Introduction

All critically ill patients receive intravenous (IV) fluids, which are given to maintain physiological homeostasis, or as a vehicle for drug administration, or as direct therapeutic administration to correct perceived haemodynamic instability [1]-[4]. In these situations, where there is a perceived reduction in venous return and cardiac output secondary to vasodilatation and/or hypovolaemia, using IV fluid to increase intravascular volume is believed to effectively compensate for these changes in vascular tone by increasing stroke volume in accordance with the Frank-Starling principle [5]-[10].

Several mechanisms for delivering IV fluids, both diagnostically and therapeutically under such circumstances, have been described. These include Weil’s central venous pressure (CVP)-guided fluid challenge technique [10]-[13], the timed and rapid infusion methods favoured by Shoemaker [7],[8],[14]-[16] and, more recently, techniques involving echocardiographic or ultrasonographic assessment of fluid responsiveness following low-volume IV infusion [17]. However, the current standard of care in the management of septic, hypotensive, tachycardic and/or oliguric patients is fluid bolus therapy (FBT), where IV fluid is rapidly administered in discrete boluses [18]-[21]. While the ideal fluid bolus would be a discrete volume of a specific fluid administered at a specified rate, accounting for individual patient features and with a defined aim (Figure 1) [11], there is no current agreement regarding exactly what defines a fluid bolus. Moreover, although strong overall consensus regarding the importance of FBT exists [18]-[20], there appears to be little randomized controlled information on the magnitude and duration of its physiological effects, or on the direct positive impact of FBT on patient outcome in sepsis as an independent intervention [22].

Figure 1
figure 1

Describing the concept of idealised fluid bolus therapy. (A) Diagram describing the key criteria defining the concept of a fluid bolus. (B) Diagram describing the idealised concept of fluid bolus therapy in critical care, including purpose, triggers, end-points and purported physiological effects of such resuscitation.

Full size image

In contrast, an expanding body of evidence suggests that FBT may contribute to a positive fluid balance, which, in turn, is independently associated with a variety of adverse outcomes in the critically ill [23]-[28]. Recent experimental evidence suggests rapid fluid infusion can also damage the endothelial glycocalyx [29],[30], a structure already at risk in patients with sepsis [31], leading to endothelial disruption and organ dysfunction [32],[33]. It appears that we need a better understanding of both the current evidence base for FBT and how best to apply it in the clinical setting [34],[35].

Accordingly, we systematically reviewed the contemporary literature to determine current practice and to identify the independent effects of FBT on both physiological and patient-centred outcomes in the management of severe sepsis and septic shock in critical care practice.

Methods

We interrogated the MEDLINE, CENTRAL and EMBASE electronic reference databases using a combination of search terms (Figure 2). The reference lists of retrieved articles were examined for additional studies of potential relevance. The search was carried out in December 2013. To achieve contemporary relevance results were arbitrarily limited to this decade (2010 to 2013) and to English language studies in humans. Paediatric studies were excluded. This search defined a set of records of studies of fluid administration or haemodynamic optimization in patients with severe sepsis or septic shock.

Figure 2
figure 2

Electronic search strategy. Diagrammatic representation of the search strategy combining terms representing fluid resuscitation, sepsis and clinical studies, along with predetermined limitations.

Full size image

The abstracts of these records were examined to identify those studies of potential relevance. These manuscripts were retrieved and examined manually in accordance with our inclusion criteria. The studies to be included in the review were checked to ensure they had not been retracted subsequent to their publication.

Study inclusion criteria

Population of included studies

We considered clinical studies of any type describing a population of patients suffering from severe sepsis or septic shock. We also included those studies of shock or circulatory failure where either the majority of patients, or a defined subgroup of patients, had severe sepsis or septic shock.

Intervention - fluid bolus administration

For the purposes of this study a fluid bolus was a defined volume of a defined fluid administered over a defined time period. We recognised that most studies do not describe FBT in ideal terms (Figure 1) and therefore studies describing at least two of the three criteria were included in the review.

