Slashdot Mirror
Home Blood Pressure Monitors Are Wrong 70 Percent of the Time, Says Study (arstechnica.com)
An anonymous reader quotes a report from Ars Technica: In a study out this week, about 70 percent of home blood-pressure devices tested were off by 5 mmHg or more. That's enough to throw off clinical decisions, such as stopping or starting medication. Nearly 30 percent were off by 10 mmHg or more, including many devices that had been validated by regulatory agencies. The findings, published in The American Journal of Hypertension, suggest that consumers should be cautious about picking out and using such devices -- and device manufacturers need to step up their game. Lead author Raj Padwal and his colleagues set out to test the accuracy of the devices themselves. Funded by the University of Alberta Hospital Foundation, they compared the home blood-pressure monitors of 85 patients with a gold-standard blood-pressure measurement technique. The patients' monitors varied by type, age, and validation-status. But they all used an automated oscillometric method, which measures oscillations in the brachial artery and uses an algorithm to calculate blood pressure. The gold-standard method was the old-school auscultatory method, which involves the arm-squeezing sphygmomanometer and a clinician listening for thumps with a stethoscope. Of the 85 home devices, 59 were inaccurate by 5 mmHg or more in either their systolic (the top number that's the maximum pressure of a heart beat) or diastolic (the bottom number that's the minimum between-beat pressure). That's 69 percent inaccurate. Of those, 25 (or 29 percent) were off by 10 mmHg or more. And six devices (seven percent) were off by 15 mmHg or more.
The ones done by doctors are off as well, it's called 'white coat hypertension'.
They might prescribe medication just because you're afraid of him.
http://www.bloodpressureuk.org...
Hmm, normal blood pressure is lt. 120mm Hg / lt. 80mm Hg.
So, we're talking a 4% error, which will only matter if you're within that 4% of a breakpoint.
Can't see this as a really big deal. I expect my blood glucose widget has a similar (if not larger) error factor, which might mean that I take an extra unit of Humalog (or take a unit less than I should) from time to time. Won't kill me....
"I do not agree with what you say, but I will defend to the death your right to say it"
I told my doctor that my blood pressure was higher when I lean forward than if I am reclined back.
He didn't believe me until he took the measurements in his office.
Turns out body position makes a statistical difference in your blood pressure readings:
https://www.ncbi.nlm.nih.gov/p...
I took both of my BP meters in to the doctor's office and compared their readings with the doctor's readings. That way I know how much deviation to expect when taking measurements.
Home Blood Pressure Monitors Are Wrong 70 Percent of the Time, Says Study
So, take 10 measurements, and look for the 3 identical ones.
Why don't they give us the most accurate devices?
They don't even put the data anywhere, so we can't even figure it out for ourselves.
It's 2017 people.
For the typical Comic Book Guy-esque behemoth that is the typical Slashdotter.
Being +/- 5mm Hg on a measurement of 140 mm is +/- 3.5%, but how accurate are the thresholds for treatment? A guideline may say that 140mm Hg needs treatment, but what is the error bar in that value? I bet it far exceeds 3.5% because of individual variation.
And more than accuracy, I'd think that being precise would be more useful to a home user -- knowing that your BP increased by 7mm over a year sounds more useful than knowing that it's close to some arbitrary value.
My home blood pressure meter was more consistent than most of the doctor measurements I have received.
I've been using a home blood pressure monitor kit to monitor my bp for two years. Been working real hard to bring my high blood pressure down. Recently felt success when my doc told me it was significantly reduced.
Now i don't know what to believe anymore
I'm a nursing aid in a hospital, we use such machines for measuring BP. I've not seen the documentation myself, but have been told that they're calibrated to be within +/- 8mmHg... which is more than the 5 for these home machines. Is 5mmHg really so alarming if a *hospital* doesn't mind an error of +/- 8?
The bigger question though, and why this research may be utterly meaningless, is: How did they do the comparisons? Per the article, they used one method (either the standard or the home machine) and followed with the other at 30-60 seconds. The problem is that BP can change by more than 5-8mmHg within a few seconds. Anyone who knows anything about hemodynamics should know this. Has anyone conducted an experiment using an artificial 'arm', with a constant pressure pulsing through it? That would be a much better way to determine relative accuracy.
https://wire.ama-assn.org/delivering-care/how-get-most-accurate-blood-pressure-measurement
What's nice is the AMA even tells you how much each nurse practitioner error can change the reading. Yes there are a lot of inaccurate home devices ... but I'd be shocked if most were off by the more than 100mmHg that a careless NP can cause when using more accurate equipment.
