1/ I'm retweeting this with interest in having a productive conversation regarding @ScepticalDoctor's suggestion. Without knowing in advance, I'm going to check this profile against NHANES data on all cause mortality and see how it compares having higher LDL, all else the same...
2/ Before doing so, I'd like to request everyone treat this conversation productively. Let's keep it about the science, not personal attacks, ridicule, etc. I think it's a worthy discussion to have on the data alone.
3/ Here's the conversion from mmol/L to mg/dL
mmol/L => Total: 3.9 | LDL: 1.9 | HDL: 1.7 | TG: 0.7
mg/dL => Total: 151 | LDL: 73 | HDL: 66 | TG: 62
And FWIW, I like this profile's HDL and TG regardless, as I think the two together are extremely relevant for CVD and all cause.
mmol/L => Total: 3.9 | LDL: 1.9 | HDL: 1.7 | TG: 0.7
mg/dL => Total: 151 | LDL: 73 | HDL: 66 | TG: 62
And FWIW, I like this profile's HDL and TG regardless, as I think the two together are extremely relevant for CVD and all cause.
4/ For NHANES IV, I have a total of 19,696 that include Total Cholesterol (TC), HDL cholesterol (HDL-C), triglycerides, (TG), and mortality status.
Now let's grab only those with HDL of 66 +/- 10 and those with TG of 62 +/- 10...
Now let's grab only those with HDL of 66 +/- 10 and those with TG of 62 +/- 10...
5/ This knocks down the total number substantially to just 1,246.
We could divide the groups by an LDL of 100...
If so, mortality overall and % per year goes to the higher LDL group in spite of their average age being 5 years greater...
We could divide the groups by an LDL of 100...
If so, mortality overall and % per year goes to the higher LDL group in spite of their average age being 5 years greater...
6/ We could further stratify to 1-79 / 80-99 / 100-119 / 120 and above. But then we get a much bigger age disparity at the high LDL side.
Ironically, it still has lower mortality than the 1-79 LDL group.
Ironically, it still has lower mortality than the 1-79 LDL group.
7/ To get us more N in each grouping, I've widened the delta in both HDL and TG to +/- 20 instead of +/-10
With a larger sample size the highest LDL appears ahead of every grouping in spite of the higher average age...
With a larger sample size the highest LDL appears ahead of every grouping in spite of the higher average age...
8/ It's worth emphasizing that many will suggest this is "reverse causality" and I think that's a fair point and worth exploring.
For example, some forms of cancer can correlate to lower LDL before death, thus potentially being the root cause of both. However...
For example, some forms of cancer can correlate to lower LDL before death, thus potentially being the root cause of both. However...
9/ With that said, one can't use this hypothesis as a catch-all assumption without establishment of the evidence. There are likewise ways in which cholesterol will increase under crisis that could also offset this correlation.
10/ So again, epidemiology isn't great at proving causation -- but it is good at knocking down claims of causation.
For example, we wouldn't expect all those living past age 100 to be three pack a day smokers.
For example, we wouldn't expect all those living past age 100 to be three pack a day smokers.
