Prehospital fluid administration in patients with severe traumatic brain injury: A systematic review and meta-analysis

Background: Prehospital management of severe traumatic brain injury (TBI) focuses on preventing secondary brain injury. Therefore, hypotension should be prevented, or if present, should be promptly treated in order to maintain optimal cerebral perfusion pressure. Fluid resuscitation is a traditional mainstay in the prehospital treatment of hypotension, however, the choice of fluid type that is to be administered in the prehospital setting is the subject of an on-going debate. This systematic review and metaanalysis was therefore performed to assess the effect of different fluid types on outcome in patients with severe TBI. Methods: PubMed, Embase and Web of Science were searched for articles up to March 2020. Studies comparing two or more prehospital administered fluid types with suspected or confirmed severe TBI were deemed eligible for inclusion. Studied outcomes were mortality and (extended) Glasgow Outcome Scale (GOS). The meta-analysis tested for differences in survival between hypertonic saline (HTS) and normotonic crystalloids (i.e. normal saline or Lactated Ringer’s) and between hypertonic saline with dextran (HSD) and normotonic crystalloids. The systematic review is registered in the PROSPERO register with number CRD42020140423. Results: This literature search yielded a total of 519 articles, of which 12 were included in the systematic review and 6 were included in the meta-analysis. Eleven studies found no statistically significant difference in survival between patients treated with different fluid types (e.g. normal saline and hypertonic saline). All studies assessing neurological outcome, measured through (extended) GOS, found no statistically significant difference between different fluid types. Meta-analysis showed no better survival for patients treated with HSD, when compared to normotonic crystalloids (overall RR 0.99, 95% CI 0.93– 1.06). Moreover, HTS compared to normotonic crystalloids does not result in a better survival (overall RR 1.04, 95% CI 0.97–1.12). Conclusions: This systematic review and meta-analysis did not demonstrate a survival or neurological benefit for one specific fluid type administered in the prehospital setting. © 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license. ( http://creativecommons.org/licenses/by/4.0/ )


Introduction
Severe traumatic brain injury (TBI) is a major cause of death among younger individuals [1] and a leading cause of disability [2][3][4][5][6][7][8] . It is therefore apparent that TBI has a substantial impact on public health, emphasizing the need for optimal treatment strategies. The prehospital phase is of great significance in the management of TBI, as this is the earliest opportunity for healthcare hypotension in the prehospital setting. However, the choice of fluids that is to be administered is the subject of an on-going debate. While isotonic crystalloids are conventionally used, a variety of fluids are available to treat patients with TBI in the prehospital setting. Prehospital guidelines state that patients in need of fluid resuscitation should be treated with isotonic fluids and that hypertonic fluids are a "treatment option" for traumatic brain injury [16] . Crystalloids, such as normal saline or Lactated Ringer's solution, are salt solutions containing small molecules that can freely move between intravascular and interstitial space [17] . Hypertonic saline causes fluid to shift from the interstitial space to the intravascular space through osmosis, which may result in a decrease of cerebral edema. Colloids, such as dextran, hydroxyethyl starch (HES) or albumin, consist of larger molecules and are assumed to stay in the intravascular compartment for a longer time and thus achieve prolonged intravascular volume replacement compared to crystalloids [18] . However, there is no evidence of a beneficial effect of colloids in the treatment of traumatic brain injury and their use is mostly limited to the hospital setting [ 19 , 20 ].

List of abbreviations
Clearly, uncertainty regarding the type of fluids used in the prehospital setting remains. Therefore, we performed a systematic review and meta-analysis on available literature to assess the effect of different prehospital fluid therapies on outcome in patients with severe traumatic brain injury [ 21 , 22 ].

Protocol and registration
This study was performed in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guideline [ 23 , 24 ]. The search strategy, study selection, bias assessment and data extraction were defined a priori and the protocol was registered in the PROSPERO register (International Prospective Register of Systematic Reviews) with number CRD42020140423.

