Cooper DJ, Myles PS, McDermott FT, Murray LJ, Laidlaw J, Cooper G, Tremayne AB, Bernard SS, Ponsford J, for the HTS Study Investigators . Prehospital Hypertonic Saline Resuscitation of Patients With Hypotension and Severe Traumatic Brain InjuryA Randomized Controlled Trial. JAMA. 2004;291(11):1350-1357. doi:10.1001/jama.291.11.1350
Author Affiliations: Departments of Intensive Care (Dr Cooper and Ms Murray) and Anaesthesia (Dr Myles), Alfred Hospital, Monash University; Neurosurgery Department, Royal Melbourne Hospital (Dr Laidlaw); Consultative Committee on Road Traffic Fatalities Victoria (Dr McDermott and Ms Tremayne); Metropolitan Ambulance Service (Mr Cooper and Dr Bernard); and Department of Psychology, Monash University (Dr Ponsford), Melbourne, Victoria, Australia.
Caring for the Critically Ill Patient Section Editor: Deborah J. Cook, MD, Consulting Editor, JAMA.
Context Prehospital hypertonic saline (HTS) resuscitation of patients with traumatic
brain injury (TBI) may increase survival but whether HTS improves neurological
outcomes is unknown.
Objective To determine whether prehospital resuscitation with intravenous HTS
improves long-term neurological outcome in patients with severe TBI compared
with resuscitation with conventional fluids.
Design, Setting, and Patients Double-blind, randomized controlled trial of 229 patients with TBI who
were comatose (Glasgow Coma Scale score, <9) and hypotensive (systolic
blood pressure, <100 mm Hg). The patients were enrolled between December
14, 1998, and April 9, 2002, in Melbourne, Australia.
Interventions Patients were randomly assigned to receive a rapid intravenous infusion
of either 250 mL of 7.5% saline (n = 114) or 250 mL of Ringer's lactate solution
(n = 115; controls) in addition to conventional intravenous fluid and resuscitation
protocols administered by paramedics. Treatment allocation was concealed.
Main Outcome Measure Neurological function at 6 months, measured by the extended Glasgow
Outcome Score (GOSE).
Results Primary outcomes were obtained in 226 (99%) of 229 patients enrolled.
Baseline characteristics of the groups were equivalent. At hospital admission,
the mean serum sodium level was 149 mEq/L for HTS patients vs 141 mEq/L for
controls (P<.001). The proportion of patients
surviving to hospital discharge was similar in both groups (n = 63 [55%] for
HTS group and n = 57 [50%] for controls; P = .32);
at 6 months, survival rates were n = 62 (55%) in the HTS group and n = 53
(47%) in the control group (P = .23). At 6 months,
the median (interquartile range) GOSE was 5 (3-6) in the HTS group vs 5 (5-6)
in the control group (P = .45). There was no significant
difference between the groups in favorable outcomes (moderate disability and
good outcome survivors [GOSE of 5-8]) (risk ratio, 0.99; 95% confidence interval,
0.76-1.30; P = .96) or in any other measure of postinjury
Conclusion In this study, patients with hypotension and severe TBI who received
prehospital resuscitation with HTS had almost identical neurological function
6 months after injury as patients who received conventional fluid.
Severe traumatic brain injury (TBI) is common in patients with major
trauma and typically involves young adult men.1 Despite
current management strategies, patients with severe TBI have a high mortality
rate (31%-49%) and a large number of survivors have persistent severe neurological
disability.1- 4 There
are 80 000 to 90 000 were cases of survivors with long-term disability
after head injury annually in the United States.5 The
mean lifetime cost of each TBI survivor with severe disability from TBI exceeds
US $2 million.6
After initial head trauma, secondary brain injury may occur due to hypoxia,
hypotension, or elevated intracranial pressure (ICP) and is associated with
a worse neurological outcome.3,7 Patients
with hypotension after severe TBI have twice the mortality rate of normotensive
patients.5 Therefore, aggressive resuscitation
with intravenous fluids is recommended in current guidelines for the management
of patients with severe TBI.8 Treatment of
increased ICP in patients with TBI is also likely to improve outcomes.3
Previous studies in unselected patients with trauma found that intravenous
hypertonic saline (HTS) increased blood pressure and decreased ICP compared
with isotonic resuscitation fluids.9- 11 Hypertonic
saline is also used for resuscitation in combination with hypertonic colloids
(usually dextran 70) to increase duration of effect. However, the combinations
are more expensive and in a randomized comparative 4-group trial, highest
survival rates were achieved with HTS alone (HTS alone, 60%; HTS with dextran
70, 56%; Ringer's lactate solution alone, 49%).11 A
meta-analysis of patients with TBI from 8 randomized trials of HTS-dextran
resuscitation reported increased survival from 27% to 38% (adjusted P = .048).12
In Europe, HTS-colloid solutions have been in clinical use since 1991
and HTS-dextran has regulatory approval in 14 European countries.13,14 Hypertonic saline or HTS-colloid
are recommended for prehospital fluid protocols for patients with head trauma.14 The Brain Trauma Foundation "Guidelines for Pre-hospital
Management of Traumatic Brain Injury" recommends HTS with or without dextran
at the "option" level.15 However, no prospective
randomized controlled trials have compared HTS with conventional intravenous
fluid resuscitation protocols in patients with TBI.
