This chapter struggles to wrap itself around the impossibly vast Section G7(iii) of the 2017 CICM Primary Syllabus, which asks the exam candidate to "describe the invasive and non-invasive measurement of blood pressure, including limitations and potential sources of error". In the First Part exam, this has appeared only once, as Question 18 from the first paper of 2018.
The attentive reader will likely have noticed that the rest of the Arterial Blood Pressure Measurement subsection deals with all the various minutiae of invasive blood pressure monitoring, as that is a quintessentially ICUish thing we do here at the ICU. For this reason, this chapter deals with invasive blood pressure measurement only superficially, by means of providing the reader with a directory of links and a summary of points, to make the revision of this complex topic bearable. Non-invasive measurement is the real star here, and most of the chapter will be used for a discussion of the common techniques and their "limitations and potential sources of error".
- Invasive (direct) blood pressure measurement
- Measures blood pressure directly by connecting the bloodstream to a pressure transducer, usually by a column of incompressible fluid (eg. saline)
- Sources of error include:
- Transducer positioning ("levelling") and calibration
- Damping and resonance, and all the things that affect it, for example length of water-filled tubing, air bubbles, clots, and the position of the catheter in the vascular tree
- Non-invasive (indirect) blood pressure measurement
- Relies on a known counterpressure to change the characteristics of downstream blood flow, which can be detected and related to the pressures in the circulation
- Methods of indirect BP measurement:
- Oscillometric (measures MAP, estimates SBP and DBP)
- Auscultatory (measures SBP and DBP, estimates MAP)
- Pulse palpation (measures SBP only)
- Flush (measures SBP only)
- Ultrasound (measures SBP and DBP, estimates MAP)
- Oscillometric measurement:
- The arterial pulse changes the volume of a limb
- This change in volume produces a change in pressure in an encircling cuff
- The cuff is deflated gradually, and the maximum amplitude of the pressure change is recorded as the MAP
- SBP and DBP are then calculated from the MAP using various algorithms
- Auscultatory measurement:
- A cuff is inflated to obliterate distal blood flow, and the distal artery is auscultated
- As the cuff is deflated, blood released into the distal limb makes characteristic sounds, which can be related to the pulse pressure range
- By this means, SBP and DBP can be measured, and MAP is calculated from these values
- Sources of error of NIBP methods
- Incorrect technique (eg. wrong cuff size, deflation speed too fast)
- Interference with measurement (eg. patient movement, AF)
- Unavoidable errors of calculation (i.e. use of equations and constants to calculate derived variables from measured values)
- Limitations of NIBP methods
- Oscillometric measurement overestimates BP in hypotension and underestimates BP in hypertension
- Auscultatory measurement underestimates BP in hypotension, and may be unable to detect BP in low cardiac output states
- Reliability of these methods rests on the correct matching of cuff width and length to the patient's arm size.
For basic scientific principles underlying the very concepts of systolic, diastolic and mean arterial blood pressure, locally some shady unreliable online material is available. There is also a chapter in the Fellowship exam section which summarises the implications of a discrepancy between invasive and non-invasive measurements. In terms of legitimate published peer-reviewed resources, the One Single Thing You Need to Read about this topic would probably have to be the article by Ward & Langton (2007), which is as brief as it can afford to be while covering everything examinable. As for non-invasive measurement specifically, the 2018 article by Meidert & Saugel gives a quick rundown (literally three paragraphs), whereas Sorvoja & Myllyla (2006) give a non-brief rundown over 26 pages, and Geddes' Handbook of Blood Pressure Measurement (2013) slowly and methodically breaks this topic down into component atoms across the whole back half of a 190-page textbook.
In the most stupid of terms, the main difference between these methods of pressure measurement is that one measures pressure directly, i.e. by interfacing its pressure sensors directly with the thing that is being measured (let's call it blood), whereas the other measures the pressure indirectly by observing some transmitted pressure, displaced volume, or behaviour of downstream vessels.