Eligibility criteria
Articles deemed eligible for inclusion were randomized controlled trials or observational studies comparing two or more prehospital administered fluid types in patients with suspected or confirmed severe TBI. In line with previous literature, severe TBI was defined as either a prehospital or admission Glasgow Coma Scale (GCS) < 9 combined with a trauma, or a Head Abbreviated Injury Score (H-AIS) ≥3 [25] . Studies reporting the outcomes mortality, Glasgow Outcome Scale (GOS) and Extended Glasgow Outcome Scale (GOSe) were eligible for inclusion. Animal studies were excluded. Studies assessing other patient populations were considered appropriate when it was possible to extract the data of the TBI population.
Articles eligible for meta-analysis were randomized controlled trials (irrespective of the score on the Cochrane Collaboration's tool) or observational studies with a Newcastle Ottawa Score of ≥6 stars and 2 awarded stars for comparability.

Information sources and search strategy
PubMed, Embase and Web of Science were searched without restrictions. This search was last updated in March 2020. The following search was used in PubMed: (("saline solution

Study selection
Publications were evaluated for eligibility by screening the abstracts of the identified studies, which was done independently by two investigators (SMB, SFB). When relevance could not be determined based on title or abstract, the full text article was retrieved. Disagreements on eligibility were solved by discussion (SMB, SFB) or by contacting a third reviewer (PS).

Data extraction
One of the authors (SFB) extracted the data using a standardized data collection sheet, which was checked for accuracy by a second author (SMB). The following data was extracted from the included studies: [1] study characteristics: study design, sample size, inclusion and exclusion criteria, time period, location of study; [2] patient characteristics: age, gender, injury severity; [3] type of fluids administered; [4] outcome variables.

Assessment of study quality and risk of bias
Two of the authors (SMB and SFB) independently assessed the quality of the included studies and in case of a disagreement, a third reviewer (PS) was consulted. The Newcastle-Ottawa scale was used to assess the risk of bias of cohort studies [26] . This scale assigns a total of nine stars per study for selection of participants, comparability of cohorts and assessment of outcome. For randomized controlled trials (RCT), the Cochrane Collaboration's tool for assessing risk of bias was used [27] .

Data synthesis and statistical analysis
A meta-analysis was performed to test for differences in survival at hospital discharge (or at 28 days when survival at hospital discharge was not reported) between patients who received hypertonic saline (HTS) versus normotonic crystalloids (i.e. normal saline or Lactated Ringer's), and for differences in survival between  hypertonic saline with dextran (HSD) and normotonic crystalloids. There were no eligible studies on HES or mannitol compared to other fluids. A random effects meta-analysis was performed with STATA 13.0 (StataCorp, Texas). As a sensitivity analysis, to gage the potential influence of lower-quality studies on the results, we re-performed the meta-analysis while excluding RCTs with a high risk of bias in any domain as well as observational studies. Testing for publication bias with funnel plots and Egger's test was planned, but was not possible due to the limited number of studies.

Study selection
In total, 519 articles were obtained through the database search. Additionally, 11 articles were found through reference lists. After removing duplicates, a total of 389 articles were screened for eligibility based on title and abstract. After excluding 375 articles (offtopic, review articles, no prehospital study, no TBI population), 14 full text articles were assessed for eligibility. Full text screening resulted in the exclusion of 2 articles. The final systematic review therefore yielded a total of 12 articles [28][29][30][31][32][33][34][35][36][37][38][39] . A total of 6 studies were included in the meta-analysis. Fig. 1 presents the PRISMA flow diagram.

Study characteristics
The 12 included studies reported data from 3253 patients. Five studies were performed in the United States, four in Canada, one in the United States and Canada, one in Australia and one in Austria. Ten studies were randomized controlled trials, one study was a retrospective cohort and one study was a prospective observational trial ( Table 1 ). Eleven studies compared hypertonic saline with or without dextran to a normotonic crystalloid, and one study compared the two crystalloids Lactated Ringer's and normal saline to each other. No studies compared other colloids (e.g., HES) or mannitol to other types of fluids.