Prehospital HTS resuscitation may decrease secondary brain injury compared
with standard resuscitation protocols alone. We therefore conducted a double-blind,
randomized controlled trial of HTS resuscitation compared with standard fluid
resuscitation in patients with severe TBI to determine whether HTS improved
long-term neurological outcomes.
This double-blind, randomized controlled trial was conducted between
December 14, 1998, and April 9, 2002, in Melbourne, Australia. This region
has a population of more than 4 million persons served by the Metropolitan
Ambulance Service and Rural Ambulance Victoria. In this region, paramedics
trained in advanced life support therapies treat patients who sustain major
trauma by using protocols based on the Advanced Trauma Life Support guidelines.16 Adult patients with major trauma were transported
by road ambulance to 1 of 12 hospitals, or by helicopter and road ambulance
to a single hospital, designated as the regional trauma center.
Patients were eligible for the study if at any time during prehospital
care all the following were present: coma due to blunt head trauma, a Glasgow
Coma Scale (GCS) score17 of less than 9 (range,
3-15), and hypotension (systolic blood pressure, <100 mm Hg). Patients
with multisystem trauma were included. Patients were excluded if they had
penetrating trauma, were younger than 18 years, were pregnant, had no intravenous
access, had a serious premorbid disease on a medical identification bracelet,
had peripheral edema, were in close proximity to receiving hospital (scoop
and run), had absent sinus rhythm, or cardiac arrest.
Patients were randomly assigned to receive a 250-mL intravenous infusion
of either 7.5% saline (HTS) or 250-mL Ringer's lactate solution (controls)
in addition to standard intravenous resuscitation fluids. This volume was
chosen because the maximum volume and concentration of HTS that is safe to
administer intravenously during prehospital resuscitation through a peripheral
catheter is 250 mL of 7.5% saline.9 All published
randomized studies of HTS and HTS-dextran resuscitation in patients with trauma
have tested the same dose. The colorless study fluids were contained in identical
Patients, paramedics, treating physicians, and study coordinators were
all blinded to treatment allocation, which was concealed. Because multiple
paramedics and hospitals were responsible for patient care after enrollment,
randomization in blocks of 4 was stratified by ambulance. Ambulances transported
patients to specific hospitals; therefore, allocation was also stratified
by hospital. Sequentially numbered, computer-randomized, externally identical
intravenous bags were packed in groups of 4 into each ambulance. After each
bag of fluid was used, paramedics packed the next sequentially numbered bag
into the equipment box.
All patients were initially evaluated and treated by paramedics. When
a patient met the eligibility criteria, the next numbered bag of study fluid
was infused as rapidly as possible. Paramedics then administered a crystalloid,
Ringer's lactate solution, or a colloid solution (Haemacell, Hoechst Marion
Roussel, Australia), or both, according to medically determined protocols.
The protocol recommended a volume of 10-mL/kg intravenous colloid or crystalloid
for hypotension after blunt trauma, and this was repeated as required if hypotension
persisted after administration of the study fluid. After hospital admission,
patient care was at the discretion of the attending physicians and generally
followed the guidelines of the Brain Trauma Foundation.8
After enrollment of each patient, a postcard attached to each fluid
bag containing the study number and patient demographic details was mailed
to the coordinating center. Each patient was followed up by a study coordinator
through the hospital stay and thereafter until 6 months after enrollment or
death, whichever came first. All data were recorded via an Access database
(Microsoft, Redmond, Wash).
Data were prospectively collected on baseline characteristics, admission
vital signs and laboratory data, and all significant events after admission.
