logoalt Hacker News

bluGilllast Wednesday at 2:07 PM1 replyview on HN

Quality of life is what everyone cares about, but is is expensive to study. We want this things like lesion growth becuase they are much easier/faster, and thus cheaper to study. There is only so many people who can do this research, and the more work they have to do in one study the less work they can do in others. You can study lesion growth with just 100 people that you follow for a year or two - this is possible for just one person (though for double blind reasons you still want a small team). To study quality of life there are so many other variables that you need tens of thousands of subjects that you follow for a decade just so you can be sure that you can statistically isolate the non-factors.

Thus last I checked statins were only proven to prevent heart attacks in males under the age of 65 who have already had a heart attack. That is a large enough group to show results, and their change of heart attacks is known to be higher than normal so you don't have to study as well. Does this really mean other people should take statins, and if so who - hard to know. Our best evidence after years of study is that high cholesterol increases your odds of a heart attack, but a few people live to their 80s with high cholestrol without ever having a heart attack and nobody really knows why. (I haven't looked at the above in about 10 years so I might not know about some more recent science) We would like better study about statins and quality of life, but those are really expensive, and that is depsite statins being cheap and a lot of people willing to take part in a study.


Replies

jmward01last Wednesday at 8:15 PM

I agree with the need for proxies, but if they are used then there needs to be strong evidence that they are valid and studies should report both the proxy results and the implied quality/quantity of life results based on the proxy results. This would help to put the true potential impact into perspective AND help with followup studies that can more easily check the actual impacts observed verses the predicted impact based on the proxy values. Keeping the focus on quality/quantity at every stage, especially if it is just mentioning that there isn't a strong connection between the study objective and real utility, is deeply important.