Introduction

Fluid administration in the intensive care unit (ICU) is one of the most common and disputed interventions triggered at the bedside by several clinical variables [1, 2].

Fluid therapy aims to increase stroke volume (SV) and cardiac output (CO) to optimise systemic blood flow and tissue perfusion. As with any therapeutic intervention, the final clinical effect elicited may vary because of a complex interplay between the patient's intrinsic conditions and the therapy itself.

Fluid responsiveness can occur only if both ventricles work on the ascending, steep part of the Frank–Starling curve, i.e. in cases where CO is preload dependent [3, 4]. Preload dependency is assessed using a diagnostic test performed by infusing a fixed aliquot of fluid, the fluid challenge [5,6,7]. From a clinical perspective, this approach also allows titration of fluid administration (when the patient becomes no longer responsive to the fluid challenge) and reduces the risk of fluid overload, which worsens the outcome of ICU patients [8, 9].

Several variables defining the characteristics of the fluid challenge have been further investigated in studies adopting continuous haemodynamic monitoring, showing that the amount of fluids given, the rate of administration, and the threshold adopted to define fluid responsiveness impact the outcome of a fluid challenge [10,11,12]. Moreover, despite conflicting results on shock reversal efficacy between crystalloids and colloids, crystalloids are now recommended as the first-line fluid type in patients with septic shock, being inexpensive and widely available. Also, the administration of colloids compared to crystalloids has not demonstrated any clear benefit in the literature [13, 14].

However, neither the nature, mode of administration, and method to assess the effectiveness of the fluid challenge are standardised in current clinical practice, and the definition of fluid challenge responsiveness is also variable among studies [15,16,17,18].

Whether or not these findings have modified the modalities of fluid challenge and the definition of fluid responsiveness in published studies is uncertain. To address this issue, we systematically reviewed existing literature from the year 2000. We appraised the characteristics of fluid challenges in critically ill patients (i.e., amount and kind of fluid administration, time of infusion, hemodynamic variables, and thresholds for fluid responsiveness) enrolled in research studies receiving continuous haemodynamic monitoring and assessed the relationship between the reported fluid responsiveness and predefined independent variables. Secondarily, we compared data from studies published in 2011–2021 versus those published in 2000–2010.

Material and methods

We adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis-Protocols (PRISMA-P) guidelines (Additional file 1: Table S1). The study protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO) in November 2021 (CRD42021284761).

Search strategy

A systematic literature search was performed, including the following databases: PUBMED®, EMBASE®, and the Cochrane Controlled Clinical trials register. The following keywords and their related MeSh terms were used: “fluid challenge”, “fluid responsiveness”, “stroke volume variation”, “pulse pressure variation”, “dynamic indices OR indexes”, “passive leg raising”, OR “passive leg raising test”, “functional haemodynamic test OR tests”. Included papers were also examined to identify other studies of interest missed during the primary search.

Study selection and inclusion criteria

Articles enrolling at least 20 adult critically ill patients, written in English and published from 1st January 2000 to 31st December 2021 in indexed scientific journals, were considered. Editorials, commentaries, letters to the editor, opinion articles, reviews, and meeting abstracts were excluded. Studies enrolling paediatric or obstetric populations were excluded. References of selected papers, review articles, commentaries, and editorials on this topic were also reviewed to identify other studies of interest missed during the primary search. When multiple publications of the same research group/centre described potentially overlapping cohorts, the most recent publications were selected.

A fluid challenge was defined as an infusion of a definite quantity of fluid (expressed as a volume in mL or ml/kg) in a fixed time (expressed in minutes), whose outcome was defined as a change in one of the following haemodynamic variables above a predetermined threshold: CO, cardiac index (CI), SV, SV index (SVI), or surrogate of SV, i.e., velocity–time integral (VTI) in the left ventricular outflow tract and aortic blood flow (ABF), as assessed by transthoracic, transoesophageal echocardiography or oesophageal Doppler. We included studies adopting both a specific (i.e., Ringer lactate, saline, etc.) and a broad definition (i.e., crystalloids, colloids, etc.) of the fluid used for the fluid challenge. Studies adopting changes in systemic arterial pressure to define fluid responsiveness were excluded. Finally, we considered the predefined clinical reasons and triggers to start fluid challenge infusion.