In patients who survived to hospital admission, computed tomography scans
were reviewed and graded for severity by a single neurosurgeon (J.L.) who
was blinded to treatment allocation and used a standard scoring system.18
The Glasgow Outcome Score (GOS) is the most widely accepted method of
analyzing outcome in patients with severe head injury and has been recently
refined by using a structured questionnaire and an 8-point scale (extended
Glasgow Outcome Score [GOSE])19 whereby 1 indicates
dead; 2, vegetative; 3, lower severe disability; 4, upper severe disability;
5, lower moderate disability; 6, upper moderate disability; 7, lower good
recovery; and 8, upper good recovery. Advantages of the GOS are simplicity,
wide recognition, and that differences in disability are clinically meaningful.20 Interrater reliability of structured interviews for
the GOS and GOSE was high (κ = 0.89 and κ = 0.85, respectively).20,21
At 3 and 6 months after injury, all surviving patients were interviewed
by the same research officer (L.J.M.) who visited each patient individually.
The GOSE was recorded using a standardized scoring system.19 Six
months after injury is considered an optimal assessment time because most
neurological outcomes have stabilized and patient loss to follow-up may be
problematic at later time points.19 The research
officer was trained by an experienced neurosurgeon (J.L.) and prior to study
commencement, interrater reliability of the outcome scoring by the research
officer was assessed against the neurosurgeon for 10 patients and was excellent
(κ = 0.78; P = .001).
Secondary outcomes included the first ICP and cerebral perfusion pressure
(CPP) recorded after ICP catheter insertion; duration of ICP elevation and
of inadequate CPP, worst oxygenation expressed as lowest PaO2/Fio2 ratio; and duration of inotropic support and mechanical ventilation.
During each patient interview, the research officer also measured the Functional
Independence Measure (score range, 1-7),22 a
well-validated measure of disability measuring physical and cognitive independence
that is highly predictive of patient need for supervision and assistance after
TBI,23,24 and the Rancho Los Amigos
score (range, 1-8),25 which measures cognitive
function in 8 categories and has been shown to have good interrater reliability.
The study was approved by the human ethics committees for all 12 receiving
hospitals and by the medical standards committee of the Metropolitan Ambulance
Service. Prehospital informed consent was waived. Patients were enrolled in
the study by paramedics, and then delayed written consent for participation
and continuation in the study was obtained from the next of kin, while the
patient was in intensive care. If the patient recovered sufficiently to provide
written informed consent for continuation in the study, then the patient's
consent was also obtained. The Alfred Hospital ethics committee supported
public disclosure and accordingly the study was publicized in the community
through print and radio media channels. In Australia, National Health and
Medical Research Council guidelines support delayed consent for appropriate
clinical research in emergency situations.26 There
are no restrictive federal regulations.
The study was managed by a steering committee comprising specialists
in trauma intensive care, emergency medicine, surgery, neurosurgery, a metropolitan
ambulance service manager, a neuropsychologist, statistician, and the project
manager. Site principal investigators (including D.J.C. and S.S.B.) managed
local study issues and ethics requirements.
A single interim analysis for efficacy was planned after recruitment
of 100 patients, using the 6-month GOSE as the primary outcome measure and
a stopping rule of P<.001, according to the study
protocol. After the statistician reviewed these interim results, the steering
committee was advised to continue the trial.
The study was designed with 80% power to detect a 20% improvement in
the conventional 5-level GOS at 6 months after injury; this improvement was
considered clinically significant. With a type I error of .05, a type II error
of .20, and allowing for nonparametric testing, 220 patients were required.
The primary outcome measure was the GOSE at 6 months. Secondary outcome
measures included serum sodium and systolic blood pressure at hospital admission,
initial measurement of ICP, hospital mortality rate, and GOSE at 3 months.
The analysis was performed by using a modified intention-to-treat basis,
with all patients who were enrolled and who correctly met study entry criteria
included in the primary and secondary analyses. Baseline characteristics of
the 2 groups were tabulated by using appropriate summary statistics. Analysis
of the principal outcome of GOSE at 6 months was performed by using the Mann-Whitney
test. Additional results are expressed as proportions with their P values or risk ratios (RRs) with 95% confidence intervals (CIs).
Numerical variables that approximated a normal distribution are summarized
as mean (SD) and the groups are compared with t tests;
variables that were not normally distributed are summarized as median (interquartile
range) and the groups are compared with Mann-Whitney tests. All reported P values are 2-sided with .05 set as the level of significance.
Statistical analyses were performed with SPSS for Windows, version 11.1 (SPSS
Inc, Chicago, Ill). Two a priori subgroups of patients with shorter prehospital
times (<1 hour) and less severe brain injury (GCS of 5-8) were identified.
