Monday, January 18, 2010

Saturated Fat and Heart Disease

I'm on a low-carb diet. I believe in LC diets for people with diabetes.

However, I also have an open mind. It's possible that new evidence will show that LC diets, although they improve blood glucose (BG) levels in people with diabetes, also make something else worse.

Richard Bernstein, the physician and author of LC diet book The Diabetes Solution, has lived with type 1 diabetes for many decades, most of those years on a LC diet. And the fact that he is in excellent health in his 70s argues against this possibility. However, Bernstein has type 1 diabetes, and very little insulin resistance. There's some evidence that fat increases insulin resistance. Hence, for those of us for whom insulin resistance is a big problem, perhaps fat of any kind, or maybe only certain kinds of fat, is not a great idea.

So, I have an open mind. But unfortunately, many people in the LC community seem not to. Many of them don't have diabetes, and they have gotten great results losing a lot of weight with LC diets. So they think the LC diet with a lot of fat is the answer for everyone.

And unfortunately, the LC world is just as guilty of spinning the news as the popular science writers who blame red meat for all our problems when some study showed that people eating red meat, hot dogs, french fries, no vegetables, and sweet desserts don't fare so well on some health factor.

A good example is the blogosphere response to this recent study, a meta-analysis of the association between saturated fat and cardiovascular disease (CVD). A meta-analysis is a study in which researchers combine the results from a lot of studies, some of which aren't statistically significant because of their small size, so that the overall results are statistically significant because of the larger populations in the combined studies.

Meta-analyses are notoriously questionable, because the researchers have to decide which studies to include. If you did a meta-analysis of the percentage of the population that watched the Super Bowl (assuming lots of people had studied this fascinating question) but excluded everyone who shaved every morning, the results wouldn't be very accurate.

Nevertheless, sometimes meta-analyses can suggest possible conclusions that other scientists can then investigate more thoroughly.

And that is what this study, titled Meta-analysis of prospective cohort studies evaluating the association of saturated fat with cardiovascular disease, did.

The authors' conclusion was that "there is no significant evidence for concluding that dietary saturated fat is associated with an increased risk of CHD [coronary heart disease] or CVD. More data are needed to elucidate whether CVD risks are likely to be influenced by the specific nutrients used to replace saturated fat."

Two things are important here.

First, the fact that there's no significant evidence for something doesn't mean it's not true. It just means no one has proved that it's true. Several studies have concluded that there's no significant evidence that BG testing in people with type 2 diabetes results in lower A1c's, but most of us know that it does when patients are educated about how to use the results from their meters to change their diets and their exercise patterns. But no one has done the study that would show this.

And second, this study was about association, not cause. Something can be associated with something else but not be the cause of it. For example, coffee drinking is often associated with smoking, but drinking coffee doesn't make you smoke, and vice versa.

The types of studies this meta-analysis looked at were not the types of studies that can show cause.

What the authors found was that some studies showed that saturated fat consumption was associated with higher rates of CVD (heart attacks and strokes), and other studies showed that saturated fat consumption was associated with lower rates of CVD. When you combined the higher rates and the lower rates, you got rates that weren't significantly different.

However, they also noted another recent study that showed that when saturated fat was replaced by polyunsaturated fat, CVD rates went down. When saturated fat was replaced by carbohydrates (what dieticians have been recommending that we all do), CVD rates went up. They said there was some evidence that the ratio of unsaturated to saturated fats was more important than the amount of saturated fat. Hence they suggest that studies are needed that would investigate whether the other elements of the diet have more effect on CVD than the saturated fat.

The authors of Meta-analysis of prospective cohort studies evaluating the association of saturated fat with cardiovascular disease never say that saturated fat definitely doesn't cause CVD. They also say that "the available data were not adequate for determining whether there are CHD or stroke associations with saturated fat in specific age and sex subgroups." In other words, the jury is still out.

Nevertheless, the Internet is awash in blogs with titles like "Two major studies conclude that saturated fat does NOT cause heart disease" and "Saturated Fats Are Not Harmful."

The following are just my opinions, and I won't cite studies to back them up. I suspect that saturated fat is fine in moderation. If you want to put a couple of teaspoons of something on your vegetables, I suspect it doesn't matter if it's butter or olive oil. People on LC diets can probably eat more saturated fat because they're burning fats instead of carbohydrates for energy. I don't think eating a lot of polyunsaturated fats (vegetable oils), which are easily oxidized (damaged), is healthy.

