You are on page 1of 10


AOPXXX10.1177/1060028016628893Annals of PharmacotherapyBurgess et al

Review Article
Annals of Pharmacotherapy

A Systematic Review of Randomized

© The Author(s) 2016
Reprints and permissions:
Controlled Trials Comparing Hypertonic
DOI: 10.1177/1060028016628893

Sodium Solutions and Mannitol for

Traumatic Brain Injury: Implications for

Emergency Department Management

Sarah Burgess, PharmD1, Riyad B. Abu-Laban, MD, MHSc1,2,

Richard S. Slavik, PharmD1,3, Erik N. Vu, MD1,4, and Peter J. Zed, PharmD1

Objective: To comparatively evaluate hypertonic sodium (HTS) and mannitol in patients following acute traumatic brain
injury (TBI) on the outcomes of all-cause mortality, neurological disability, intracranial pressure (ICP) change from baseline,
ICP treatment failure, and serious adverse events. Data Sources: PubMed, EMBASE, CENTRAL, Cochrane Database of
Systematic Reviews,, and WHO ICTRP (World Health Organization International Clinical Trials Registry
Platform) were searched (inception to November 2015) using hypertonic saline solutions, sodium chloride, mannitol, osmotic
diuretic, traumatic brain injury, brain injuries, and head injury. Searches were limited to humans. Clinical practice guidelines
and bibliographies were reviewed. Study Selection and Data Extraction: Prospective, randomized trials comparing
HTS and mannitol in adults (≥16 years) with severe TBI (Glasgow Coma Scale score ≤8) and elevated ICP were included.
ICP elevation, ICP reduction, and treatment failure were defined using study definitions. Data Synthesis: Of 326 articles
screened, 7 trials enrolling a total of 191 patients met inclusion criteria. Studies were underpowered to detect a significant
difference in mortality or neurological outcomes. Due to significant heterogeneity and differences in reporting ICP change
from baseline, this outcome was not meta-analyzed. No difference between HTS and mannitol was observed for mean ICP
reduction; however, risk of ICP treatment failure favored HTS (risk ratio [RR] = 0.39; 95% CI = 0.18-0.81). Serious adverse
events were not reported. Conclusions: Based on limited data, clinically important differences in mortality, neurological
outcomes, and ICP reduction were not observed between HTS or mannitol in the management of severe TBI. HTS appears
to lead to fewer ICP treatment failures.

traumatic brain injury, hypertonic saline, mannitol, intracranial pressure, emergency medicine, trauma

Introduction perfusion, poor neurological outcome, and mortality.1 In

one review, the incidence of death was 18.4% for patients
Traumatic brain injury (TBI) is associated with significant with an ICP <20 mm Hg but rose to 55.6% for those with an
morbidity, mortality, and economic impact.1 The annual ICP greater than 40 mm Hg.1 There are various methods of
incidence of severe TBI in Canada is estimated at 11.4 per controlling ICP; however, one of the main pharmacological
100 000 people. In 2003, the incidence of TBI in the United interventions in severe TBI is hyperosmolar therapy.
States was estimated to be 1.5 million cases, of which 78%
were treated in the emergency department (ED) alone and
3% were fatal.2 In 2005, an estimated 3.2 million people in 1
University of British Columbia, Vancouver, BC, Canada
the United States were living with a long-term disability VCHRI Centre for Clinical Epidemiology and Evaluation, Vancouver,
BC, Canada
following TBI.3 The direct and indirect economic cost of 3
Interior Health, Kelowna, BC, Canada
severe TBI in the United States in 2010 was estimated to be 4
British Columbia Emergency Health Services, Vancouver, BC, Canada
76.5 billion dollars.4 A key component in the management
Corresponding Author:
of severe TBI is prevention of secondary brain injury asso- Peter J. Zed, Faculty of Pharmaceutical Sciences, University of British
ciated with elevated intracranial pressure (ICP). Elevation Columbia, 2405 Wesbrook Mall, Vancouver, BC, V6T 1Z3, Canada.
of ICP to 20 mm Hg or greater can result in impaired brain Email:

