In Part 1, we looked at ScvO₂: a physiological signal that gradually became a treatment target.
The physiology was real. Oxygen is delivered to tissues in blood. Tissues extract some of it. What comes back in the venous circulation therefore contains information about the relationship between oxygen delivery and oxygen consumption.
The problem was what happened next. A downstream result of flow, haemoglobin, arterial oxygenation, oxygen consumption, extraction and venous sampling began to acquire a much more definite meaning. A low ScvO₂ became inadequate resuscitation. A value above 70% became a target.
The CO₂ gap approaches the circulation from another direction.
Tissues continuously produce carbon dioxide. Blood carries it away. If blood flow falls relative to the amount of CO₂ being produced, the venous–arterial difference increases. This is also real physiology and, in some respects, the CO₂ gap is a cleaner signal of flow than ScvO₂.
Then we start asking it to do more. We combine it with ScvO₂ to classify haemodynamic states. When that still leaves uncertainty, we divide the CO₂ signal by an oxygen-content difference and try to identify anaerobic metabolism.
Eventually we have two blood gases, several differences, a ratio and a four-box diagnostic matrix.
The question is whether we have actually learnt anything that could not have been established more directly.
CO₂ is mostly a product of aerobic metabolism
One misconception needs removing before we start. Most metabolic CO₂ is produced during aerobic metabolism. Anaerobic glycolysis itself does not generate CO₂. So a large venous–arterial CO₂ difference is not, by itself, evidence of anaerobic metabolism.
There is one caveat that becomes relevant later. During severe metabolic acidosis, hydrogen ions are buffered by bicarbonate, generating additional CO₂. That can alter the relationship between CO₂ production and oxygen consumption during profound oxygen-delivery failure.
But the main reason the CO₂ gap widens when cardiac falls is not enough blood moving past the tissues to carry away the CO₂ they are already producing.
Flow and CO₂
We can describe that using Fick’s principle:
VCO₂ = blood flow × (venous CO₂ content − arterial CO₂ content)
VCO₂ is the amount of CO₂ produced by the body each minute.
In words: The amount of CO₂ carried away from the tissues each minute equals how much blood passes through them multiplied by how much extra CO₂ each unit of blood acquires.
Rearrange it:
Venous–arterial CO₂ content difference = VCO₂ / blood flow
So:
For a given rate of CO₂ production, less blood flow means each unit of blood must leave carrying more CO₂.
This is what clinicians usually mean by poor CO₂ “washout”. A better way of thinking about it is that CO₂ production imposes a transport load on the circulation. If fewer litres of blood pass through the tissues each minute, each litre has to carry more of that load. Therefore:
The CO₂ difference reflects blood flow relative to CO₂ production.
But in a patient whose metabolic rate, haemoglobin, oxygenation and acid-base state are reasonably stable over the period being assessed, the useful shorthand is even simpler:
flow falls → CO₂ gap rises
flow rises → CO₂ gap falls
That is why it is reasonable to regard the CO₂ gap as an inverse signal of flow, provided we remember what has been simplified.
Why clinicians became interested in it
it was observed that during cardiac arrest, arterial PCO₂ could remain relatively unimpressive while venous PCO₂ became enormous. Very little blood was moving through the tissues, but those tissues continued producing CO₂. The small amount of blood passing through them accumulated a large CO₂ load before eventually returning to the lungs.
Subsequent studies showed an inverse relationship between the venous–arterial CO₂ difference and cardiac output. Teboul and colleagues demonstrated the same principle using dobutamine: as cardiac output increased, the CO₂ gap fell. When further increases in flow were accompanied by greater metabolic activity and CO₂ production, that fall stopped.
So the gap was never simply a cardiac-output meter. It reflected flow relative to CO₂ production. The bedside variable is usually:
P(v-a)CO₂ = PvCO₂ − PaCO₂
A value above roughly 6 mmHg is commonly considered abnormal.
But notice the P. We have already changed variables.
CO₂ content is not PCO₂
CO₂ travels in blood mainly as bicarbonate, with smaller amounts dissolved in plasma and bound to proteins as carbamino compounds.
Only the dissolved component directly determines PCO₂. So PCO₂ is not the total amount of CO₂ carried in blood.