As your doctor gets older your blood pressure will go down because your doctor's hearing will get worse. A good automated machine should be able to do better than a human.
This is my world.
The wrist ones are crap. Extrapolate to the finger one yourself...
The GOLD STANDARD is only calibrated to within +/- 2mm.....ASSUMING they used a mercury unit, they did NOT specify this it seems.
An typical manual aneroid sphyg is rarely maintained to even that level. They give no detail on their base unit, it could be off.
Being off a few is normal.
9 readings in 45 min is using the bare MINIMUM time between readings. Some people recover faster than others...
Even doctors arent consistant. from TFA:" If they disagreed on a measurement by more than 4 mmHg, they tossed the data and took the measurement again. "
Just 2 doctors taking it can be off that much.
** Use BP readings from one machine by one person and look for changes. Use a good doctor with a nice mercury unit for real number (such as it is) **
I have been repairing BP units as a side job for almost 40 years now!
Human blood pressure should be measured in centimeters of mercury, not millimeters. Saying a patient's blood pressure is say 135/78 is misleading, since blood pressure is DYNAMIC and continuously changes MOMENT TO MOMENT. Whoever decided it should be measured this way had either very limited knowledge of basic human physiology or a grave misunderstanding of how significant figures work.
When you have an event like a concert, people say "28,000 people attended" or whatever. They don't generally claim 28,351 people attended because the one implies that you know down to a single person how many were in attendance. The five, in the absence of the one, implies that the attendance is known to within +/-5 people, and the three means it is known to +/-50, if all trailing digits after that are zero, (and by convention, there's no decimal point used as a marker, such as 300. mg. That decimal implies that all the zeros to the left of that decimal point are known to be zeros).
So... the number of decimal places implies to what degree the number is known.
28351 means exactly that many were there, 28351 to a man.
28350 means 28345 X = 28355 ppl
28300 means 28250 X = 28350 ppl
28000 means 27500 X = 28500 ppl
& 30000 means 25000 X=35000 ppl. Note that each of these ranges includes the value 28351, the actual number who were there.
Similarly, if you asked how many people are at a festival, someone MIGHT be able to say at any given moment what the exact population there is, but as people are constantly leaving while others (and not generally in a one-to-one ratio or basis, any exact number given for any given time might be true, precise AND accurate, AT THAT MOMENT but could EASILY not be a good representation of attendance at any other time or during any arbitrarily chosen long period of time.
The same is true with blood pressure. (And I speak as a former Army medic who has taken a LOT of blood pressure readings in my day, and yeah, I knew what I was doing and I know what I'm talking about.
Even if you measure a patient's blood pressure perfectly, down to a nanometer of mercury (equivalent weight,) in a few seconds it'll change enough for your reading to be not just no longer true, but laughably inaccurate and worthless. Measureing like that would only be effective in watching for trends and detecting blood pressure at either extreme and you don't generally need to know (because the information is fundamentally worthless,) causing alarm where none's warranted.
We should measure BP in cmHg, rather than mmHg. This, a patient's blood pressure would be listed as 14/8 cmHg, and it would be true and accurate and useful in a way mmHg readings aren't.
If I may be permitted one more illustrative analogy... imagine if your car'sor motorcycle's speedometer doesn't tell you feet per hour, or even inches per hour which would be even more hilariously egregious.
"Do you know how fast you were going, sir?"
"Um... 343,200 feet per hour, officer?"
"Well, sir, I clocked you at 343,315 fph, and the speed limit is 316,800 fph. I'm going to have to issue you a citation."
"Ah man... how much is the fine?"
"Pretty steep, considering this is a construction area... it's going to cost you 22,000 cents, unless you wish to contest it in court, in which case follow the instructions on the back, to indicate your intentions within the next 259,200 seconds."
See how absurd that is?
Using cmHg would mean less time wasted trying to get it EXACT, when doing so is, as I've illustrated, POINTLESS!
