How do we get the word out about accuracy:

[Protimenow]

I think that most of us here know that tests using a meter usually don't match the results of labs. Many of us know that if the values are less than 20% apart, the values should be considered accurate.

I don't think most people at labs (and probably doctors) are aware of this.

I recently had a blood draw for many things, including INR. The phlebotomist asked me if I had a meter that uses fingersticks. I told him that I do.

He went to a shelf and pulled down a CoaguChek XS box - he said that a patient returned it because it was 'inaccurate.' I assume that he wanted the result of this meter to match the result of the lab. To him, I'm assuming, if the two values didn't match, the meter must have been inaccurate.

I also assume that because giving it up was so easy for him, there was no investment made in the meter -- it was probably provided by insurance or Medicare.

(I'll see if I can get the meter, see if the batteries have leaked and ruined the meter, and do some tests on it).

This is a waste of what's probably a fully usable meter.

It would be good if the people at labs, and the medical professionals that are involved in INR testing were aware of the 20% range of acceptable values, both the prevent tossing what are perfectly good meters, and declaring values to be within testing accuracy limits.

But HOW (and who) can we educate these people, or the public at large?

 

[jeffp]

can we educate these people, or the public at large?

The answer would be “no”; especially when dealing with “the public at large”.

There’s a limit to how close one should expect lab and meter tests to align. It’s not a simple measure like blood glucose. The ISI factor of the reagent has to be taken into the INR calculation. The ISI can vary with temperature and age and the PT can vary with that well as sample technique (how much and if the finger is milked, time from sample to test, etc.). I’ve had lab tests taken from the same sample vary by 0.2 or 0.3 INR units. The Coaguchek in serial tests has usually been within 0.1 or 0.2. I always do back to back tests when changing to a new batch of strips. There’s always a variance between the tests (yes, I change the chip). I think the 20% acceptable variance Protimenow discussed is a good ballpark guideline. Hence, I target my INR more to the higher end of the range than the lower.

 

[nobog]

The industry I was in (biomedical) they calibrated everything (even rulers), I am not personally familiar with how the INR meters work but there has to be something that the manufacturers do when they set these things up. Seems like there should be a calibration interval for these meters or a self test that determines their accuracy.

 

[pellicle]

The answer would be “no”; especially when dealing with “the public at large”.

sadly I simply can't argue with this (and yet I'm writing a book on INR management )

 

[pellicle]

The industry I was in (biomedical) they calibrated everything (even rulers), I am not personally familiar with how the INR meters work but there has to be something that the manufacturers do when they set these things up. Seems like there should be a calibration interval for these meters or a self test that determines their accuracy.

I think roche are pretty good and generally speaking the vast majority of people I discuss this with get within 0.2 INR units from labs.

Further discussions usually (not always) reveals a user error (there are many possible) but there are definitely cases where blood types. For instance antiphosopholipid syndrome is one I'm familiar with:

https://onlinelibrary.wiley.com/doi/full/10.1002/jha2.522

Point of care testing (POCT) to monitor INR is discouraged in patients with APS as interactions between antiphospholipid antibodies and thromboplastin used for INR testing may influence results. ​

I'm sure there are others.

My first few tests were ~ 0.5 away from labs, however I realised that (after re rereading the documentation) I was not adhering to the actual guidance strictly. I honed my method and presently find much better accordance (not always with the same lab):

date​	Coagu INR​	lab INR​	dose​
14/12/2012​	3.2​	3.4​	7.0​
17/12/2012​	2.7​	2.8​	7.0​
20/12/2012​	2.6​	​	7.0​
21/12/2012​	2.8​	2.8​	6.7​
28/12/2012​	2.5​	2.6​	7.0​
21/06/2014​	2.3​	2.5​	7.6​
31/01/2015​	3.0​	2.9​	7.0​
3/08/2015​	2.4​	2.6​	7.3​
6/02/2016​	3.0​	2.8​	7.3​
8/11/2017​	2.7​	2.6​	7.0​
20/04/2021​	2.3​	2.3​	7.0​
16/05/2022​	3.2​	3.0​	7.0​

for the US audience we write Day / Month / Year

 

[Protimenow]

My question wasn't about accuracy or meters or blood draws.

I consider the CoaguChek XS (and, for others, the Coag-Sense) to be accurate meters - within the 20% variance that is an accepted deviation.