One might have expected the CICM examiners to never ask for anybody to draw a direct comparison between invasive and noninvasive blood pressure measurement, as this assessment item would occupy some lower tier in Bloom's taxonomy. However, their cruelty knows no limit. Question 18 from the first paper of 2018 asked the candidates to compare and contrast invasive arterial pressure measurement with the oscillometric method. As such, trainees should prepare for a future where questions like this are commonplace. In preparation for this new dark age, here is a tabulated answer all ready to go:
|Domain||Invasive (arterial catheter)||Non-invasive (cuff manometer)|
|Physical principles and method||
|Sources of error||
Now, I know what you must be thinking: how could the ability to regurgitate this list be mistaken for a desirable characteristic of an intensive care specialist? Reader, it can. As history has repeatedly demonstrated, the ability to reproduce lists from memory is one of the most essential qualities for a successful exam candidate in medicine, particularly at the early stages of training. Fortunately, examples of good assessment design also exist. One such example is Question 13 in the first fellowship exam paper of 2001, which presented the candidates with a very realistic clinical scenario, were the arterial line and the cuff disagree, and asked them to interpret it on the basis of what they know about the limitations of invasive and non-invasive blood pressure. After multiple attempts to arrange and rearrange this chapter, it seemed most fitting to mention this in the early stages, and then to come back to it after all of the information has been delivered.
Without going into any excessive detail:
Several methods are usually listed together, and their discussion is usually structured in a way which is clearly borrowed from Geddes' Handbook of Blood Pressure Measurement (2013), where the following classic table can be found:
All of these are different variations of the theme of counterpressure, i.e. they generally involve the use of some sort of pressurised cuff which is used to compress a branch of the arterial circulation (classically, the brachial artery, though more interesting vascular structures have been used). The methods, in brief, are:
You'd have to make the argument that for completeness an ICU trainee should know something about all of these methods of blood pressure measurement, but that realistically only the auscultatory and oscillometric methods are ever likely to be in any sort of exam, as it would be insane to ask anything detailed about the others.
This is a quintessential "counterpressure" measurement technique, as it involves the application of some known external counterpressure, and its adjustment to achieve some sort of response from the patient's circulatory system. It was asked about in Question 18 from the first paper of 2018, where the college examiners wanted some sort of comparison between this NIBP method and the invasive method. Geddes (Geddes!) published on this in 1998, and it is absolutely worth reading, even though it has no relevance to exams. Interestingly, even this mossy cornerstone of haemodynamic literature does not contain a clearly articulated statement of the physical principles involved. At risk of sounding as if he knows something about those principles, the author has himself made an attempt to synthesise such a statement, but the trainees are reminded to use their own judgment in their use of such unofficial material for exam purposes.
In short, the main point is this:
Etienne Jules Marey, somewhere around 1876, performed such a measurement for the first time by shoving his assistant's hand into a jar filled with water, and measuring the pressure in that water, observing that the arterial pulse was transmitted to the water and was recordable. Here's a picture of this early experiment from the Geddes paper, which is presumably one of the original pieces of art from Marey, but which is not otherwise referenced.
In case you're wondering, that big cylindrical thing on the right is a smoked drum kymograph. It was covered in smoked paper and made to rotate with a known velocity (the popular models had gear and clutch mechanisms capable of eight different speeds); as the manometer pressure changed, the needle scratched a pale line in the smoked paper, creating some of the first tracings of the arterial pulse.
Sans kymograph, this concept can be represented as follows:
As Marey increased pressure in the water jar, the pulsations decreased, and then completely disappeared, with the assistant's hand becoming "blanched" as all blood was driven out of the hand. From this, Marey was able to conclude that the pressure in the water chamber exact point where pulsation had disappeared was the systolic blood pressure.
Even more interestingly, Marey found that the amplitude of the pulse waveform increased as the counterpressure increased, before it was completely obliterated and the hand was blanched. He graphed the amplitide of the waves against the pressure being tested and produced this graph (again, stolen shamelessly from Geddes), which is generally viewed as the origin of oscillometric measurement:
From this foundation, we have the modern automated method of oscillometric blood pressure measurement, which has become refined to the point of excellent reliability and user-friendliness, such that cheap domestic devices can be deployed and laypersons trained to use them. The best explanation of how these modern devices work is probably the article by Lewis (2019). Their function can be described as follows:
In theory, that sounds pretty solid, but in practice there is actually no distinct transition above the systolic or below the diastolic, i.e. the overinflated or underinflated cuff will still measure some oscillations. This severely doctored recording from Lim et al (2015) illustrates the point:
As you can see, everything is pulsating, and all the time. Looking at the cuff pulse amplitude recording, you really couldn't intuitively tell where the systolic and the diastolic pressures were. The only thing you can be directly sure of is the maximum amplitude (as that's not subjective), which corresponds pretty well with the mean arterial pressure (Posey et al, 1969).
That's not bad, as the MAP is a rather important parameter to accurately measure, but the systolic and diastolic also have their relevance, so it would be nice to have at least some estimate. That is exactly what these automated cuffs tend to do: produce "some estimate" of the systolic and diastolic pressure. Each device manufacturer seems to have slightly different proprietary algorithms which are used to calculate these variables, and it is probably unnecessary to go into these in any great detail (but if that is what you want, Chandrasekhar et al from 2016 will have you covered). For CICM exam purposes, it will suffice to know that this means each device will produce a slightly different measurement from the same oscillometric data. These proprietary algorithms range from brutally simple (a percentage of the maximum amplitude) to highly sophisticated ones (where the "envelope" of amplitude waveform distributions is mapped and the area beneath it is integrated, etc). The choice of method or its mathematical complexity does not seem to matter in the slightest, as all of these devices seem to have the same margin of error as compared to other methods. Wong et al (2006) found an error of something around ± 5mmHg comparing the automated cuffs to the mercury sphygmo, and Bur et al (2003) found ± 12 mmHg error when comparing them to the invasive arterial monitors.
The technique of measuring blood pressure using the automatec oscillometric cuff device obviously owes its accuracy to the cuff and the device. Moreover, in any scenario where clinical data is generated by an algorithm, the calculation also becomes a source of error, as whatever coefficients are used to multiply the measured numbers can magnify the error. To summarise, the following factors can become sources of error in the automated oscillometric measurement of blood pressure:
Of these, the cuff thing is probably the biggest issue. Bur et al (2003) found that fiddling with the algorithm is probably not going to achieve any appreciable increases in accuracy unless the cuff is well matched to the upper arm circumference.
Nikolai Korotkoff, who published his description of this method in 1905, was originally listening to the sounds made by a traumatic aneurysm (he was a surgeon in the Russo-Japanese war of 1904-1905). He was able to describe the use of this method in healthy arteries before dying of tuberculosis in 1920. At its most basic level, the method requires for an upstream artery to be occluded, and for the occlusion to be gradually released as the auscultator listens for The Sounds. Those are, to borrow from Geddes verbatim:
- Phase I_"a loud clear-cut snapping tone."
- Phase II_"a succession of murmurs."
- Phase III-"the disappearance of the murmurs and the appearance of a tone resembling to a degree the first phase but less well marked."
- Phase IV- [the tone] "becomes less clear in quality or dull."
- Phase V-"the disappearance of all sounds."
Again from Geddes, the relative intensity of the Korotkoff sounds and their position in the sequence of events can be summarised by this diagram:
This method was so simple and so easy to deploy in the clinic that it became an immediate favourite, and various official societies adopted it remarkably quickly (thirty years later, which is quick for the AHA). However, almost immediately as it was described, people started debating the physiological meaning and origin of those sounds. Where do they come from, and how do we know that they are related to blood pressure?