Data extraction

Three couples of examiners independently evaluated titles and abstracts. The articles were then subdivided into three subgroups: “included” and “excluded” (if the two examiners agreed with the selection) or “uncertain” (in case of disagreement). In the case of “uncertain” classification, discrepancies were resolved by further examination performed by one of the three expert authors (A.M., X.M., or M.C.). We used a standardised electronic spreadsheet (Microsoft Excel, V 14.4.1; Microsoft, Redmond, WA) to extract data from all included studies, recording: the study setting (type of study, geographical area and time, where and when the study was carried out, and sample size), patient characteristics (gender, age, reason for admission, underlying diseases, ICU scores of severity, mode of ventilation, and inotropic/vasopressor support), criteria for haemodynamic instability, fluid challenge characteristics, pre- and post-fluid challenge haemodynamic variables. When necessary, the corresponding authors of the included studies were contacted to obtain missing data related to trial demographics, methods, and outcomes (Additional file 1: Table S2).

Statistical analysis

Statistical analysis was conducted on the summary statistics described in the selected articles (e.g., means, medians, proportions) and, therefore, the statistical unit of observation for all the selected variables was the single study and not the patient. Due to the discrepancy between the overall patients enrolled in the trials over the two considered decaders, the comparisons were not weighted for study size.

Fluid challenge was the exposure variable, and clinical and haemodynamic characteristics were considered outcome variables. Descriptive statistics of individual studies used different statistical indicators for central tendency and variability, such as means and standard deviations (i.e., age, tidal volume, fluid responders, severity scores), whereas absolute and relative frequencies were adopted for qualitative variables. To show one indicator for the quantitative variables, we collected means with standard deviations (SD) or medians and inter-quartile ranges (IQR).

Student's t test or Mann–Whitney test in case of parametric or nonparametric distributions, respectively, were used to assess the difference in mean values between responders and non-responders.

Statistical analyses were conducted using GraphPad PRISM® 8 (GraphPad Software Inc., San Diego, CA, USA) and STATA®15 (StataCorp, College Station, TX, USA). For all comparisons, we considered p values < 0.05 significant.

Results

The electronic search identified 3,963 potentially relevant studies. Figure 1 and Additional file 1: Table S3 provide a detailed description of the selection process flow. After evaluating 160 full-text manuscripts, the inclusion criteria were met by 124 studies, 32 (25.8%) published in the period 2000–2010 and 92 (74.2%) in the period 2011–2021. Ten studies (8.1%) required revision by senior examiners because of disagreement regarding inclusion criteria between the initial examiners. We did not find any further relevant publications by reviewing the bibliography of the selected studies.

Fig. 1
figure 1

Flow of the studies; *Reasons for studies' exclusion are reported in the Supplementary materials

Full size image

The general characteristics of the patients are reported in Table 1. We included 6,086 patients, with a median (IQR) of 38 (30–59) patients enrolled in each study. Six studies (4.8%) [20,21,22,23,24,25] were retrospective, while the others were prospective. The median (IQR) period of enrolment [reported in 66 (52.8%) studies] was 12 (6–18) months. At baseline, 2,985 (49.0%) patients received norepinephrine, 179 (2.9%) dopamine, 416 (6.8%) dobutamine, and 177 (2.8%) epinephrine.

Table 1 Patient characteristics at enrolment
Full size table

The reliability of a functional haemodynamic test in predicting fluid responsiveness was assessed in 46 (37.1%) studies. Comparing the two considered decades, no difference was found in the rate of FC administration [17 min (17–33) vs. 33 min (17–50); p = 0.39), in the percentage of fluid responders [52% (43–67) vs. 53% (45–60); p = 0.91], in the percentage of studies adopting crystalloids over colloids [63.6% vs. 67.9%; p = 1.00), or in the threshold of increase in CO or surrogates adopted to define fluid responsiveness (10% over 15%) [18.2% vs. 24.1%; p = 1.00).

Forty-four studies (35.4%) investigated the reliability of a dynamic index in predicting fluid responsiveness. Comparing the two considered decades, no differences were found in the rate of FC administration [17 min (17–25) vs. 29 min (13–33) p = 0.42), or in the rate of fluid responders [53% (41–62) vs. 50% (44–56) p = 0.81), or in the threshold of increase in CO or surrogates adopted to define fluid responsiveness (10% over 15%) (78.5% vs. 66.67 p = 0.42), as compared to studies in the decade 2000–2010. On the contrary, in the decade 2010-2021 we adopted more frequently crystalloids (21.4% vs. 60.0% p = 0.024).

Fluid challenge characteristics and haemodynamic monitoring

Overall, the included studies infused 6,333 fluid challenges. The median (IQR) proportion of fluid responders was 52 (44–62)% (Table 2).

Table 2 Fluid challenge characteristics and haemodynamic monitoring in the included studies
Full size table

In 19 studies (15.3%), the volume of the fluid challenge was reported in mL/kg, with a median (IQR) of 7 (6–8) mL/kg (Table 2). A fixed volume of 500 mL was administered in 95 (76.6)% of the included studies. The median (IQR) of the dispensed volume of fluid was 500 (500–500) mL, infused in a median (IQR) of 18 (11–30) min. Then, the median (IQR) infusion rate was 25 (17–33) mL/min.

CO/CI was used as target variables in 78 (62.9%) studies, while SV/SVI was used in 40 (32.2%) studies. The other six studies (4.8%) adopted SV surrogates (ABF in 4 studies and VTI in two studies). In 88 (70.9%) studies, the threshold adopted to define the fluid responsiveness was an increase of the considered variable ≥ 15% from baseline (Table 2).

Three studies (2.4%) [25,26,27] did not report the type of fluid used for the fluid challenge. Among the others, crystalloids were used in 68 (56.6)% studies, colloids in 52 (43.3) %, and blood in one (0.8)% (Table 2).

The majority of the studies [49 (39.5%)] used transpulmonary thermodilution/dye dilution calibrated haemodynamic monitoring; 22 (17.7%) studies adopted the pulmonary artery catheter monitoring. Echocardiography (either transthoracic or transoesophageal) was used in 31 (25.0)% of studies, and 5 (4.0%) used oesophageal doppler monitoring. Uncalibrated pulse wave analysis monitoring was used in the other 14 (11.2)% studies (Table 2). Finally, bioreactance was adopted in three studies (2.4%). Haemodynamic pre–post-fluid challenge variables in responders and non-responders populations are reported in Table 3.

Table 3 Haemodynamic parameters before and after fluid challenge administration in responders and non-responders
Full size table

Trigger of fluid challenge administration.

Hypotension (i.e., systolic or mean arterial pressure below a fixed value or reduced by a fixed percentage from baseline) was used in 68 (62.4)% of studies. Oliguria (i.e. a drop in urine output below 0.5 mL/h for 2 or 3 consecutive hours) was used in 54 (49.5)% studies, skin mottling or peripheral hypoperfusion in 47 (43.1)% studies, tachycardia (i.e. an increase in heart rate above 100–110 beats/min) in 43 (39.4)%, the need for initiating the infusion or reducing the dose of vasoactive drugs in 41 (37.6)% studies, an increase in blood lactate in 34 (31.2)% studies, a diagnosis of sepsis/septic shock in 12 (11.0)% studies, and renal or hepatic dysfunction in seven (6.4)% studies. Fifteen studies (12.1%) did not report any trigger to start fluid challenge administration.

Comparison of publication periods 2011–2021 versus 2000–2010

The comparison between the 2000–2010 and 2011–2021 decades is reported in Table 4. The percentage of fluid responders (52% for both the decades) and the volume infused (500 mL) were comparable. On the contrary, the infusion time was lower in the last decade (a median of 15 (10–30) min vs 30 (15–30) min, p = 0.03). Crystalloids were used in 61.9% of studies published between 2011–2021 and 34.3% in the 2000–2010 decade (p = 0.007) (Figs. 2 and S1 in the Additional file 1).

Table 4 Comparison between 2011–2021 and 2000–2010 decades regarding the modality of fluid challenge administration
Full size table
Fig. 2
figure 2

Percentage of studies in the two decades adopting different infusion timings. Fluid challenge, fluid challenge

Full size image

CO/CI was used in 67% of the studies published in 2011–2021 and in 60% of those published in 2000–2010 (p = 0.51). The threshold adopted was an increase in CO or surrogates ≥ 15% in 67.4% of the studies of the 2011–2021 decade and in 81.2% of the studies published in 2000–2010 (p = 0.17) (Additional file 1: Figure S1).

Discussion

The results of this review, including research studies investigating the fluid challenge effect in critically ill adult patients receiving haemodynamic monitoring, may be summarised as follows: 1) fluid challenge is usually performed infusing a bolus of 500 mL of fluid, most often a crystalloid, in less than 20 min; 2) the response to fluid challenge is usually defined as a CI or CO increase ≥ 15% as compared to baseline; 3) positive response to fluid challenge is reported in about 50% of ICU patients; 4) the most common trigger for fluid challenge administration is usually the occurrence of hypotension, followed by oliguria and clinical signs of hypoperfusion; 5) the comparison between the 2000–2010 and 2011–2021 decades of publication showed no difference in the percentage of fluid responders (52% on average for both the decades), the volume infused (500 ml), and the criteria defining fluid responsiveness. On the contrary, compared to the 2000–2010 decade, in the period 2011–2021, the fluid challenge infusion time was lower, and crystalloids were more frequently used.

Fluid challenge characteristics

Among the included studies, the fluid challenge usually consisted of a median volume of 500 mL administered over a 20-min period and defined as a positive response by an increase ≥ 15% of CO or surrogate. These characteristics and responsiveness definition are to be considered good practice, for the response of CO to a fluid bolus is poorly followed by the simultaneous changes in arterial pressure [28, 29] or heart rate [30]. However, this is not the case in clinical practice, where the fluid challenge effect is often assessed by a rise in arterial blood pressure [16].

Interestingly, 500 mL was also the median volume fluid challenge used in the FENICE study (an observational study including 311 centres across 46 countries) [16], whereas a fluid challenge of 250 mL is usually adopted in high-risk surgical patients undergoing goal-directed therapy optimisation [31]. The use of large volumes for fluid challenge optimisation should be balanced to the detrimental risk of fluid overload [9], primarily if safety limits (i.e., increase in CVP) dynamically indicate fluid non-responsiveness are rarely used [19]. Since fluid challenge volume should be at least 4 mL/kg [32], smaller fluid challenge volumes may be considered for repetitive tests.

Moreover, the FENICE study reported a median of 24 min of infusion time and a rate of 17 mL/min [16]. Hence, the volume and rate of administration seem comparable between clinical and research settings. On the contrary, the infusion time was lower in the last decade (a median of 15 min vs 30 min, p = 0.03), indicating a trend towards the increase in the infusion rate in more recent studies. This global inception cohort study evaluated the clinical use of the fluid challenge in daily practice, whereas our review considered only research papers adopting the fluid challenge as a part of a protocol, limiting the comparison with the results of the FENICE. Moreover, in contrast with a previous metanalysis, including ICU studies up to 2014 [19], crystalloids are used in most studies. Crystalloids have been used in two-thirds of the studies from 2011 to 2021, compared to one-third from 2000 to 2010. These data indicate an alignment between research studies, recent guidelines, and metanalyses [13, 14].

Limitations

Limitations of our review have to be considered when extrapolating the results to clinical practice. First, the present study does not report any outcome endpoints. A recent large randomised-controlled trial showed no difference in mortality rate among ICU patients receiving different fluid bolus infusion rates [33]. However, the faster rate adopted in this study (5.5 mL/min) is below the median rate found in the studies included in the present review (25 mL/min) [33]. The administration of aliquots of fluids at a slow rate should not probably be indicated as a fluid challenge. Moreover, all the included studies are research papers whose aim was to evaluate the haemodynamic changes after the fluid challenge infusion or assess the reliability of indexes or functional haemodynamic tests in predicting the response to a fluid challenge. We did not include studies on the fluid challenge clinical use in ICU patients.

Another potential source of bias is related to the different haemodynamic monitorings used to assess fluid challenge responsiveness. When considering the median cut-off value identifying responders from non-responders, the accuracy of measurement of the changes in CO, or its surrogates, is undoubtedly relevant. Additionally, the reliability of different monitorings in tracking the dynamic trends of CO may not be consistent and may be below the boundaries of accuracy and precision of the Critchley–Critchley criteria [34]. Hence, the reproducibility of CO measurements obtained by the different monitoring systems may be limited. Moreover, cut-off values and measurement techniques potentially induce heterogeneity in response to the fluid challenge administration. As confirmed, responders ranged from 23 to 100% across the included studies (Table 2). The use of echocardiography is associated with high proportions of fluid responders compared to other haemodynamic monitoring devices. The operator-dependent bias may affect the evaluation of SV changes after fluid challenge.

We excluded studies in which the fluid challenge response has been assessed without haemodynamic monitoring and, hence assessing changes in systemic arterial pressures, potentially limiting the whole comparability of the technique in the two considered decades. Finally, the overall number of patients enrolled in the trials of the two considered decades was considerably different. This could bias the comparisons between the two groups if weighted for study size.

Conclusions

This systematic review, including research studies on fluid challenge use in critically ill adult patients receiving haemodynamic monitoring, showed a positive response in 52% of the patients. This test was usually performed infusing a bolus of 500 mL fluid, more often a crystalloid, in less than 20 min, and fluid responsiveness was generally indicated as a CI or CO increase ≥ 15% compared to baseline. Fluid challenge administration is usually triggered by hypotension. In the 2011–2021, the infusion time was shorter, and crystalloids were more frequently used than in the 2000–2010 decade.