My question was how we (or anyone) can educate professionals who do INR testing (including the phlebotomists) or INR management, and the self-testers that there is no 'exact' INR value, and that they should accept the fact that the values WILL deviate from an ideal 'accurate' measurement. Again, this was prompted by my experience with a phlebotomist who showed me that a customer 'abandoned' his metet because it was 'inaccurate.'

It would be good if these people realize that, when it comes to INR, 'accurate' is actually values within a certain range.

 

[Unicusp]

With the Coag-Sense meter every new lot of test strips include control test strips, both low and high control. The test confirms the performance of the meter vs control samples. It is a validation of the calibration.

 

[tom in MO]

The industry I was in (biomedical) they calibrated everything (even rulers), I am not personally familiar with how the INR meters work but there has to be something that the manufacturers do when they set these things up. Seems like there should be a calibration interval for these meters or a self test that determines their accuracy.

I believe the test strip has chemistry that assures the machine is operating properly. Calibration and operating information is contained on the chip that comes with the strips to control the measurement and meter.

 

[Protimenow]

This is off target.
CoaguChek XS uses a chip that sets the reagent value, so tests are 'accurate.'

Coag-Sense also tells the meter what value to use for the reagent. This is put onto an NFC chip attached to the box and is programmed into the meter when a new box is started - there is also a strip with the information printed on each test strip.

My question isn't about 'why' are meters accurate - it was about how to educate people so they know that meter results aren't supposed to be exact matches for labs - that lab values vary, and that a 20% margin of error is acceptable.

I'm not sure if ANYONE has given ideas about THIS.

 

[pellicle]

With the Coag-Sense meter every new lot of test strips include control test strips, both low and high control. The test confirms the performance of the meter vs control samples. It is a validation of the calibration.

I think you need to check the details of this, as a recent discussion with a coag-sense (well now an ex) user who is a forum member who is a medical professional (no, I realise that's impossible, we don't have any doctors on there because "none of us here are doctors") has been told in a discussion with CS about massive inaccuracies he was increasingly getting that the test strips and test reagent do not then go on to effect the calibration of the meter. I would recommend that you take this point up with CS themselves but more to the point I'd recommend (if you aren't already) getting tests now and then (say every six months) to validate the readings against an actual vein draw from a lab you trust.

Best Wishes

 

[LondonAndy]

I am not sure this issue arises here in the UK. As we have a National Health Service and a National Institute of Clinical Excellence which has published a report specifically recommending the CoaguChek devices as best practice, readings from the machines are simply accepted as sufficiently accurate. I have been self-testing and managing my INR for 8 years now, and every six months I am required to attend my local anticoagulation clinic and compare my machine's results with their test. So admittedly this is a comparatively small sample, but such is the faith in the CoaguChek accuracy that initially the comparison was with a blood draw test, and these days it is compared against their own CoaguChek device, albeit the multi-patient version. I have never had a result that is more than 0.2 different from their test. On a therapeutic range of 2.5 to 3.5, this is approximately +/- 6% variation.

 

[LondonAndy]

"for the US audience we write Day / Month / Year"

@pellicle In the UK we also write it the correct way

 

[pellicle]

@pellicle In the UK we also write it the correct way

well, given that we are an ex-colony that's hardly surprising(no .. wait, so are they ;- )

I think its worth mentioning that in Japan (and other asian countries they write it YYYY/MM/DD which actually sorts nicely even as ASCII text when you have a spreadsheet exported into a .CSV (or some other common database table dump formats.

I can only offer you the following diagram for which to classify me as you see fit

ta taa for now ;-)

 

[Unicusp]

I think you need to check the details of this, as a recent discussion with a coag-sense (well now an ex) user who is a forum member who is a medical professional (no, I realise that's impossible, we don't have any doctors on there because "none of us here are doctors") has been told in a discussion with CS about massive inaccuracies he was increasingly getting that the test strips and test reagent do not then go on to effect the calibration of the meter. I would recommend that you take this point up with CS themselves but more to the point I'd recommend (if you aren't already) getting tests now and then (say every six months) to validate the readings against an actual vein draw from a lab you trust.

Best Wishes

Good idea. I do need to get a lab blood draw and compare to meter results. Have not done in about a year now. I often wonder about the accuracy of lab results too. They take a blood draw and then send the sample to a lab for testing which can be 24 hours after the blood has been drawn. Just doesn't sound right to me. I wonder how much the time delay affects coagulation results.

 

[Protimenow]

I think that a bit of this has gone to comparing apples to eggplants (not even oranges).