Downloaded from at Middle East Technical Univ on February 2, 2016

2 Annals of Pharmacotherapy 

Mannitol and hypertonic saline or sodium (HTS) solutions Systematic Reviews and Meta-Analysis (PRISMA).11
are commonly used hyperosmolar agents for this purpose PubMed (1947 to November 2015), EMBASE (1947 to
despite a lack of high-quality clinical trials. Predicated on November 2015), CENTRAL, Cochrane Database of
fluid shifts and the Monro-Kellie Doctrine, both agents are Systematic Reviews,, and the WHO
believed to lower ICP predominantly by creating an osmo- ICTRP (World Health Organization International Clinical
lar gradient that leads to mobilization of fluid from the brain Trials Registry Platform) were searched using combina-
to the intravascular compartment.5 The beneficial effects of tions of the following search terms: hypertonic saline solu-
both these agents has also been postulated around their neu- tions, sodium chloride, mannitol, osmotic diuretic,
roprotective effects at a cellular level, based on prevention traumatic brain injury, brain injuries, and head injury.
of oxidative stress and secondary insults from inflammation Searches were limited to humans, with no restriction on
and cytokine-mediated cellular pathways.6 Historically, language of publication. Clinical practice guidelines and
20% mannitol at a dose of 0.25 to 1.4 g/kg infused intrave- reference lists of retrieved articles were also reviewed.
nously over 20 minutes has been considered the reference Authors of trials were contacted for additional data or clar-
hyperosmolar agent to acutely lower ICP and is a level II ification, as required.
recommendation in both the Brain Trauma Foundation and Eligibility criteria for study inclusion consisted of pro-
the European Brain Injury Consortium guidelines.7,8 More spective, randomized clinical trials that compared HTS and
recently, interest has increased for the use of HTS at con- mannitol solutions administered intravenously in adult
centrations ranging from 3% to 23.4% in this setting, humans with evidence (clinical or measured) of elevated
although no formal recommendations exist for its use.7,8 ICP secondary to acute severe TBI defined as Glasgow
The lack of consensus over the preferred hyperosmolar Coma Scale score ≤8. Cross-over studies were eligible for
agent has led to practice variation around the globe. A 2007 inclusion, provided the initial treatment was randomized.
survey in the United Kingdom found that 50% of neuro– Studies that included only pediatric patients (<16 years old)
intensive care units used HTS either as sole therapy or in and those who had intracranial hemorrhage, subarachnoid
cases refractory to mannitol, whereas the remaining 50% hemorrhage, and/or ischemic stroke were excluded. The
used mannitol only for elevated ICP following TBI.9 primary outcome was all-cause mortality. Secondary out-
Similarly, a study published in 2014 reported a survey of comes included morbidity, defined as functional neurologi-
emergency physicians in Québec, Canada, that found that cal recovery or disability as measured by a validated scale
25% and 13% of respondents reported using mannitol or (Disability Rating Scale, Glasgow Outcome Scale); the
hypertonic saline, respectively, often or always in the man- magnitude of ICP change from baseline; ICP treatment fail-
agement of presumed elevated ICP in patients with severe ure, defined as failure to obtain ICP target and/or requiring
TBI.10 additional intervention; and serious adverse events. Time
Most patients with acute severe TBI are first evaluated in frames chosen by individual study authors to determine ICP
the ED before undergoing surgery or being transferred to a change from baseline and ICP treatment failure were con-
neuro–intensive care unit. As a result, emergency physi- sidered. When multiple time frames were used, only the
cians typically manage acute TBI and make initial decisions time frame that corresponded to the study’s primary out-
regarding hyperosmolar therapy for confirmed or presumed come was reported. Magnitude of ICP change was assessed
ICP elevation. To our knowledge, there are no randomized by comparing the mean change in ICP from baseline to end
controlled trials comparing hyperosmolar therapies for ele- of study period (ranged from immediately following treat-
vated ICP and acute severe TBI in the ED setting. As a ment to 120 minutes) with HTS or mannitol and taking the
result, data from other settings must be extrapolated and difference. If means and SDs were not presented, authors
applied. The objective of this systematic review is to evalu- were first contacted for these data. A HTS solution was
ate the efficacy and safety of HTS solutions and mannitol in defined as a solution with a concentration of sodium chlo-
adults with presumed or measured elevated ICP secondary ride equivalent to or greater than 3%. One reviewer (SB)
to acute severe TBI by examining the data from randomized screened titles and abstracts of studies identified by the
controlled trials. The primary outcome is all-cause mortal- search strategy to exclude irrelevant studies. Two indepen-
ity. Secondary outcomes included morbidity defined as a dent reviewers (SB and PJZ) confirmed the eligibility of
measure of neurological function and disability, magnitude each study, with discrepancies resolved by consensus. A
of ICP change from baseline, ICP treatment failure, and standardized data collection form was developed to extract
serious adverse events. the following information: study design and setting, patient
characteristics, intervention characteristics (concentrations,
dosages, and administration), mortality data, neurological
outcomes, magnitude of ICP change from baseline, and
The design and results of this systematic review follow the treatment failure as defined in each study and serious
recommendations from the Preferred Reporting of adverse events. Dichotomous and continuous outcomes

Downloaded from at Middle East Technical Univ on February 2, 2016

Burgess et al 3

Figure 1.  Trial flow diagram.

Abbreviations: ICP, intracranial pressure; RCT, randomized controlled trial.

were entered into RevMan 5.3 Software to generate sum- Results

mary statistics using differences in means and risk ratios
(RRs) with 95% CIs. If data were not available, authors of Study Characteristics
respective trials were contacted. All meta-analyses were Seven randomized trials enrolling a total of 191 patients
performed using a random-effects model. Heterogeneity were identified that met the inclusion criteria (Figure 1).
was assessed graphically using a forest plot and quantified Study characteristics are presented in Table 1. Most trials
statistically using the χ2 test and I2 statistic. Significant het- were performed in European centers and none specifically
erogeneity was predefined as P < 0.1 for the χ2 test. The evaluated the use of HTS or mannitol in the management of
degree of heterogeneity was defined a priori as low (I2 ≤ TBI in the ED setting. All trials had a small sample size; the
25%), moderate (I2 ≤ 50%), and high (I2 ≥ 75%).12 Potential largest trial enrolled 47 patients.14 Four trials14-17 included
reasons for detected heterogeneity were explored. No pre- severe head trauma/TBI patients with a total of 130 patients,
specified subgroup or sensitivity analyses were planned. whereas 3 trials18-20 included a mixed population of neuro-
The internal validity of each trial was evaluated using The surgical patients, yielding 33 more patients with TBI. To be
Cochrane Collaboration’s tool for assessing risk of bias in included in the studies, patients required a GCS ≤8 and/or
randomized controlled trials.13 Publication bias was required ICP monitoring as part of their management. HTS
assessed graphically with a funnel plot. solutions ranging in concentrations from 7.2% to 15% were

Downloaded from at Middle East Technical Univ on February 2, 2016

Table 1.  Randomized Trials Comparing Hypertonic Saline/Sodium Solutions and Mannitol for Elevated ICP in Traumatic Brain Injury.
Comparators Outcomes

