An introduction: What is the BP reading?

July 2, 2015 in abstract

This is a blog about abstracting in medicine. How we summarise. What do we mean and what do we actually know. It is all too often that we say things, the meaning of which we have forgotten.

When we listen to a patient we are abstracting from what they say, collating the salient points. We can leave things out and we can give undue importance to parts of their speech ( perhaps). Summarising a patient’s history is an art in abstracting information.

We also need to impart information. To do this we need to first learn facts from the “evidence base”. How good is the data in this “ evidence base”? Is the data honest and does it reflect a true representation of the disease process in humans? We draw information in , process it and then we have to present it to our patient. We usually do this in a very few words. This is abstracting too.

I will go as far as to say that the essential art of being a doctor is about how you abstract information, use it and manipulate it. It therefore goes without saying that we need to be very aware of how we record and relay information.

So, to start with ( we have to start somewhere) I will talk about some simple concepts that we use every day. I will start off with a common scenario : patient goes to her GP and has her BP measured. What could be simpler than that?

Let’s talk about Alice’s blood pressure reading.

Alice goes to her GP in Aberdeen and has her blood pressure measured. Her doctor uses an automatic blood pressure machine. Her GP tells her that her systolic BP is 140 mmHg (  for simplicity’s sake I wont deal with the diastolic readings).

Q : Is Alice’s BP 140mmHg?

A : No, it isn’t.

Alice’s friend Bob, goes to see his GP in Birmingham. His GP measures his BP as well, using the same make of machine that Alice’s GP used in Aberdeen ( I shall call this machine, machine X). His GP tells him that his BP is 140mmHg.

Q : Is Bob’s BP the same as Alice’s?

A : No, it isn’t.

Alice then goes on holiday to Cardiff, where another GP takes her BP again. He uses the same machine ( machine X) that her GP used in Aberdeen. Her BP is 140mmHg.

Q : Is Alice’s BP the same in Cardiff that it was in Aberdeen?

A : Yes, it is.

Alice then goes to Daventry where another GP takes her BP, but this time he uses a different type of automatic machine, Machine Y. Her BP , she is told, is 140mmHg.

Q : Is Alice’s BP in Daventry the same as it was in Cardiff and Aberdeen?

A : No, it isn’t.


Ok, by now I think you will be thinking that I am not thinking straight, what is going on here?

The problem lies with the usage of oscillometric BP machines. These are very user friendly, and can be used by anyone. This is one of their main marketing strengths. They have opened up the field of BP taking to the whole population. But there are problems with them. To find out what these are we need to look a little deeper into how these machines are made and how they work.

A blood pressure monitor is constructed to pass the AAMI SP10 protocol. if its measurements error has a mean value of less than 5 mmHG with a SD of no more than 8 mmHG then it will pass. Manufacturers these days also want to pass the BHS protocols ( click here) which grade devices into 4 different categories. For ease of explanation I am using the AAMI standard in this discussion

So, with oscillometric devices, using the AAMI protocol, 68% of the BP readings have to be within 13mmHg ( 5+8) of the actual BP, and 95% have to be within 21mmHg ( 5 + 8+8) of the real reading.

So, if Alice’s BP reading is 140mmhg, as taken with an oscillometric machine, her real BP ( as measured with a mercury sphygmomanometer) is anything between 119 and 161 mmHg, 95% of the time. Hence it is very much more likely that Alice’s BP is anything other than 140mmHg, rather than being 140mmHg.

Let’s take a look at how oscillometric machines are made, to give us an idea of how this variation in readings occurs.

Oscilloemtric machines record the amplitude of the “oscillations” in the cuff, and example of which is shown here ( Ref 1)


The top graph shows how the pressure in the cuff decreases, and the middle graph shows the corresponding “oscillations” recorded as a result, from the cuff. The amplitude of these oscillations are plotted onto  the lower graph.

The problem that the manufacturer now has, is to work out where on this lower graph you place the points for where the systolic and diastolic BP measurements are read.

This takes us into a rather deeper place, as this information is not easily available. But, not to fear, here is how they do it.

Below is a graph from a paper by Sorjova ( ref 2) and it relates to the personalised oscillometric data needed to build and calibrate a machine for a specific person, in this case a  66 year old woman who was 172 cm tall, weighed 75.2 kg and who had an arm circumference of 33cm, using an arm cuff of width 13cm.


So, lets take a little time explaining this graph. This graph shows how you can take an oscillometric device and calibrate it specifically for one person.  The plot on the right is the raw plot of the amplitude of the oscillations versus the actual cuff pressure ( the pressure in the cuff). This raw data is however not good enough to measure this person’s BP though. The maximum of the amplitude curve is the point at which MAP occurs ( mean arterial pressure) yet this maximum is not aligned with the actual MAP of this person. This is where the left hand plot comes in. The left hand plot is the plot where the peak of the amplitude curve is aligned with this persons MAP. To achieve this the data has to be adjusted by a “ scaling multiplier” of 0.85, to shift the plot leftwards. ( a little detail you may not need to know is that it is not precisely the maximum of the amplitude plot that is aligned with the MAP, is the 95% point of the maximum that is aligned).

To calibrate for systolic and diastolic BP you first need to know these for this person. You then read off the corresponding points of the curve and obtain the ratios ( of the maximum oscillation).

You now have the 3 important factors that have enabled you to specifically calibrate this oscillometric BP machine to a particular person:

  • the sliding multiplier factor,
  • the systolic coefficient and the
  • diastolic coefficient ratios.

With these 3 factors you can happily wear this oscillometric machine and it will faithfully record your BP variations accurately.