Patient characteristics
The patients included had a median or mean age of around 36 to 46 years and were mostly male (60-86.4%). In accordance with our inclusion criteria, patients were severely injured (head AIS, ISS) ( Table 2 ). One study explicitly reported their study population as being isolated TBI [36] , meaning that patients were excluded if they suffered from life threatening injury in organs other than the brain.

Results of individual studies
Our main outcome of interest was survival. Seven studies reported survival at hospital discharge. Four studies reported survival at 28 or 30 days and two studies reported survival at 3 or 6 months. Eleven studies assessing survival found no statistically significant difference in survival between patients treated with different fluid types, such as normal saline and hypertonic saline with or without dextran. One study compared the two crystalloids Lactated Ringer's and normal saline to each other. In that study by Rowell et al. (2016), treatment with Lactated Ringer's was associated with higher 30-day mortality compared with normal saline [39] .
Five studies assessed the Extended Glasgow Outcome Score (GOSe). All three studies assessing GOSe as a primary outcome measure did not find a statistically significant difference in patients treated with different fluid types. A total of three studies assessed the Glasgow Outcome Score (GOS), all of which found no statistically significant difference between different fluid types. Table 4 presents the outcome measures of the individual studies.

Meta-analysis
After excluding lower quality studies (observational studies with a Newcastle Ottawa Score of < 6 stars and < 2 awarded stars for comparability), 2 meta-analyses were performed. The metaanalysis of 5 studies comparing treatment with hypertonic saline with dextran (HSD) to crystalloid fluids (normal saline in all studies) [34][35][36][37][38] does not show a better survival rate for one specific fluid type (overall RR 0.99, 95% CI 0.93-1.06) ( Fig. 2 ). The metaanalysis of 3 studies comparing hypertonic saline (HTS) with crystalloids (normal saline in two studies and Lactated Ringer's in one study) [ 31 , 35 , 38 ] does also not show a better survival rate for either of the groups (overall RR 1.04, 95% CI 0.97-1.12) ( Fig. 3 ). The sensitivity analysis excluding two randomized controlled trials with high risk of bias provided virtually identical results (data not shown).

Summary of evidence
This systematic review and meta-analysis was performed to assess the effect of different prehospital fluid therapies on outcome in patients with severe traumatic brain injury. The included studies did not demonstrate a survival or neurological benefit for one certain fluid type.

Strengths and limitations
This systematic review contains an extensive search strategy in three databases and was performed in accordance with prespecified guidelines and recommendations of PRISMA [ 23 , 24 ]. Search strategy, selection of studies, data extraction and quality assessment were double-checked for accuracy.
The quality assessment of the included studies shows varying quality. Only one of the ten RCTs shows low risk of bias in all domains in the Cochrane Risk of Bias Tool. One of the two cohort studies shows low risk of bias through the Newcastle Ottawa Scale. To present a complete overview of all the articles addressing this subject, we included all articles that met our inclusion criteria. However, to limit bias in the quantitative analysis, we performed a sensitivity analysis which excluded lower quality studies.
Due to expected clinical heterogeneity between studies, we had planned a random-effects meta-analysis to account for this heterogeneity. This analysis was performed on a limited number of studies, which can be problematic due to the limited precision in the estimate of the between-study variance, and it has been proposed that a fixed-effect analysis should then be performed instead [40] . However, as the estimated between study variance ( τ 2 ) was 0 in our meta-analysis, the random-effects and fixed-effect metaanalysis give virtually identical results.
Our main outcome of interest was survival, because of its high clinical relevance. Even though most of the studies reported survival, the timing of measurement of survival varied between studies. Seven of the included studies reported survival at hospital discharge. Other studies reported mortality at 30 days, at hospital arrival and at 3 or 6 months. We chose mortality at hospital discharge for meta-analysis (or at 28 days when survival at hospital discharge was not reported) because this was the time point that was being assessed most often. Other outcomes, such as (Extended) Glasgow Outcome Scale, are of great relevance as well. However, we could not perform a meta-analysis of these results, because these outcomes were not consistently reported.
In our systematic review we were also interested in the use of mannitol in the prehospital setting. Only one study assessed the prehospital use of mannitol for TBI [41] . This very small study on 44 patients found no evidence for a difference in mortality between mannitol and normal saline. However, we did not include this study in our systematic review, as this study included patients with a GCS < 12, as opposed to GCS < 9.
We did not specifically review the quantity of the fluids as a detailed description of the type and quantity of "standard resuscitation fluids", which were administered in addition to the fluids studied, was often lacking.