ICP Reduction and Treatment Failure Serious Adverse

Study Design  Setting Patient Population Inclusion Criteria HTS Mannitol All-Cause Mortality Neurological Outcomes (as defined and reported in trial) Events

Cottenceau et al,14 2 ICUs, France and •• n = 47 •• n = 22 •• n = 25 At 6 months: 6/22 6 Month GOS (HTS vs •• Mean change in ICP at 30 NR
2011; prospective, Israel •• >16 years old •• 7.5% saline •• 20% Mannitol (27.3%) HTS mannitol) minutes (mmHg): −5.7 HTS vs
RCT •• Severe TBI •• 2 mL/kg IV over 20 minutes •• 4 mL/kg IV over 20 vs 6/25 (24.0%) •• Good recovery: 3 vs 7 −5.8 mannitol; P = NS
•• GCS ≤8 •• Median GCS: 5 minutes mannitol, P = NS •• Moderate disability: 2 vs 4 •• Mean change in ICP at 120
•• Baseline ICP: 18.7 mm Hg •• Median GCS: 7 •• Severe disability: 6 vs 7 minutes (mmHg): −4.0 HTS vs
•• Baseline ICP: 17.7 mm Hg •• Vegetative: 6 vs 1 −2.7 mannitol, P = NS
Treatment failure not reported
Vialet et al,15 2003; Single trauma center, •• n = 20 •• n = 10 •• n = 10 At 90 days: 4 (40%) 90-day GOS (severe disability): Treatment failure defined by NR
prospective, open- France •• >18 years old •• 7.5% saline (2400 mOsm/kg/H2O) •• 20% Mannitol (1160 HTS vs 5 (50%) 6 (60%) HTS vs 5 (50%) inability to ↓ ICP <35 mm Hg with 2
label, RCT •• Severe head trauma •• 2 mL/kg IV over 20 minutes mOsm/kg/H2O) mannitol; mannitol; P = NS consecutive infusions: 1 (10%) HTS vs
•• GCS < 8 •• May repeat 10 minutes after end •• 2 mL/kg IV over 20 P = NS 7 (70% mannitol; P = 0.01
•• Refractory to other ICP-lowering of infusion minutes
interventions •• Initial GCS: 4 •• May repeat 10 minutes
•• Baseline ICP:NR after end of infusion
•• Initial GCS: 6
•• Baseline ICP: NR
Ichai et al, 2008; Single center, France •• n = 34 •• n = 17 •• n = 17 NR 1-Year GOS •• Mean (SD) change in ICP at 120 NR
prospective, open- •• ≥18 to <65 years old •• sodium lactate solution (1100 •• 20% Mannitol (1160 •• Good recovery + minutes (mm Hg): −5.9 (4.1)
label, RCT •• Severe head injury GCS ≤8 or a mOsm/L) mOsm/L) moderate disability: 11 sodium lactate vs −3.2 (3.7)
rapid worsening of neurological •• 1.5 mL/kg IV over 15 minutes •• 1.5 mL/kg IV over 15 (68.8%) sodium lactate vs mannitol
status •• Rescue therapy to alternate minutes 5 (31.2%) mannitol •• Treatment failure defined as ICP
treatment if ICP still not controlled •• Rescue therapy to •• Severe disability + decrease by less than 5 mm Hg
•• Initial GCS: 6 alternate treatment if ICP vegetative + dead: 6 or absolute value of >20 mm Hg
•• Baseline ICP: NR still not controlled (35.3%) sodium lactate vs following treatment: 5 (29.4%)
•• Initial GCS: 4 11 (64.7%) mannitol sodium lactate vs 8 (47%) mannitol
•• Baseline ICP: NR P = 0.055 P = 0.053
Sakellaridis et al17 Single ICU, Greece •• n = 29 •• 15% Saline •• 20% Mannitol NR 3-Month GOS: did not •• Mean (SD) change in ICP (mm NR
2011; randomized, •• Mean age = 36 years •• 0.42 mL/kg IV bolus •• 2 mL/kg IV over 20 differentiate between groups Hg): −8.43 ( 6.65) HTS vs −7.96
alternating treatment •• Severe head injury •• Initial GCS: 5.4 minutes (5.79) mannitol, P = 0.59
protocol •• GCS ≤8 •• Baseline ICP: NR •• Initial GCS: 5.4 •• Treatment failure not reported
•• Baseline ICP: NR
Harutjunyan et al,18 Single center, •• n = 32 •• n = 17 •• n = 15 •• ICU mortality: 7 •• NR •• Median change (range) in ICP at NR
2005; prospective, Germany •• >18 years old •• 7.2% saline/hydroxyethyl starch •• 15% Mannitol (41.2% ) HTS vs 60 minutes (mm Hg): −11 (5-18)
open-label, RCT •• Severe brain damage 200/0.5 •• Infused IV until ICP <15 9 (60%) mannitol HTS vs −14 (7-20) mannitol; P
•• GCS <8 with cerebral edema on •• Infused IV until ICP <15 mm Hg mm Hg •• P = NS < 0.0001
CT scan •• Median dose used = 1.4(0.3-3.1) •• Median dose used = 1.8 •• Treatment failure defined by
•• (SAH, n = 9; infarct, n = 7; trauma, n mL/kg (0.5-6.5) mL/kg inability to attain ICP <15 mm
= 10; ICH, n = 4; other, n = 2) •• Initial GCS: 5.8 •• Initial GCS: 6 Hg (percentage of elevated ICP
•• Baseline ICP: 22 mm Hg •• Baseline ICP: 23 mm Hg episodes): 2 (3.5%) HTS vs 4

Downloaded from at Middle East Technical Univ on February 2, 2016

(7.5%) mannitol
Battison et al,19 2005; Single ICU, Scotland n=9 •• n=9 •• n=9 NR NR •• Median change (95% CI) in ICP NR
prospective, open- ≥16 years old •• 7.5% saline/6% dextran-70 solution •• 20% Mannitol (1245 between HTS vs mannitol before
label, randomized Brain injury (TBI = 6, SAH = 3) requiring (2498 mOsm/kg) mOsm/kg) and after administration (mm
cross-over trial ICP monitoring •• 100 mL IV over 5 minutes •• 200 mL IV over 5 minutes Hg): −5 (−10.8, −3.0); P = 0.014
•• Initial GCS: NR •• Initial GCS: NR •• Treatment failure defined by
•• Baseline ICP: NR •• Baseline ICP: NR inability to attain ICP <20 mm
Hg (percentage of elevated ICP
episodes): 2 (11.1%) HTS vs 4
(22.2%) mannitol
Francony et al,20 2008; 2 ICUs, France •• n = 20 •• n = 10 •• n = 10 NR NR •• Mean (SD) change in ICP at 120 NR
prospective, open- •• ≥18 years old •• 7.45% saline (2548 mOsm/L) •• 20% Mannitol (1100 minutes (mm Hg): −6 (3) HTS vs
label, RCT •• Severe brain trauma (TBI, n = 17; •• 100 mL IV (255 mOsm/dose) over mOsm/L) −10 (4) mannitol; P < 0.01
ICH, n = 2; ischemic stroke, n = 1) 20 minutes •• 231 mL IV (255 mOsm/ •• Treatment failure defined as
•• GCS NR •• Mean GCS: 7 dose) over 20 minutes a reduction in ICP <20% of
•• Sustained ICP >20 mm Hg for >10 •• Mean ICP: 27 mm Hg •• Mean GCS: 8 baseline values at 60 minutes: 1
minutes •• Mean ICP: 31 mm Hg (10%) HTS vs 0 (0%) mannitol

Abbreviations: GCS, Glasgow Coma Scale score; GOS, Glasgow Outcome Scale; HTS, hypertonic saline or sodium solution; ICH, intracranial hemorrhage; ICP, intracranial pressure; ICU, intensive care unit; IV, intravenous; NR, not
reported; NS, not statistically significant; RCT, randomized controlled trial; SAH, subarachnoid hemorrhage.
Burgess et al 5

Figure 2.  Funnel plot studies reporting difference in ICP treatment failure between HTS and mannitol.
Abbreviations: ICP, intracranial pressure; HTS, hypertonic saline or sodium solution; RR, risk ratio; SE, standard error.

compared with 15% or 20% mannitol. Five of the studies Neurological Outcomes. Four trials reported neurological
compared equimolar solutions.14,16,17,19,20 Figure 2 does not outcomes using the 5-category Glasgow Outcome Scale
reveal overt evidence of publication bias; however, the (GOS).14,15,16,17 In the trial by Vialet et al,15 6 of 10 patients
accuracy of the funnel plot is limited given the small num- in the HTS arm and 5 of 10 patients in the mannitol arm
ber (<10) of trials.21 were classified as having severe neurological disability at
90 days. Ichai et al16 assessed neurological outcomes at 1
year after TBI. Patient outcomes were compared according
Main Results to the initial randomization (hypertonic sodium lactate or
mannitol), regardless of rescue cross-over treatments; 11
All-Cause Mortality. Three trials reported mortality
patients (68.8%) in the hypertonic sodium lactate arm com-
outcomes.14,15,18 Cottenceau et al14 compared 6-month mor-
pared with 5 patients (31.2%) in the mannitol arm had a
tality between patients who received 7.5% HTS or 20% good recovery or moderate disability, and 6 patients (25.5%)
mannitol. Mortality rates were 27.3% and 24.0% for the in the hypertonic sodium lactate arm compared with 11
HTS and mannitol arms, respectively (P value calculated as patients (64.7%) in the mannitol arm had the reported out-
0.8). Vialet et al15 reported 90-day mortality between 7.5% come of severe disability, vegetative state, or death. The dif-
hypertonic saline and 20% mannitol in 20 patients with TBI ference between the 2 groups failed to reach statistical
resistant to standard therapy. Four of 10 (40.0%) patients in significance (P = 0.055). Sakellaridis et al17 reported
the HTS arm compared with 5 of 10 (50.0%) patients in the 3-month neurological outcomes for 29 patients with TBI
mannitol group died; the difference was not statistically sig- using the GOS; however, because patients received both
nificant [RR = 1.25 (0.47-3.33)]. Harutjunyan et al18 interventions (HTS and mannitol), we are unable to attri-
assessed ICU mortality between patients with severe brain bute the outcome to one specific treatment.17 An analysis of
damage (including TBI, subarachnoid hemorrhage, isch- neurological outcomes at 6 months by Cottenceau et al14
emic infarct, and intracranial hemorrhage) who received found a statistical trend for better outcomes in the mannitol
either 7.2% saline/hydroxyethyl starch (HES) 200/0.5 solu- group, with numerically more patients having a good recov-
tion or 15% mannitol to decrease ICP to <15 mm Hg. Seven ery or only moderate disability.
patients (41.2%) assigned to hypertonic saline and 9 patients
(60.0%) assigned to mannitol did not survive (χ2 statistic = ICP Control. All 7 included trials reported a measure of
1.13; P = 0.29). change in ICP from baseline, which was variably defined

