Ask a doctor you trust: “Can you tell CFS and hypothyroidism apart by clinical symptoms rather than by blood test?”
I have no idea if he could, but even assuming he can’t, that wouldn’t prove anything. “Have similar symptoms” isn’t the same thing as “are similar”. Many symptoms are produced by lots of different conditions.
I’m not sure why you think the blood test shouldn’t work. If you mean that it shouldn’t be able to tell apart two different conditions, of course it should—being able to tell apart two conditions is the whole point.
No, I mean that the blood test shouldn’t be able to tell the difference between hypothyroidism (shows hypothyroid symptoms, gets better when given thyroid hormones) and not-hypothyroidism.
The TSH test measures one variable in a very complicated system full of feedback loops. It’s a bit like if your aeroplane keeps crashing, and you take it to the mechanic and he says “The rudder wires are fine, so there’s nothing wrong with it”. And you say “But why do you think the rudder wire is all you need to test?”, And he says, “Whenever an aeroplane is good, and doesn’t crash, we check the rudder wires and they’re always fine”.
Except the TSH test isn’t even that good! Because for forty years they’ve been arguing about what blood levels of TSH fine people have, and the latest answer is “such a narrow range that huge numbers of people who aren’t ill are outside of the range”.
So it diagnoses lots of people who seem perfectly healthy as ill, and it tells lots of people who are obviously ill that they’re fine. I call bullshit!
But the real problem is that they never checked whether it worked or not. You’re supposed to check these things. And they didn’t.
Where is the evidence that (shows hypothyroid symptoms, gets better when given thyroid) is the same as (TSH in whatever the normal range is)?
Why did they ever believe that? Why did they not check to see whether it was true?
for forty years they’ve been arguing about what blood levels of TSH fine people have, and the latest answer is “such a narrow range that huge numbers of people who aren’t ill are outside of the range”.
Rephrased a bit, this says “Healthy people have such a narrow range of TSH that huge numbers of healthy people are outside of this range”. I trust you see the problem.
the real problem is that they never checked whether it worked or not
What do you mean by “worked”? As you yourself point out, it’s just one variable in a complicated system with feedback loops. If you mean “necessary and sufficient to give a clinical diagnosis”, then no, it never “worked”. Doesn’t mean the number is not useful.
Absolutely I see the problem! Thank you for phrasing it more elegantly than I did. And absolutely TSH is a useful number. It’s a very fine measure of the hormone thought to control the T4->T3 conversion rate, as well as the thyroid output.
Now can you tell me (a) What the AACB did wrong in their report, or (b) why, when someone goes to their doctor with many of the symptoms of hypothyroidism, they do this test, and they find the number to be within a very broad range, and then they say “your problems can’t be anything to do with thyroid hormones”.
Especially in the context of diabetes, another endocrine disorder, where we know that you can have normal levels of hormones, but for some reason they aren’t working, and you have high blood-sugar that is killing you anyway? No one in their right mind would tell a type-2 diabetic that they weren’t diabetic because they had a normal amount of insulin.
I don’t know, haven’t looked at it. I suspect that the underlying issue is that for one variable in a complicated process the concept of unconditional “normal range” might be too simplistic.
Because for forty years they’ve been arguing about what blood levels of TSH fine people have, and the latest answer is “such a narrow range that huge numbers of people who aren’t ill are outside of the range”.
Tests are Bayseian evidence. If people outside the range disproportionately have a condition, a test is evidence for that condition, even if not every person outside the range has the condition, and not every person with the condition is outside the range.
Absolutely agree, what I’m saying is that this test isn’t very strong evidence. And that they decided to use it without considering how strong it was. They did check specificity, by measuring TSH levels in some healthy people. As far as I can tell, they didn’t check sensitivity, and they never have.
I think you’ve just nicely summed up my argument for me. Thank you.
If you can find the bit where they checked sensitivity, and it’s sound, then I’m flat wrong and I’ll shut up!
They say:
(? (hypothyroid symptoms, gets better with T4) [in the Aristotelian sense]
I say they’ve shown:
(unambiguously healthy) ⇒ (0<TSH<2.5) [In the sense of plausible reasoning]
And that’s all they’ve shown. Ever.
I am avid for evidence that they’ve done more than that.
My counsellor tells me that he literally cannot believe that they would have forgotten, it must be buried in the literature somewhere, and that’s why we can’t find it. I am perfectly happy to believe that it was a careless mistake.
I have no idea if he could, but even assuming he can’t, that wouldn’t prove anything. “Have similar symptoms” isn’t the same thing as “are similar”. Many symptoms are produced by lots of different conditions.
Yes indeed! And if the only way to tell them apart is a blood test that shouldn’t work, then....
I’m not sure why you think the blood test shouldn’t work. If you mean that it shouldn’t be able to tell apart two different conditions, of course it should—being able to tell apart two conditions is the whole point.
No, I mean that the blood test shouldn’t be able to tell the difference between hypothyroidism (shows hypothyroid symptoms, gets better when given thyroid hormones) and not-hypothyroidism.
The TSH test measures one variable in a very complicated system full of feedback loops. It’s a bit like if your aeroplane keeps crashing, and you take it to the mechanic and he says “The rudder wires are fine, so there’s nothing wrong with it”. And you say “But why do you think the rudder wire is all you need to test?”, And he says, “Whenever an aeroplane is good, and doesn’t crash, we check the rudder wires and they’re always fine”.
Except the TSH test isn’t even that good! Because for forty years they’ve been arguing about what blood levels of TSH fine people have, and the latest answer is “such a narrow range that huge numbers of people who aren’t ill are outside of the range”.
So it diagnoses lots of people who seem perfectly healthy as ill, and it tells lots of people who are obviously ill that they’re fine. I call bullshit!
But the real problem is that they never checked whether it worked or not. You’re supposed to check these things. And they didn’t.
Where is the evidence that (shows hypothyroid symptoms, gets better when given thyroid) is the same as (TSH in whatever the normal range is)?
Why did they ever believe that? Why did they not check to see whether it was true?
Rephrased a bit, this says “Healthy people have such a narrow range of TSH that huge numbers of healthy people are outside of this range”. I trust you see the problem.
What do you mean by “worked”? As you yourself point out, it’s just one variable in a complicated system with feedback loops. If you mean “necessary and sufficient to give a clinical diagnosis”, then no, it never “worked”. Doesn’t mean the number is not useful.
Absolutely I see the problem! Thank you for phrasing it more elegantly than I did. And absolutely TSH is a useful number. It’s a very fine measure of the hormone thought to control the T4->T3 conversion rate, as well as the thyroid output.
Now can you tell me (a) What the AACB did wrong in their report, or (b) why, when someone goes to their doctor with many of the symptoms of hypothyroidism, they do this test, and they find the number to be within a very broad range, and then they say “your problems can’t be anything to do with thyroid hormones”.
Especially in the context of diabetes, another endocrine disorder, where we know that you can have normal levels of hormones, but for some reason they aren’t working, and you have high blood-sugar that is killing you anyway? No one in their right mind would tell a type-2 diabetic that they weren’t diabetic because they had a normal amount of insulin.
I don’t know, haven’t looked at it. I suspect that the underlying issue is that for one variable in a complicated process the concept of unconditional “normal range” might be too simplistic.
For the usual reasons :-/ Find a better doctor.
Except that all the better doctors have been struck off! Who takes hypothyroidism to a chiropractor?
Tests are Bayseian evidence. If people outside the range disproportionately have a condition, a test is evidence for that condition, even if not every person outside the range has the condition, and not every person with the condition is outside the range.
Absolutely agree, what I’m saying is that this test isn’t very strong evidence. And that they decided to use it without considering how strong it was. They did check specificity, by measuring TSH levels in some healthy people. As far as I can tell, they didn’t check sensitivity, and they never have.
I think you’ve just nicely summed up my argument for me. Thank you.
If you can find the bit where they checked sensitivity, and it’s sound, then I’m flat wrong and I’ll shut up!
They say:
(? (hypothyroid symptoms, gets better with T4) [in the Aristotelian sense]
I say they’ve shown:
(unambiguously healthy) ⇒ (0<TSH<2.5) [In the sense of plausible reasoning]
And that’s all they’ve shown. Ever.
I am avid for evidence that they’ve done more than that.
My counsellor tells me that he literally cannot believe that they would have forgotten, it must be buried in the literature somewhere, and that’s why we can’t find it. I am perfectly happy to believe that it was a careless mistake.
… maybe the reasoning that led you to the conclusion that the blood test “shouldn’t work” was wrong.
g, see answer to Jiro above. I agree that I am probably wrong. I would like to see evidence that I am wrong.