[nonameiguess]

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.

[sfink]

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.