My doctor got a 170/90 off me from their machine. When tested with a trained nurse and a spigometer, I was 124/80.
I've never had them *underreport* but I've had them over report many times in my life.
She was like chocolate when she drank... semi-sweet at first and then increasingly bitter.
Why not say 100% MAY be inaccurate. For that matter, maybe NONE are inaccurate. Maybe....
If the FDA is not requiring these devices to be accurate and precise then could not some private entity offer validation services of their own? I'd think a seal of approval for accuracy would be worth something for both makers and consumers, that can translate to real money.
Think of the Underwriters Laboratory, it is a private company that tests and approves electrical devices for safety. People that try to sell something not approved by them will have problems. Stores are not likely to even put them on shelves if not UL approved.
I've also seen a bunch of comments on how physicians and drug makers make a lot of money on blood pressure medications and so they are not really interested in an accurate blood pressure meter. I don't agree as I'd think that there is a lot of liability in prescribing medications for those that do not need it.
Let's assume that this is true that since physicians and drug makers are set on handing out meds for money then would not market forces work in offering alternatives? Do not insurance companies play a role? It's rare for people to pay for medical care out of pocket. The insurance companies see a payout for high blood pressure as a cost. I'd think that it would be in their best interest to make sure that those that truly have high blood pressure have it treated and those that don't are not given medications that can have harmful side effects.
I think that if this a real problem then some insurance companies should get together and make sure these meters work like they should. I would think that fixing the meters could save them a lot of money in payouts for medications and care. It would also have the side effect of a healthy public. If appealing to profit doesn't work then appealing to principles should.
I am armed because I am free. I am free because I am armed.
I don't know if this has anything to do with the Ars article but here's a comparison study comparing oscillometry and auscultation for BP measurements.
https://bmccardiovascdisord.biomedcentral.com/articles/10.1186/s12872-017-0521-6
From the conclusion, in part; "the systolic BP value measured by oscillometry may be 27 mmHg higher or 17 mmHg lower for patients with SR [sinus rhythm] and 24 mmHg higher or 12 mmHg lower above the BP values measured by auscultation for patients in AF [atrial fibrillation]. The wide range of random error rates is a questionable topic in clinical practice, as it could possibly affect the treatment of arterial hypertension in patients with AF.
Ouch, that's pretty significant. Also, as an aside, the measurements were all done with the patients supine. I wonder how much of an effect that has with measurements taken in a doctor's office?
godel_56 with mod points
It's less than 5% in a system that easily varies 20% in the course of an hour, and than vary nearly 50% in moments under stress. It's also easily affected by flexing the arm sitting in a different chair.
They compared home machines with gold standard. They should have also done doctor's office machine with gold standard. Decisions based on home measurements rarely take into account 5mmHg on the other hand, at doctor's office, decisions are based on 5mmHg.
on a 5mmHg difference in individual measurements reported by a patient. Your circadian rhythm double amplitude in sys and dia are each more than that and most people don't even factor those in.
As a former EMT-Cardiac, and having worked in a number of emergency departments, I can say that the blood pressure obtained by many clinicians is off by more than that, and I'll explain why. When using a sphygmomanometer and auscultating for blood to begin flowing through the veins, on the systolic (the first number / highest pressure value) you will only hear the sound of the blood flow on the heart beat. So the rate in which you are letting air out of the cuff determines the accuracy, and further, the slower the patient's heartrate, the greater the inaccuracy will be.
So let's say they are letting air out of the cuff at 20 mmHg per second (thus from full inflation at 200 mmHg to a normal diastolic of 70 it would be 130 mmHg = 6.5 seconds), and a patient's heartrate is 60 beats per minute. The heart is beating once each second and the needle is moving 20 mmHg per second, thus the number they see when the heart beats could be as much as 20 mmHg lower than the actual blood pressure. I'm sure you have had nurses take your blood pressure and they took way less than 6.5 seconds to measure it - in that case the error margin would be even greater.
For the systolic value the inaccuracy will be a lower value than actual, and for the diastolic the inaccuracy will result in a higher value than actual.
Better known as 318230.
There has been a lot of news about scientific articles being bunk.
Is this a good, de-bunking article, or a bad article with questionable science?
