Hacker News Reader

Comment by kens | Hacker News Reader

There's a lot of discussion here about getting tested for the gene, but the gene isn't as important as you might think from the title. The article states that the gene is associated with "1.6-fold higher odds of developing long COVID". So it's just one factor among many.

View on HN

The language there is weird ("fold" is always an ambiguous way to express relative increases, but it's especially weird here). I assume that means a 60% higher chance?

It’s the standard term for such things. I think it makes sense.

Consider a 140% increase. Or a 1400% increase. It’s a bit hard to internalize what those mean.

2.4x and 15x is easier

I can't think of any other way to interpret it. A x-fold increase of y is x times y.

Seems Wikipedia agrees, so that's something:

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

In which case a 60% increased probability is a very significant increase. Seems like a valid thing to test for when assessing someone for long COVID, given it still remains somewhat of a diagnosis of exclusion.

it's a relative risk, and the underlying absolute risk was very small, so even if 60% sounds big, testing everybody isn't the right solution.

IIRC the absolute risk of long Covid is 7-15% per infection, which is anything but very small.

if that's true, and people get reinfected periodically, then basically everybody will have long covid :)

but not really. what you're seeing is that some fraction of the populace is susceptible to long covid (risk factor or genetic).

"Increase" is ambiguous here. If I said a 160% increase, that is typically taken to mean that the new value is 260% of the original. So "1.6-fold" could mean the same, or it could mean 160% of the original.

No, the "-fold increase" form specifically means multiplying by that value, not adding that multiple of the value. A "3-fold increase" always means increasing to 3 times the original value, not adding 3 times the original value (which would be a 4-fold increase). See [1], the wiki link in the sibling comment to yours and many other results from googling "fold increase".

The percentage variant, in contrast, is indeed ambiguous, the "60% increase" mentioned in the comment I responded to ironically only being unambiguous because the purely multiplicative interpretation (going to 60% of the value) would be a decrease. Such ambiguous use of percentage changes is common and annoying, and it would be good to see more use of the unambiguous "x-fold increase" wording.

[1] https://www.collinsdictionary.com/dictionary/english/fold-in...

That sounds like a pretty decent reason to get tested for it?

Plus modern tests can sample hundreds of gene in one batch. So they will only become better as time goes on, as more linked genes are discovered.

If it changes your odds from "1 in 100" to "1.6 in 100", you might have better things to spend your time worrying about. Absolute risk is important to keep in mind, not just relative risk.

Depends on things like other risk factors, though. And like you say, if you can isolate ten genes that each have a similar effect, then you're getting somewhere.

Getting in a car today probably increases your odds of brunt force trauma by 1000x that day, but in the end the odds of dying that day in a car are still very low.

Reply on news.ycombinator.com