Practical use of central and mixed venous blood gases

This chapter is only very slightly relevant to Section G4(vi)  of the 2017 CICM Primary Syllabus, which asks the exam candidate to "describe the factors that affect venous oxygen saturation". No questions have ever been asked specifically about the practical uses of the SvO₂, and they are highly unlikely to be asked in the future because the practice of measuring this parameter has gone out of fashion.

The one question from the First Part papers which asked about this topic was Question 8 from the first paper of 2019. The specific instructions were to compare and contrast the measurement and interpretation of central and mixed venous oxygen saturations. A suggested answer to that question is offered here in lieu of a summary, as this could potentially have some value to the revising exam candidate, in contrast to everything else that follows.

Measurement of central venous blood composition

For this, you have several options, some of which are even continuous. In other words, rather than sampling the venous blood intermittently and risking blocking the lumen or introducing contamination, you can suspend a sensor in the bloodstream and continuously monitor the parameter of your choice.

That's relatively easy, provided the parameter of your choice is venous oxygen saturation. Popularised by the EGDT cult in the early 2000s, ScvO₂ measurement attracted substantial industry attention, and gave rise to the development of various products which promised to monitor the ScvO₂ continuously. These were so heavily marketed that the early versions of Surviving Sepsis Guidelines were widely criticised for being essentially an advertising pamphlet for Edwards Life Sciences (although the pork-barrelling by Eli Lilly was probably more egregious). The taint of industry whoring still remains, unfairly discouraging the use of these instruments. In contrast, the results of large-scale modern trials ( ProCESS,  ARISE,  ProMISE) discourages their use totally fairly, from the standpoint of overwhelming evidence. 

Anyway: in Question 8 from the first paper of 2019, the examiners insisted that 
"methods of measurement such as co-oximetry and reflectance spectrophotometry needed to be explained", and not just listed. To explain spectrophotometry and co-oximetry in detail is somewhat ambitious for a 40% answer (or this brief chapter) and so only an outline is offered here, to simplify revision. 

Infrared oximetry (reflection spectrophotometry) is the gentle art of bouncing a near-IR light beam off erythrocytes, first described by Landsman et al in 1978. Because deoxygenated haemoglobin and oxyhaemoglobin reflect completely different wavelengths,  you can calculate the concentration of each from the analysis of the reflected light. One could go on, but they say a picture is worth a thousand cliches:

For those who are not visual learners, 

• The catheter consists of two bundles of optic fibres
• One of these fibres carries light from the light source to the bloodstream.
• The light source usually flickers with a frequency of 200-300 Hz, to compensate for the effect of blood flow on the measurement. 
• The other carries the light reflected off the red cells back to the detection unit.
• The detection unit contains a beam splitter which separates the beam into a dichroic mirror, transmitting most of the light with/l < 800 nm and reflecting most of the light with 2 > 900 nm, thus splitting the light into two beams
• These pass through interference filters with nominal wavelengths of 640 and 920 nm respectively.
• Oxygenated red cells reflect mainly 640 nm light, and deoxygenated red cells reflect mainly 920 nm light. The signal strength of reflected light of both wavelengths is measured and stored electronically (as R₆₄₀ and R₉₂₀)
• The relationship of log (R₆₄₀ / R₉₂₀) and oxygen saturation is represented by a slightly curved line, which is derived empirically from a calibration sample, and which is stored in the oximeter device as a reference table of values.
• The oximeter can therefore look up the measured signal in the table of values, and produce a saturation measurement.

Co-oximetry is upheld as the gold standard of oximetry, as it measures the concentration of oxyhaemoglobin and deoxyhaemoglobin in as direct a fashion as is reasonably possible. It is described in detail elsewhere, as the average ICU trainee usually contacts this technology in the course of learning about dyshaemoglobin species and ABG machine innards. For a good peer-reviewed source, a reader may also refer to the excellent short paper by Elizabeth Mack (2007). In brief:

• Instead of the reflection of light, co-oximetry relies on the absorption of light in the sample
• As each haemoglobin species absorbs light of different wavelength to a different extent, their concentration can be determined by shining a light of several known wavelengths through the sample
• The absorption of each wavelength can be related to the concentration of the substance in the sample using the Beer-Lambert Law.
• From the ratio of concentrations of oxyhaemoglobin and deoxyhaemoglobin, the percentage saturation of haemoglobin can thus be calculated.

Calculation of sO₂ from PO₂, PCO₂ and pH is a possible third way of getting am SvO₂ value (or any sort of sO₂) and should be mentioned here because it exists, but not dwelled upon because it is an anachronism of an era when blood gas analysers were not uniformly outfitted with haemoglobin spectrophotometry. It was dependent on the measurement of other values, and had the effect of amplifying their measurement error. Moreover, it would be confused by the presence of dyshaemoglobin, and could not be done continuously. Nierman & Schechter (1994) compared this technique (unfavourably) to co-oximetry of venous blood

Interpretation of central and mixed oximetry data

To the question, "how do you interpret SvO₂ data", most reasonable people would respond, "with a grain of salt". However diminished the role of this strategy, for exam purposes the CICM trainee still needs to be aware of these techniques; and there is every possibility that one day they may work with a старове́р consultant who incorporates them into their routine practice. A good article by Keith Walley (2010) summarises some of them. In short, the following applications for SvO₂ and ScvO2 monitoring have been historically supported, however briefly, by the then-available evidence:

• Cardiac output estimation
• Efficacy of tissue perfusion 
• Estimation of shunt fraction

Established uses which remain essential in practice include:

• Estimation of recirculation during VV ECMO
• Calculation of shunt from cardiac output measurements

Estimation of cardiac output from SvO₂ using Ficks equation

With a rearrangement of Fick's equation, we can relate cardiac index (CI) to some known and measured parameters:

Now, with some suspended disbelief, we can convince ourselves that some of the above parameters will not change much in the course of our daily work. For instance, the SaO₂ would rarely be outside the 90-100 range - as it is not routine for an ICU patient to be kept at an SaO₂ 60% for long. Similarly, we can take a bit of a guess at what the VO₂I should be - for most people, its 3.5ml/kg/min, or 125ml/min.  That leaves only two variables in this equation which are subject to change: cardiac index and SvO₂.

Now, if we rearrange the equation even more, by substituting the assumptions we have made, we have an even simpler form of the formula:

So... lets plug in some numbers: if our patient has a Hb of about 70, SaO₂ of 96% and an SvO₂ of 70%, the cardiac index is 1.42 x (1 / 0.26), or about 5.5 L/min/m². If the SvO2 drops to 50%, the CI generated by this equation is 3L/min/m². And so forth.

Mind you, this is not what one could call a precise and meaningful measurement. Many assumptions are made on the way to deriving this formula, to the point that many respected experts have called to abandon its use. A particularly excellent takedown may be seen at PulmCrit. In short:

• Single measurements of SvO₂ cannot describe the cardiac output. From looking at the SvO₂ alone, it is impossible to determine whether the cardiac output is high or low. For example, an SvO₂ of 50% may be due to a low cardiac output, or a high tissue metabolic demand, or to a low arterial oxygen content.
• Serial measurements of SvO₂ cannot describe changes in cardiac output. Following from the above, it is clear that a change in SvO₂ does not necessarily mean a change in cardiac output, as the SvO₂ can change dramatically for a variety of already mentioned reasons. 
• Multiple estimated values are sources of error: for example, where in the paragraphs above we "take a bit of a guess" at what the VO₂ should be. Clearly the metabolic rate of one ICU patient is going to be significantly different to that of another, and there is usually no way of measuring the VO₂ directly. This is a major source of error.
• Use of ScvO₂ as a surrogate for SvO₂ adds further inaccuracy. As is discussed in the chapter on mixed venous oxygen content, there is often a difference between the central venous and mixed venous oxygen saturation, which is unpredictable and which could be as large as 8-10%. 

Use of SvO₂ to assess tissue perfusion

Elsewhere, the relationship of SvO₂ and tissue oxygen consumption is discussed. Without reproducing much of the contents of that page, it will suffice to say that SvO₂ can be said to reflect the oxygen consumption by the tissues, and therefore their perfusion. This conclusion can be reached by the following mental acrobatics:

• Blood contains some mls of oxygen, and flows through tissues at a rate of ml per unit time.
• Tissues consume oxygen at a rate of (x) ml per unit time.
• The oxygen consumption rate is therefore (x) × (time spent by blood in the tissues)
• Tissue blood flow, as it changes the time spent by blood in the tissues, is therefore a determinant of the total amount of oxygen which ends up being removed from the blood.
• If blood flow is too slow, most of the oxygen from the blood will be extracted by the tissues, and as a result the tissues may become hypoxic.
• Thus, the final oxygen content of venous blood can be used to assess whether the tissue blood flow is adequate.
• Below a certain oxygen extraction ratio (about 60-70%, corresponding to an SvO₂ of 30-40%), critically ill patients seem to develop lactic acidosis, which  supports this hypothesis

However, one might point out that:

• It is possible to have tissue ischaemia and lactic acidosis without impaired blood flow (eg. by having a very low arterial oxygenation)
• It is possible to have a normal measured SvO₂ or extraction ratio and profound regional hypoperfusion (eg. abdominal aortic cross-clamp)
• Certain disease states where perfusion is decreased may actually present with a low oxygen extraction ratio, because their pathophysiology involves some level of mitochondrial dysfunction (for example, sepsis).

