Yes, I understand your indignation and I should not have been as quick to spurt that out without more information. I've tried to justify everything I've said above with data from Framingham and the Cochrane Collaboration, but I hope you'll forgive me if I have to lapse into a few sources from less hallowed publications once in a while.

"Eating cholesterol doesn't cause high blood cholesterol."

Since only citing things from Framingham or Cochrane is a hard constraint to keep I am forced to commit the minor sin of citing a work not published in a peer-reviewed journal and refer you to the Framingham Diet Study, a subpart of the Framingham Heart Study whose methodology was published appropriately but whose results for some reason weren't. A guy who tracked down the results reports on them here and finds that

"With one exception there was no discernible association between reported diet intake and serum cholesterol level in the Framingham Diet Study Group. The one exception was a weak negative association between caloric intake and serum cholesterol level in men. [As to] coronary heart disease–was it related prospectively to diet. No relationship was found."

If you'll allow me to step out of Framingham and Cochrane for a moment I can also link to a review on egg consumption and LDL which found that in most of the population there's no relationship.

"Eating saturated fat probably doesn't cause higher blood cholesterol."

This claim can also be supported by the Framingham data above finding no link between diet and serum cholesterol.

You may also be interested in a Cochrane review that finds no effect reducing dietary fat consumption on cardiovascular events, cardiovascular mortality or total mortality, and finds that changing the type of fat slightly decreases cardiovascular events (look how close that confidence interval gets to 1!), but not cardiovascular mortality or total mortality. And here's a non-Cochrane meta-analysis saying pretty much the same thing.

If those are true, either the fat -> blood cholesterol or the blood cholesterol -> CVD connections have to be flawed; I'd say at this point it's more likely the former.

"High blood LDL probably doesn't cause heart disease."

That was irresponsibly speculative as written; I have now edited it out of the original post lest it cause exactly the confusion you worry about. I am impressed by arguments suggesting that the association between cholesterol and heart disease may be more correlational than causal, or at least super-complicated and modulated by weird factors we don't understand like subtype of LDL, but at this point I can't support that with Framingham or Cochrane and it would be irresponsible to say so too loudly.

But I think it's weird how randomized trials keep finding that a lot of drugs that change cholesterol don't change disease risk. The ENHANCE trial of ezetimibe finds it successfully reduced LDL cholesterol but didn't decrease (and may increase) cardiovascular events. The AIM-HIGH trial of niacin finds it successfully raises HDL cholesterol and lowers LDL cholesterol but didn't decrease cardiovascular events. Fibrates may prevent CVD but have no effect on all-cause mortality. The only lipid-lowering drugs that seem to be consistently awesome are, of course, the statins, but they seem to work equally well in high and low cholesterol populations leading some to think they also have non-cholesterol-related effects.

I also just think LDL is a lousy biomarker. A majority of heart attack patients have LDL levels considered normal; this is especially impressive considering the high results of high cholesterol even in the general population.

But in general you're right and I apologize for asserting this one until I can back it up better.

"High blood cholesterol levels are protective against cancer and the mortality gain here probably outweighs any mortality loss from cardiovascular disease."

Framingham found that mortality increased with increasing cholesterol in people younger than 50. In people older than 50 (ie 90% of heart attack victims) it found no relationship (other sources say low cholesterol led to higher mortality in these age groups, but I can't access the paper to check.)

In other studies, the relationship between cholesterol and noncardiovascular mortality has indeed been clearly protective.The article I linked to tries to explain this as people with hidden catabolic diseases (eg cancer) having decreased cholesterol levels. This has become less plausible as longitudinal studies show low cholesterol long precedes disease; see for example Schatz 2001.

Overall I am not as confident in my claim above as I was when I wrote it. I wrote it right after reading Good Calories, Bad Calories, but before double-checking everything in it to avoid effects of confirmation bias. I continue to think the statements above are mostly true, in the vague and weasel-ish way that medical statements are true (true in most subgroups, if you hold risk factors constant, so on). But I suggest reading Good Calories, Bad Calories and doing some more research and checking for yourself: I would be really interested to know what another rationalist who's interested in medicine thinks of it.

I would suggest (not "recommend", I'm nowhere near high-status or high-information enough to "recommend") that you use a better source of monitoring cardiovascular risk than LDL (ApoA/ApoB ratios seem to work), that you suggest an Obvious Healthy Diet Without Twinkies Or Coca-Cola and leave it at that unless you want to wade into the monster-infested radioactive swamp that is nutrition science, and that you continue prescribing statins where indicated.