Comparator - alternatives to fluid administration

Any studies comparing FBT with the initiation of vasoactive medication, the increase of such medication or observation as an alternative to the administration of FBT were included in the review.

Between groups analysis

Where studies included in the review assigned patients to multiple treatment arms, each treatment group was treated as an individual group.

Outcome - physiological effects of bolus administration

Subsets of studies were selected from those describing FBT. The first included those reporting changes in cardiac output, heart rate, mean arterial pressure, central venous pressure, venous oxygen saturation, blood lactate concentration, urine output or haemoglobin concentration following FBT; for the purposes of inclusion, studies could describe changes in any or all of the haemodynamic parameters listed, but the direction, magnitude and duration of the change had to be extractable from tables or figures contained in the paper. The second group included those reporting non-physiological, patient-centred outcomes. Our primary outcome of interest was mortality at all reported time points. Secondary outcomes of interest included duration of ICU and hospital stay, duration of mechanical ventilation, and need for continuous renal replacement therapy (CRRT). We did not contact authors for additional information or individual patient data.

Data collection

We collected data on study type, study setting and location, study population and the aims of the study. Due to our acceptance of multiple types of study, we chose not to adopt a methodological scoring system. We examined the definition of a fluid bolus in each study fulfilling our criteria and recorded the type and volume of fluid used, as well as the rate of administration. We identified the trigger and end-points for fluid bolus administration, the number of boluses administered and the use of red cell transfusions and vasoactive medication as part of the experimental protocol. We identified the demographic group in which subsequent observations were recorded. In those studies describing the physiological effects of bolus administration, we recorded the absolute change in cardiac output, heart rate, mean arterial pressure, venous oxygen saturation, blood lactate concentration, urine output and haemoglobin concentration. In those studies reporting patient-centred outcomes we recorded mortality at all reported time points, duration of ICU and hospital stay, duration of mechanical ventilation, and need for CRRT.

Statistical analysis

We expected grossly heterogeneous results across different study types and study protocols. A meta-analysis approach could not be applied. Results are therefore presented as crude medians with full ranges. These exclude alternative units of measure, which are reported separately - for example, the median may be given in millilitres, followed by individual reporting of ml/kg.

Results

Electronic search

Our search strategy identified 2,956 articles over the period 2010 to 2013. Of these, 2,875 were excluded as duplicates, irrelevant, paediatric research or having been published in a language other than English. Of the 81 potentially relevant publications identified, 33 met our inclusion criteria (Figure 3) [36]-[68]. In total, 17 of these described the physiological changes occurring following FBT [36],[39],[40],[45],[46],[48],[50],[53]-[55],[57],[59],[60],[62],[63],[65],[66] and seven studies described patient-orientated outcome measures [37],[42],[43],[49],[58],[59],[64].

Figure 3
figure 3

Study selection. Flow diagram of the study selection process and detailed description of study exclusions. FBT, fluid bolus therapy.

Full size image

Relevant contemporary studies

The study details, population, size and aims are presented in Table 1. We identified 22 prospective observational studies, four retrospective observational studies, two quasi-experimental studies, and five randomised controlled trials (RCTs). Of the five RCTs, none compared FBT with a control intervention; two actually reported the impact of blood volume analysis on protocolized resuscitation [64],[67]; two compared hypertonic versus isotonic fluids [51],[65]; and one actually compared two vasopressors and reported fluid data as an addendum [38]. Additional study data can be found in the electronic supplemental material (Additional file 1: Table S1).

Table 1 Study settings, size, population and aims
Full size table

Pre-fluid bolus therapy fluid administration

Fluid resuscitation prior to study recruitment and FBT was described in 10 studies. In the five studies describing finite volumes of resuscitation fluid, the median volume administered was 2,200 ml (range 1,000 to 5,060 ml) [38],[47],[51],[53],[58]. The five remaining studies reported weight-dependent volumes of between 20 and 30 ml/kg of resuscitation (Table 2) [41],[43],[49],[56],[57].