Clinical implications
In the prehospital care of patients with severe traumatic brain injury, a single event of hypotension worsens outcome [15] . Therefore, hypotension should be prevented and optimal cerebral perfusion pressure should be maintained. Fluid resuscitation is regarded as an important therapy in the treatment of patients with TBI in the prehospital setting.
Various types of fluids are available for fluid resuscitation, with different pharmacological characteristics and with specific advantages and disadvantages. Traditionally, normotonic crystalloids are predominantly used in the prehospital setting. However, intravascular volume expansion is limited as only about 20% of the infused volume remains in the intravascular compartment [42] . Moreover, side effects such as edema and hyperchloremic acidosis (for normal saline) or electrolyte imbalances limit the usefulness for volume resuscitation. Hypertonic saline may be advantageous as it causes fluid to shift from the interstitial space to the intravas-cular space through osmosis, which may result in a decrease of cerebral edema. Moreover, hypertonic saline may have an antiinflammatory effect which reduces cerebral edema [43] . However, the hyperosmolarity may also lead to pulmonary edema and heart failure as a result of rapid volume expansion. Mannitol, an osmotic diuretic, is not useful for volume resuscitation but is yet regularly used in the prehospital setting to reduce intracranial pressure [44] . However, mannitol raises concerns with regard to its diuretic effect, which may cause hypotension and decrease cerebral perfusion [44][45][46] . In Emergency Departments and Intensive Care Units, mannitol, as well as hypertonic saline, are widely used to treat intracranial hypertension [45][46][47][48][49][50][51][52][53][54][55][56][57][58] . In the prehospital setting, however, no definitive guidelines regarding the treatment of suspected intracranial hypertension exist. Colloids, such as dextran, albumin or HES, should be effective for prehospital volume resuscitation, as the large molecules and oncotic pressure prevent rapid redistribution to the extravascular compartment, at least when the vasculature is intact. However, potential detrimental effects on coagulation, thrombocyte function or renal function raise concerns and may contribute to morbidity and mortality.
A previous review of Tan et al. published in 2011 found no evidence to support the use of hypertonic saline or colloid solutions over isotonic crystalloid solutions in patients with TBI in the prehospital setting [59] . Tan et al., however, did not perform a metaanalysis and a number of potentially relevant studies have been published in the meantime. We therefore re-evaluated the current evidence and also performed a quantitative data synthesis. Consistent with previous results from Tan et al., our meta-analysis showed no evidence for a higher survival in patients receiving hypertonic saline or hypertonic saline with dextran, when compared to normotonic crystalloids. Moreover, our qualitative analysis of functional outcome did not show evidence for a beneficial effect of any fluid. Hence, current data do not allow a recommendation on which type of fluid should be preferred for volume resuscitation in the prehospital setting for patients with severe TBI. Healthcare providers should thus individually balance the potential advantages and risks for each patient. More research is needed to further assess whether other fluid types, such as mannitol and HES, would confer benefit in this setting.

Conclusions
For the treatment of patients with severe traumatic brain injury in the prehospital setting, this systematic review did not demonstrate a survival or neurological benefit for one particular fluid type. All but one study compared a hypertonic with an isotonic solution.

Data availability
All relevant data are within the paper. Data in this systematic review are abstracted from previously published studies, which can be obtained from the respective publishers.

Declaration of Competing Interest
The authors report no conflict of interest.