High sensitivity troponin. What am I missing?

[kwt8]

---

Members do not see this ad.

This topic keeps coming up in various FOAM articles. I understand that they are getting so sensitive that the specificity is suffering. What I don't get is why we can't use different cutoffs to guide management and take advantage of each of these properties. Example:

Patient comes in with chest pain. We think he is low risk, do the normal workup including a troponin.

If it comes back lower than some small number, x, he goes home.

If it's >x but lower than some other number, y, he stays for a second troponin, obs admit, etc.

If it's >y he gets NSTEMI management.

It seems to me that cutoff x takes advantage of high sensitivity, y is specific. Is there something wrong with this logic?

 

[KarlPilkington]

Truly a great idea.
It would take a TON of work to get this study off the ground, it needs to happen, but its gonna take a lot of $$, a lot of time, and some really smart folks to make it happen.
Even if your results are great, folks will refuse to accept your results initially (See all Canadian research; Head CT, C-spine, SAH...)
So then you're stuck waiting for someone else to validate your results.
Most folks are just going to start ordering high sensitivity troponins and admitting/working up way too many folks who are in actually NO not LOW risk for having ACS.
Reasonable AHA guidelines would help.
Giving an acceptable miss rate (see 2% PE miss rate) would help.
Tort reform would help.
Unfortunately if the culture for chest pain continues to have no tolerance for any missed cases of ACS, there is very little room for doing anything with these high sens. trops, other than admit anyone who isn't stone cold normal.

 

[xaelia]

Most folks are just going to start ordering high sensitivity troponins and admitting/working up way too many folks who are in actually NO not LOW risk for having ACS.
Reasonable AHA guidelines would help.
Giving an acceptable miss rate (see 2% PE miss rate) would help.

I.e., "Doctor, did you not order a troponin because you were worried about a heart attack? Look at this test result ... it even displays as a critical high in your EHR! How is it possible you ignored this result, and now my client is dead?!"

 

[Old_Mil]

Tort reform would help.

This. Most of what we do is driven by lawsuit avoidance. Until this happens chest pains are still going to get two troponins, two EKGs, plus or minus an obs stay and a stress regardless of what they do with the Troponin assay. Changing the assay to increase the number of false positives is not going to change anything.


Sent from my Z10 using Tapatalk

 

[Birdstrike]

This "acceptable miss rate" stuff is whistling past the graveyard. 1%, 2%, or 50% does you zero good in court because no one patient thinks they're an "acceptable miss" and no malpractice attorney is going to acknowledge his client was an "acceptable miss."

The standard and acceptable miss rate is, and in all of our lifetimes will always be, 0%.

That's why you buy the insurance. You're going to get sued, so you do the best you can, and when the papers come, the papers come. You're covered. It is, what it is. It means nothing. It's a trip and fall on your snowy doorstep by the mailman.

It's absolutely nothing. It's like paying the mob for protection. It's not a great system, in fact it's a terrible one, and you do it because you have no choice, but it is a system of protection.

Know that a certain percentage of your salary covers the inevitable lawsuit or two, or three. And allow yourself not to worry about it, or at least not as much as you would otherwise. Don't torture yourself over it because it will happen, and guess what? Life goes on.

A lawsuit doesn't have to ruin your life or career, and you should not let it. It's normal for a doctor to be sued. Own it. And welcome to the big leagues.

 

[deleted109597]

This topic keeps coming up in various FOAM articles. I understand that they are getting so sensitive that the specificity is suffering. What I don't get is why we can't use different cutoffs to guide management and take advantage of each of these properties...Is there something wrong with this logic?

Do you have another test that uses this method for us to base our design on?

 

[kwt8]

So I was thinking that the studies to "validate" the higher level, NSTEMI cutoff (y I mentioned above) would be the studies that validated the old "low sensitivity" troponin tests. The studies that would validate the low level (x from above) would be the ones validating the new assays.

As for other tests using similar logic, it seems similar to how we react to lots of quantitative tests:

Outpatient BP management: 120<SBP<140 we recommend lifestyle changes, SBP>140 we start antihypertensives
Potassium replacement in DKA: k<3.3 hold insulin and start K replacement. If 3.3<K<5.3 start K and insulin. If K>5.3 start insulin with K to follow.

These are just examples I came up with off the top of my head and I realize there are differences between these and troponin. But, if we reduce the lower limit of detection on a quantitative test, shouldn't we be able to continue to use all if the information we had on the old limit?

For example, we have all this info that says a BAC > 0.08 is bad for driving. Now some guys invents a test that can measure BAC as low as 0.001 and levels above this make you bad at neurosurgery. Wouldn't we still let people at 0.02 drive? We wouldn't claim, that just because someone came up with this more sensitive test, now the driving limit is 0.001.

