preamble: historical perspective

For centuries, medical experts practiced bloodletting for a variety of ailments.  This was widely believed to rid the body of evil humors.  When patients didn’t respond well, this was believed to reflect an inadequate or delayed bloodletting.  Practitioners competed to see who could partake in the most rapid and aggressive bloodletting. 

Some centuries later, bloodletting has been replaced by fluid bolusing.  The River’s trial on septic shock pushed this into the fore in 2001.  A key treatment for septic shock became early and aggressive bolusing with large volumes of crystalloid.  This practice grew in popularity and metastasized to other areas of medicine (e.g. pancreatitis).  Nary a patient left the emergency department without having received a fluid bolus. 

Currently, most practitioners would scoff at bloodletting, while undoubtedly accepting the virtue of fluid boluses.  However, for most disease processes, the two treatments may be similarly misguided.  Iatrogenic swings in volume status probably aren’t the ideal treatment for most acute illnesses. 

lack of diagnostic value

Traditional teaching has been that the administration of fluid boluses provides diagnostic value regarding the patient’s volume status:

• If the patient responds to fluid administration, this reveals that they were volume depleted.
• If the patient fails to improve following volume administration, the patient isn’t hypovolemic.

This seems very logical, and I’ve used these principles for years.  Unfortunately, however attractive this concept may be, it lacks evidentiary support. 

Two provocative RCTs show that the temperature of the fluid has a dominant effect on the patient’s hemodynamic response, with only cold fluid increasing the blood pressure.^​1,2​   Thus, it’s possible that an improvement in blood pressure following fluid administration could have little to do with hypovolemia!  Instead, it’ appears that hypothermia is triggering an endogenous sympathetic nervous system response, which increases the blood pressure.  So instead of bolusing cold fluid, we could obtain the same hemodynamic response with a cooling blanket (a benefit which has indeed been demonstrated in an RCT!).^​3​

Another study looked at the effect of two small sequential fluid boluses on hemodynamics.  If bolusing with fluid is a useful evaluation of volume status, then it must be reproducible.  However, the hemodynamic impact of two sequential boluses of fluid were widely disparate.^​4​   Thus, if a patient responds to one small bolus of fluid, this doesn’t necessarily predict that they will respond well to additional fluid.  

The last reason that fluid boluses are poor indicators of hemodynamic responses is that we usually evaluate the wrong parameter.  The ideal parameter to assess following a fluid bolus might arguably be cardiac output or tissue perfusion.  However, these are hard to assess – so we instead will often look at blood pressure.  This is a poor endpoint (for example, fluid therapy could improve cardiac output, triggering a reflexive reduction in endogenous vasoconstriction, and thereby cause no improvement in blood pressure). 

The net result for everyday clinical practice is that giving fluid boluses and then trying to determine if the fluid has been helpful is a dubious diagnostic test. 

lack of therapeutic benefit

There are numerous intermediary steps which are required between providing a patient with fluid and obtaining clinical improvement.  Let’s walk through these steps individually…

(1) Fluid must increase the cardiac output (“fluid responsiveness”)

In order for fluid administration to improve cardiac output, both the right ventricle and the left ventricle must be functioning on the steep part of the Starling curve (wherein increased filling improves cardiac output).  This depends on the volume status and the function of both ventricles.

Fluid responsiveness will vary depending on the patient population and clinical scenario.  Overall estimates often suggest that only half of critically ill patients are fluid responsive.  So right off the bat, lots of patients won’t obtain benefit from a fluid bolus. 

(2) Increase in cardiac output must translate into an increase in oxygen delivery (DO2 responsiveness)

“Fluid responsiveness” is generally defined based on achieving an increase in cardiac output by >15%.  Fluid responsiveness is defined that way due to convenience, because measuring changes in cardiac output is easily achieved in clinical research.  However, what actually matters to the body isn’t the cardiac output, but rather the amount of oxygen delivered to the body (the DO2):

DO2 = (constant)(Cardiac output)(Hemoglobin)(Oxygen saturation)

The average blood volume of an adult is ~5 liters.  A liter bolus will cause hemodilution, with a decrease of the hemoglobin concentration by ~1/6th (16%).   If a fluid bolus causes an increase in cardiac output of 15% with a simultaneous decrease in the hemoglobin concentration by 16% then overall there is no change in the DO2!

