A few things I see on reading through it :
Disclaimer - I haven't read any of the studies mentioned so I don't know what was actually reported (vs. what is reported about what was reported), what choices were justified and how, etc.
No mention of the data Keys excluded from the 7 countries trial. This is also the first time I realized that the 7 countries trial involved a recall element, that is, part of the data is based on what people remembered and reported having eaten. Self-report data such as this can certainly be useful and in some cases is unavoidable/the only way to get the data you're looking for, but it's also less reliable because it depends on people's memory and on their willingness to disclose.
Correlation is implied as causation in almost every single instance where they cite data. "X group eats more dietary fat and has more heart disease than Y group" doesn't prove that fat causes heart disease. What other factors may be implicated? Carb quantity? Carb type (refined vs. whole grains)? Activity level? Hours spent sitting? Stress levels? Pollution? Other nutritional deficiencies?
None of the statin trials reported appear to have been done on normal populations. This brings up two points. One - therefore we can only generalize results at best to "high risk" populations. Which is already an assumption of a sort (that said populations are homogenous and normalized and there aren't other factors involved). And when it comes to determining high risk, again, "other factors" becomes an issue. What is it about being diabetic that increases heart disease risk, and whatever "it" is, would statins effectively treat that? What is it about having had a prior heart attack that makes another one more likely, and do statins effectively treat that? (it appears so, at least, this is the one population that we have significant information about). There is no guarantee that what increases risk in one population is the same as what increases risk in another. Same way as treating high blood sugar can require very different approaches depending on what is really at the heart of it (autoimmune, IR, other metabolic disorders, something else?)
Two - how is "normal" defined? Not only "at what level" but literally how did we come to that decision? I notice in some places they cite certain levels as being "near normal" and in other places similar levels are reported as being "at risk", ex. 5.5 for the highest level of mortality in the Japanese studies, vs. 5.5 as "near-normal" in the AFCAPS and ASCOT studies cited further down. Also, are we defining normal as being "what generally healthy members of the population have" or are we defining it based on "what we usually see in studies" or "what we usually see in studies minus a little bit because we assume the study participants are diseased?" Are we defining normal or healthy cholesterol as something that is realistically achievable without intervention for the majority of healthy people, or are we defining it on paper based on conclusions we have already drawn about what healthy cholesterol ought to be? If we put "normal" at 4, and most of the population is at 5, then every time someone with a "high" (but actually quite average) cholesterol of 5 dies from heart disease that adds to the statistics that say that "high" cholesterol contributes to heart disease and heart disease deaths.
I also note that there is very little attention given to whether lowering cholesterol in general, and statins in particular, have other harmful effects on the cardiovascular or other systems. In other words there is no cost-benefit analysis mentioned. Yes, for a drug to be released, a cost-benefit study is required, however what are the "costs" that were taken into account (typically only mortality or serious side effects during the period of the study) and what other things might have been overlooked? For example what if statins reduce inflammation and lower cholesterol counts (it appears that they do) but then lead to whatever cholesterol IS produced being of lower quality and therefore more damaging to the cardiovascular system? What is the overall reduction to risk in that case? Plus, "safety" and "efficacy" as statistical terms don't cover the entirety of a drug's effects - that's why we see so many drugs that make it to market only to be recalled. A larger swathe of the population generates more data over more time than a concentrated study, and things pop up that were missed before. But in many cases these additional effects aren't made the focus of subsequent studies, so they don't get cited in the literature, and thus they don't get included in education programs.
Of course the same critical analysis can be applied to the other side of this debate. No study is without unconsidered factors (you can never control for everything, on the one hand, and on the other hand new knowledge is updating what we consider as a "factor" all the time, for example we now know that total cholesterol isn't the whole story and we have to look at breakdowns and ratios to be able to draw meaningful conclusions). And every study must make methodological choices, which means that it could also have been done a different way and perhaps gotten different results. In the end what I always draw from these kinds of debates is that we are certain, in fact, of very little in any scientific field. And a good scientist will agree that certainty is an illusion. However by the time it has been filtered through the public health machine and reached the level of patient education/general knowledge, all these things are presented as facts, with the idea that it is possible to know exactly what the right answer is, and going against that right answer is potentially a mortal sin. In the end even science involves a certain amount of personal belief.