I like browsing through some vegan discussion groups looking at the papers, publications and studies they use to support their belief system. McDougall's forum is particularly useful because of their tenacity in trying to use science to justify themselves. Experience taught me that such studies are always ambivalent and very often prove a completely opposite views to those of the vegan believers who posted them.
The following study from 2003, did not disappoint me:
Dietary fats and the risk of incident Alzheimer disease.
The abstract looked very foreboding, for example, quote:
Persons in the upper fifth of saturated-fat intake had 2.2 times the risk of incident Alzheimer disease compared with persons in the lowest fifth
Not all is lost fortunately, because of the widely known practice in the medical "science" to print only the politically correct (i.e. false) information in the abstracts while hiding the true albeit inconvenient facts in the full text.
The facts are that the saturated fat data and most of the other results are generally not statistically significant!
The facts are that even when one takes the trend line across saturated fat quintiles and makes it appear statistically more significant than an individual datum (the so-called "p value for the trend") - the resulting p is still much greater than 0.05 and thus is still not statistically significant! Best illustration is the following (upper) portion of the Table 3:
The facts are that if you take the most basic age-adjusted only data (look at the first row called "Age-adjusted+"), there is no clear trend at all since the middle and the second highest saturated fat columns (quintiles) have exactly the same Alzheimer's risk as the lowest reference quintile of saturated fat! The second and the third row (headed by "Multivariable" and "Multivariable... other fats") above are the more processed data, that seem to exhibit a weak rising trend, albeit also not statistically significant! One shall always keep in mind that multivariable-corrected trends are dependent upon some specific model-dependent assumptions that may or may not be correct.
It gets more intereseting as one reads down the table. If one takes the above discussed saturated results on faith, beliving that the weak statistics may be reflecting some real underlying trend rather than being some artefacts of the data gathering and processing methodology (as I suspect is the case), then one should also take a notice and state that the rest of the data "proves" (also not statistically significanly) that the total fat consumption, dietary cholesterol intake, animal fat consumption and vegetable fat consumption all seem either not to correlate or to correlate NEGATIVELY (protectively) with the Alzheimer's risk! See the lower portion of the Table 3:
For example, the total fat consumption seems to be protective against Alzheimer's! The second lowest and the middle quintile in consumption of animal produce (indicated by dietary cholesterol!) also have higher Alzheimer's risk than the two highest quintiles! Animal fat consumption seems to show no correlation to Alzheimer - the trend curve is pretty flat except the third row ("Multivariable adjusted for vegetable fats and trans-fats") which shows a weak NEGATIVE (i.e. protective) trend!
Last but not least the bottom group Vegetable fats shows the strongest correlation in the whole study. That correlation is strongly NEGATIVE (protective), that is the more vegatable fat the less Alzheimer's cases! In fact that result, after "Multivariable adjusted for vegetable fats and trans-fats" - is the only one, alongside the omega-6 fat result ( which is basically its subset) that does exhibit a statistically significant trend!
Unfortunately for those who believe in the low fat dogma, this is not a good news! Especially when compared with other sources, see for example this and that.