Type-1 and Type-2 errors... :-)

https://en.wikipedia.org/wiki/Type_I_and_type_II_errors

Medical testing is very, very murky. There was a time a routine medical test for me came back, indicating I possibly had Lupus. It turns out; this test has a false positive rate of about 5%.

I am a white male, and the rate for lupus for my group is astronomically low. The rate for the worst group (black females) is like 500 in 100,000. There's some evidence the rate around white males might be six times lower.

Had I known any of this at the time, or had my doctor explained it, I wouldn't have spent weeks worrying.

The fact is, medical tests just update the probability you're sick or well. And this is why they have to be well understood.

I had a problem a couple years ago where my test showed that I had hypothyroidism. My doctor wanted to immediately put me on sythetic thyroid hormone to adjust this. The problem is that I'm thin, athletic, and have no other indicators for risk of hypothyroidism.

I said no to the doctor, and had them do another test, which said I was just fine.

The story is not so remarkable: there is no harm from taking synthetic thyroid hormone, and your TSH is monitored while taking it such that I think the mistake would have been discovered later.

>there is no harm from taking synthetic thyroid hormone

Where are you getting this crazy idea? Have you ever known anyone who took it? I have two relatives (not biological) on synthroid and even small changes in doses have massive effects on their metabolism, tiredness, etc. Taking thyroid hormone when you don't need it is not harmless, just like any drug or hormone.

Hm, okay. Apologies for the misinformation.

there's a particularly sad story about HIV testing, which is told as a cautionary tale to medical students: during the HIV outbreak, the tests had something like a 2% false positive rate. the result letter made the mistake of telling patients there was only a 2% chance the test was wrong. people committed suicide after discovering their test came back positive.

but they neglected conditional probability. so many people were being tested that the probability they had HIV was much lower than 98%. they changed the letters to say "inconclusive" instead of "positive" and had them take the test again, which reduced the needless suicides.

From reports, the Chinese tests have fairly high false negatives, but they’re still deploying them en masse because catching as much as is possible better than not catching anything at all. I’ve seen nothing on false positives.

Yeah at the end of the day it seems by far the best outcome. I'm just thinking there may be some sort of red tape surrounding the liability aspects of that outcome. Like maybe there's a waiver of liability for passing a sick person off as healthy, except it only applies to such and such equipment, etc. I'm just using my imagination here, I have no experience in the medical industry.

If you're using an officially published primer sequence? Possible (due to supplier or user error), but _highly_ unlikely. We're pretty good at molecular biology at this point.

Isn’t the idea of review to confirm that “highly unlikely?”

Without review, unscrupulous (or incompetent) vendors could make a test with unknown usefulness.

I'm not talking about making an entire test kit, but rather labs with existing RT-PCR abilities (ie the vast majority of molecular biology labs) having the appropriate primers and probes synthesized by the standard (non-clinical) vendors that they order from every day.