Downloaded from at Middle East Technical Univ on February 2, 2016

6 Annals of Pharmacotherapy 

Figure 3.  Forest plot of risk of ICP treatment failure between HTS and mannitol.
Abbreviations: ICP, intracranial pressure; HTS, hypertonic saline or sodium solution.

and included the mean change in ICP, median change in secondary end point of percentage of successfully treated
ICP, maximal change in ICP, time to ICP below a specified episodes of ICP. At 4 hours, the median change in ICP from
target, duration of ICP lowering effect, and percentage of baseline was −7.0 mm Hg with the sodium lactate solution
successfully treated elevated ICP episodes. Because of the compared with −4.0 mm Hg with mannitol (P = 0.03).
heterogeneity of ICP outcome measures and failure to Episodes of elevated ICP were successfully treated 90.4%
retrieve mean difference (±SD) in ICP between treatment and 70.4% of the time in those receiving sodium lactate
groups from all authors, we decided against pooling the solution or mannitol, respectively (P = 0.053). Sakellaridis
data in a meta-analysis, and the results are presented quali- et al17 compared the ICP-lowering effects of 15% HTS and
tatively. It was felt that there were sufficient data in ICP 20% mannitol in 29 patients with severe head injury and a
treatment failure to pool these results. total of 199 elevated ICP events. Mean change in ICP was
In patients with elevated ICP refractory to standard inter- reported as −8.4 mm Hg (±6.7) with HTS compared with
ventions, Vialet et al15 found that 7.5% HTS compared with −8.0 mm Hg (±5.8) with 20% mannitol (P = 0.59). The
20% mannitol significantly reduced the mean daily number mean duration of the ICP-lowering effect was 257 minutes
of episodes in which ICP was greater than 25 mm Hg: 6.8 with HTS and 213 minutes with mannitol, respectively (P =
(±5.5) versus 13.3 (±14.2; P = 0.02. HTS also decreased the 0.4). Cottenceau et al14 compared 7.5% HTS and 20% man-
mean total duration in minutes of episodes of ICP >25 mm nitol in 47 patients with severe TBI and elevated ICP. At 30
Hg: 62 (±81) versus 95 (±92); P = 0.04. Treatment failures and 120 minutes after infusions, the mean change in ICP
defined as inability to decrease ICP <35 mm Hg with 2 con- was −5.7 and −4 mm Hg with HTS and −5.8 and −2.6 mm
secutive infusions occurred in 1 patient in the HTS arm Hg with mannitol, respectively (P > 0.05). The risk of ICP
compared with 7 patients in the mannitol arm (P = 0.01). treatment failure as defined by each study reached statisti-
Harutjunyan et al18 found that compared with mannitol cal significance in favor of HTS compared with mannitol
15%, a solution containing 7.2% HTS/HES 200/0.5 reduced (Figure 3; RR = 0.39; 0.0.18-0.81; I2 = 0%).
the median time to obtain an ICP <15 mm Hg: 6 minutes
(1.2-15) versus 8.7 (4.2-19.9); P < 0.0002. The maximal Serious Adverse Events.  None of the included studies assessed
decrease in ICP from baseline was 57% with HTS com- or reported the occurrence of serious adverse events (Table 1).
pared with 48% with mannitol; P < 0.01. Battison et al19
compared the median difference in ICP before and after Risk of Bias Assessment.  All the included studies were ran-
administration of 7.5% saline/6% dextran-70 solution or domized; however, the randomization process was not well
20% mannitol. The HTS solution significantly decreased described. Allocation concealment was also not well
ICP compared with mannitol by a median of 5 mm Hg described and clearly present in only 3 studies (all with the
(range: −10.8 to −3.0; P = 0.014) and maintained an ICP use of sealed opaque envelopes).14,16,20 Only one study
<20 mm Hg for a longer duration: 148 minutes (interquar- reported blinding of patients, personnel, and outcome
tile range [IQR] = 124.5-227.8) versus 89.5 minutes (IQR = assessors17; however, the outcome of ICP measurement was
75.5-205.5); P = 0.044. Francony et al20 compared the mean objectively quantified in all studies, leading to a lower risk
change in ICP from baseline at 2 hours following a single of performance and detection bias. There did not appear to
IV infusion of either 7.45% HTS or 20% mannitol in 20 be any evidence of incomplete outcome data or selective
patients with severe brain trauma. HTS decreased ICP by 6 outcome reporting. Three studies used crossover designs,
mm Hg (±3), whereas mannitol decreased ICP by 10 mm which could potentially lead to over- or underestimates of
Hg (±4); P < 0.01. Ichai et al16 compared the effect of an IV the true intervention effect as a result of carryover effects
infusion of sodium lactate solution with 20% mannitol on from the previous treatment.16,17,19 Risk of bias assessment
the primary end point of ICP reduction at 4 hours and the for the included studies is summarized in Figure 4.

Downloaded from at Middle East Technical Univ on February 2, 2016

Burgess et al 7

Cottenceau √ √ ? ? √ √ √
et al.
Vialet et al. X X ? √ √ √ √
Ichai et al. √ √ ? ? √ √ ?
Sakellaridis et ? X ? √ √ √ ?
Harutjunyan √ ? ? ? √ √ √
et al.
Battison et al. ? X ? ? √ √ ?
Francony et al. ? √ ? ? √ √ √

Random Allocation Blinding of Blinding of Incomplete Selective Other

sequence concealment participants outcome outcome outcome bias
generation (selection and personnel assessment data reporting
(selection bias) (performance (detection (attrition (reporting
bias) bias) bias) bias) bias)
X = high risk of bias; √ = low risk of bias; ? = unclear risk of bias

Figure 4.  Risk of bias assessment of included studies.