 

[WilcoWorld]

McNinja - The VQ scan is a test that is appropriately analogous to what kwt8 is suggesting. However, I think it's (lack of) utility proves your point.

kwt8 - Your examples don't work because the diseases you're testing for exist on a spectrum, so of course the test results also exist on a spectrum. ACS, however, is binomial (as is PE). Sure, some PE's are small, and some heart attacks are worse than others, but in either case you've got it our you aint.

 

[deleted109597]

McNinja - The VQ scan is a test that is appropriately analogous to what kwt8 is suggesting. However, I think it's (lack of) utility proves your point.

I would argue the "small, medium, large" qualitative ketones is closer, since VQ is overlaying ventilation and perfusion and guessing if the difference is due to a clot or not. Agree that it again is more of a spectrum, which is why I would argue that no test has clinically different sensitivity and specificity limits. While a laudable goal, @xaelia has a definite point about any cutoff above "negative" being a huge risk medicolegally. All this is going to do is force us to admit more people or increase our risk threshold.

 

[Arcan57]

I would argue the "small, medium, large" qualitative ketones is closer, since VQ is overlaying ventilation and perfusion and guessing if the difference is due to a clot or not. Agree that it again is more of a spectrum, which is why I would argue that no test has clinically different sensitivity and specificity limits. While a laudable goal, @xaelia has a definite point about any cutoff above "negative" being a huge risk medicolegally. All this is going to do is force us to admit more people or increase our risk threshold.

I'm confused about this point since none of my shops have had UHSTs. I can understand that they detect miniscule amounts of troponin, but do they have a different cut-off for AMI than regular troponins? Our current troponins still have a significant range of not undetectable but not MI (0.02 to 0.4 for our current assay). Are the UHSTs different in that it flags every troponin that's detectable as an MI? Otherwise it seems like it adds a group that you can send home immediately (undetectable trop), dumps most into a group that you need serial markers to r/o (our current standard of care), and everybody with an NSTEMI is still classified as an NSTEMI.

 

[deleted109597]

Nope. Now the <0.02 range is measureable. And if you get two, and the second one goes up 0.001ng/dL, what are you supposed to do?

 

[kwt8]

McNinja - The VQ scan is a test that is appropriately analogous to what kwt8 is suggesting. However, I think it's (lack of) utility proves your point.

kwt8 - Your examples don't work because the diseases you're testing for exist on a spectrum, so of course the test results also exist on a spectrum. ACS, however, is binomial (as is PE). Sure, some PE's are small, and some heart attacks are worse than others, but in either case you've got it our you aint.

But I would argue ACS is a spectrum. We treat it as such. Some people get immediate cath lab, others get conservative treatment +/- delayed cath. But I agree that troponin is built (currently) to give you a yes/no answer about whether a person is anywhere on the spectrum.

But, with the new troponin tests, we're finding normal people have a baseline troponin level. So it seems like the Step I question with overlapping normal curves. (Example picture on this site http://omerad.msu.edu/ebm/Diagnosis/at_roc.html)

Below a low cutoff we can say there is a miniscule risk of disease (probably have to move that point to the left) and you go home. Between the low and high cutoffs (again for our purposes probably move 2 left a lot) you have a greater likelihood of disease but still more likely than not to be normal, so we need more tests. Above that high cutoff risk is so high we need to treat you.

 

[deleted109597]

But I would argue ACS is a spectrum. We treat it as such. Some people get immediate cath lab, others get conservative treatment +/- delayed cath.

Just because it is treated that way doesn't make it a spectrum. Immediate cath is for STEMI (which doesn't require troponin values). NSTEMI doesn't require immediate cath or delayed cath, and doing such is simply to line the cardiologist's pockets. Lots of data shows medical management is better than cath for NSTEMI, and newer data shows stents for NSTEMI are inferior to CABG. Medical management=/=conservative treatment by any measure.
We know how cutoffs work. Again, show another test that uses your separate cutoffs for diagnosis.

 

[Arcan57]

Just because it is treated that way doesn't make it a spectrum. Immediate cath is for STEMI (which doesn't require troponin values). NSTEMI doesn't require immediate cath or delayed cath, and doing such is simply to line the cardiologist's pockets. Lots of data shows medical management is better than cath for NSTEMI, and newer data shows stents for NSTEMI are inferior to CABG. Medical management=/=conservative treatment by any measure.
We know how cutoffs work. Again, show another test that uses your separate cutoffs for diagnosis.

Synovial fluid WBC count. Although the rise of MRSA is screwing that up.

 

[deleted109597]

Synovial fluid WBC count. Although the rise of MRSA is screwing that up.

Hey, you found one. And it's terrible.
Hard and fast cutoffs for WBCs are irresponsible at best. Other than orthopods, how many people say "nope, no chance of infection with only 79K WBCs in there."