Table 2 Description of fluid boluses, triggers, physiological end-points and primary confounders
Full size table

Initiation and cessation of fluid bolus therapy

Across the 33 studies, 19 predetermined clinical or physiological features triggered FBT. In the remaining 14 studies, FBT was triggered by clinical judgment in eight, by ‘hypotension’ in two, simply by the diagnosis of severe sepsis or septic shock in two, and remained unspecified in two (Table 2).

In the majority of studies (18 of 33) FBT ceased at the end of the bolus in question; 10 studies used predetermined immediate changes in physiological variables as end-points; four studies did not define the physiological end-points of fluid resuscitation (Table 2).

Defining fluid bolus therapy

Overall, 41 forms of FBT were described, fully or in part, in 33 studies. They are presented in Table 2. In 20 studies, the fluid type was fixed; in 13 more than one fluid type was used. In six studies the fluid type was not identified beyond the generic ‘crystalloid or colloid’. The fluid most commonly used as a bolus was 0.9% saline (17 studies), followed by 6% hydroxyethyl starch (eight studies). On the other hand, 4% albumin was used in only four studies [38],[53],[59],[65], 4% gelatin in only three [38],[48],[66], physiological lactated solutions in only two [59],[61], and 20% albumin and blood products in only one [38].

The median amount of fluid administered as a finite volume was 500 ml (range 100 to 1,000 ml). However, 20 ml/kg and 7 ml/kg were individually reported as weight-dependent boluses. The median number of boluses (24 studies) was 1 (range 0.68 to 10). Rates of administration were defined for 31 of 41 boluses with a median rate of 30 minutes (range 10 to 60 minutes).

Haemodynamic changes after fluid bolus therapy

Comparing different interventions

No RCTs compared the haemodynamic changes induced by FBT with ‘observation’ or ‘vasopressor administration’ or ‘inotropic drug administration’ or ‘continuous low dose IV fluid infusion’ or any combination of the above. The only study comparing FBT with an alternative intervention was a single, non-randomized, prospective, observational study that compared acute circulatory failure patients treated with FBT (500 ml of saline) or with increased norepinephrine dose according to clinician preference [55]. The two groups had clearly different baseline characteristics and were not directly compared.

Temporal trends in physiological changes following fluid bolus therapy

The temporal change in physiological parameters following FBT is described in 31 different groups across 17 studies (Table 3).

Table 3 Physiological effects grouped by measurement time
Full size table

Immediately post-infusion

Ten studies reported the physiological state after bolus administration in 18 groups immediately post-administration. In the six studies describing changes in cardiac index immediately post-FBT, cardiac index increased by a median of 800 ml/minute/m2 (range 0 to 1,300 ml/minute/m2). The median reduction in heart rate at the end of a fluid bolus (eight studies) was 2 bpm (range 10 to 0 bpm reduction) and the median increase in mean arterial pressure (eight studies) was 7 mmHg (range 1 to 15.2 mmHg). The median increase in CVP across five studies was 3.2 mmHg (range 2.3 to 5.2 mmHg). Only a single study reported the effect on venous oxygen saturation, blood lactate concentration or haemoglobin concentration. No study reported the effect on urine output.

Thirty minutes post-administration

Five studies reported the physiological effects of FBT 30 minutes after administration. Cardiac index increased by a median of 300 ml/minute/m2 (range -400 to 600 ml/minute/m2) in three studies. The median reduction in heart rate (five studies) was 2 bpm (range 11 bpm reduction to 0.3 bpm increase) and the median increase in mean arterial pressure (five studies) was 7.5 mmHg (range 3 to 11 mmHg). The median increase in CVP across four studies was 3 mmHg (range 2 to 5.25 mmHg). There was a median increase in central venous saturation of 2% (range 4% reduction to 8% increase) across two studies. Changes in other indices are reported in Table 3.

Sixty minutes post-administration

Only three studies reported the physiological effects of FBT 60 minutes after administration (Figure 4) [36],[57],[65]. Cardiac index increased by a median of 300 ml/minute/m2 (range -300 to 400 ml/minute/m2) in two studies. The median reduction in heart rate 60 minutes after a fluid bolus (three studies) was 1 bpm (range 11 bpm reduction to 2 bpm increase) and the median increase in mean arterial pressure (three studies) was 3 mmHg (range 2 to 7 mmHg). The median increase in CVP across three studies was 2 mmHg (range 1 to 3 mmHg). There was a median increase in central venous saturation of 1% (range 0.4% to 2% increase) across two studies.

Figure 4
figure 4

Physiological effects of fluid bolus therapy over time. Multi-panel figure of the haemodynamic effects of fluid bolus therapy (FBT) as reported in studies with observation periods of 60 minutes or more. (A) Changes in heart rate over time. (B) Changes in cardiac index over time. (C) Changes in mean arterial pressure over time. (D) Changes in central venous pressure (CVP) over time. Each solid black line represents a patient group and the average physiological response to FBT over the observation period. Lines terminate when measurements were discontinued in the study from which the group was taken.

Full size image

Beyond 1 hour post-fluid bolus therapy

Only one study reported the effects of BFT at 120, 180 and 240 minutes after administration (Figure 4) [65].

Comparing responders and non-responders

Overall, 10 studies compared the physiological responses to FBT administration between groups defined by changes in a physiological variable. Patients were defined as either responders or non-responders depending on the response exhibited. Different variables are used in different studies: stroke volume index (five studies), cardiac index or output (three studies), increase in oxygen consumption (one study) or aortic blood flow rate (one study). All reported changes only within 30 minutes of FBT completion (Additional file 1: Table S2).

In the six studies describing changes in cardiac index, cardiac index increased by a median of 850 ml/minute/m2 (range 600 to 1,300 ml/minute/m2) in fluid responders compared with 200 ml/minute/m2 (range 0 to 1,000 ml/minute/m2) in non-responders. The median increase in mean arterial pressure (10 studies) in responders was 9.5 mmHg (range 7 to 15.2 mmHg) versus 4.8 mmHg (range 1 to 13 mmHg) in non-responders. Similarly, the median increase in central venous pressure (six studies) was 3 mmHg (range 2.6 to 3.4 mmHg) in responders versus 3.7 mmHg (range 2 to 5.2 mmHg) in non-responders. The median decrease in heart rate (nine studies) was 3.3 bpm in responders (range 1.5 to 10 bpm decrease) and 1.2 bpm in non-responders (range 0 to 4 bpm decrease). Information on changes in venous oxygen saturation, blood lactate concentration, and blood haemoglobin concentration in the few studies reporting such data are presented in Additional file 1: Table S2.

Additional comparisons

The physiological effects of FBT grouped by speed of FBT delivery (Additional file 1: Table S3) and by class of fluid administered (Additional file 1: Table S4) have also been presented. There is no consistent pattern demonstrated across or between groups.

Relationship between physiological changes after fluid bolus therapy and clinical outcome

Overall, seven studies described clinically orientated outcomes [37],[43],[44],[49],[58],[59],[64]. All reported the effects of complex interventions, such as early goal-directed therapy. No studies examined the relationship between FBT and outcome directly (Tables 4 and 5).

Table 4 Clinically orientated primary outcomes
Full size table
Table 5 Clinically orientated secondary outcomes
Full size table

Discussion

We examined the contemporary literature on FBT in severe sepsis and septic shock and identified 33 original studies describing the characteristics of a fluid bolus, 17 of which also describe the associated physiological changes. We found heterogeneity of triggers, amount, fluid choice and speed of delivery for FBT, which was administered to achieve heterogeneous physiological targets. We similarly found heterogeneity of physiological changes after FBT. In addition, no RCTs compared FBT with an alternative intervention. Finally, no study related physiological changes after FBT to clinically relevant outcomes.

FBT is a widespread intervention in the management of the critically ill septic patient, despite lack of a consistent definition or use of terminology. Our study demonstrates that no contemporary RCTs exist that compare FBT with alternative interventions. The only study comparing FBT to an alternative intervention was a single, non-randomized, prospective, observational study that compared acute circulatory failure patients treated with FBT (500 ml of saline) or with increased norepinephrine dose according to clinician preference. The two groups had clearly different baseline characteristics and were not directly compared [55]. Alternative interventions to FBT may include a diagnostic low-volume FBT [17], classic fluid challenge [11],[12], low-volume FBT and low-dose vasopressor therapy, or cardiac output-guided therapy. Despite the availability of such strategies and the availability of non-invasive cardiac output monitoring, these alternative approaches have not been studied.