A total of 262 patients were enrolled in the study, including 27 patients
who were subsequently excluded because they did not fulfill study entry criteria
(Figure 1). These included 9 patients
in whom paramedics incorrectly measured the prehospital GCS score, 1 patient
whose systolic blood pressure was more than 100 mm Hg, 8 patients who had
cardiac arrest before receiving study fluid, 6 patients with penetrating trauma,
and 3 patients who did not have trauma. In addition, 6 surviving patients
declined consent for further participation in the study, leaving 229 randomized
patients. Two patients who received an incorrect fluid volume, including 1
patient who received 2 bags and 1 who received 125 mL of study fluid, were
included in the outcome analysis.
Of the 229 correctly enrolled patients, 114 were randomly assigned to
the HTS fluid group and 115 to the control group. The treatment groups had
equivalent baseline characteristics (Table
1). Most patients with TBI were young (mean [SD], 38  years)
and male (66%). The preenrollment GCS score and systolic blood pressure were
equivalent. Apart from the fluid therapy, there were no differences in intubation
rates, scene times, or transport times between the groups. The total intravenous
colloid and crystalloid resuscitation fluids received in addition to the study
fluid (median, 1250 mL; Table 1)
and the body temperature on arrival at hospital (35°C) were the same in
The median injury severity scores in both groups were 38, indicating
severe injury, and the maximum abbreviated injury score was the same in both
groups for the score relating to the head injury. There were also no differences
between groups with respect to probability of survival, as measured by trauma
injury severity scale (TRISS; range, 0%-100%).27
Patients treated with HTS had a significant increase (P<.001) in serum sodium and chloride concentrations compared with
the patients receiving Ringer's lactate solution at hospital admission. This
difference was present on arrival in the emergency department and persisted
for about 12 hours (Table 2 and Figure 2). Prehospital hypotension had been
corrected in both groups and on arrival at hospital there were no significant
differences in systolic blood pressure between the groups (Table 2).
There were no significant differences between the groups with respect
to ICP (P = .08), CPP (P =
.40), duration of CPP of less than 70 mm Hg (P =
.06), gas exchange (PaO2/FIO2 ratio), or duration of
mechanical ventilation (Table 2).
The duration of inotropic support was less in patients receiving HTS than
in those receiving Ringer's lactate solution (P =
The outcome of the patients is shown in Table 3. Of the 229 patients who were enrolled in the study, 8 (3.5%)
died before hospital arrival and 47 (21%) died either in the emergency department
or in the operating room. A total of 174 patients were admitted to the intensive
care unit and 120 patients (53%) survived to hospital discharge. The proportion
of patients surviving to hospital discharge was similar in both groups (n
= 63 [55%] for HTS group and n = 57 [50%] for controls; P = .32). The proportion of patients surviving at 6 months was n =
62 (55%) in the HTS group and n = 53 (47%) in the control group (P = .23; RR, 1.17; 95% CI, 0.9-1.5).
The GOSE at 6 months for each group is shown in Figure 3. At 6 months, a total of 2 patients (1%) were lost to follow-up
and 1 patient had withdrawn consent. Therefore, 228 patients (116 survivors)
were assessed for neurological outcome at 3 months and 226 (115 survivors)
at 6 months after injury. There were no significant differences between the
groups with respect to the primary study end point, the GOSE, or other measures
of functional neurological status (Table
3) at either 3 or 6 months after injury. There was also no difference
in the rate of favorable outcome, defined by a GOSE of 5 or more, for HTS
vs control (RR, 0.99; 95% CI, 0.76-1.30; P = .96).
Rates of return to work were not significantly different between the groups.
Predetermined exploratory analyses were conducted to identify possible
subgroups that may benefit from initial resuscitation with HTS, which revealed
no significant benefit from HTS for patients with a less severe brain injury
according to a baseline GCS score of 5 to 8 (n = 101, P = .48); for those patients with a shorter (<60 min; n = 95, P = .26) or longer (>60 min; n = 122, P = .86) injury-to-hospital time; or for those patients treated with
intravenous crystalloid fluids alone (n = 96, P =
.85), with respect to an improved GOSE score 6 months after injury.
In patients with severe TBI, prehospital hypotension is strongly associated
with poor outcome.3,7 Accordingly,
more effective resuscitation with intravenous fluids should improve cerebral
perfusion, decrease secondary brain injury, and improve neurological outcomes.
Although the role of intravenous fluid therapy in prehospital trauma care
is controversial,28 current guidelines recommend
that hypotension be urgently treated in patients with severe TBI.7 However, the choice of fluid is controversial.