But I don't think eating gargantuan amounts of fat of any kind is healthy, even on a LC diet. I once measured my triglyceride levels after eating an extremely high fat breakfast. You can see the results here. The triglyceride levels were astronomical.

People with diabetes probably have a disturbed lipid metabolism, so it's possible that nondiabetics would not have such astronomical triglyceride levels after pigging out on fats (for example, eating half a pizza). But headlines proclaiming that saturated fat isn't harmful will be interpreted by many people to mean that fat isn't harmful. They won't stop eating all those carbohydrates, the doughnuts and french fries and white bread. They'll just add more fat because they remember that they saw headlines saying fat doesn't cause heart disease.

The study showing no association between saturated fat consumption and CVD, despite its many limitations, is important. It should lead to more studies that will attempt to show causation or lack thereof.

I just hope the misinterpretations don't result in more unhealthy eating.


  1. I agree that meta-studies are open to misinterpretation. My understanding is that prospective studies are used to generate a hypothesis about cause and effect. Controlled studies are then conducted to test the hypothesis. This meta-analysis tells me that researchers are beating a dead-horse trying to prove that excessive saturated fat in the diet causes disease. My hope is that researchers will generate better a hypothesis. In the mean-time, pass the butter please. It is clear that we have been lied to regarding healthy eating. The "meta-analysis showed that there is insufficient evidence from prospective epidemiologic studies to conclude that dietary saturated fat is associated with an increased risk of CHD, stroke, or CVD."

  2. If you don't eat carbs & you don't eat fat, what do you eat? It would be difficult to get a large proportion of one's calories from protein.

    And also, if one eats real food & doesn't eat carbs, then one eats a fair amount of saturated fat, because real food that doesn't have carbs (e.g. meat, eggs, dairy, nuts, seeds) contain a fair amount of saturated fat.

    And also, if one eats fat, then the TGs go up after the meal. If one eats carbs, then they go up later (and stay up longer).

    It seems to me that pretty much any food has someone who thinks it is bad for health. For myself, I intend to eat real food & avoid carbs -- that seems to me to be the best bet available.


  3. Anonymous, I get the feeling that what they're trying to prove is that substituting sat fat with polyunsaturated fat will reduce CVD.

    I agree we've been lied to. They say there's evidence that substituting carbs for sat fat increases CVD.

  4. Beth, I can understand your frustration. Even lean meat contains some fat. I wasn't suggesting that we avoid all saturated fat, just excessive amounts.

    Fatty cuts contain a lot more fat than lean cuts. Eating the meat from a lamb chop is different from eating the chop plus all the fat on the outside.

    I think we should aim at reasonable amounts of real foods. As long as you're not gaining weight, you're probably eating reasonable amounts.

    I'm not sure how you're defining "carbs." Spinach and lettuce contain carbs, and I eat LC vegetables, so I eat carbs. Again, it's a question of how much. You can raise your BG with lettuce if you eat enough of it.

    Nuts and seeds don't have a lot of saturated fat, although any kind of food contains some.

  5. One thing I haven't clarified. I'm on a LC diet. When I've calculated nutrients, I find I'm eating about 60% fat.

    Some non-LC people would be horrified. But I'm actually eating a lower *amount* of fat than I did when I was on a "standard American diet," meaning I was eating lots of carbs as well as fat.

    If you're eating a lot of carbs plus fat and you cut way back on the carbs, the percentage of fat will increase as your total calories decrease and the amount of fat and protein remain the same.

    In other words, the *amount* may be more important than the percentage, but most analysts focus on the percentages of nutrients.

    If you were eating 10,000 calories a day, you could be eating only 10% fat and still be eating a heck of a lot of fat.

    Combined with a lot of carbs, that would not be a good thing

    As Richard Feinman of the Metabolism Society ( says, “The deleterious effects of fat have been measured in the presence of high carbohydrate. A high fat diet in the presence of high carbohydrate is different than a high fat diet in the presence of low carbohydrate.”

  6. Gretchen, great post

    Disagree with your point oh your triglyceride levels after eating an extremely high breakfast.