Pierrakos et al evaluated changes in cardiac output and DO2 in 71 critically ill patients following a fluid bolus (figure below).  Of these patients, only 27 were fluid responsive as defined by an increase in cardiac index by >15%.  Among these 27 patients, only 21 patients experienced an increase in DO2 by >15%.^​5​ So, only a fraction of patients who are fluid responsive will see a meaningful increase in tissue oxygen delivery.

(3) Improved oxygen delivery must cause an increase in oxygen utilization (VO2 responsiveness)

Increasing the oxygen delivery is necessary, but not sufficient to cause clinical improvement.  Increasing oxygen delivery to the tissues is beneficial only if this leads to an increase in tissue oxygen consumption (something which might be termed “VO2 responsiveness”). 

Unfortunately, it’s common to encounter situations wherein increasing the systemic oxygen delivery doesn’t increase the tissue oxygen utilization.  This frustrating conundrum may be explained by the following phenomena:

• Mitochondrial dysfunction impairs oxygen utilization at the cellular level.
• Microcirculatory dysfunction leads to shunting, which causes oxygenated blood to bypass areas of tissue which are deprived of oxygen.  
• Tissues are already receiving adequate amounts of oxygen and don’t actually need any additional oxygen.

Lack of VO2 responsiveness partially explains why interventions such as increasing the hemoglobin to 10 mg/dL or targeting a supra-normal oxygen delivery have failed to improve clinical outcomes.  Simply cranking up the systemic oxygen delivery won’t necessarily help the patient – and may actually cause harm. 

There doesn’t seem to be any evidence regarding whether fluid bolus therapy in humans improves tissue oxygen consumption.  In the absence of any evidence, the only thing we have is the following physiological equation: 

(Improvement in VO2) < (Improvement in DO2)

The tissues will never utilize 100% of the oxygen delivered to them. This means that increases in oxygen delivery (DO2) may over-estimate any true clinical benefits from such increases. 

(4) Even if improvement in oxygen utilization occurs, this is transient due to fluid extravasation.

The greatest limitation to benefit from fluid boluses is that they usually cause only a transient hemodynamic improvement.  Numerous clinical and animal studies have found that any hemodynamic benefit from a crystalloid bolus will disappear within about two hours.^​6–13​

Lack of any sustained benefit from a fluid bolus is explained by rapid fluid extravasation out of the vasculature and into the tissues.  Even among healthy volunteers who have been rendered hypovolemic, ~70% of a fluid bolus will extravasate into the tissues within merely 30 minutes!^​14​

Among sick patients with leaky vasculature due to endothelial dysfunction, fluid extravasation is an even greater problem.  For example, among septic patients <5% of an infused bolus may remain intravascular after one hour.^​15​ Many critically ill patients will have pre-existing endothelial dysfunction (e.g. due to systemic inflammation).  In other cases, it’s possible that the fluid bolus itself may contribute to endothelial damage (more on this below).  Regardless of the mechanism, fluid usually will rapidly redistribute into the tissues. 

Redistribution of a fluid bolus may lead to a vicious cycle of repeated fluid bolusing as shown above.  This illustrates how liberal application of the concept of “fluid responsiveness” may be dangerous. 

potential harm

(1) Fluid boluses may damage the vascular endothelium

Rapid bolusing of fluid may damage the vascular endothelium.  For example, Mathru et al showed that a prophylactic fluid bolus of only 750 ml (~9 ml/kg) prior to spinal anesthesia for elective C-section could cause shedding of the endothelium.^​16​ This effect may be partially driven by transient intravascular hypervolemia, caused by rapid infusion of fluid.^​17​

(2) Volume overload is bad (either intravascular or extravascular) & probably worse than hypovolemia.

There is increasing awareness of the harm caused by volume overload.  Harm from volume overload may occur regardless of whether the fluid stays in the vasculature or extravasates into the tissues:

• Intravascular congestion increases the central venous pressure, thereby decreasing the perfusion pressure of vital organs (which is equal to the MAP minus the CVP). 
• Extravascular fluid overload causes tissue edema which may directly cause organ failure (e.g. pulmonary edema, bowel wall edema). 

On the whole, the human body evolved to cope with hypovolemia (e.g. due to hemorrhage or starvation).  Our physiology is poorly designed to deal with volume overload, suggesting that when in doubt it may be better to leave patients a bit on the dry side. 

lack of evidentiary support

Fluid boluses have been widely adopted in medical practice because they make sense and cause transient clinical benefit.  As such, there isn’t much evidence regarding this practice. 

