In case you haven’t been paying attention (although this one was hard to miss as it was all over the airwaves and blogosphere) a meta-analysis was recently published which concluded that saturated fat is harmless when it comes to risk for coronary disease (CD).  To quote the authors conclusion, “Current evidence does not clearly support cardiovascular guidelines that encourage high consumption of polyunsaturated fatty acids and low consumption of total saturated fats.”  

While I would heartily agree that imbibing more polyunsaturated fats will not put us on the path to better health, I do question the latter part of their conclusion.  In light of the fact that we’ve been told to reduce intakes of saturated fat (at least for as long as I’ve been a student of nutrition), this comes across as a rather bold and sweeping statement.  Have we been given bad information all this time?  Has the science finally weighed in on this question once and for all? Can we now eat saturated fat-laden foods with impunity?

It’s been said that “Extraordinary claims require extraordinary evidence.”  I would contend that this falls into the category of an extraordinary claim that fails to provide extraordinary evidence.  

Before I weigh in, let me point out that this paper has been widely criticized by some prominent researchers in the field of diet-heart.   Jim Mann, a long-time lipid researcher, said that if he had refereed the paper, he would have rejected it.  Dr Walter Willett, chair of the nutrition department at Harvard’s School of Public Health, was quoted as saying “I think a retraction with similar press promotion should be considered.” But alas, the cat is out of the bag and the press will invariably report on nutrition studies which go against the norm (this is what makes it newsworthy after all).  The damage has been done.  

Before I offer my comments, let me say that there are more thorough critiques than what I will provide.  If you want a deeper dive into why this analysis fails to provide extraordinary evidence, then I encourage you to check those out (1, 2, 3).  

Of interest is that a meta-analysis by Siri-Tarino et. al., published in 2010, also had a similar conclusion, i.e. that increased saturated fat intake was not associated with increased risk of cardiovascular disease.  Also of interest is that many of the prospective studies included in that analysis were also included in this one.  Well-reasoned critiques of that study were similarly ignored.  One such critique by Jeremiah Stamler published in the American Journal of Clinical Nutrition received little attention.  Many of the same criticisms would certainly apply to this paper, some of which I will include in my comments.

A meta-analysis is a useful tool when analyzing findings on a particular question or research topic.  It combines the results of different studies by different researchers.  Yet, the conclusions of the meta-analysis are only as valid as the papers included in the analysis.  The papers used in this analysis were a weak subset of many studies which have looked at the relationship between saturated fat intake, diet and CD risk.  Yet, from the author’s conclusion and the media commentary, a naive public is led to believe that this paper provided a definitive, all-encompassing review of the research.  Alas, this is not the case.   This is why I find it odd that the authors were so quick to dismiss the recommendations to reduce saturated fat intake.  

For a very thorough review of the relevant research, I would recommend the book, Cholesterol and Beyond by Stewart Truswell.  It is a review of “The research on diet and coronary heart disease from 1900-2000.”  Of relevance to this discussion, the author cites Kritchevsky, author of the first monograph on cholesterol.  He wrote, “One thing must be made clear at the outset of this discussion:  the literature relating to atherosclerosis is so extensive that it is possible to find conflicting views on practically every aspect of the disease”.  This statement was made over 50 years ago!  I guess some things don’t change.  

But back to a short (and incomplete) summary of the flaws of many of the studies included in this paper.