Now, and here is the point about these machines. They are never, commercially, calibrated to any individual. I don’t know why they aren’t. I would be more than happy to adjust a machine’s parameters to any person who wanted to have their ambulatory BP measured at home. All I would have to do is to take their BP auscultatorily, work out their MAP, and set their machine.

This doesn’t happen. What happens is that the manufacturer calibrates the machines with an average scaling multiplier factor, and average systolic and diastolic coefficients, for the population that he wants the machine to cover.

Take a look at the plot for amplitude and you will see that the diastolic coefficient ratio ( CR) and systolic coefficient ratios are 32 and 49 respectively ( 32% and 49% of the maximum amplitude achieved). These 2 ratios are specifically for this individual person( and interestingly are much lower than what is reported to be in normal usage). They will be different for each individual. These two coefficients are dependent upon the pulse pressure for any individual. This is the reason why oscillometric machines cannot be used on patients with abnormal pulse pressures ( you will note that each machine is designated for a particular type of person, usually people with essential hypertension). If you have a valvular abnormality, aortic stenosis or incompetence, or if you are in heart failure, or if you are unwell with volume loss, then the algorithms designated for that particular machine will be wrong.

It is for this reason that oscillometric machines are not recommended for pregnant women ( high vascular volume and effectively a large aorto-venous shunt passing through the placenta). They are also not recommended for patients who are diabetic.

Now, lets look at why Alice’s and Bob’s BP. Although they have both been taken with machine X and both have a reading of 140mmHg, it would be a wrong assumption to say that they are identical. Look  at the normal distribution graph.

normal dist

The reading obtained , 140mmHg , is represented by the mean reading. The real BP is within the curve described by the distribution, and there is a 95% chance it will lie within 21mmHg of this ( mean error 5mmHg and the standard deviation of 8 mmHg). Alice’s real BP can be anywhere between 119 and 161 mmHg, and it is unlikely that it will be the same as Bob’s, whose real BP will also lie within these values. Again we can say it is far more likely that they have different BPs rather than having the same BP.

So, what happens when Alice goes to Cardiff, and has her BP taken with the same oscillometric machine, machine X. Here we can say that her BP, taken at 140mmHg, is exactly the same as it was when she first had it taken in Aberdeen. Because it was taken on the same machine, and these machines, using the same algorithm, are very repeatable.

However, when she travels to Daventry, and has her BP measured with machine Y, it can’t be said that her BP is the same as it was when recorded in either Aberdeen or Cardiff. As she has it measured on a different machine. Each machine uses a different algorithm to calculate BP. I have above given a description of the very basic “ amplitude” algorithm used by manufacturers. Each manufacturer uses a different algorithm, tweaked to enable them to achieve “validation” using either the AAMI SP10 standard ( mean error 5 mmHg and SD 8 mmHg) or the BHS standard ( which just frames the data in a different way).

Each manufactuer has a commercially sensitive algorithm, and these are not made available for anyone to contemplate. At the very least they will use variable sliding multipliers and coefficient ratios.

Because these oscillometric machines are repeatable though, they can be used for repeated measurements on any individual, enabling them to be used as ambulatory devices, which show the variability of a person’s blood pressure throughout the day.

So, to finalise my thoughts, here is a Bland- Altman plot of nurse recorded auscultatory BP readings compared with oscilloetric readings using an MAA ( maximum amplitude algorithm) algorithm. ( ref 2)


You can see here how the results are spread out over a large distribution of BP values.

So, to summarise. Oscillometric BP readings are strictly not interchangeable, if they are taken on different models of machines, and not comparable between individuals when taken on the same machine. A set of patient notes usually treats all BP readings the same, and yet there are not.  If a BP has been made with an auscultatory method, then these results are interchangeable and standardised between patients and machines.

So, what ought we to do about this?

Personally, I tend to use a mercury sphyg to diagnose and I take a long time to confirm a BP reading. I correlate it with a oscillometric device and I take great care in accepting home and ambulatory readings. I don't use an oscillometric device in patients with arythmias, or with valvular disease, those who are acutely unwell and those with diabetes or who are pregnant. I take notice when my oscillometric device  yields an unexpectedly low or high result.

What device do I use?

I use an ADC e-sphyg 2. This is really excellent device that I bought from the USA. It has 2 modes, auscultatory and oscillometric, and can easily be switched between the two. It has the useful advantage that you can listen to the korotkoff sounds as it takes an oscillometric reading, which alerts you to any great disparity in readings, and it is easy to calibrate ( in auscultatory mode) against a mercury standard.





Ref 1 Molecular and Quantum Acoustics vol. 26, (2005) 235 ACCURACY COMPARISON OF OSCILLOMETRIC AND ELECTRONIC PALPATION BLOOD PRESSURE MEASURING METHODS USING INTRA-ARTERIAL METHOD AS A REFERENCE . Hannu SORVOJA, Risto MYLLYLÄ, Päivi KÄRJÄ-KOSKENKARI*, Juha , KOSKENKARI**, Mauno LILJA*, Y. Antero KESÄNIEMI*University of Oulu, Department of Electrical and Information Engineering, Optoelectronics and Measurement Techniques Laboratory and Infotech Oulu, PO Box 4500, FIN-90014 University of Oulu, FINLAND

Ref 2 :Sensors 2013, 13(10), 13609-13623:  On Using Maximum a Posteriori Probability Based on a Bayesian Model for Oscillometric Blood Pressure Estimation: Soojeong Lee 1,*, Gwanggil Jeon 2,* and Gangseong Lee 

1 response to An introduction: What is the BP reading?

Leave a reply

You must be logged in to post a comment.

Skip to toolbar