Understanding which patient will be fluid responsive is a vital part of rationalising fluid therapy [69]. However, there are multiple different definitions of fluid responsiveness, each dependent on different interventions and different measurements. It would appear that there is little evidence to suggest a consistently different response to FBT based on pre-intervention physiology, as fluid responsiveness is often tautologically and retrospectively defined by participants’ responses to the therapy. A full review of this topic is beyond the scope of this review, though this information is available elsewhere [69],[70].

The contribution of FBT to a positive fluid balance remains poorly understood. In a recent observational study, Bihari and colleagues [36] found that a median of 52.4% of fluid balance on the first, 30.8% on the second and 33.2% on the third study day consisted of FBT. In the Fluid and Catheter Treatment Trial [27] and Sepsis Occurrence in Acutely Ill Patients [71] studies, increasing fluid balance was associated with increased risk of acute kidney injury and mortality. In a retrospective study of septic shock patients in a North American university hospital, non-survivors had a significantly greater positive net fluid balance than survivors over the first 24 hours from onset [34]. Our study also shows little or no evidence for any persisting beneficial physiological changes following FBT. These observations suggest the need for RCTs comparing FBT with alternative interventions and well-defined triggers and physiological outcomes.

This review has several strengths. To our knowledge this is the first review of the contemporary literature on FBT in critically ill patients with severe sepsis.

We are the first to explore the contemporary features of a FBT, and the first to produce a summary of the physiological changes associated with FBT in septic, critically ill patients, including data from RCTs, and observational and quasi-experimental studies. Our wide search criteria, use of three separate sources and hand searching references reduced the risk of inclusion bias and makes it unlikely that we missed relevant studies.

Our study also has some limitations. Our assessments of physiological changes are necessarily limited to the measures of central tendency provided in tables and graphs in the studies identified. We have only provided crude median results in an attempt to provide a rough estimate of possible effect. We limited our search to the present evolving decade. It is unlikely that current clinical practice is better reflected by earlier studies. Indeed, in comparing our results with similar, earlier studies, the reported physiological changes are similar [14],[71]-[75]. We did not account for the effect of vasoactive medications beyond noting their administration. It appears obvious that the mixed and differential inotropic/vasopressor/lusitropic/chronotropic effects of different vasoactive medications are likely to have an effect on the physiological changes reported, as would the administration of blood products. Inadequate information was provided in the studies to make such adjustments possible. FBT is normally part of a complex intervention - the resuscitation of the critically ill patient. As well as the initiation and manipulation of vasoactive medications, analyses must contend with the impact of the use of mechanical ventilation, CRRT, and antibiotic administration. These confounders were not reliably reported in the studies identified and could not be evaluated. In addition, the perceived haemodynamic success of an intervention often depends on the trajectory of the patient’s clinical course. Unfortunately no such information was available from the studies reviewed.

Conclusion

FBT in severe sepsis and septic shock is described in 33 articles in the contemporary literature. Only 17 of these studies report the physiological changes associated with FBT. Evidence regarding the efficacy of FBT compared with alternative interventions is lacking. Crucially, no studies relate the physiological changes after FBT to clinically relevant outcomes. In light of recent studies highlighting the association between FBT and fluid administration in general and harm, there is a clear need for at least obtaining randomised controlled evidence for the physiological effects of FBT over the immediate (0 to 4 hours) post-intervention period in patients with severe sepsis and septic shock.

Author’s contributions

NJG: study design, electronic search design, literature search, study selection, data extraction, data handling/analysis, manuscript preparation, manuscript revision, and manuscript submission. GME: literature search, study selection, manuscript revision, and manuscript submission. RB: study design, electronic search design, data analysis, manuscript preparation, manuscript revision, and manuscript submission. All authors read and approved the final manuscript.

Additional file