1.  Over adjustment.  A number of the studies used in both the Chowdhury and Siri-Tarino analysis adjusted for serum total and LDL cholesterol levels.  As saturated fat has been shown in both human and animal feeding studies to increase serum cholesterol, this is problematic and would serve to dilute the correlation of saturated fat intake and CD risk.  This is a bit like looking at the impact of obesity on the risk of diabetes and then adjusting for waist circumference.
2.  Lack of meaningful differences between dietary groups.  Many of the studies used in the analysis were done on western populations where diets tend to be relatively homogeneous, i.e. there were only small differences between those eating a diet “low” in saturated fat vs. those eating a diet high in saturated fat.  It is virtually impossible to have a “low” saturated fat group in a typical western population as average intakes are high to begin with.  Trust me, the low fat groups in these studies were not plant-eating vegans.  Without meaningful differences in saturated fat content, there is less of a chance of finding any relationship to CD risk.
3. Measures of food intake.  A number of the studies (if not most) used a 24 hr dietary recall or a food frequency questionnaire to determine saturated fat intake.  They then correlated this with disease outcomes 5-20 years later without further analyzing the diets.  The reliability of this method is questionable for obvious reasons.  To quote one researcher in a letter to the editor, “Such methods [24 hr dietary recalls] cannot reliably rank individuals by their long-term intake, especially within populations with a uniformly high saturated fat intake.”  In the massive Cornell-China study (where saturated fat intake, and in particular animal-sourced fat, was strongly associated with increased risk of CD) researchers went into people’s homes and recorded dietary habits, weighed food and took food samples.  This is considered the gold standard when doing diet-disease research at the population level.  
4.  Measures of lifestyle factors.  Similar to dietary measures, this was assessed at the beginning of the study but not during or post.  Considering that some of the studies were started in the 80s (prior to statin therapy and when smoking rates were just starting to decline), one has to wonder how many people subsequently stopped smoking and/or started cholesterol-lowering or BP medication which would confound the outcomes.  
5. Reductionism.  Whenever population studies look at an isolated nutrient (in this case fat) and disease outcomes, I think the results are somewhat suspect.  Free living humans don’t eat isolated nutrients.  Reductionism serves us better when looking for mechanisms which help to explain relationships.  A better approach is to look at the type of foods eaten.  It has been consistently shown that dietary patterns which contain high amounts of meat (or animal- derived foods) and processed foods contribute to increased risk for obesity, diabetes, heart disease, certain cancers and Alzheimer’s disease.  In fact, I have yet to see a study where increased meat consumption is associated with improved health outcomes. In the European Prospective Investigation into Cancer (EPIC) study, for example, the number one predictor of weight gain over 5 years was meat (chicken included) consumption (this after adjustment for calories).  Personally I think that looking at food types and disease outcomes is far more informative.
6. In some of the studies used in this analysis, the researchers didn’t differentiate between polyunsaturated fats from oils or margarines vs from whole food sources like nuts and avocados.  As it turns out, that matters.  Again, reductionist research does little to inform us as to the best diet for long-term health.  

Meta-analysis of prospective and case-controlled studies is useful and can serve to inform and raise questions, but they should not be used as conclusive evidence and their conclusions are only as sound as the studies used in the analysis.  Determining the best diet for preventing coronary disease should be based on a wide variety of study designs, including epidemiological, population studies (both prospective and retrospective), metabolic ward feeding trials, animal studies and intervention studies.  To date, the only diet shown to reverse CD (based on intervention feeding studies) is one that is high in nutrients and fiber (whole plant foods) and very low in saturated fat and cholesterol (i.e. animal foods).  Until such a time that there is an intervention study showing regression of CD with high intakes of saturated fats, I will continue to severely limit my intake of saturated fat and cholesterol-laden foods.  

Having said all that, trying to make a case for single nutrients or isolated food compounds as either causing our health problems or solving them will invariably lead us down the wrong path.  To quote David Katz of Yale University, “we seem to have an insatiable appetite for mere grains of truth about diet and health then the complete recipe.” 

Not all carbs lead to weight gain, nor will going gluten-free solve our health problems.  Just because jelly beans and black beans are both high in carbs, does not mean they are created equal.  Olives and olive oil are two very different foods, one has nutrients and fiber, the other is well, mostly calories.  Going vegan can be just as unhealthy if we substitute animal foods for junk plant-foods.  Demonizing or promoting various intakes of fats, carbs or proteins is an exercise in futility.  True we need all three to some degree, but we need to eat them well packaged with a full complement of fiber, water and micronutrients.  Kale and steak are both protein rich foods but steak is devoid of the health promoting nutrients and antioxidants found in kale.  

After 15+ years of reading nutrition research it quite simply boils down to this: eat mostly whole plant foods (vegetables, fruits, beans, whole grains, mushrooms, nuts and seeds), leave dairy for what nature intended (growing small baby cows to large ones) and limit meat consumption for those rare special occasions (or avoid it altogether), or if you must, use it as a condiment.  The same is true for sugar.  This way of eating will go a long ways towards resolving our current health problems.