Fick’s principle describes:
venous CO₂ content − arterial CO₂ content
At the bedside we usually measure:
venous PCO₂ − arterial PCO₂
While related, they are not identical. There are therefore two stages:
Flow and metabolism determine how much additional CO₂ venous blood must carry.
Then:
Blood chemistry determines how much PCO₂ must change to carry it.
The relationship between content and PCO₂ is influenced by pH, haemoglobin, oxygenation, temperature and the Haldane effect — the increased ability of deoxygenated haemoglobin to carry CO₂.
It would be easy to stop here and declare the CO₂ gap hopelessly confounded. That would be unfair.
Across much of the physiological range, CO₂ content and PCO₂ track one another sufficiently well for the pressure difference to act as a useful surrogate for the content difference. Even the original investigators who later developed the CO₂/O₂ ratio explicitly recognised both the approximation and its limitations.
More recent arguments have pushed the effects of acidosis, anaemia and the Haldane effect harder, and those effects are real. Dubin and Pozo have produced particularly detailed critiques of the assumption that a PCO₂ difference simply behaves as a CO₂-content difference. But even their review accepts that cardiac output remains an important determinant of the gap, especially in low-flow states.
So I would not throw the signal away because PCO₂ is not content. In many patients, the useful clinical approximation remains:
If everything else is reasonably stable, the CO₂ gap is mainly telling us about flow.
Flow is not hypoxia
This distinction was demonstrated beautifully by Vallet and colleagues. They created tissue hypoxia in an isolated canine limb in two different ways.
In one condition they reduced blood flow. Oxygen delivery fell because less blood reached the tissue.
In the other they maintained blood flow but reduced arterial oxygenation. Oxygen delivery again fell, this time because each unit of blood contained less oxygen.
Both could produce profound oxygen supply dependency.
But the CO₂ gap rose mainly when flow fell.
When oxygen delivery fell with flow maintained, the gap did not rise in the same way.
This separates two ideas that are often blurred together:
low flow
and:
tissue hypoxia
A tissue can be hypoxic without generating a large venous CO₂ gap if blood flow remains sufficient to carry CO₂ away. And low flow can widen the gap through stagnation without proving anaerobic metabolism.
The 2002 Mekontso-Dessap paper explicitly acknowledged both problems: high CO₂ gaps could occur without evidence of tissue hypoxia, while normal gaps could coexist with profound hypoxia when flow was preserved. So the simple conclusion is:
The CO₂ gap is mainly a signal of flow relative to CO₂ production. It is not a measurement of tissue hypoxia.
Fair enough. If only we stopped there.
When ScvO₂ looks reassuring
The CO₂ gap becomes particularly attractive when ScvO₂ is normal or high.
ScvO₂ depends on cardiac output, haemoglobin, arterial oxygen saturation, oxygen consumption and oxygen extraction. It also depends on how flow is distributed and which venous territories are sampled.
Sepsis makes this particularly troublesome. Flow becomes heterogeneous and oxygen extraction can be impaired. Blood can therefore return from poorly extracting vascular beds with a high oxygen saturation despite serious abnormalities in tissue perfusion.
So a high ScvO₂ can mean excellent oxygen delivery. Or it can mean that oxygen was not extracted. Those are very different states.
The CO₂ gap is less dependent on oxygen extraction. Metabolically produced CO₂ can still enter blood even when extraction of oxygen is abnormal. This is one reason modern proponents of CO₂-derived variables argue that the gap can retain useful information when ScvO₂ becomes difficult to interpret.
Vallée and colleagues tested this in 2008. They studied 50 patients with septic shock who had already achieved an ScvO₂ above 70%. ScvO₂ looked satisfactory in all of them.
Yet patients with a CO₂ gap above 6 mmHg had a mean cardiac index of about 2.7 L/min/m², compared with 4.3 L/min/m² in those with a low gap. ScvO₂ itself was very similar. The high-gap patients also had a smaller subsequent fall in lactate and less improvement in SOFA.
This is probably the strongest argument for measuring the CO₂ gap. ScvO₂ had lost much of its ability to distinguish flow. The CO₂ gap had not.
Two complicated numbers must be better than one
The obvious next move was to combine them. The standard interpretation goes something like this: