End-tidal capnometry

This chapter is at least loosely associated with Section F12(ii) from the 2017 CICM Primary Syllabus, which expects the exam candidates to be able to "describe the principles of capnography, including calibration, sources of errors and limitations".  The interpretation of abnormal end-tidal capnogram patterns is explored elsewhere. The objective of this resource is to describe the physical principles important for the measurement. A more precise objective is difficult to delineate considering the brevity of the syllabus entry and the lack of coherent examiner's comments to the single past paper question on this topic, Question 10 from the second paper of 2019. One basically has to assume that the college would want their trainees to know something about the physical principles which allow CO₂ detection, the design characteristics of end-tidal capnometers which use these principles, and the potential sources of error. 

In summary:

 A good introduction to the topic is offered by Ward & Yealy (1998), and it would serve the purposes of most people whose time and interest in capnometry are finite. For the rest, Capnography by Gravenstein et al (2011) is an entire 490-page textbook dedicated to the subject. The latter resource was used to structure this chapter. Omissions of important material can be defended on the basis of the almost total irrelevance of this subject.  

Definitions 

To be clear:

• Capnometry is the measurement of the concentration of CO₂ 
• Capnography refers to the graphic display of this measurement over time.

To be even more clear, essentially everything CO₂-related one can see in the ICU will be capnography of some sort, but only the measurement and display of the very last CO₂ value before inspiration is actual end-tidal capnography. However, for the purpose of most bedside conversations, "end-tidal capnography" is a term used to describe the whole process of measuring expired CO₂.

Methods of measuring CO₂ in expired gas

There are numerous methods of measuring CO₂ in air:

• Infrared spectroscopy
• Mass spectroscopy
• Raman spectroscopy
• Chemical colourimetry
• Acoustic capnometry

There is no universe in which a CICM First Part exam candidate could ever be expected to know these in any great detail, but to have some superficial understanding of the physics and chemistry would probably be harmless. For this, the relevant chapter by M.B. Jaffe from Capnography (Ch. 37, p.381) is summarised and distilled here into easily digested point-form. Where possible, the limitations and disadvantages of each method are listed alongside the method.

Measurement of CO₂ by infrared spectroscopy

Without recapitulating large sections from the chapter on pulse oximetry, the scientific foundation of infrared CO₂ can be summarised as follows:

• Using the Beer-Lambert law, it is possible to calculate the concentration of a substance in some medium by measuring the amount of light absorbed by that substance
• Carbon dioxide absorbs light at specific wavelengths; in fact in the near-infrared region (wavelength of 4.26 μm) it has a nice convenient absorption peak.
• Thus, by passing that specific wavelength of near-IR light through the respiratory gas mixture, one can estimate the concentration of CO₂ in that mixture.

That CO₂ absorption peak at 4.26 μm is convenient because it is well separated from other absorption peaks of other substances which might be present in the respiratory gas mixture. Observe this lovely image from Wikimedia Commons: 

As you can clearly see, at the near-infrared wavelength of 4.26 μm the CO₂ is very much on its own. In fact, the near-IR range of the electromagnetic spectrum is quite good for separating CO₂ from the rest of the pack. In order to absorb IR radiation, a molecule has to be asymmetrical and polyatomic, because IR absorption occurs as a consequence of atomic vibration in a molecule with a dipole moment. Without going into completely irrelevant (but fascinating) detail, it will suffice to say that monoatomic helium and nice symmetric O₂ molecules do not have those characteristics and consequently do not absorb light in the IR spectrum. Molecules which are strongly asymmetrical, and particularly ones with a double bond (eg. C=O), absorb IR avidly, which accounts for the height of the CO₂ peak in the diagram above. 

Anyway, back on topic. The only gas which mimics the absorption characteristics of CO₂ is nitrous oxide (N₂O), which can occasionally fool the older sensors. Various spectrum-narrowing IR filters have been built into anaesthetic gas monitoring equipment which - these days - is reasonably good at discriminating between these substances. In any case even in the bad old days the error was not very great - for example, Kennell et al (1973), in the Jurassic period of anaesthesiology, reported a maximum error of 0.7% CO₂ (or approximately 1.12 mmHg) when a gas mixture containing 70% N₂O was in use. At worst, one would have to apply a correction factor to one's CO₂ measurement, which is usually something like [corrected CO₂ = observed CO₂  × 0.90].

This being the method of measurement, one might be able to imagine that a device designed to perform it would consist of the following components:

• A light-emitting diode, ideally producing near-IR light in a narrow spectrum
• A photosensitive IR detector which measures the absorption of light at this spectrum
• Some sort of circuit which relates the IR absorption signal to a CO₂ concentration, using a lookup table of calibration values
• An output device designed to represent the absorption data, ideally as a graph of CO₂ concentration over time

There are two main ways of doing this, which are usually described in terms of "mainstream" and "sidestream", the latter ironically being the more mainstream of the two (i.e. in common use). The mainstream monitor places the detector directly into the path of moving respiratory gases, whereas the sidestream monitor samples these gases by removing a small fraction of them via an additional set of tubing, to be analysed by a detector array at some distance from the patient (i.e somewhere in the monitoring rig). It's one of those things where a diagram helps:

The sidestream system entrains some variable volume of gas per minute from the main circuit, usually no more than 200ml/min. Though negligible from a clinical standpoint, strictly speaking this represents a circuit leak, and the ventilator flowmeters will usually report it as such (at least insofar as it affects the difference between delivered and expired volumes, on the volume-time waveform).

Each method has its advantages and disadvantages:

• Mainstream devices add dead space, though usually only a few cubic centimetres. Additionally, as objects exposed to the feral zoo of human lungs, these things become contaminated and need to be either replaced or sterilised between patients, increasing the total cost of the monitoring. However, they have superior accuracy, and immediate response time in terms of changing respiratory conditions. 
• Sidestream devices are less accurate, particularly at low expiratory flow rates or low volumes, because the entrained gas from the circuit ends up being diluted with some fresh gas contributed by the bias flow of the ventilator. Also, there is usually a delay between the change in the patient's condition and the detection of a CO₂ change because the changed gas takes a few moments to make its way up the sample tubing. A lower than usual rate of aspiration (eg. a faulty gas pump) can give rise to situations where the end-tidal CO₂ graph appears to be slurred, giving the spurious impression of bronchospasm. Fortunately, that tubing is cheap and requires minimal maintenance while in use. Moreover, one can shove one of those tubes through the vent holes of a Hudson mask or even directly into somebody's nostrils, making this modality much more versatile (in contrast, the mainstream devices are mainly designed to sit at the ends of endotracheal tubes).

IR spectroscopy of CO₂ also has various limitations:

• It may be confused by (massive quantities of) nitrous oxide
• Measurement becomes impossible with secretions or precipitation blocking the light sensor
• The signal produced by the sensor has to be calibrated. i.e. the direct interpretation of the signal is impossible
• Accuracy of the sensor depends on the narrow bandwidth, which means the sensor (or the detector) can't be the cheapest possible versions of themselves.
• Ambient infra-red light can confuse the monitor (though theoretically the patient's bed space should not be inundated with near-IR light under any normal circumstances).

Measurement of CO₂ by chemical sensor colourimetry

Though this technique is called "colorimetric" (or "colourimetric" if you're in the colonies), the exact definition of the term describes analytical chemistry by the quantitative analysis of colour, which is not exactly what happens when you roughly eyeball the colour change indicator after an intubation. Therefore, the application of "colourimetry" to the use of these devices perhaps drapes an unfairly scientific facade over something which essentially a urine dipstick applied to the airway.

Anyway. The technique usually takes advantage of the fact that CO₂ tends to increase the concentration of the hydrogen ions in solution when it dissolves in it, lowering the pH. One can therefore employ a number of different chemical indicators to detect pH, and use that as the surrogate for CO₂. Under most earthly circumstances, your patient will not be exhaling anything else that is likely to be highly acidic. 

With a premise this simple, one would expect this sort of capnometry to be the earliest available form, with all the others to come later. That's probably true. The first recorded instance of anybody measuring expired CO₂ in this way comes from Mariott (1916), who collected expired air and tested its CO₂ levels by the visual identification of a colour change:  "The reaction of such a solution may be determined by adding to it an indicator such as phenolsulphonephthalein, which shows over a considerable range of reaction definite color changes." That would probably have been a rich burgundy-purple. The first use of this to detect correct placement of an endotracheal tube was probably by Berman et al (1984), although the authors for some reason settled on the use of cresol red, usually used as an aquarium pH indicator. 

All these options required you to bubble the expired gas through some puddle of indicator solution, which is messy and inconvenient. Modern devices such as this Medtronic product use some porous material (usually foam or paper) which has been impregnated with pH-sensitive indicator and which ships as a dry product, relying on the humidity of exhaled gas to provide the water required for the aqueous reaction. This also extends the shelf life of the device.

The limitations of these devices are as follows:

• Small tidal volumes or low cardiac output states can produce a colour change which is difficult to detect, leading the operator to believe that their endotracheal tube is displaced.
• Ambient light conditions can make it difficult to see the colour change
• These devices, though small and of minimal volume, do still increase ther resistance to airflow and the dead space, increasing the work of breathing
• Strictly speaking, this technique is neither capnometry nor capnography, as the CO₂ is not measured qualitatively, and the graphic representation (the colour change) is in response to pH, not CO₂ concentration directly.
• These devices are cheap (you can bed a box of 24 for $15.99 US) but the drawback is that they are not sophisticated, eg. for example they do not come with audible alarms.
• Prolonged use is impossible for most of the makes and models, as the pH indicator ends up becoming saturated with expired moisture and the colour change stops
• The pH change indicator is highly sensitive to low CO₂ values, which means that potentially it could detect gastric CO₂ and make you think you've intubated the trachea, whereas in fact your tube's in the oesophagus. Punterwall et al (2002) intentionally intubated fourteen oesophaguses (oesophagi?) and gloated over the obvious colour change in five of the patients. "May cause confusion and dangerously delay the detection of an oesophageal intubation", they said.