After about 100 years, we still have no idea. Babbs (2015) thought they were produced by the oscillation of the arterial walls and derived a purely mathematical explanation based on wall motion mechanics. Drzewiecki et al (1989) created an alternative model which attributes the sounds to fluid turbulence, a change in the frequency of the normal pulse pressure which ends up being lifted into the audible range by the change of flow characteristics in the collapsed segment of the brachial artery. Nobody is particularly convinced by these models, and neither are widely accepted. It is safe to surmise that nobody will ever be asked about this in any sort of exam scenario, as even experts tend to stammer and babble when questioned on this subject. Probably the more important aspect here would be an appreciation of just how flimsy it is, in terms of accuracy and reliability.
There's obviously some serious limitation to any method of measurement which relies on the human senses to directly detect some physiological variable. For one, your doctor might be going deaf from old age. The ambient environment might be noisy. The stethoscope might be of a poor quality. In short, there are numerous reasons to mistrust the measurement. The following listed sources of error were extracted from the excellent paper by Kallioinen et al (2017), who performed a systematic review of studies quantifying BP measurement inaccuracy.
Perhaps revealing something about his local environment, the author will recall numerous episodes from his own practice where, confronted with some preposterously low or high blood pressure measurement on the automated cuff, the nursing staff will dust off an old sphygmomanometer and perform a manual auscultatory measurement amid the din and bustle of a medical emergency response. To most people, this might give the impression that auscultatory measurements are viewed as more accurate than automated oscillometric measurements. That impression is also clearly shared by people who write guidelines, as they make statements such as this:
"If the automated blood pressure reading is outside the patient’s usual range, in the yellow or red zone of the standard adult general observation chart, then a manual reading should be obtained."
- SESLHD protocol, 2020
At face value, basic logic supports the idea of checking the accuracy of one measurement by using a different device and method to collect the same measurement. That's a sensible rationale for doing a manual blood pressure when the automated measurement throws you some silly-sounding values. But is the auscultatory method more accurate? Is it somehow closer to the One True Pressure?
In short, no. For standard non-critically-ill patients out in the community, the two methods are usually in close agreement. Landgraf et al (2010) found a discrepancy of only around ± 5-7 mmHg between them (the manual was usually higher). So, for a normal stable patient, the difference is probably not going to be eyebrow-raising. It's going to be 120/80 or 125/85, or 115/75. Still, there is a discrepancy, and systematic reviews of measurement techniques (eg. Skirton, 2011) generally conclude that there is a small error associated with noninvasive measurement.
With worsening haemodynamic derangement, the oscillatory measurement is thought to become less accurate than invasive measurement. Caramella et al (1985) found that the oscillometric devices became progressively less accurate with MAP values below 70 mmHg. These were similar findings to Meidert et al (2020), who found that in shocked ED patients oscillometric MAP was always higher than invasive MAP with a mean difference of 13 mmHg (and a range of ± 15 mmHg, i.e. your MAP could be off by 27 mmHg). "In 64% of readings, values obtained by the upper arm cuff were not able to detect hypotension", the authors complained.
So, the oscillometric method of measurement might become inaccurate with shock, and tends to overestimate the blood pressure. What about the auscultation method? Turns out, it's also useless. Cohn (1967) performed a study on shocked patients and determined that this method generally underestimated the invasive measurements, often by a massive margin (33 mmHg). The auscultation of dulled, feeble Korotkoff sounds suggested that both the systolic and diastolic were much lower than they really were (according to the arterial line), and in some of the patients the manual blood pressure was actually unrecordable while the invasive trace gave values around 100/40. The authors surmised that this may be because of increased peripheral vascular resistance in shock, and proved this hypothesis by reproducing the same results in healthy volunteers - the infusion of vasoconstrictor drugs into their brachial arteries produced a premature loss of Korotkoff sounds and increased the discrepancy between the different measurement methods. Similarly, in a group of hypertensive subjects, Nielsen et al (1983) demonstrated a similar effect.
