What’s the point of estimating fluid responsiveness?

The Thoughtful Intensivist navigates dangerous waters

I REMEMBER¹ a peculiar case I saw in an ICU rotation during the second year of my Internal Medicine residency in 1999. A patient had cardiac surgery and had mediastinitis. His sternum had been removed with some portions of his pericardium. I know this is not accurate but this is how I recall it. The result is he was awake, on spontaneous breathing, and had daily cleaning, pus removal, and dressing by the ICU nurses. During the procedure, we could see his heart beating while the nurse delicately removed pus and debris. We noticed that as the days and weeks passed, his right atrium wall appeared more bruised.

One morning it happened, during our round. The patient had a coughing attack and suddenly the mediastinal dressing became red. We ran to the bedside, removed the dressing, and saw the blood coming from his right atrium. The blood ran swiftly and started pouring to the bed, and then to the ICU floor.

I think no one is trained for this. At least, in my second year after graduation, I wasn't. We thought of compressing the atrium and immediately realized one can't and shouldn't do it. We started trying to suture the torn wall and started pumping fluids. Someone took a peripheral IV line, one placed a central line², and someone else intubated. A digestive surgeon appeared in the ICU and we grabbed him by his arms, urging him to do something. The scene resolved in a few minutes. The patient's blood turned into a light red and soon assumed a shade of pink. In the final moments, parts of his heart still moved in gentle agony, but it bled normal saline.

In the following years, I was trained in hemodynamic support. I learned the formulas for cardiac output and blood oxygen content and learned to combine them to calculate oxygen delivery (DO2).

I spent the best part of my twenties chasing pulmonary capillary wedge pressure goals to improve DO2 in shock. We tried to maximize cardiac output by infusing fluids to keep the patient at the plateau of the Frank-Starling mechanism.

The idea is old. Search the oldest biomedical journals and you will see that by the end of the 19th century, the role of the blood pumped by the heart in the delivery of oxygen and food for the cells was common sense. Leaping from this point to “the more DO2 the better” was almost natural. The idea is deeply ingrained in our brains.

However, I never forgot the pink blood, and finally normal saline, pouring from the leak in that patient's right atrium. It resonated. I saw neither crystalloids nor colloids transport oxygen, implying they increase cardiac output to the same extent they dilute hemoglobin, exacting a neutral effect in oxygen delivery. Look at DO2 formulas and prove me wrong.

CO = cardiac output

HR = heart rate

SV = systolic volume

Hb = hemoglobin level

SaO2 = oxygen saturation

I don't expect a dispute based on DO2 formulas. However, I anticipate receiving a far-fetched reasoning about blood viscosity in the capillary, circumventing DO2. It's time to recall the most important math in critical care: the Hagen–Poiseuille equation for the flow of an incompressible fluid through a long cylindrical tube, i.e., the arteries of a vascular segment or whole body circulation.

∆P = pressure difference between the two ends

µ = fluid viscosity

L = tube length

R = tube radius

A = tube cross-sectional area

Q = flow

First, let's practice using Hagen-Poiseuille to dispel the fluids increase blood pressure ghost. In this case, ∆P is mean arterial pressure (MAP) minus central venous pressure (CVP) and Q is cardiac output. Arterial pressure is the pressure blood exerts against the arterial walls. I am interested in augmenting MAP, so I administer some fluids. However, the fluid will reduce blood viscosity (µ) causing ∆P reduction. Of course, it is counterbalanced by an increase in Q through augmented systolic volume. If the patient's heart is good enough to endure the fluid challenge, cardiac output increases as blood viscosity reduces, and MAP oscillates mildly. However, if the patient's heart is not strong enough, CO doesn't increase, and ∆P reduces by augmenting CVP, satisfying the formula.

Please explain it to your friend who orders 500 mL of fluids over the telephone instead of evaluating the hypotensive patient

Remember that the widespread idea of predicting fluid responsiveness is based on a few assumptions: (1) more fluid increases cardiac output by expanding blood volume, (2) fluid infusion keeps blood volume expanded for enough time to (3) improve clinical outcomes.

The fluids increase CO assumption will require CO to rise by reducing viscosity, i.e. diluting red cells, without augmenting CVP. It works until CVP increases and CO plateaus, satisfying Hagen-Poiseuille and Frank-Starling.

The second assumption is that infused fluids remain in the vascular beds. I think this one is the easiest to dismiss. No math is needed. Infuse 1 liter of a crystalloid on yourself and observe it being dumped in the toilet during the next hour. Take a final look while flushing, and say goodbye to fluid responsiveness discussions as they find their way.

Nature made it clear: no extra fluids are allowed in your vascular beds. It will be urinated or leave the circulation to the interstitial space, to hydrate the tissues. Every day several liters of fluids leave the vasculature into the interstitial space and part is reabsorbed via lymphatic circulation.

The amount of fluid transferred from the vascular to the interstitial space is a function of differences in hydrostatic pressure and colloid osmotic pressure between the compartments, attending to the Starling Principle. The extra fluid will increase vascular hydrostatic pressure and decrease vascular colloid osmotic pressure (by dilution), causing fluid to shift to the interstitial compartment until the forces equalize. The extra fluid wasn’t transporting oxygen. Nature is wise.

The patient might benefit if dehydrated, of course, otherwise the patient will need to handle the extra volume. In septic shock, capillary leakage, glycocalyx loss, and the catabolic state (resulting in lower oncotic pressure) all contribute to shifting fluid from the vascular compartment to the interstitial space creating edema.

The infusion of large fluid volumes, large enough to cause abdominal compartmental syndrome, was praised as a display of good practice in the nineties. Later on, following the 2001 Early Goal-Directed Therapy study, I became a volume liberal. In the next decade, the idea of de-resuscitation emerged, and everyone turned early liberal and late restrictive / de-resuscitative. After some nice RCTs retested the liberal hypothesis and found no effect, everyone is becoming restrictive. Although not explicitly, or maybe instinctively, we spent the last decades testing the limit of the “less is more” approach.

To wrap it up, neither of these strategies improved outcomes. It suffices to say that no fluid resuscitation strategy reduces mortality in distributive shock, debunking the third premise. Don't get me wrong. Patients in septic shock should receive full circulatory support including vasopressors, inotropes as needed, and maybe 20-30 mL/kg of fluids. However, if there is any benefit in making the heart pump diluted blood for a while it is completely lost in marginal utility.

Please give fluids to a patient who looks dehydrated or on the brink of oliguria without further reasoning. Use your echo skills to ensure you won’t cause congestion.

While I wait for someone to falsify the brave positions I took in the preceding paragraphs, I will keep pretending I believe in this dogma and keep teaching my Med students to infuse fluids in septic shock based on the embarrassing Oxygen Choo-Choo Train model. I am thankful for my students’ condescendence.

I hear the whistle blowing! It is time to stop daydreaming and embark again on the train, the inconsistencies of critical care science well sealed in my baggage. If my fellow intensivists would rather cling to fluid responsiveness sciences, I have only one request. Let’s update the train metaphor. How about Tattoo driving intensivists around the Fantasy Island?

1. My Substack has a therapeutical role, as disclaimed in the About page.

2. We were f* good and fast in the pre-ultrasound era