    I just overlayed the two graphs in photoshop, and the total area under the curve for triglycerides is maybe 20-25% more with the low carb diet. I didn't integrate the data, but I think visually it's pretty clear it's not a whole lot.

    I would speculate the net insulin difference from low fat to low carb is of greater magnitude.

  7. Chris, let me clarify. I'm not recommending that anyone switch to a low-fat diet to reduce PP triglycerides. If I'd started at the higher fasting level after 24 hours of low fat, the PP peak would have been even higher the next day.

    What I'm saying is that perhaps the very high peak after the very high fat breakfast (about 50 g of fat) isn't a good idea. Note the smaller second peak. This was after what I'd consider a reasonable amount of fat: some fatty salmon (about 10 g of fat) and probably some LC vegetables and a salad with oil and lemon juice. (I'm too lazy to find that day's log to see exactly what).

    I agree about the net insulin difference, although I'm probably not producing much of my own insulin, which is why it took so long for the peak to come down on the low-fat day.

    I happen to believe that there's no Ideal Diet that would work best for everyone. We have to find what works for us.

  8. Yeah, I love the discussion that nothing's optimal, centrist paleo approaches probably work well for 70% of people (1 standard deviation).

    Insulin didn't change much in LC, maybe up 15% max. In HC insulin beyond doubled.

    PP Triglycerides went up 6x on HC, 10x on LC.

    The magnitude of the change of insulin is huge here, like, give us a reason to care about TG's stacking really high from a fat low carb breakfast.

    Intermittent fasting (which prob should be core paleo) takes care of PP triglyceride stacking between meals.

  9. Hi Gretchen,

    In Gretchen's Post-prandial Diet Experiment, how many grams of fat did you eat in your low-carb breakfast?

    I'm trying to work out where the serum TGs went. Some went into muscle cells at a rate of ~60kcals-worth/hour (~6.7grams of fat/hour) and the rest went into fat cells.


  10. Nige, pretty much none. Spaghetti, white rice, low-fat tomato sauce.

  11. Chgris, are you referring to my graphs? I wasn't measuring insulin, but blood sugar.

    "Insulin didn't change much in LC, maybe up 15% max. In HC insulin beyond doubled."

    "centrist paleo approaches probably work well for 70% of people"

    I'm not sure if you're referring to people in general or people with diabetes. What bothers me about paleo diets for people with diabetes is that you're permitted fruit, which is not great for many people with diabetes.

    I know someone who gained weight on a paleo diet because he was eating too much dried fruit.

  12. Nigel, I answered it. What else do you want to know?

  13. I asked "how many grams of fat did you eat in your low-CARB breakfast?" That's the high-FAT breakfast.

  14. Nigel, Sorry. I misread it. I ate about 50 g of fat in the LC breakfast, as I noted in my response to Chris on Jan 19. I think that's why I misread your question, which I thought hadn't been answered.

  15. Because several people have expressed interest in exactly what I ate these days, I've gone back and found my food logs from spring 2001.

    Caveat: I was not planning on publicizing these results, which I thought were for my own edification. So I didn't weigh most of the food or measure portion sizes, as I normally don't. If I write down "green beans," I know what a usual (for me) portion of green beans is. I usually eat 3 to 4 oz of meat/fish per meal.


    9 a.m. 1 oz spaghetti (dry weight) cooked al dente, 3/4 C white rice, canned tomato sauce (Classico)

    1 p.m. chicken breast (no skin), pepper mushroom, black bean sauce, ginger, garlic, 1/4 C white rice, 1 teaspoon SF jam.

    4 p.m. (Very hungry) Lettuce, turkey breast chunks, mushroom, 1 teaspoon olive oil and lemon juice, 1/4 C white rice with 1 teaspoon olive oil.

    9:30 p.m. Telapia with tomato sauce, 1/4 C white rice, green beans, bit of LC muffin (homemade) with SF jam, walnut


    10 a.m. Kelly sausage, mushrooms fried in butter with 2 slices bacon, 1/4 C cream from the top of creamy full-fat yogurt. (Burp!)

