Well, the sample size is still tiny. But even so, the case-control you're suggesting would definitely be open to survivorship bias, because you're selecting people who have lived a very long time. More appropriate to start at, say 50, and see how many make it to 100
-
-
Replying to @GidMK @ethanjweiss and
Technically, the sample size is 124 of those 84/85 in the NHANES dataset. The five are the ones who have categorically survived to 100 and are still alive at last check.
1 reply 0 retweets 0 likes -
Replying to @DaveKeto @ethanjweiss and
Mmmm but for this kind of study 124 is still far too few. You'd want at least that many cases, and then link to 2-3x as many matched controls. Even then, it's going to be biased, but at least you'd be able to make some statistical comparisons
1 reply 0 retweets 1 like -
Replying to @GidMK @ethanjweiss and
I don’t disagree. But as an epi guy, you know what it means when correlations are running counter to expectations of a given hypothesis of causality. If we’re seeing more and more people who test decades earlier with the triad showing greater longevity, this would be powerful.
1 reply 0 retweets 0 likes -
Replying to @DaveKeto @ethanjweiss and
Not necessarily. I mean, firstly, depends on what you mean by "more and more". Another 5 would be an interesting statistical blip, another 100 might be worth investigating
1 reply 0 retweets 1 like -
Replying to @GidMK @ethanjweiss and
Can I clarify? -- If both numbers doubled, which is to say we had 248 of which 10 survived to age 100 or more, and the vast majority of them had high LDL, high HDL, and low TG from a decade and a half earlier, you'd consider that a statistical blip?
1 reply 0 retweets 0 likes -
Replying to @DaveKeto @ethanjweiss and
Potentially. How many of the 248 had those same results? How many of them developed the results at/before age 85? What's the breakdown by gender, SES, ethnicity etc? Like I said, it depends on a lot of things!
1 reply 0 retweets 1 like -
Replying to @GidMK @ethanjweiss and
I'm not disputing that it depends on a lot of things, but to be sure, LDL gets attention in singularity. Even in addressing it in combination with HDL and TG I get pushback. Certainly I'd want an even more robust dataset of inflammation markers, fasting insulin, thyroid, etc...
2 replies 0 retweets 1 like -
Replying to @DaveKeto @ethanjweiss and
Oh man, I'm not even talking about the specifics of the biomarkers here, I'm just saying the minimum I'd want before analyzing the dataset to make any conclusions other than "this is interesting"
1 reply 0 retweets 1 like -
Replying to @GidMK @ethanjweiss and
If the five surviving centenarians were all two pack a day smokers at 85 and this were unusual from their likewise cohorts, I'd find this more than interesting.
1 reply 0 retweets 0 likes
I wouldn't, to be honest. Large samples make unlikely things a given, there are lots of statistically unlikely people out there. If we're talking the US, there are at least 300 1-in-a-million people out there! That's what controlled studies are for imo
-
-
Replying to @GidMK @ethanjweiss and
Wow -- okay. We definitely differ there. Again, it is more than interesting not that the only survivors to that age all have something in common, it's that they all have something in common that is considered dangerous and life threatening.
1 reply 0 retweets 1 like -
Replying to @DaveKeto @ethanjweiss and
Yeh but that's the issue with big datasets. Maybe 99% of people like this die at 65 but the 1% remaining are really hardy. Maybe we all develop this biomarker set eventually, so it's just a function of age
1 reply 0 retweets 1 like - Show replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.