Pellicle's comment was SPECIFICALLY about the Coag-Sense system. After the company changed its executive team (and possibly its ownership), my results were almost consistently lowher than the CoaguChek, and usually the lab results.

I was concerned that the errors were somehow only mine - especially after getting reports from others on this forum that the lab results were close to their blood draw results. Calls and messages to Coagusense either went unanswered, or without adequate responses.

On one occasion, when I reported that my results were about 1.0 below the labs or CoaguChek, I was told by the person at Coagusense that I should just add 1.0 to the Coag-sense to get the right value. Huh? It's like saying that if my clock is always an hour behind, I should just add an hour to the result. This doesn't work when the tester's life hangs in the balance.

True - blood draws can't always be relied on. The blood may be mishandled between the time it is drawn and the time it is actually tested. The person drawing the blood can mishandle it. There are many things that can effect the accuracy. The lab may even do it wrong, or use the wrong reagent values.

FWIW - drawn blood is supposed to be put into a tube with a lavendar (or purple) colored cap. With proper handling, it should produce a fairly accurate result.

But - back to my original question - HOW do we educate people to the fact that if a meter and blood draw vary by 20%, both should probably be considered to be accurate?

 

[nobog]

Everything has a tolerance - as long as in big bold letters on the meter that says +/- 20% - so be it. That may be "the state of the art" for these meters.

 

[Protimenow]

THAT'S THE POINT. My question is HOW do we get the word out that a certain level of tolerance IS acceptable.

I know of one patient to stopped using his meter because it was 'inaccurate. ' I would be very surprised if it wasn't within the 20% tolerance.

 

[pellicle]

Everything has a tolerance ... That may be "the state of the art" for these meters.

it may even be "state of the art" with resect to how you can determine exactly when that clag has set (assuming that jiggling it influences how it sets).
I love this quote from here:
https://www.ahajournals.org/doi/10.1161/01.atv.0000193624.28251.83

We chose the title, “What does it take to make the perfect clot?” because of our sense of wonder that a process as complex as hemostasis ever works as it should. ​

nice little hat tip to Rachel Carson in there.

This 2005 article makes this important point (among others)

Cells Control the Coagulation Process In Vivo and Regulate the Amount and Tempo of Thrombin Generation​

It is widely acknowledged that the coagulation reactions occur on specific cell surfaces in vivo rather than on phospholipid vesicles as they do in the prothrombin time and activated partial thromboplastin time assays. About 15 years ago, our group hypothesized that the key to understanding the hemostatic process was the correct incorporation of the roles of cells into a conceptual model.​

So basically what we have is at best and estimation of an approximation.

I say again and again here: we are not measuring steel with a micrometer, which is why we should consider carefully our target (not our range) because INR is a rubbery measurement.

 

[Unicusp]

This thread has provided me the incentive to check the accuracy of my Coag-Sense meter that I've been using for about 1.5 years now. I last compared results about a year ago. Recently received a new lot of test strips and successfully ran the low and high control. Yesterday, I went to my local Quest Diagnostics lab for a blood draw and then came home and ran my test 30 minutes after the blood draw at the lab. The results fall within the parameters of what others have stated here.
Coag-Sense: INR = 2.4 / PT = 26.1
Quest: INR = 2.9 / PT = 27.7
Difference: INR = 18.9% / PT = 6%
Now I wonder if there is a consistency with the meter always understating true INR or if is variable. Last year when I compared the values, the meter did understate INR by 17% difference.

 

[pellicle]

Now I wonder if there is a consistency with the meter always understating true INR or if is variable

welcome to the rabbit hole ;-)

CS owners seem to want to find some sort of "conversion" that they can apply to their meters.

I think if you didn't read them the posts by ETC901 are worth reading

https://www.valvereplacement.org/th...een-pt-and-inr-lab-vs-home.888470/post-914837
actually after you read that you should go to the start of the thread and read it all (if you didn't already read it)

I know that some people here (lets call them idomatic types) say "none of us here are doctors" .. well except for the ones who actually have doctor in their fukken name how would you know who's what?
The answer is we do actually have doctors here, further we have people who are not called doctors generally because they are specialists in a field. Further we have professional researchers here.

So when I say ETC901 knows enough to know what he's talking about I say that from the perspective of having worked with him in helping him start on my INR management system. I know what he does, I know I respect him.

Personally I don't have much respect for the CS product.