    2 p.m. Skinless chicken breast, snowpeas, mushroom, 3 LC muffins with LC jam

    4:45 p.m. 2 Brazil nuts and 1 LC muffin

    8 p.m. Salmon, spinach, artichoke with oil and vinegar, 2 squares of chocolate (don't remember what kind, if LC or not), almonds

  16. Gretchen, 3rd time lucky! I missed your post of 19th Jan possibly due to time-zone difference (or possibly due to me not seeing it).

    50g of fat contains 450kcals. If you burn ~60kcals/hr at rest (people at rest burn ~1kcal/min) and 100% of energy came from fat, it would take ~7.5hrs to clear the fat from your blood. If 66% of energy came from fat (the average for sedentary people), it would take ~11hrs to clear the fat from your blood.

    As it took ~6hrs to clear the fat from your blood, this suggests that some fat entered your fat cells without your serum insulin being raised.

  17. The health authorities are fully aware of the serious flaws and omissions in this meta-analysis. This study was funded by the National Dairy Council, dairy being the number one source of saturated fat in the U.S. and many other parts of the world. It was also conveniently published just before the USDA lowered the dietary recommendations of saturated fat for the first time in 20 years, from 10% to 7% of total calories.

    Below is a section from the statement released by the European Heart Network in regards to their opinion of this meta-analysis, titled “European Heart Network position piece: Impact of saturated fat on cardiovascular disease obscured by over‐adjustment in recent meta‐analysis”

    “However, the meta‐analysis (and an accompanying opinion piece by the same authors (4)) is compromised by a number of serious flaws and omissions. These are enumerated and discussed in detail in an editorial from Jeremiah Stamler (5). The most serious of these flaws is an over‐adjustment for serum cholesterol levels. The meta‐analysis involves data from 16 studies that evaluate the impact of saturated fat intake on CHD incidence or mortality, and 8 studies that evaluate the impact of saturated fat intake on stroke incidence or mortality. The authors state that ‘wherever possible, risk estimates from the most fully adjusted models were used in the estimation of the pooled [relative risks]’. It is well‐established that saturated fat intake is associated with increased level of serum cholesterol (6), and that serum cholesterol levels are associated with CHD and CVD (7). Therefore, serum cholesterol levels lie on the causal chain between saturated fat intake and CHD and CVD, and to adjust for serum cholesterol levels in a meta‐analysis will obscure the impact of saturated fat intake on these health outcomes. Yet 7 of the 16 studies included in the meta‐analysis of CHD events, and 4 of the 8 studies included in the meta‐analysis of stroke events were adjusted for serum cholesterol levels. These studies accounted for nearly half of all CHD and CVD events included in the meta‐analyses. Adjustment for serum cholesterol levels will inevitably bias the results of the meta‐analyses towards finding no association between dietary saturated fat intake and cardiovascular disease, but the authors do not mention this limitation in their article. As Jeremiah Stamler asserts in his editorial, what was actually found by the meta‐analysis was ‘a statistically non‐significant relation of SFA [saturated fat] to CHD... independent of other dietary lipids, serum lipids, and other covariates’ (5). A more appropriate and informative analysis would have included non‐adjusted associations between saturated fat and cardiovascular disease. An examination of the forest plots provided in the article shows that those cohort studies that did not adjust for serum cholesterol levels were more likely to find a positive association between saturated fat intake and cardiovascular disease, suggesting that a meta‐analysis of unadjusted data would likely produce positive results. “

    References 5-7
    (5) Stamler J. Diet‐heart: a problematic revisit. American Journal of Clinical Nutrition, 2010; 91: 497‐499.
    (6) Clarke R, Frost C, Collins R, Appleby P, Peto R. Dietary lipids and blood cholesterol: quantitative meta‐analysis of metabolic ward studies. BMJ, 1997; 314: 112.
    (7) Prospective Studies Collaboration. Blood cholesterol and vascular mortality by age, sex, and blood pressure: a meta‐analysis of individual data from 61 prospective studies with 55,000 vascular deaths. The Lancet, 2007; 370: 1829‐1839.

    The full statement from the European Heart Network can be found here:

    Below is a published study showing reversal of severe heart disease backed up with angiogram evidence.

  18. t, your references concern primarily diet and heart disease. Those of us with diabetes also need to worry about the effect of any diet on our blood glucose levels. It's very difficult to do a low-carb diet eating mostly plants when you have diabetes, and low-fat diets tend to be high in carbs, which make blood sugar rise.