Ratio of arteriovenous CO₂/O₂ differences

An interesting alternative to the use of raw untreated SvO₂ has been proposed by Mekontso-Dessap et al (2002). Under their proposal, one calculates the arteriovenous differences in CO₂ and  O₂ content, and uses the ratio of these differences to determine how much oxygen is being used for aerobic metabolism. Consider:

• During aerobic metabolism, CO₂ is being produced as O₂ is being consumed
• As aerobic metabolism decreases, CO₂ production and O₂ consumption should decrease together, i.e. at the same rate
• During anaerobic metabolism, additional CO₂ production occurs, mainly as the result of buffering of tissue acidosis by bicarbonate.
• Thus, during anaerobic metabolism, CO₂ production should decrease less than the O₂ consumption
• This means the VCO₂/VO₂ ratio should increase in anaerobic metabolism

By this theory, if one encounters a lactic acidosis and the VCO₂/VO₂ ratio is normal, one must conclude that this is a Type B lactic acidosis. Of course, detecting anaerobic metabolism does not necessarily mean knowing the management strategy which is required to make it aerobic. Is the ratio low because flow to the issues is poor, or are the patient's mitochondria just not using the oxygen properly, or is there microvascular shunting occurring which prevents proper oxygen use?   Su et al (2018) used this to guide therapy (comparing it to SvO2, of all things), and found that there was no difference in outcome. It remains to be seen whether this turns into a valuable bedside instrument.

Use of SvO₂ to estimate shunt fraction

If you ignore the negligible contribution from dissolved oxygen, the shunt fraction equation can be rearranged into this:

Thus, if you have an arterial saturation of 99% and a venous saturation of 70%, your shunt fraction is 3%; if the arterial saturation is 90% and venous saturation is 60%, the shunt fraction is 25%. And so on.

Of course, anybody who has been paying attention will realise that the contribution of dissolved oxygen is often far from negligible (for example, where Hb is low and FiO₂ is high, it could contribute up to 20% to the total oxygen content of whole blood). To actually calculate shunt, SvO₂ is still essential 

Use of SvO₂ in guiding resuscitation of septic shock

Rivers famously used an SvO₂ of 70% as one of the goals of early goal-directed therapy. Though the trial has its critics, the Surviving Sepsis people had taken it on board as part of their recommendations, and for a period suggested that everybody needs to aim for a SvO₂ of 65-70% (Grade 1C). 

The use of SvO₂ in sepsis resuscitation was subsequently  tested in the ProCESS,  ARISE and ProMISE studies. The use of ScvO₂ was one of the main differences between groups. These studies did not demonstrate any survival benefit from protocolised care, in which the use of ScvO₂ was one of the major components. To be fair, many of these patients were not particularly unwell (critics of ARISE will point out that most of the patients improved after a couple of litres of fluid), but even subgroup analysis of truly ICUish patients did not reveal any effect.

Use of SvO₂ in assessing readiness for extubation

A study published in 2010 has reported on changes in SvO₂ in response to a spontaneous breathing trial. These investigators do things slightly differently to the general norm of Australian ICU; the patients were subjected to a 2-hour trial of being intubated but not ventilated - on a T-tube blow-over. This was essentially a test of their respiratory muscle power. If you can suck on a straw for two hours without building up a sweat, then you can probably breathe without a straw.

The SvO₂ was used as a measure of whole-body oxygen demand (given a stable level of oxygen delivery). If you are physiologically crippled, with poor gas exchange and poor cardiac output, the increased metabolic demands of breathing though a hellishly narrow tube will increase your SvO₂. The abovelinked study showed that if during such a 2-hour trial your SvO₂ decreases by more than 4.5mmHg, you are at greater risk of re-intubation over the coming 48 hours.

This study has its flaws. Apart from doing T-piece spontaneous breathing trials (which many don't bother with any more, even though they are probably equivalent in effectiveness to PSV) the small studied groups were heterogenous, and the institutional reintubation rate was very high, even accounting for the fact that the study selected only "difficult to wean" patients (i.e. those who had already failed the T-piece torture). Their reintubation rate was 42%, whereas in large Australian ICUs the rate of re-intubation of electively extubated patients seems to be 1.8%