They did not publish their raw results, only statistics for strange combinations of the data.
They did say they excluded one patient because the reference measurements were not consistent between two observers.
That seems like bias against the automatic measurements.
I think from the data they published, it's impossible to tell if the article is fud or good information.
It's a sad state of affairs when an article does not provide enough information to understand what they did.
It smells of picking a how to present the statistics to support something to make the news and hiding the raw data to prevent actual understanding of what is happening in these machine and human measurements.
And told me to get a home unit, and bring it in to test it. His assessment differed from this in that he said that "most home blood pressure monitors are accurate". I brought in mine, the nurse tested it, and said it was accurate. The readings weren't 100% the same as what she got, but then they normally vary second to second anyhow.
Also of note is that she was much more careful with the test when testing my unit as opposed to normal. For both their and my unit she had me sit quiet and still, she made sure to place the cuff in the same place, and she took the reading slowly on their manual unit. On a normal physical she places the cuff over my shirt and and drops the pressure quite fast.
Now I would guess this is because they aren't that worried. My BP is normal to the high end of normal, but is normal, when measured at home and is on the high end of normal up to just at the bottom of the pre-hypertension range at the office. So I would guess she's not that concerned with having it right down to the mmHg, with in 10 is probably good enough. If it hasn't changed much since last time, no need to worry and no need to spend a bunch of time being super precise.
Now maybe my doctor is just lax and stupid, but he doesn't seem that way (and his background.credentials don't indicate that). However maybe this journal has a bit of a bias in wanting to over-diagnose hypertension and/or push that physician measurements are the One True Way(tm).
To me it seems silly to worry about 5% or less error on a test like this. The fact that BP ranges neatly line up on clear decimal lines should tell you that the specific numbers are guidelines only, not maxims. It isn't like the did some measurements and said "My god at precisely 140mmHg blood pressure becomes unhealthy and at precisely 120mmHg it becomes a complete non-factor!" Of course not, rather based on medical knowledge they established the normal, pre-hypertension, hypertension, and hypertensive crisis ranges and set them along base 10 boundaries because we like that.
It is a guide to trained professionals, not a stress point past which there is a sudden failure. Your doctor isn't going to treat it radically different if your BP is 141/91 vs 137/89. They'll evaluate what kind of treatment (if any) they think you should have based on a number of factors about you.
Just who makes clinical decisions on a 10mmHg spot measurement?
a) home blood pressure readings aren't used to make clinical decisions.
b) doctor blood pressure readings aren't used in isolation to make clinical decisions.
c) one off blood pressure readings aren't used to make clinical decisions.
d) doctors manually taking readings are likely to be off by more than 10mmHg.
e) depending on the time of the day your blood pressure readings are likely to be off by more than 10mmHg.
f) depending on which arm you take the reading from will affect your by 3-10mmHg.
g) depending on how long you've been sitting in the chair at the doctors office will affect your reading by 3-10mmHg
I still remember my last doctors visit. The doctor looked shocked and said I have hypertension. Then he told me to sit and relax a for a while. We did some other checks, then back to blood pressure. Well I fell 11mmHg down to pre-hypertension. He said to come back 3 times a week for the next 2 weeks preferably at the same time of day. Final diagnosis: Bloodpressure was normal.
Has anyone actually read the AJH article in question:
"In summary, although the mean BP differences between home BP monitors and auscultation were within 5 mm Hg, over two-thirds of devices tested exhibited a systolic or diastolic BP difference of 5 mm Hg, a degree of BP difference considered to be clinically important."
So, in average most of the tested BP monitors were off by less than 5 m Hg.
Nuff said ...
we-go-we-fly
I take my BP meter into the doctor's office, I let them measure my BP with their gear and then I use mine to take a reading for comparison. The readings are (usually) very close.
Note that your BP can vary considerably over the course of just a few minutes depending on your environment and how you're reacting to it. Stress, eating, movement, etc can make the results vary significantly.
Never use any single BP reading as an "accurate" measurement. You normally need multiple readings over an extended period of time to get any real sense of what your BP actually is.
If you take several readings a day under similar conditions at roughly the same times of the day, then you can start to see what your actual BP is. Infrequent, isolated readings are nearly worthless (unless they show an exceptionally high BP, then there's probably cause for alarm).
