Hacker News Reader

Comment by kbos87 | Hacker News Reader

I looked into a full body scan about a year ago after hearing from someone I trust that they had one done.

While I didn’t end up moving forward with one, I was struck by how weak the arguments against having one done were. They amounted to:

- There’s a strong chance you’ll be sent on a wild goose chase or two after seeing something on the scan that is meaningless

- They only find something that actually turns out to be concerning in a small percentage of people

- This isn’t a cost effective approach to early diagnosis for most people

I’m sorry, but if I’m at a place in my life where this is a reasonable expense, even 1% odds such a scan turns up something genuinely troublesome is very worth it. It’s a life we are talking about.

View on HN

> I’m sorry, but if I’m at a place in my life where this is a reasonable expense, even 1% odds such a scan turns up something genuinely troublesome is very worth it.

That's actually not how diagnostic tests work.

Because of Bayes theorem, tests that have a high likelihood of returning a positive result in a high-incidence population are totally useless when applied to asymptomatic random people in the general public.

Your likelihood of being harmed by one of these wild goose chases exceeds the chance that an early diagnosis will help you. First, you don't know there's a 1% chance that it'll turn something up - you don't know the chance at all. And you also don't know the likelihood and magnitude of harm chasing down false positives.

That's why we don't test people randomly for things they likely don't have.

It's just a logical fallacy and cognitive bias to think that doing something, anything, is better than doing nothing. Often that isn't the case.

> It’s a life we are talking about.

Which is why we shouldn't let our biases pull us in unhelpful directions.

No one says you have to follow up with anything invasive. I know someone who had a sinus polyp turn up on an MRI. Knowing what it was let him live with the symptoms it was causing without fearing that it was something much worse. I don't think he ever got it removed.

If you don't have symptoms and a scan showed an incidental something, you don't have to act. Personally, I'd rather know. And, if something does go wrong in the future you can say "I have an X in my left Y." Might give the doctor a place to start debugging.

bit off thread, I have a sinus polyp that seems to be under control when using nasal corticosteroids, with the massive risk of developing a dependency.

at my wits end on other alternatives

Basically every human is significantly different inside, and they look nothing like the medical literature. Any scan will reveal all sorts of things that may or may not be at all clinically relevant.

And to your original point: No one needs to do anything about those differences just because they got an MRI.

Assuming you know which ones are problematic and which one's aren't. We don't. All the test provides is noise, not signal, when used in a low-incidence population.

The fact is there's no evidence such testing improves outcomes - if there was we'd just be doing it as part of the standard course of care. Folks are just demanding access to noise. Just go flip some coins instead, and save the medical resources for the people who need them.

Talk to a radiologist and ask them about the stuff they’ve found incidentally in “healthy” people. The average person is very much as described in the medical literature, with some small variations. This isn’t a good use of my time. I’m done replying.

Would you mind explaining the bayes theorem scenario in simpler terms?

This is a pretty good explanation [1].

Let's say you have a test that's 80% accurate in telling you that you have cancer given you have cancer. Let's say the false positive rate is 10%.

- If 100% of the population has cancer, then a positive test means you're 80% likely to have cancer. This makes sense.

- If 10% of the population has cancer, then a positive test means you're only 50% likely to have cancer because of all the false positives in the general population.

- If 1% of the population has cancer, then a positive test means you're only 7% likely to have cancer.

- If 0.1% of the population has cancer, then a positive test means you're only 0.83% likely to have cancer.

Basically, the false positives dominate the results once you start applying the test to a low-incidence population. Which is why we don't.

[1] https://betterexplained.com/articles/an-intuitive-and-short-...

Out of curiosity, why didn't you go for it?

Reply on news.ycombinator.com