Comment by timr | Hacker News Reader

Comment by timr | Hacker News Reader

This is not a scientific paper. It is a "narrative review", which is another way of saying "editorial". It superficially looks like science, but doesn't do any of the methods, controls or (correct) statistics that you would expect in a legitimate meta-analysis. Do not take it seriously.

The "dedication" at the top should clue readers into what is going on -- it's not even trying particularly hard to look legitimate -- but alas.

Edit: this line, in particular, made me laugh. It is one of the more egregious examples of "pretending to do statistics" that I have seen recently:

> Table 1 summarizes the studies in the last decade examining a potential link between neuropsychiatric reactions and finasteride exposure. When prescribed mainly for AGA, all reports suggest that finasteride can cause depression, anxiety, suicide ideation, and suicides. Assuming a null hypothesis (finasteride does not affect mood) and a 50% chance of 1 result against this hypothesis, the probability of getting all 8 studies concluding against the null hypothesis by chance is 0.58 = 0.0039.

From your description it sounds like a review article? Which is common and normal. I haven't read it though because I'm not bald.

It isn't a review article, though the difference is more subtle, because a review article is also editorial in nature.

The difference here is that this article is someone with an axe to grind (again, see the "dedication", which is never done in a legitimate review; and the clownish application of statistics, which is so completely absurd that it implies incompetence, malice or both). I have no faith that this writer made a legitimate attempt to impartially weigh the evidence on this question.

Doesn’t mean this isn’t correct. These problems are definitely real.

> These problems are definitely real.

Maybe they are, maybe they aren't. Proof is required.

IMHO proof of safety of the drug should be required from the maker before they profit from selling it.

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

Indeed. This is why the drug passed phase 1 (safety) trials when it was first approved in 1992 for treatment of prostate enlargement. Moreover, it subsequently passed additional rounds of clinical trials for the hair loss indication in 1998, and has been involved in more than 30 different clinical trials overall.

Now, I'm not going to argue that there can't be rare side effects that are only discovered with time. I'm also not going to claim that the original trials were perfect, or that research into the question isn't justified. But you're trying to assert that they didn't test for safety, and that's just factually incorrect.

I should take back my statement- I had a knee jerk reaction to someone saying that proof is required of harm for a drug, when I've seen so many cases of drugs being inadequately tested and then causing harm, and I think the precautionary principle is often not followed anywhere close to adequately when it comes to new chemical stuff we do to our bodies. We probably mostly agree in principle. I'm not saying they didn't do safety testing. I would suspect that the safety testing was flawed as it has been in every other case that I have looked into, and failed to catch possible harms that may now be happening.

Whether those harms outweigh the benefits overall remains to be seen and likely will never be known unless it's really really bad, which is likely not the case here.

I'd agree more research is probably justified, but there's likely little profit in it for anyone.

TFA cites eight peer-reviewed studies finding a link. Sounds like decent evidence to me.

> TFA cites eight peer-reviewed studies finding a link. Sounds like decent evidence to me.

There have been over 30 randomized controlled trials of this drug, and the author picked only eight papers, none of which were randomized, none of which were controlled, and all of which were based on mining self-reported data from patient databases.

Come now.

I checked closer and you're correct. Good call, best to stick with meta studies (eg, https://doi.org/10.1080/09546634.2021.1959506).

I’ve personally experienced it. There’s so much data out there it now comes with a warning label.

There's an FDA-mandated warning label for every drug. The current label for Finasteride [1] includes depression only as a "postmarketing experience", which is based on the same data you're reading here (self reports), and is not reliable for determining causality. [2]

There is no current listing for suicide, suicidal ideation, etc.

[1] https://www.accessdata.fda.gov/drugsatfda_docs/label/2012/02...

[2] "Because these reactions are reported voluntarily from a population of uncertain size, it is not always possible to reliably estimate their frequency or establish a causal relationship to drug exposure"

"Misery loves company" the paper

Reply on news.ycombinator.com