Take a sitting reading, then a standing reading. Do this a few times a day and record the results (most meters will do this for you). After a week you'll have a much more realistic and dependable sense of what your actual BP is.
Just cruising through this digital world at 33 1/3 rpm...
5 or 10 points ain't nothing. You've clearly never had high blood pressure issues.
This "ballpark figure" for home use is about knowing when you should contact your doctor because something is well wrong (20, 30, 50 points or more) out of your normal range. Yes, really shitty ones should be banned, but doctors measure their own at the office professionally for clinical decisions.
I've had doctors come in and personally do my high blood pressure multiple times themselves because the nurse's measurements seemed odd.
(-1: Post disagrees with my already-settled worldview) is not a valid mod option.
It's not just Blood Pressure monitors. Blood sugar testing devices also are inaccurate. I can try 3 different machines or the same one several times and each time get a significantly different reading.
I have a few things to say about this and I hope you'll bear with me. This is an issue of importance to me and although I'm not a physician or clinician I have some background and have done a great deal of research on the issue in part because I've developed a free (with no advertisements!) mobile/desktop web-app to try to help people take control of, manage, track and report their BP over time. The app works well and I'm about to roll it out for general use (just testing with friends for now). But to the article... First, the sample was 85 patients. That's a rediculously low number for the study to be of any serious value. Honestly, I'm not even a clinician or researcher and I could get a couple of hundred people in a controlled study pretty quickly! Given that there are millions of inexpensive home BP monitors out there, they should be looking at at least 2,000+ samples for this to be considered "meaningful" IMHO.
... absolutely nothing. It is not only well within the margin of error for *gold-standard* methods but considering the variability of BP due to even the slightest change in physical position it's far below the "noise". Try this: use a calibrated automatic sphygmamomometer and take FOUR readings (the kind found in hospitals - the article is talking about home machines). Then, while the cuff is inflating, tense your body up. Hold your breath and push out. The systolic pressure will rise way, way more than 5mm Hg and likely the diastolic too. Similarly, "white-coat" syndrome or being nervous for any reason will do much the same. Also, keep in mind that the device might give an inaccurate high reading but again, if that's within 5% it is NOT significant no matter what the study concludes. Seriously, no physician does anything different if your avg. systolic is 145 versus 140.
That said, the notion that home machines can be/are very inaccurate is widespread and this study just reinforces a strong bias against inexpensive home machines by many physicians who just don't seem to believe that spending less than a few hundred dollars can get you a machine that can provide meaningful readings.
Well - as others have pointed out - a reading that's off by 5mm Hg for a typical borderline hypertensive patient with a nominal at-rest systolic of 140mm Hg (a value that physicians call "borderline" or "stage 1" hypertension and are only "slightly worried" that the patient has a problem) means an error of about 3.5%. This is
Note that for an *actual* hypertensive patient with an at-rest systolic above 160mm Hg, an error of 5mm Hg low - even 10mm Hg - is unlikely to change the clinical diagnosis of "classic hypertension" at all. Any phyisician presented with a log of consistent 150mm Hg +/- 5% at-rest systolic readings will give the same advice to the patient (though, sadly, the advice is often suspect and poorly understood by the physician) and diagnose them with "classic hypertension".
The fact is that even if your machine is 5% out on average, chances are it's going to give you an important indicator as to where you are with your BP and more particularly how it changes over time and with activity. More importantly, instead of throwing out the device because it's accuracy is +/- 5% *per reading*, that does not factor much and people should educate themselves as to how to obtain meaningful results under similar conditions from one day to the next. They should DEFINITELY compare their machine against a gold-standard test in their doctor's office; any responsible physician will let you bring your machine in and run a few tests against his manual sphygmamomometer "gold standard". If not, find one who will. If yours is out by 5%, it either doesn't matter (your are - or you are borderline - hypertensive) or it's of marginal importance.
My biggest concern with an article like this is that people will say: "well, fuck-it, this is not totally accurate and so it's meaningless" and avoid buying a monitor. That is not true in any sense and really, while manufacturers should strive to build more accurate machines, the fact is that regularly using a machine that's 5-10% inaccurate is far, far better than not monitoring at all.