Only one study evaluated the effect of infusing crystalloid at different rates.^​18​ Sankar et al. performed an RCT of pediatric septic shock wherein children were randomized to receive fluid boluses of 20 ml/kg rapidly (over 5-10 minutes) or more slowly (over 15-20 minutes).  Children in both groups received nearly identical volumes of fluid.  However, children receiving fluid more rapidly required intubation more frequently (table below).  Unfortunately, the study was stopped prematurely, leaving it statistically fragile.  Nonetheless, it suggests that rapid volume shifts may cause a greater degree of endothelial dysfunction and fluid leak than more gradual fluid infusion.  

Other seminal trials also suggest harm from fluid boluses. 

• FEAST trial:  This study involved the use of fluid boluses for children in Africa with shock.^​19​  Fluid boluses increased mortality due to delayed cardiovascular collapse – which makes sense for all of the reasons explored above.  Interestingly, investigators of this non-blinded study thought the fluid boluses were beneficial because they caused improvements in the short term.^​20​  More on this trial here by TheBottomLine. 
• Andrews et al. 2017 performed an RCT of septic shock in Zambia which centered around early administration of fluid boluses.^​21​  The study was stopped early due to harm, with more aggressive fluid boluses increasing mortality.  More on this trial here by TheBottomLine. 
• The CLASSIC trial:  following initial sepsis resuscitation, patients were randomized to a more liberal versus more conservative fluid strategy.^​22​  This was a feasibility trial, so the primary endpoint was simply the ability to alter the difference in fluid administration.  However, patients treated with a conservative fluid strategy (who received fewer fluid boluses) were less likely to develop worsening renal failure.  This is a statistically weak finding (p=0.03) in a secondary endpoint, but it does support other data suggesting a signal of harm from liberal fluid use.  More on this trial here by TheBottomLine. 

where do we go from here?

Traditionally, the approach to hemodynamic instability has always started with fluids and only resorted to vasopressors if fluids failed (a fluid-first strategy).  This strategy is nearly unavoidable if vasopressors may be administered only through a central venous catheter.

Over the past few years, there is increasing recognition that peripheral administration of vasopressors may be safe.  This makes it logistically feasible to use early vasopressors, rather than using solely fluids initially.  Compared to fluid boluses, vasopressors have greater speed and efficacy at achieving an adequate blood pressure.  Subsequently, gradual fluid infusions may be considered to repair any true underlying hypovolemia. 

Ongoing research will hopefully clarify this over time.  For now, it may be sensible to avoid fluid boluses when possible. Of course, there are certainly some situations where patients are in extremis and require aggressive fluid resuscitation. 

• Fluid boluses don’t necessarily provide reliable information about the patient’s hemodynamics (especially if they are provided casually and without precise hemodynamic monitoring).
• As shown below, the majority of fluid boluses don’t lead to sustained clinical benefit.
• Risks of fluid bolus therapy include damage to the endothelial glycocalyx and volume overload.
• Available RCTs involving fluid bolus treatment suggest that it is harmful. 
• Overall, fluid boluses may lead to transient improvements in hemodynamics, which reinforces the practice of providing them.  However, a more detailed evaluation of the evidence and physiology suggests that most fluid boluses probably lead to harm. 
• The widely-recommended practice of treating anyone with potential infection and possible “sepsis” with a 30 cc/kg fluid bolus is likely dangerous and not evidence-based. 

Related

• Petition to retire the surviving sepsis campaign guidelines
• The surviving sepsis campaign 1-hour bundle is… back?
• Six myths promoted by the new surviving sepsis guidelines
• Abdominal compartment syndrome (IBCC)

• Author
• Recent Posts

Social Me

Josh Farkas

Josh is the creator of PulmCrit.org. He is an associate professor of Pulmonary and Critical Care Medicine at the University of Vermont.

Social Me

Latest posts by Josh Farkas (see all)

• PulmCrit Wee – A better classification of heart failure (HFxEF-RVxEF) - August 26, 2024
• PulmCrit Wee: Rational selection of infusion rate based on loading dose - June 25, 2024
• PulmCrit: PPIs are safe and effective for GI prophylaxis… the end. - June 18, 2024