, specifically age, GCS, and pupil reactivity (Table 3). We created four different models. The first
was a non-robust logistic regression with unfavorable
outcome as the dependent variable. Age was a strong
positive predictor of poor outcome (OR 1.09, CI 1.021.17,
P = 0.011). GCS was a strong negative predictor of poor
outcome (OR 0.78, CI 0.620.98, P = 0.031) meaning that
a high GCS on admission reduced the likelihood of having
a poor outcome. Metabolic crisis was a strong positive
predictor of poor outcome (OR 2.16, CI 1.054.45,
P = 0.036), meaning that for each 12 h spent in metabolic
crisis the odds of having a poor outcome more than
doubled.
The data were not normally distributed and logistic
regression is well-known to be sensitive to violations of
normality, so we also performed three robust regression
analyses. Age and metabolic crisis were significant predictors in all three robust models, and GCS in only the
weighted likelihood model. Number of pupils reactive on
admission was not a significant predictor in either the
logistic or robust regressions.
Discussion
The main findings in this study were: (1) Most patients
with TBI received adequate hemodynamic resuscitation
within 24 h after TBI. (2) The majority of patients demonstrated persistent metabolic crisis as measured by
cerebral microdialysis despite adequate resuscitation and
control of ICP. (3) The persistence of metabolic crisis
during the initial 72 h after TBI is a strong independent
predictor of neurological outcome, after controlling for
classical predictors.
The initial triage and resuscitation of TBI is focused on
accurate diagnosis, stabilization, and life saving treatments.
The goals of resuscitation after TBI are cardiopulmonary
endpoints such as heart rate, blood pressure, and oxygen
saturation, and later intracranial pressure [2, 5, 16]. While
these endpoints are well accepted, they may not best reflect
the state of cerebral resuscitation. In the present results,
most patients were hemodynamically resuscitated yet had
cerebral metabolic crisis, as reflected by low extracellular
glucose and elevated LPR. This finding suggests that direct
monitoring of the brain rather than systemic monitoring
alone is required to determine the clinical treatment endpoints for resuscitation. One could envision the use of
cerebral microdialysis or some other less invasive cerebral
monitor against which resuscitation could be better gauged,
while retaining the use of systemic hemodynamics for
evaluation of other organ systems. A fundamental message
from these data is that the brain compartment appears to be
distinct from that of the systemic circulation. Future
55
123
56
Limitations
The study is based on an observational dataset and therefore can only show relationships and correlations, without
definitive evidence that chemistry directly influences outcome. The data were collected over more than a decade
which could introduce bias due to gradually changing
monitoring and clinical practices. Most patients did not
123
Summary
After TBI, resuscitation in the ER bay and neuro ICU was
successful in nearly all patients: 13 of 89 (14.6 %) patients
failed resuscitation and/or had intractable ICP. Nevertheless, metabolic crisis occurred in the first 72 h post injury
in 74 % of patients despite successful resuscitation and
controlled ICP. The duration of metabolic crisis was substantially longer for those patients with an unfavorable
outcome at 6 months. In four iterations of multivariate
models including age, GCS, pupil reactivity, and metabolic
crisis in the first 72 h, metabolic crisis was the strongest
predictor of unfavorable outcome. This research provides
further support for the utility of microdialysis to predict
outcome in TBI patients, and calls for further research in
controlled studies to determine whether specific treatment
interventions can influence early cerebral biochemistry and
thereby improve outcome.
Acknowledgments
by NS-058489.
Conflict of interest
report.
References
1. Bhardwaj A, Mirski MA, editors. Handbook of Neurocritical
Care: Second Edition. doi:10.1007/978-1-4419-6842-5_18,
Springer Science+Business Media, LLC 2011.
2. Hesdorffer DC, Ghajar J. Marked improvement in adherence to
traumatic brain injury guidelines in United States trauma centers.
J Trauma. 2007;63:8418.
3. Chesnut RM, Marshall LF, Klauber MR, et al. The role of secondary brain injury in determining outcome from severe head
injury. J Trauma. 1993;34:21622.
4. McHugh GS, Engel DC, Butcher I, Steyerberg EW, Lu J,
Mushkudiani N, et al. Prognostic value of secondary insults in
traumatic brain injury: results from the IMPACT study. J Neurotrauma. 2007;24:28793.
5. Stiefel MF, Udouteuk JD, Spiotta AM, Gracias VH, Golberg A,
Maloney-Wilensky E, Bloom S, Le Roux PD. Conventional
neurocritical care and cerebral oxygenation after traumatic brain
injury. J Neurosurg. 2006;105:56875.
6. Fletcher JJ, Bergman K, Bolstein PA, Kramer AH. Fluid balance,
complications, and brain tissue oxygen tension monitoring
7.
8.
9.
10.
11.
12.
13.
14.
15.
57
16. Brain Trauma Foundation. Guidelines for the management of severe
traumatic brain injury, 3rd ed. J Neurotrauma. 2007;24(Suppl 1):
S713.
17. Hebert PC, Wells G, Blajchman MA, et al. A multicenter, randomized, controlled clinical trial of transfusion requirements in
critical care. Transfusion Requirements in Critical Care Investigators. Canadian Critical Care Trials Group. N Engl J Med.
1999;340(6):40917.
18. Clifton GL, Miller ER, Choi SC, Levin HS. Fluid thresholds and
outcome from severe brain injury. Crit Care Med. 2002;30(4):73945.
19. Lam TS, Mak PSK, Siu WS, Lam MY, Cheung TF, Rainer TH.
Validation of a Modified Early Warning Score (MEWS) in
emergency department observation ward patients. Hong Kong J
Emerg Med. 2006;13:2430.
20. Wilcox R, Schoendbrodt F. Robust Statistics. 2012. http://dornsife.
usc.edu/labs/rwilcox/software/index.cfm. Accessed 6 Mar 2012.
21. Wilcox R. Introduction to robust estimatino and hypothesis
testing. 2nd ed. New York: Elsevier Inc; 2005.
22. Agostinelli C. wle: Weighted Likelihood Estimation. R package
version 0.9-4. 2010. http://CRAN.R-project.org/package=wle.
Accessed 6 Mar 2012.
23. Saatman KE, Duhaime A, Bullock R, Maas A, Valadka A,
Manley GT. Classification of traumatic brain injury for targeted
therapies. J Neurotrauma. 2008;25:71938.
24. Sander A. The Extended Glasgow Outcome Scale. The Center for
Outcome Measurement in Brain Injury. 2002. http://www.
tbims.org/combi/gose. Accessed 30 Mar 2012.
25. Vespa PM, Nuwer MR, Nenov V, Ronne-Engstrom E, Hovda
DA, Bergsneider M, Kelly DF, Martin NA, Becker DP. Increased
incidence and impact of nonconvulsive and convulsive seizures
after traumatic brain injury as detected by continuous electroencephalographic monitoring. J Neurosurg. 1999;91:75060.
26. Vespa PM, McArthur DL, Xu Y, Eliseo M, Etchepare M, Dinov
I, Alger J, Glenn TP, Hovda D. Nonconvulsive seizures after
traumatic brain injury are associated with hippocampal atrophy.
Neurology. 2010;75:7928.
27. Menon DK, Coles JP, Gupta AK, Fryer TD, Smielewski P,
Chatfield DA, Aigbirhio F, Skepper JN, Minhas PS, Hutchinson
PJ, Carpenter TA, Clark JC, Pickard JD. Diffusion limited oxygen delivery following head injury. Crit Care Med. 2004;32:
138490.
123