Discussion significant reduction in ICP in favor of HTS compared with

mannitol.15,16,18,19 Two trials14,17 found no statistically sig-
Both HTS solutions and mannitol are commonly used in the
nificant difference, and the results of 1 trial favored
management of elevated ICP in the setting of severe TBI
mannitol.20 The heterogeneous results are not surprising
despite the lack of randomized controlled trials comparing
given methodological differences among the trials, includ-
these 2 interventions. In this review, there were no statisti-
ing various definitions of ICP treatment thresholds and
cally significant differences between HTS solutions or
treatment failure, sampling time frames to determine ICP
mannitol for the outcomes of mortality and neurological
change and treatment failure, formulation and osmolar
function, but the few small trials included were underpow-
ered to detect clinically significant difference in these out- loads of solutions, and diverse study populations. The total-
comes. In the studies that did report neurological outcomes, ity of the data suggests a trend toward greater ICP-lowering
it is possible that differences in initial GCS, a marker of effects from HTS, but the clinical significance of this is
baseline severity of illness, may have accounted for some of unknown. We did, however, detect a significant decrease in
the numerical disparities. For example, Cottenceau et al14 the risk of ICP treatment failure favoring HTS solutions (RR
reported numerically higher rates of mortality and severe = 0.39; 0.18-0.81), which is concordant with previous find-
neurological disability for the patients receiving HTS com- ings. Although the difference is statistically significant, there
pared with mannitol; however, this difference could be par- is variability in the precision, and it likely suffers from a
tially accounted for by differences in the degree of severity small study effect and low number of events. Without a con-
of injury at the time of admission in favor of mannitol sistent and meaningful definition of ICP treatment failure, the
(median baseline GCSs were 5 and 7 for HTS and mannitol, clinical relevance and importance of this finding is unclear.
respectively). Similarly, Ichai et al16 reported worse neuro- In the absence of overwhelming evidence favoring one
logical outcomes in a mannitol group with an initial GCS of hyperosmolar agent, choice of therapy could be decided
4 compared with a HTS group with an initial GCS of 6. based on safety. However, none of the included trials ade-
Furthermore, if the effects of HTS solutions and mannitol quately assessed or reported serious adverse events. There is
on the reduction of ICP truly are similar, it is arguably concern that mannitol may precipitate acute renal failure if
unlikely that significant differences in clinical outcomes serum osmolarity exceeds 320 mOsm/L, through intrarenal
such as functional neurological recovery and mortality vasoconstriction combined with intravascular volume
would be observed. depletion.22 In addition, the diuretic effect can lead to elec-
For the outcome of magnitude of ICP change from base- trolyte disturbances, hypovolemia, and hypotension which
line, defined in many of the studies as a mean or median may be undesirable in trauma patients. Neither of these
decrease in ICP following a prespecified time period, there adverse events were identified or reported in the included
appeared to be no consistent or significant advantage of one trials. HTS solutions may be preferred in situations requiring
hyperosmolar therapy. Four of 7 trials found a statistically rapid cardiovascular resuscitation of associated hemorrhagic

