Hacker News Reader

Comment by pakitan | Hacker News Reader

Yes, and future spaceships will travel 100x times the speed of light.

You really think we're up against fundamental limits of medicine? I'm struggling to understand your meaning here

I'm just making fun of the certainty with which the poster assumes that just because we had humongous progress in all areas of knowledge for the last 100 years, it's somehow guaranteed that the progress will continue at the same rate. Fundamental limits or not, we've already picked the lowest hanging fruit and further progress is painfully incremental, slow and expensive and Star Trek-like devices seem extremely unlikely.

I think you're reading it backwards. If you look closely at how medicine is done today, you will see that there are many areas where it is wildly divorced from reality. So, the point was not "we'll be vastly better soon", it's more "we're in a bad place now".

The current most wildly successful, heavily prescribed medicines today are statins. They help 1 in 104 people in terms of preventing heart attacks, 1 in 154 people in terms of preventing stroke. (Those are people without known heart disease, but they are the vast majority of people taking statins.) They harm 1 in 10 by causing muscle damage, 1 in 50 by causing diabetes. [1] That's the success story. (Sure, you can debate the details. Do they really cause diabetes? Unclear. Do they help anyone, ever, to not die sooner? Unclear.)

It seems like the main reason they're considered so successful is that they do indeed lower an intermediate metric, namely blood cholesterol level. I am sure that bloodletting was successful at removing blood, and if you have an infection, you could even say at removing bad blood.

And yes, I'm cherrypicking my definition of success. Modern medicine can indeed dramatically improve outcomes for a large set of problems (eg cancer). But doctors were successfully setting bones back in the bloodletting days, too.

[1] https://thennt.com/nnt/statins-for-heart-disease-prevention-...

There is a serious problem with that site's analysis. The meta cited on statin death prevention covered an average trial length of 3.74 years per person. That means they can give you, at best, your 3-4 year probability of having a fatal heart attack. For most age cohorts, that probability is very near 0 no matter what you do, so no intervention whatsoever can prevent cardiac event death by this metric. But this metric isn't what people care about. They're not trying to reduce the risk of having a heart attack in the next few years. They're trying to reduce the risk of ever having a heart attack.

Note this is exactly why we actually use the studies of people with prior cardiovascular disease that this meta excludes. Those people are sufficiently likely to actually have another heart attack within the time horizon of the study that you can get useful data!

The other option is to only conduct 60 year trials. It should be obvious why that isn't a viable option.

The limited time duration is a big deal, I agree. It's an extrapolation from insufficient data. (Though the studies were evidently powerful enough to come up with a number, so the probability is not that near 0.) But that also means insufficient data to provide evidence for net benefit from an intervention, and an intervention really needs to prove its worth before you go about tempting fate by taking something biologically active. Where is the evidence that statins "reduce the risk of ever having a heart attack"?

I'm going to disagree about the cohort. That only means that if you have prior heart disease, you should not be looking at an NNT derived from a population without prior heart disease. The site's conclusions are mostly irrelevant for you, and should not factor into a rational decision.

If you don't have prior heart disease and are weighing your options, then those data are relevant to you. The vast majority of people who are deciding whether to take statins are in this category.

People deciding whether to try to remove a bullet from their abdomen, and who have no reason to believe that they have ever been shot, should not be weighing the outcomes of test subjects who had been shot before participating in the trial. (It would really suck to be in the control group...)

I'm not saying you shouldn't take statins, with or without prior heart disease. An individual would have more to go on than the existence or absence of a prior heart disease diagnosis. Exact cholesterol readings, for example, might create more or less urgency.

But if I were in the situation of deciding for myself, I'd want better evidence for them than I have seen presented so far. I am suspicious of an industry for which this is a big success story.

Reply on news.ycombinator.com