In critically ill patients, systematic reviews have reported colloid
resuscitation29 and albumin therapy30 to be associated with increased mortality. In patients
with major trauma, another systematic review31 found
colloid resuscitation was associated with adverse outcomes. In patients with
trauma, there has been considerable interest in the possible role of hypertonic
crystalloids for prehospital fluid resuscitation. A meta-analysis of patients
with severe TBI from randomized trials of HTS-dextran for prehospital trauma
resuscitation reported an 11% absolute increase in survival compared with
standard resuscitation fluids.11 Furthermore,
no adverse effects of HTS were detected in more than 600 patients with trauma
receiving prehospital HTS in clinical trials.9 Thus,
HTS and HTS-dextran are increasingly recommended for initial resuscitation
of patients with hypotension and trauma, particularly those with head injuries.13,15 However, although our double-blind,
randomized trial of prehospital intravenous HTS compared with standard resuscitation
fluids in 229 patients with severe TBI and hypotension found a small trend
toward greater survival for HTS patients, neurological outcomes were almost
identical 6 months after injury.
Unexpectedly, both study groups received the same prehospital volume
of intravenous resuscitation fluids in addition to the study fluids (median,
1250 mL). It is likely therefore that paramedics ran all resuscitation fluids
at maximal rates regardless of fluid type during the prehospital period. Patients
in both groups also received a similar volume of colloid (median, 500 mL for
HTS patients and 250 mL for controls; Table
1). Because HTS expands intravascular volume 4 to 10 times greater
than the infused volume,9 it was expected that
HTS would significantly improve CPP in patients with severe TBI. However,
prehospital hypotension had been corrected by hospital arrival in both groups.
Although HTS resuscitation is likely to have been faster, conventional resuscitation
protocols were equally effective for prehospital resuscitation of these patients.
Intravenous HTS also decreased ICP in patients with TBI32 and
in our study, the ICP was lower when first measured in hospital in the patients
with HTS than in controls. This decrease was not significant (P = .08), perhaps because intracranial hypertension is usually not
problematic in the first hours after TBI. Interpretation of ICP and CPP outcomes
is limited because patients who improved quickly or who died quickly were
unlikely to have ICP measured.
Two a priori subgroups were investigated. First, our total prehospital
times were relatively long (median, 60 minutes) and it has been proposed that
HTS has an advantage of more rapid resuscitation than isotonic crystalloids
and may therefore be beneficial when prehospital times are shorter. Analysis
of patients with short prehospital times, however, showed no benefit in this
group. Second, some patients with low GCS values may have had severe primary
injuries with little or no potential for cerebral recovery and improved resuscitation
fluids may benefit only patients with less severe primary brain injuries.
However, our subgroup analysis of patients with less severe brain injury (GCS
score range, 5-8) did not support this hypothesis.
This study has a number of strengths. It was, to our knowledge, the
first randomised prehospital trial of HTS resuscitation in hypotensive patients
with trauma with severe brain injury. Allocation was concealed and paramedics,
patients, physicians, and outcome assessors all were blinded to treatment
allocation. The randomization was stratified by ambulance and receiving hospital
to minimize between-hospital management differences. Accordingly, baseline
characteristics were well-balanced between groups. Unlike many head injury
trials, patient loss to follow-up at 6 months was only 1%. Finally, this was
the first prehospital resuscitation fluid trial to measure long-term neurological
function as the primary outcome in patients with TBI.
This study also has several limitations. First, unlike some previous
studies of HTS, we did not combine HTS with dextran. Our study was designed
to test HTS alone, because a previous randomized 4-group trial found the greatest
survival benefit for prehospital patients with trauma after HTS alone (without
dextran).11 Recent meta-analyses had reported
increased mortality after colloid resuscitation,29,30 particularly
in patients with trauma31 and furthermore there
seemed little benefit from adding dextran when the paramedic protocols included
an optional colloid solution. Finally, the addition of dextran to HTS increases
costs and the potential risk of adverse reactions.
Second, our study included only 229 patients. However, the study had
80% power to identify a 1-grade change in the GOS following HTS. This difference
would have been clinically meaningful in terms of long-term quality of life.19 Other neurological scoring systems also were not
different between the groups.
Third, the study population predominantly (90%) included patients with
multisystem trauma. Patients with isolated head injury may respond differently
than those with multiple injuries. However, in both study groups, survival
was substantially better (mean survival, 60%) than predicted by the calculated
probability of survival using TRISS (mean TRISS, 45%). This suggests that
paramedic and hospital protocols, including vigorous prehospital fluid resuscitation,
were as good or better than standard benchmarks.
We found that prehospital HTS and conventional resuscitation protocols
alone resuscitated hypotensive patients with TBI equally well. In an established
trauma system with effective paramedic resuscitation protocols, prehospital
HTS did not improve long-term neurological function compared with conventional
resuscitation fluids alone.