Downloaded from at Middle East Technical Univ on February 2, 2016

8 Annals of Pharmacotherapy 

shock with small-volume infusion, given the volume expan- 1.0-1.33). Similarly, a meta-analysis by Mortazavi et al26
sion and lack of a diuretic effect. Some concerns and poten- that included 8 prospective RCTs comparing HTS and man-
tial complications with HTS include volume overload, nitol for the management of elevated ICP in a mixed trauma
severe hypernatremia (>160 mEq/L), acute kidney injury, and nontrauma population found lower rates of treatment
and the osmotic demyelination syndrome. Similarly, these failure with HTS (OR = 0.36; 0.19-0.68; P =0.002). A 2014
were not reported in any of the trials. review by Rickard et al27 compared mannitol and HTS in
It is difficult to generalize the available evidence of patients with TBI and evidence of elevated ICP. The pri-
hyperosmolar therapy for elevated ICP in TBI to settings mary outcome was mean reduction in ICP. Pooled data from
beyond critical care settings. All the trials included patients 6 RCTs found no statistically significant difference in the
with severe brain injury who were intubated and mechani- mean ICP reduction but reported a trend in favor of HTS
cally ventilated in an ICU and had access to continuous ICP (−1.39 mm Hg, −3.35 to 0.74 mm Hg). There was also no
monitoring. In the ED setting, insertion of an intracranial significant difference in ICP treatment success (RR = 1.05;
device is almost never performed, and patients are treated 0.94-1.19). The most recent review by Li et al,28 including
based on the clinical suspicion of elevated ICP or other sur- similar studies, concluded that HTS is more effective than
rogate markers of elevated ICP (eg, CT scan findings, optic mannitol for reducing ICP in cases of TBI, with a pooled
nerve sheath diameter). Another area of uncertainty is the difference in means of −1.69 mm Hg (95% CI = −2.95 to
optimal HTS concentration. Various concentrations of HTS −0.44 mm Hg). We acknowledge the similarities in design
solutions were administered in the trials, the most common and included studies between this review and previous
being 7.5% hypertonic saline. This concentration along reviews; however, we believe that this review differs in
with 7.2% HTS/HES 200/0.5 solution, 7.5% saline/6% dex- important respects. First, mortality was chosen as the pri-
tran-70 solution, and hypertonic sodium lactate may not be mary outcome, and data on neurological outcomes were
currently available in many EDs for various reasons. This sought rather than the surrogate outcome of ICP reduction
includes a lack of awareness of the data supporting their use alone. Unfortunately, these outcomes were not reported in
and potential safety concerns. For example, the Institute for many of the studies, and given the small number of patients
Safe Medication Practices classifies HTS solutions greater in which these outcomes were reported, we are unable to
than 0.9% to be high-risk medications and recommend lim- draw any confident conclusions. Second, we attempted to
iting their availability in areas outside a pharmacy.23,24 summarize data on serious adverse events to assess the
Given the data that support the efficacy of HTS solutions in safety of hyperosmolar therapies. Serious adverse events
documented severe TBI with elevated ICP, this should not were not reported for any of the studies. Even data on any
preclude their use in the ED setting for this purpose. Both adverse events were sparse and underreported, which high-
HTS solutions and mannitol are effective in reducing ICP in lights a gap in the knowledge base regarding the safety and
the setting of TBI and appear to have an acceptable risk efficacy of these agents. If additional studies are performed
profile; therefore, the choice of first-line agent may come in the future, these outcomes will need to be well docu-
down to practical considerations such as local availability mented and reported to inform clinicians of the potential
of solutions, ease of administration, and physician familiar- impact of administering these interventions. Finally, we
ity. Because administration of hyperosmolar agents in the believe that this review adds a unique contribution to the
ED can have downstream implications, the choice and indi- current literature on this topic. Given the clinical equipoise
cations for the use of such agents should not be made in a over which is the superior hyperosmolar therapy, we have
vacuum. Rather, site-specific multidisciplinary guidelines attempted to provide clinicians with practical consider-
or tiered protocols, developed by emergency medicine, neu- ations and implications when choosing an agent, including
rosurgery, critical care, and pharmacy services, should be preferred clinical settings, potential safety concerns, and
developed and used where possible. Standardization of considerations regarding administration and availability.
goal-directed care in severe TBI will also facilitate future The main limitation of this systematic review is the reli-
outcome-based analyses. In smaller centers without such ance on a small number of randomized trials with method-
resources, consultation with a receiving neurosurgeon may ological limitations. Overall, the existing data provide no
help identify trigger points, therapeutic thresholds, and consistent and clinically important differences in mortality,
treatment goals prior to and during transfer. neurological function, or ICP control between HTS and
Previous systematic reviews and meta-analyses have mannitol for elevated ICP in acute, severe TBI. All trials
attempted to delineate the comparative benefits of HTS were unblinded and failed to report adverse events. In addi-
solutions and mannitol.25-28 A meta-analysis of 5 RCTs tion, many patients received multiple doses of the study
comprising 112 patients with 184 episodes of elevated ICP drug or more than 1 study drug in a crossover fashion, fur-
by Kamel et al25 suggested that equiosmolar HTS solutions ther complicating interpretation of the results. In addition,
(3% to 7.5%) may be more effective than mannitol in con- most studies did not report on whether cointerventions were
trolling acute elevations in ICP (ICP control: RR = 1.16; similar between groups. The lack of obvious superiority of

Downloaded from at Middle East Technical Univ on February 2, 2016

Burgess et al 9

one hyperosmolar agent is congruent with the apparent clin- 8. Maas AIR, Dearden M, Teasdale GM, et al. EBIC-guidelines
ical equipoise that currently exists regarding which agent is for management of severe head injury in adults. Acta
preferred in the setting of acute ICP elevation. Large, pro- Neurochir. 1997;139:286-294.
spective, methodologically strong randomized trials would 9. Wenham T, Hormis A, Andrzejowski J. Hypertonic saline
after traumatic brain injury in UK neuro-critical care prac-
be required to determine the effects of hyperosmolar ther-
tice. Anaesthesia. 2008;63:558-559. doi:10.1002/14651858.
apy on survival and neurological recovery, rather than effect
on ICP alone. Another unanswered question to address 10. Berger Pelletier E, Émond M, Lauzier F, Savard M, Turgeon
would be the optimal concentration of HTS that should be AF. Hyperosmolar therapy in severe traumatic brain injury:
administered. Future research should evaluate these thera- a survey of emergency physicians from a large Canadian
pies in the ED or prehospital settings because this is the province. PLoS One. 2014;9:e95778. doi:10.1371/journal.
initial contact point for virtually all patients presenting with pone.0095778.
acute TBI with evidence of elevated ICP. Currently, for the 11. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred

management of elevated ICP following TBI, emergency reporting items for systematic reviews and meta-analyses:
physicians must extrapolate from the critical care settings, the PRISMA statement. PLoS Med. 2009;6:e1000097.
and practical considerations such as availability of solu- doi:10.1371/journal.pmed.1000097.
12. Higgins JPT, Thompson SG, Deeks JJ, Altman DG. Measuring
tions, ease of administration, and physician familiarity
inconsistency in meta-analyses. BMJ. 2003;327:557-560.
should be considered in a multidisciplinary fashion when
13. Higgins JPT, Altman DG, Gøtzsche PC, et al. The Cochrane
site-specific, goal-directed guidelines or protocols are Collaboration’s tool for assessing risk of bias in randomised
developed. trials. BMJ. 2011;343:1-9. doi:10.1136/bmj.d5928.
14. Cottenceau V, Masson F, Mahamid E, et al. Comparison of
Declaration of Conflicting Interests effects of equiosmolar doses of mannitol and hypertonic saline
The authors declared no potential conflicts of interest with respect on cerebral blood flow and metabolism in traumatic brain
to the research, authorship, and/or publication of this article. injury. J Neurotrauma. 2011;28:2003-2012. doi:10.1089/
15. Vialet R, Albanèse J, Thomachot L, et al. Isovolume hyper-
Funding tonic solutes (sodium chloride or mannitol) in the treatment
The authors disclosed receipt of the following financial support of refractory posttraumatic intracranial hypertension: 2 mL/
for the research, authorship, and/or publication of this article: This kg 7.5% saline is more effective than 2 mL/kg 20% man-
manuscript was prepared without any financial support. nitol. Crit Care Med. 2003;31:1683-1687. doi:10.1097/01.
References 16. Ichai C, Armando G, Orban J-C, et al. Sodium lactate ver-
sus mannitol in the treatment of intracranial hypertensive
1. Treggiari MM, Schutz N, Yanez ND, Romand JA. Role of episodes in severe traumatic brain-injured patients. Intensive
intracranial pressure values and patterns in predicting out- Care Med. 2009;35:471-479. doi:10.1007/s00134-008-
come in traumatic brain injury: a systematic review. Neurocrit 1283-5.
Care. 2007;6:104-112. doi:10.1007/s12028-007-0012-1. 17. Sakellaridis N, Pavlou E, Karatzas S, et al. Comparison of
2. Rutland-Brown W, Langlois JA, Thomas KE, Xi Y. Incidence mannitol and hypertonic saline in the treatment of severe
of traumatic brain injury in the United States, 2003. J Head brain injuries. J Neurosurg. 2011;114:545-548. doi:10.3171
Trauma RehabilTrauma. 2006;21:544-548. /2010.5.JNS091685.
3. Zaloshnja E, Miller T, Langlois JA, Selassie A. Prevalence of 18. Harutjunyan L, Holz C, Rieger A, Menzel M, Grond S,

long-term disability from traumatic brain injury in the civil- Soukup J. Efficiency of 7.2% hypertonic saline hydroxy-
ian population of the United States, 2005. J Head Trauma ethyl starch 200/0.5 versus mannitol 15% in the treatment
Rehabil. 2008;23:394-400. of increased intracranial pressure in neurosurgical patients:
4. Centers for Disease Control and Prevention. Severe trau- a randomized clinical trial [ISRCTN62699180]. Crit Care.
matic brain injury. 2005;9:R530-R540. doi:10.1186/cc3767.
severe.html. Accessed January 11, 2016. 19. Battison C, Andrews PJD, Graham C, Petty T. Randomized,
5. Mokri B. The Monro-Kellie hypothesis: applications in CSF controlled trial on the effect of a 20% mannitol solution and
volume depletion. Neurology. 2001;56:1746-1748. a 7.5% saline/6% dextran solution on increased intracranial
6. Mojtahedzadeh M, Ahmadi A, Mahmoodpoor A, et al. pressure after brain injury. Crit Care Med. 2005;33:196-202.
Hypertonic saline reduces the oxidative stress responses doi:10.1097/01.CCM.0000150269.65485.A6.
in traumatic brain injury patients . J Res Med Sci. 2014;19: 20. Francony G, Fauvage B, Falcon D, et al. Equimolar doses of
867-874. mannitol and hypertonic saline in the treatment of increased
7. Bratton SL, Chestnut RM, Ghajar J, et  al. Guidelines intracranial pressure. Crit Care Med. 2008;36:795-800.
for the management of severe traumatic brain injury: II. doi:10.1097/CCM.0B013E3181643B41.
Hyperosmolar therapy. J Neurotrauma. 2007;24(suppl 1):S14- 21. Higgins J, Green S,eds. Addressing reporting biases. In: Cochrane
S20. doi:10.1089/neu.2007.9994. Handbook for Systematic Reviews of Intervention. Version 5.1.0

Downloaded from at Middle East Technical Univ on February 2, 2016

10 Annals of Pharmacotherapy 

[updated March 2011]. The Cochrane Collaboration; 2011:chap 26. Mortazavi MM, Romeo AK, Deep A, et al. Hypertonic saline
10. for treating raised intracranial pressure: literature review with
22. Visweswaran P, Massin E, Dubose T. Mannitol-induced acute meta-analysis. J Neurosurg. 2012;116:210-221. doi:10.3171/
renal failure. J Am Soc Nephrol. 1997;8:1028-1033. 2011.7.JNS102142.
23. ISMP Canada. ISMP Canada Safety Bulletin. http://ismp- 27. Rickard A, Smith J, Newell P, Bailey A, Kehoe A, Mann C. Salt or sugar for your injured brain? A meta-analysis of
11NaCl.pdf. Accessed January 11, 2016. randomised controlled trials of mannitol versus hypertonic
24. World Health Organization. Control of concentrated electro- sodium solutions to manage raised intracranial pressure in
lyte solutions. traumatic brain injury. Emerg Med J. 2014;31:679-683.
patientsafety/PS-Solution5.pdf. Accessed January 11, 2016. doi:10.1136/emermed-2013-202679.
25. Kamel H, Navi BB, Nakagawa K, Hemphill JC, Ko NU. 28. Li M, Chen T, Chen S, Cai J, Hu Y. Comparison of equi-
Hypertonic saline versus mannitol for the treatment of ele- molar doses of mannitol and hypertonic saline for the treat-
vated intracranial pressure: a meta-analysis of randomized ment of elevated intracranial pressure after traumatic brain
clinical trials. Crit Care Med. 2011;39:554-559. doi:10.1097/ injury. Medicine (Baltimore). 2015;94:e736. doi:10.1097/
CCM.0b013e318206b9be. MD.0000000000000736.

Downloaded from at Middle East Technical Univ on February 2, 2016