In general, authors (Perloff et al, 1993; Ribezza et al, 2014) recommend against relying on manual measurements (or noninvasive measurements in general) for patients with shock. In this group, all the indirect methods of blood pressure measurement generate values which are unacceptably far from the truth. This is even more important if you consider that an error of 10 mmHg is going to have greater clinical relevance in the 60-80 mmHg blood pressure range than in the 100-120 blood pressure range. In short, an entire morning of looking for evidence in support of a manual blood pressure "double-check" in shock did not produce a single study. The only logical reason is the real need to average multiple repeat measurements from several devices, which does help reduce measurement error.
So, that brings us to the question: if there is a discrepancy, and you need to titrate your management to a blood pressure endpoint, which blood pressure should you choose?
Let's say you have a patient with both an arterial line and a blood pressure cuff. Let's say these devices are giving you completely different blood pressure measurements. Which of them is wrong? Which is the least wrong? As mentioned above, this has come up as a Fellowship question (Question 13 from the first paper of 2001). This was a grim dark age before CICM even had a primary exam, which probably means that this (very basic science-y) question will probably reappear in the First Part papers, if it is ever seen again in the future. Knowing what we know about the bottomless darkness that resides in the hearts of CICM examiners, it would be wise to expect the worst.
So, which measurement is most accurate? On one hand, you might point to the wide discrepancies between invasive and non-invasive measurements of blood pressure, and complain that their reproducibility is poor. Because of the way they work, repeated non-invasive measurements of blood pressure will necessarily return a range of different values, even when the conditions are stable. This already makes you suspicious of their accuracy, even though within the normal blood pressure range they tend to agree between themselves and with invasive measurements. Moreover, as the patient becomes more unstable, the accuracy of noninvasive techniques will decrease, making them less trustworthy.
On the other hand, one might point out that invasive monitoring is not without its caveats. Specifically, the measured pressure depends on the site of measurement; as discussed in the chapter on normal arterial line waveforms, the values generated by invasive arterial pressure monitoring are subject to variation depending on how far distally the catheter is inserted. To borrow and modify a classic diagram from an ancient copy of Gedde's Handbook of Blood Pressure Measurement (1981):
So as you can see, all the values (systolic, diastolic, mean) are going to be different, depending on where you put the probe. To ask which of these values is more "accurate" is meaningless, as all of them are presumably measured accurately within the limits of the transducer device and recording software. In other words, those are the actual pressure measurements from those arteries. The real question is, which of these pressures is the most important one for clinical use. You, standing there at the noradrenaline pump, will want to know whether to increase or decrease the infusion rate.
Even though most people in positions of authority seem to take the stance that aortic root pressure is the most "important" pressure, this is still a rather difficult question to answer. The actual clinical importance of the aortic root pressure values is open for debate, as the aortic root pressure is not the pressure that perfuses your renal medulla, or your cerebral hemispheres, for example. However the fact remains that pressure at the aortic arch and carotid glomus (which are basically the same) is what your vasomotor autoregulatory centres monitor, and so it has at least physiological importance. The same cannot be said about the radial artery: your body does not care overmuch about the pressure there, and no important systemic cardiovascular decisions are made on the basis of it. Most patients in ICU end up with a radial arterial line not because that site is of clinical importance, but because it is safe and convenient. Wherever the aortic root is accessed (eg. where there is an IABP), aortic pressure measurements are preferred (Knippa, 2019).
Following from this, pressure which is collected closer to the aortic arch is more meaningful and "truer" than distal pressure, even if the proximal pressure is measured non-invasively. From Wax et al (2011):
"When a discrepancy between NIBP and ABP is seen, a practitioner may question which of the two is the “real” pressure upon which clinical decisions should be based. Because maintaining adequate blood pressure at vital organs is the usual goal of therapy, central pressure should probably be of more interest than peripheral pressure. ...Thus, brachial pressure measured by NIBP cuff may be a better measure of central pressure."
But, as we have already discussed, with worsening hypotension both the oscillometric and the auscultatory methods become increasingly less and less accurate. So, which measurement will you instruct the staff to go from? The following take-home message can be distilled from everything discussed so far:
This simple series of statements can be represented by a graph, probably with no additional educational benefit: