This review will be conducted in accordance with The Cochrane Collaboration [
15] principles for Systematic Reviews and reported following the PRISMA guidelines [
16]. This protocol was drafted in accordance with PRISMA-P guidelines (see checklist in Additional file
1) and has been registered with the PROSPERO International Prospective Register of Systematic Reviews (#CRD42016042960).
Search strategy and data sources
Our search strategy will be conducted using Epub Ahead of Print, In-Process & Other Non-Indexed Citations, Ovid MEDLINE(R) Daily and Ovid MEDLINE(R), EMBASE Classic + Embase, Web of Science BIOSIS Previews® and EBM Reviews (including Cochrane Central databases) from inception to the moment of review. EMBASE also includes the abstract publications from major international conferences including the International Stroke Conference, Neurocritical Care Society Meeting, Society of Critical Care Medicine, and the International Symposium on Intensive Care and Emergency Medicine. A comprehensive search strategy will be constructed and implemented by a health information specialist with systematic review experience, in collaboration with the research team. MeSH terms will be used to capture each of the principal elements of the research question. We will restrict our strategy to focus on population (acute brain injury) and intervention/exposure (hypotonic crystalloid solutions and hypertonic solution alternatives—see Eligibility criteria) and will not be limited by outcome studied to increase our yield. A sample search strategy is presented in Additional file
2. Upon completion, identified citations will be exported to a cloud-based citation manager (DistillerSR v2: Systematic Review and Literature Review Software) for study selection. Manual review of the reference lists of all included studies and previous systematic reviews will be conducted. A final gray literature search will be conducted using “Google Scholar” as well as a review of the trial register (clinicaltrials.gov, WHO ICTRP) for any ongoing and unpublished studies. Duplicate citations will be removed. The search strategies will be kept up to date to the time of the end of the review.
Study eligibility
For the purposes of this review, we will refer to normal saline as an isotonic solution and those with lower sodium concentrations (e.g. Ringer’s Lactate, Plasma Lyte®) will be referred to as “relatively hypotonic.” Those crystalloid solutions with higher sodium concentrations (e.g., 3% saline) will be referred to as hypertonic. We will include pre-clinical studies that compare a relatively hypotonic crystalloid resuscitation fluid (Ringers Lactate, Hartmann’s or Plasma Lyte®) to an isotonic (normal saline (0.9%)) or hypertonic (i.e., hypertonic saline (3–23.4%)) crystalloid resuscitation fluid. No overtly hypotonic solution (e.g., 5% dextrose in water, 5% in 0.45% saline) or colloid solutions will be included given their very different physiological properties that would have them behave significantly differently as compared to crystalloid solutions of interest in this review.
For clinical studies, we will include observational studies that have a control group for comparison, whether prospectively or retrospectively conducted, and intervention studies (e.g., randomized controlled trials). An iterative process for study selection will be followed using the criteria set out in Table
1.
Table 1
Eligibility criteria for the systematic review (expanded detail provided in Additional file
3)
Population | We will include all studies examining acute brain injury (human and in vivo animal brain injury models) with treatment initiation within the first 7 days post-injury. We are targeting acute primary neurological diagnoses such as traumatic brain injury, stroke, hemorrhage, or post-neurosurgical care. |
Intervention | The intervention of interest is the intravenous use of relatively hypotonic crystalloid resuscitation fluids. These include Ringer’s lactate, Hartmann’s or Plasma Lyte® fluids. These can be administered as bolus or maintenance infusions. There will be no limits on dose or frequency of administration. |
Comparator(s) | The main comparator of interest is an isotonic crystalloid resuscitation fluid, specifically normal saline (0.9%), which will serve as the prototypical control. Other crystalloid resuscitation fluids (e.g., hypertonic saline (3–23.4%)) will also be included as an additional comparator of interest. These can be administered as bolus or maintenance infusions. There will be no limits on dose or frequency of administration. |
Outcome(s) | The primary outcomes of interest are: • Intracranial pressure (ICP; mmHg) and cerebral perfusion pressure (CPP; mmHg) Secondary outcomes include: • Cerebral edema (mL H 2O/g dry), serum and/or brain electrolyte concentrations (Na+, Cl−; mmol/L) and osmolarity (mOsmol/L [ 29]), clinical outcomes as assessed by any of modified Rankin Scale (mRS), Glasgow Outcome Scale (GOS), extended Glasgow Outcome Scale (eGOS), mortality; and Tertiary outcomes include: • Any adverse events as defined by the study authors |
Study design | We will include all completed publications reporting the intravenous use of crystalloid fluids in acute brain injury. • Clinical: randomized controlled trials, quasi-randomized trials, and retrospective and prospective studies that include a control group for comparison • Pre-clinical: randomized laboratory studies There will be no date or language restrictions applied. In-progress studies and letters to the editor identified will be included in a qualitative analysis. Case reports, case series, editorial reviews, and guidelines will be excluded. |
Study selection process
All records will first be screened by title and abstract. All citations clearly not relevant to the review (e.g., wrong population and narrative review) will be excluded. This process will be performed in duplicate by two independent reviewers. Any citation in which an abstract is not available and where suitability for inclusion is questioned will proceed to the next stage. All citations not excluded in the first screen will have full articles retrieved for a second review, in duplicate by independent reviewers, and the selection criteria applied. Any differences in classification between the two independent reviewers will be reviewed and consensus decision made. A third independent senior reviewer will be used in any instance in which consensus is not reached.
A data extraction form will be prepared a priori in MS Excel (Microsoft Corporation, Seattle, Washington) and piloted prior to duplicate extraction by two independent reviewers. The data extraction form will be designed to capture information regarding study characteristics, design, and methods (i.e., title, authors, journal/source, year of publication, country, type of study, study period, total number of subjects, case ascertainment and/or inclusion/exclusion criteria, randomization, allocation concealment and blinding methods (where applicable)); study population characteristics (i.e., clinical studies: age, sex, primary neurological diagnosis, injury severity/characteristics, comorbidities; pre-clinical studies: species, species detail, age, sex, weight, model type and experimental conditions, injury severity/characteristics); interventions and co-interventions [crystalloid resuscitation fluid type, dose, frequency, adjunctive fluid administration, use of a management protocol, use of co-interventions for intracranial pressure management strategies (including hyperosmolar therapy, paralysis, hyperventilation, barbiturate coma, decompression)]; and outcome (our pre-defined primary, secondary and tertiary outcomes).
Outcome data will be extracted based on a priori specified time points and dosage format. Given fluid resuscitation is the intervention of interest, data will be extracted for early (0–6 h; at hourly time points) and delayed resuscitation (up to 7 days post-injury; at 12, 24, 48, etc. hours) in order to evaluate both the immediate and delayed effects of treatment. Data will then be pooled by dosage format [means of administration (bolus vs maintenance infusion) and total dose] for the purposes of analysis.
Disagreements in data extraction will be resolved by consensus or by a third reviewer with methodologic and clinical expertise as required. If there are missing data, the corresponding authors of the studies will be contacted and if the data is not located, the remaining available data will be analyzed. The possible impact of the missing data will then be discussed as a limitation.
Risk of bias
Risk of bias will be assessed using the Newcastle-Ottawa Scale [
17] for observational studies, the Cochrane Collaboration tool for assessing the risk of bias in randomized controlled trials (RCTs) [
18] and SYRCLE’s risk of bias tool for animal studies [
19]. Bias risk assessment will be completed in a similar fashion as the study selection process: in duplicate by two independent assessors. Cases of discordance not resolved by consensus will be reviewed by a third senior assessor. Risk of bias assessment of all included studies will be summarized and presented in table format. Low risk of bias will be defined as those studies with score of ≥ 7 using the Newcastle-Ottawa Scale, those deemed low risk across all domains of the Cochrane Collaboration’s tool for assessing risk of bias, or those receiving an overall “yes” judgment across all signaling questions used in SYRCLE’s risk of bias tool. The authors recognize that no formal cut-offs exist to define low or high risk of bias with the Newcastle-Ottawa Scale; however, we deem that in order for low risk of bias to exist in an observational study, there must be excellent reporting, high internal and external validity with little risk of confounding such that high scores in each of these domains are necessary to meet low-risk criteria.