Exotic methods for CO₂ detection

• Photoacoustic spectroscopy uses the combination of infra-red light and an acoustic detector. When a pulse of IR light shines through the gas mixture, it excites the absorbing molecules (in the case of 4.26 μm, that would be the CO₂). These molecules then vibrate more, collide with neighbouring molecules, and generally increase the pressure in the sample transiently. The light is pulsed, and so during the "dark" period, the molecules relax again and the pressure decreases. If the light is pulsed at regular intervals, there will be a cyclical change in pressure. A sound detector which is sensitive enough can pick up these changes in pressure can therefore be tuned to "listen" to a specific frequency, which (according to Rosencwaig, 1980) is specific for every gas. To be clear, this technique represents something that can be done, but not something that is routinely done, and as far as a few minutes of Googling can reveal, there are no commercially available photoacoustic spectroscopy end-tidal CO₂ monitors designed for human use. 
• Mass spectrometry relies on the concept that charged particles move in different ways depending on their mass and charge when they are subjected to an electromagnetic field. Thus, gas can be aspirated into a vacuum chamber, ionised by an electron beam, and allowed to separate into distinct streams according to mass within a magnetic field. A detector at the end can measure these streams and report the data in a way which allows one to detect the molecules involved. The advantage of this method is its precision. The mass spectrometer will not care whether your molecules are asymmetrical or not, or whether your gas is monoatomic. Theoretically one could measure all of the gases in the expired gas mixture, in case one is really interested in the patient's end-tidal argon concentration. Moreover, truly minuscule gas samples can be accurately measured, making this method ideal for experiments on small animals and preterm infants. The disadvantage is the fact that one usually needs to waft the sample through a vacuum chamber, whereas expired gas is usually under atmospheric pressure. Ingenious methods of overcoming this limitation have yielded to the effortless availability of cheaper IR spectrometers which require minimal fiddling; and even in its heyday this technique never really achieved commercial popularity.
• Raman spectroscopy is an ancient and depreciated variant of spectrometry, which relies on the phenomenon of Raman scattering to detect specific gases. When a gas sample is exposed to a high-intensity light source (eg. a monochromatic laser), the molecules of the gas become excited into unstable energy states. They then rapidly collapse back into a more stable state, emitting the excess energy as light which is the Raman scatter radiation. You can measure this low-intensity radiation, which would then be characteristic for each emitting molecule. Westenkow et al (1989) compared the accuracy of this technique favourably to a mass spectrometer, complaining only that the cooling fan of the argon laser "generates enough noise to be noticed in the operating room". The prototype device they tested weighed 35kg and consumed 8 amps of current at 110V. Over the subsequent decades, the free market killed this technique, as consumers chose smaller lighter devices without military-grade lasers in them.  

Problems with expired CO₂ measurement as a concept

Yes, each specific technique has its various drawbacks, but overall one can say that capnography in general has some limitations:

• The end-tidal CO₂ value is not pathology-specific or diagnostic.  For example, a perfusion deficit may lower EtCO_2, while a ventilation deficit may raise it. Theoretically, you could have a very disturbed total body perfusion and ventilation (peri-arrest, near-apnea) with a relatively normal-looking EtCO₂ value. In short, a normal value may be spurious, and an abnormal value is not diagnostic. 
• The measured value relies on the availability of alveolar gas. If CO₂-rich alveolar gas does not flow past the sensor, you get abnormal low readings and weird waveforms. This can happen if the tidal volumes are very low, or if the expiratory flow is minimal. 
• Bias flow can dilute the sample. If there is a vigorous flow through the circuit (eg. where the ventilator does not have a solenoid expiratory PEEP valve and instead relies on some sort of crude mechanical barrier for generating PEEP), even a mainstream capnometer module may return an inappropriately low value because there will be enough fresh gas mixing to dilute the expired gas.
• False-positive CO₂ measurements can occur;  for example, in oesophageal intubations as mentioned above. For some reason, many authors and textbooks make reference to situations where the expired capnometer is confused because "the patient ingested carbonated beverages" shortly before an intubation attempt, which makes one think of various sorts of horrific frathouse scenarios. However, intestinal and gastric gases have no need for any help from beer, as their CO₂ content is already heroically high (it usually exceeds the PCO₂ of venous blood). If one were to accidentally intubate the rectum instead of the trachea, one would be greeted by a gas mixture with 5-30% CO₂ content by volume, depending on the diet.  In short, the respiratory tract is not unique in being able to produce CO₂-rich gas.