Visit Us
Follow Me

There seems to be a lot of confusion about the meaning of the term “lipid hypothesis” in pop science books, blogs, and other places.

Often times, the lipid hypothesis is confused with the diet-heart hypothesis.  The two are very different.  The lipid hypothesis concerns the role of lipids in the blood.  The diet-heart hypothesis concerns the role of lipids in the diet.
The term “diet-heart,” as best as I can tell, was coined in the 1960s to describe the National Diet-Heart Study, which was designed to test the effect of reductions of dietary saturated fat and increases in dietary polyunsaturated fat on blood cholesterol and heart disease risk.  In 1969, a panel of the American Medical Association headed by Edward Ahrens Jr. called this “the diet-heart question.”
In 1976, Ahrens first popularized* the term “lipid hypothesis” to refer specifically to the hypothesis that increased levels of cholesterol in the blood increase the risk for heart disease.  He published a paper in the Annals of Internal Medicine entitled, “The Management of Hyperlipidemia: Whether, Rather than How.”  Here's what he wrote:

What is The Lipid Hypothesis?

The Lipid Hypothesis is the postulate, based on Framingham (8) and similarly derived data, that reducing the level of plasma cholesterol in an individual or in a population group will lead to a reduction in the risk of suffering a new event of coronary heart disease.  It is a premise based on the undisputed fact that people with higher plasma cholesterol levels have more and earlier coronary heart disease than do others with lower cholesterol levels; but the premise has not yet been proved true to the satisfaction of epidemiologists and biostatisticians or of the medical community at large.  The Lipid Hypothesis, then, is simply an inference derived from accepted facts (Figure 1); though the hypothesis has been put to the test repeatedly in the past two decades, completely satisfactory evidence has not yet been advanced either pro or con.

His figure is meant to demonstrate the difference between association and causation:


On the left, we see the facts as they had been gathered at the time in Framingham, that blood cholesterol is correlated with the risk of heart disease.  On the right we see a depiction of the hypothesis used to explain those facts.  The arrow within the curved line represents a treatment used to lower the level of blood cholesterol.  The arrow on the left side of the picture represents a hypothetical decrease in the risk of heart disease that results from the treatment.  This hypothesis is the lipid hypothesis.
I have been writing about the need to distinguish between the lipid hypothesis and the diet-heart hypothesis for three years, since first reviewing Daniel Steinberg's The Cholesterol Wars (you can read my review here).
I credited Steinberg with making that proper distinction (indeed, I was not even aware of it until reading Steinberg's book), and criticized Uffe Ravnskov for conflating the two hypotheses in The Cholesterol Myths (you can read that review here).
I also emphasized this distinction in my last Wise Traditions lecture, “Heart Disease and Molecular Degeneration: The New Paradigm.”
Whether we view the lipid hypothesis as true, partly true, or false, I think it is important to make this distinction clear.  The alternative is to jumble up various hypotheses and make it difficult to find the truth.
*I say “popularized” because Ahrens was not the first to use the term, even though this paper quickly became referenced as the first to define the lipid hypothesis.  For example, see Epstein's 1977 paper, “Preventative trials and the diet-heart question: wait for results or act now.”  Indeed, Ahrens' paper is the first indexed for pubmed that is searchable by the term “lipid hypothesis,” and appears to be the first paper ever written specifically devoted to defining the lipid hypothesis.  However, Daniel Steinberg used the term in 1974 as a presentation to the Drugs Affecting Lipid Metabolism conference in Milan.  Steinberg was the Chair of the planning committee for the Corornary Primary Prevention Trial, and described the committee as being charged to “to design a feasible study to test the ‘lipid hypothesis,' i.e., the hypothesis that intervention to reduce serum cholesterol levels does reduce risk of clinically manifest coronary artery disease.”  Reference (provided by Dr. Steinberg): Steinberg, D. Planning the Type II Coronary Prevention Trial of the Lipid Research Clinics (U.S.A.) in Advances in Experimental Medicine and Biology 63: 417-426, 1975. Also cited as a book: (in) Lipids, Lipoproteins and Drugs (eds. D. Kritchevsky, R. Paoletti,and W.L. Holmes) Plenum Press, New York, 1975.  Note that Steinberg's definition is the same as Ahrens'.  It is possible that earlier print references exist; however, none appear to have popularized the definition until Ahrens' paper.  The actual hypothesis, however, dates to Anitschkov's cholesterol-fed rabbit in 1913.
Visit Us
Follow Me

You may also like


  1. I recently watched Fat Head a documentary by Tom Naughton and it seems that he didnt know the differnece between the lipid hypothesis and the diet-heart hypothesis. He made the lipid hypothesis seem to be about the affect of saturated fats on cholestrol levels and the impact that that chloesterol level has on the risk of heart disease.

  2. "The arrow on the left side of the picture represents a hypothetical decrease in the risk of heart disease that results from the treatment. This hypothesis is the lipid hypothesis."

    No, that's not the lipid hypothesis. The lipid hypothesis predicts such, but as a special case of a more general idea.

    "The "lipid hypothesis" postulates that hypercholesterolemia is a major causative factor in atherosclerosis and coronary heart disease." p. 1 of Steinberg's _The Cholesterol Wars_.

    Note that we have an "and" there indicating that the lipid hypothesis concerns coronary heart disease (redundantly), *as well as* atherosclerosis. Steinberg's own glossary on p. 215 says "Atherosclerosis=hardening of the arteries, now used synonymously with "arteriosclerosis": a chronic disease of the arteries that worsens progressively with age and can ultimately lead to heart attacks and/or strokes and/or deterioration of blood flood to the legs (peripheral vascular disease, PVD)."

    Not all arteries exist inside the heart.

    In terms of substances in a diet, the lipid hypothesis predicts that for any substances which end up causing elevations of serum cholesterol levels, a correlation (once properly adjusted) between those substances and atherosclerotic events exists. One could argue that this implies *several* diet-atherosclerosis ideas instead of just one. As I understand it, food substances that can or do raise serum cholesterol levels aren't just saturated fats and dietary cholesterol (the relationship here isn't always linear), but also trans-fats, concentrated sources of sugar… such as soft drinks and energy drinks, as well as other foods/food substances which cause surges of insulin such as refined carbohydrates like semolina pasta.

  3. I feel less embarrassed about being confused about this after re-reading a post on Mike Eades' site on the "lipid hypothesis" where he wrote:

    In simplistic terms the lipid hypothesis is as follows:
    a) cholesterol and/or fat in the diet leads to cholesterol and/or fat in the blood;
    b) cholesterol and/or fat in the blood causes plaque formation in the arteries and, consequently, heart disease; and, therefore
    c) cholesterol and/or fat in the diet causes heart disease.

    Interestingly, I originally followed a tweet of his to this post, tho he didn't have any comment.

    I'll let the two of you duke it out if need be. ;).

  4. Geoff's Dad's doctor might defend his recommendation by pointing out that he doesn't believe in the lipid hypothesis. He could say "benefits of statins are apparent even if statistically separated from cholesterol lowering (the key point of the lipid hypothesis). If the Dad has had CAD, so this is a secondary prevention situation, the case for statins conferring benefit is relatively decent even if we don't understand the mechanism, so here is the scrip."

    Now, if the Dad is North of age 60 or 65, I would respond that the benefits of statins seem to disappear in the PROSPER study if HDL is above 45 or the LDL to HDL ratio is low (less than 3.3 as I recall), and Geoff's Dad's LDL to HDL ratio is outrageously good on the Paleo diet. I wish my ratio was near that.

    By the way, I am not defending the lipid hypothesis, just agreeing that even a believer with any degree of sophistication would not be pushing statins based on the lipid panel of Geoff's Dad.

  5. Geoff's Dad's situation is interesting. If all that is known is the lipid panel, the likelihood of a high LDL particle count in spite of the small calculated LDL volume is low. The TG of 55 in particular, but the HDL of 55 in a man as well, suggest pretty good size LDL particles and, therefore, the likelihood that there are a lot of LDL particles given a calculated LDL volume of 44 (Friedewald style, which tends to overestimate LDL where TG is below 100) is low.

    My parents and mother-in-law are all getting the advice to take statins. My mother, who has become a thin and frail woman of 86 — 105 lbs — was getting 80 milligrams a day of simvastatin (along with 10mg of ezetemibe – in a discredited Vytorin). Her main health problem is feeling like crap most of the time, for which one must immediately suspect a massive dose of statins and a massive shortfall of cholesterol in her body.

    Geoff, if someone were on statins and had your Dad's panel, a responsible doctor would view him as overtreated (in the absence of a man-bites-dog NMR or VAP, which I would not bother to get based on the evidence).

  6. Poisonguy,

    I'm not sure exactly what you mean, but no, they are not ever interchangeable. The lipid hypothesis states that in an individual or population, reducing blood cholesterol will reduce heart disease risk. The diet-heart hypothesis states the same about dietary fat.


  7. Steve,

    I wasn't making a recommendation. I was saying that his doctor's judgment didn't make much sense based on the lipid hypothesis, unless of course the doctor had information not presented here. I'm not sure statins are the best way to increase LDL particle size though.


  8. Your father's doctor's actions don't even make sense from the perspective of the lipid hypothesis, because if his total cholesterol is 110 mg/dL his dose of the drug should either be titrated down or the drug should be eliminated

    Why would you recommend this without having any idea via NMR or VAP his particle levels? They could be quite high even with a low cholesterol reading( via Fredenwahl) and isn't the current state of art in assessing heart disease based upon size and amount of particles that carry cholesterol and other sterols?

  9. Aghast and amused,

    Ahrens' choice of wording is rather bizarre, for the reasons you pointed out, as well as the facet that he describes it as a 'premise' derived from observed fact, when as far as I have ever understood a "premise" is something postulated that is not derived empirically.

    However, I think that if you read Ahrens' paper his logic is actually extremely sound, and he is just misusing some words. He is scientifically conservative and describes the need for testing in various forms in different populations, and the caveats of interpretation that would follow positive results.

    And this is not the beginning of the lipid hypothesis. It's just the beginning of the popular use of that particular term. The actual hypothesis begins with Anitchkov's cholesterol-fed rabbit in 1913. Anitchkov was quite brilliant, even time has proved him wrong on a few points.


  10. That's a fascinating choice of words by Ahrens:

    "The Lipid Hypothesis is the postulate…."

    And a "postulate", of course, is "something taken as self-evident or assumed without proof as a basis for reasoning."

    OK! No need to test a self-evident fact, is there?

    And then comes something stranger:

    "It is a premise based on the undisputed fact that [blah, blah]….but the premise has not yet been proved true to the satisfaction of epidemiologists and biostatisticians or of the medical community at large."

    Amazing! Ahrens spoke of a hypothesis, which he then called a postulate and described as an undisputed fact — that he then describes as unproved and (evidently) widely doubted by scientists and physicians.

    This twisted language makes a nice starting point for the bizarre story of the "Lipid Hypothesis".

  11. Just to clarify, I agree that dogmatically clinging to a hypothesis against the evidence is harmful, or that actions based on a hypothesis with insufficient evidence can be harmful, but it's the dogmatism or the inappropriate actions and not the hypothesis itself that is harmful. Scientific debate and research should be about finding the truth, not making war for or against a hypothesis.


  12. Geoff,

    I agree that they shouldn't be thought of as completely separate. However, we should use terminology that distinguishes between them, because precision clarifies in science while imprecision obfuscates.

    The diet-heart hypothesis, in its history, does rest firmly on the lipid hypothesis.

    However, while the term "lipid hypothesis" arose after the term "diet-heart," the actual lipid hypothesis itself dates to the cholesterol-fed rabbit model in the 1910's, and therefore predates the diet-heart hypothesis and has completely independent foundations. Anitchkov was VERY clear that he was not drawing any dietary conclusions from his rabbit model.

    I disagree that a hypothesis can be "detrimental." Your father's doctor's actions don't even make sense from the perspective of the lipid hypothesis, because if his total cholesterol is 110 mg/dL his dose of the drug should either be titrated down or the drug should be eliminated.

    What is important is whether there is any truth in the lipid hypothesis, and while there is a great deal to criticize in it, there is some truth.

    And no, the lipid hypothesis is not all based on blood cholesterol measurements.


  13. I can't disagree with any of the above statements. Still, I don't think that it is a given that we should be thinking of them as totally separate. The roots of the diet-heart hypothesis rest firmly in the lipid hypothesis.

    I know you have a plan to put out a review of Good Calories, Bad Calories at some point in the future, so I'm not sure if you've read it yet or not, but there is an extensive account of both in the book, as well as extensive take-downs of both theories. The two theories originated out of each other based on cholesterol measurements in the blood, most likely because this was one of the only things that they were capable of measuring at the time, like the story of the man franticly looking for his keys under the lamp in the dark parking lot.

    It's also quite arguable as to which is more detrimental, and thus more important to take down. If the only proposed mechanism by which dietary lipids cause heart disease is via the lipid hypothesis, then a takedown of the lipid hypothesis is not necessarily a bad way to start when going after the diet heart hypothesis, although it's certainly not the only way, and there's a very good argument to be made against the diet-heart hypothesis within the framework of the lipid hypothesis. Still, I would argue that the lipid hypothesis is far more detrimental than the diet-heart hypothesis.

    To give an illustrative example, my father had high cholesterol and was on a statin, and after about a year of wearing him down, he finally agreed to go paleo for a couple of months (between thanksgiving and christmas, then a cheat on christmas and new years, and then semi-strict since then). He got his cholesterol measured recently and it's ~110 total, 55 TG, 55 HDL, but his doctor is still strongly advising against going off the statins. I saw an article recently linked from Dr. Eades' blog about putting statins in the drinking water. There was a similar story from the BBC:

    On the other side of the spectrum, someone could be deathly afraid of dietary fat and still eat a relatively healthy diet, getting most of their calories from carbs in the form of rice and potatoes and the rest from lean meats and fish, and "good fats" like olive oil and avocado.

  14. Geoff,

    I agree that if the lipid hypothesis is wrong, the diet-heart hypothesis must either be wrong, or need to be modified in order to provide an alternative mechanistic explanation.

    However, the lipid hypothesis being wrong does not necessarily imply that the diet-heart hypothesis in its core structure and dietary prediction is wrong, and the correctness or incorrectness of the diet-heart hypothesis has no bearing whatsoever on the status of the lipid hypothesis.

    Moreover, while the lipid hypothesis is almost certainly wrong according to a literal and precise reading of Ahrens' 1976 definition, it is at least arguable whether a solid theory of heart disease should retain some lipid hypothesis-derived core. In other words, the lipid hypothesis needs to be substantially modified from Ahrens' definition, but there is still an overwhelming amount of evidence that blood lipids play some role in the development of atherosclerosis.

    By contrast, there is 1) much less experimental evidence about the diet-heart hypothesis than the lipid hypothesis in general, and 2) the balance of that evidence overwhelmingly favors the rejection of the diet-heart hypothesis in its entirety.

    Therefore, using the flaws of the lipid hypothesis to take stabs at the diet-heart hypothesis is incredibly backwards.


  15. But since the diet-heart hypothesis rests on the lipid hypothesis, by asserting that the lipid hypothesis is wrong/flawed etc., by the transitive property we're also saying that the diet-heart hypothesis is wrong. As such, it doesn't seem like much of a stretch to just use the term "Lipid Hypothesis" to refer to either/or. Referring to them both is just redundant since it would seem that one (the diet-heart hypothesis) is essentially a subset of the other. At least when we're referring to the legitimacy of the lipid hypothesis itself.

  16. Doesn't the diet-heart hypothesis rest on top of the lipid hypothesis? i.e. the proposed mechanism by which dietary fat causes heart disease is by elevating blood cholesterol?

  17. Crunchy,

    I'll address this issue in an upcoming blog post.

    Andy and Donat, I agree with some of what Peskin writes, and I'll be making a more comprehensive response to his positions in my upcoming PUFA Report Part II. However, I disagree that there is any solid evidence for the physiological essentiality of linoleic acid, and I disagree that no studies have been done with oils that aren't rancid.


    1. Dear Chris,
      With Cardiolipin requiring mostly linoleic acid, I would say that it is fair to assume that this fatty acid is indeed essential, even if the diet provides plenty of EPA/DHA.

      Where did you publish your “PUFA Report Part II”? Google cannot find it.
      Kind regards,

  18. @Andy: A genuinely reasoned discussion of how the natural vs processed and the saturated (poly)unsaturated distinctions interact would indeed be extremely valuable, especially if it separated clearly what is known, what is an educated guess and what is simply unknown at this point. (Also, not always kept separate: recommendations based on knowledge of effects and prudent recommendations based on lack of it.)
    @Chris: OT? not sure. Cf. the parallel nature of the claims (which I noted earlier on a Chris Kresser thread):

    "diet SAFA increases LDL, LDL correlates with CVD" :: "diet omega 6 increases tissue ox PUFA (and 6/3 ratio), high ox PUFA (and 6/3 ratio)correlates with CVD"

    While the correlation being also disease-causation seems a lot more plausible in the second case, we really have no evidence (AFAIK) as to the effects of long term natural (unprocessed) PUFA consumption (again, parallel to Stephan's (nonTRANS) SAFA post)

  19. cont.
    The good O6 linoleic acid actually has a higher physiological requirement in the body based on tissue types, for instance there's NO omega 3 in your skin, it's all omega 6 & natural cholesterols, which your skin depends on for vitamin D synthesis when your skin is exposed to sunlight.

    "Figure 2 shows the parent omega-6 to omega 3 ratios of major organs along with the respective weights. Contrary to what many researchers think, the brain is comprised of a 100:1 parent omega 6-3 ratio, not 1:1. Most of the plasma free fatty acid and EFAs are derived from the triglycerides stored in the adipose tissue (body fat), and organs, including the brain, use these EFAs for structural incorporation."34 See the chart on page 253 here:,pdf.pdf

    On the other hand, the researchers & writers recommending high doses of omega 3 & EPA & DHA fish oil derivatives are advising the intake of what amounts to a pharmacological overdoses of EPA & DHA, which are merely very low conversion rate DERIVATIVES of the omega 3 substrate (2%) with very low physiological need in the body. The body can & does make the derivative forms on it's own, but only when needed and only in the amounts needed. Taking too much EPA & DHA fish oils can cause depression, a cascade of brain signaling & neurological problems, thins the blood too much causing bruising & excessive bleeding, which would not be good in an accident, and on & on. There is little correct understanding of these O3 oils & derivatives & their correct use as well.

    Lands entire presentation at that conference was based on the TRANS O6 in foods the military uses but conflates the good CIS forms in with the TRANS. I shudder to think how much of his 40+ career was based on this fundamental error. Even Wikipedia does a better job of explaining the benefits of O6 but even they don't make a clear distinction between the two.

    The good, unadulterated, CIS configured omega 6 linoleic acid is a vital and extremely necessary essential fatty acid, but this fundamental error in understanding and the failure to clearly distinguish between the two types is causing much confusion and even damage to the health of the public that needs every beneficial natural advantage that's available to them.

  20. Now you need to write a piece on the failure, even of PhD's like Bill Lands & Mary Enig, and the venerated Mark Sisson, to properly distinguish between the TRANS configured, oxidized, adulterated omega 6 linloeic acid vegetable poly oils that causes all the problems, and the good, pure, Un-adulterated, functional organic omega 6 linoleic acid that is the precursor to gamma linoleic acid & that is the base substrate for production of PGE1 prostaglandin, the body's most powerful anti inflammatory, & prostacyclin, the body's most powerful blood thinner & anti thrombotic.

    I can't begin to tell you how much I tire of & how much my ass drags from reading yet another article by someone that doesn't understand the difference between CIS & TRANS molecules decrying the BAD! inflammatory omega 6 linoleic acid & it's overabundance in the Standard American Diet that is often 20, 40, 800 or 10,642 times higher than omega 3, and then throws the foods containing the good CIS configured O6 into the discussion with the TRANS O6 as if there is only one form of the O6 linoleic acid molecule!

    “Omega-6 fatty acids give rise to both pro-inflammatory compounds and anti-inflammatory compounds. To say that they are bad because they produce pro-inflammatory compounds ignores the fact that they give rise to anti-inflammatory compounds as well and reflects a rather naive understanding of the biochemistry.”59,pdf.pdf page 269

    Talk about conflation & confusion. The other night I was watching the YouTube video of Bill Lands' presentation at the NIH D.O.D conference on Warfighter Nutrition, he goes off on peanuts having 4400 milligrams of O6, like that's a bad thing. IT'S NOT!! The oils in peanuts are fine, carry much oxygen & impart much energy, that's why we use it. But conflating the good sources of CIS configured O6 oils with the BAD, toxic TRANS configured oils creates MUCH confusion & precludes any proper understanding of the powerfully beneficial effects of supplementing the diet with the good CIS configured omega 6 oils.

    If the oil in the peanuts was rancid, oxidized & toxic YOU'D TASTE IT. I've eaten a bad walnut in times past, everybody in the nut eating world has, they are *not* pleasant. But you spit it out & carry on. That peanuts have 4400 mg's. of good, organic **CIS configured** O6 linoleic acid simply means you don't have to take your 3 grams of EPO, cold pressed sunflower or O6 safflower oil for that day, but you still need to take your 1 mg. of O3 flax oil to balance the two oils at around 2.5 to 1 because O6 & O3 work synergistically, you have to take both of them so you don't create an imbalance & because the body actually has a *MUCH* greater physiological need for omega 6 than omega 3 based on tissue types & requirements. cont.

  21. I am one of those individuals with FH, currently trying to weed through the information and determine how best to care for myself. Would you be open to sharing some of your advice, as based on your research, with me? I would be happy to send you specifics of any information that would be helpful for you.

    Thanks in advance –
    Crunchy Pickle

  22. I have adopted the stance of using "conjecture" in place of hypothesis, whenever I judge it tired, inadequate or in Pauli's exclusionary phrase "not even wrong"!

    Thus , the Lipid Conjecture or Diet-Heart Conjecture, or dare I say, the Cholesterol Conjecture.

    Hours of harmless amusement!

  23. Jeanne,

    People with low cholesterol do get heart attacks. There is nevertheless and overwhelmingly documented correlation between plasma cholesterol and heart disease, so that it can predict relative risk. In other words, a higher number of people with high cholesterol get heart disease than people with low cholesterol, even though both groups have people who do and don't get heart disease. That is what Ahrens was referring to.


  24. Melissa,

    To my knowledge, Keys didn't use the term "lipid hypothesis." I agree that there are other hypotheses about lipid, and that in some of them some people might use the term "lipid hypothesis," but if we accept that there are a group of different lipid hypotheses, the term "THE lipid hypothesis" becomes so useless that it shouldn't be used at all.

    No statements such as "the lipid hypothesis is true," or "the lipid hypothesis is false," or "the lipid hypothesis is flawed," can be meaningful if "the lipid hypothesis" doesn't refer to one specific hypothesis.

    People could resolve this by using the terms "the lipid hypothesis of atherogenesis," "the lipid hypothesis of schizophrenia," and "the lipid hypothesis of osteoporosis." In that case, "the lipid hypothesis of atherogenesis" would still be about blood lipids, and not dietary lipids.

    One could solve this issue by using "the dietary lipid hypothesis of atherogenesis," to make it clear that one is not talking about the original lipid hypothesis, although that is rather cumbersome.


  25. Oh never mind the lipid thing. I just looked it up and it seems it's a broader term than just "fat" or "alcohol" but encompasses several different classes of chemicals.

    I feel silly now…

    Just the same. *What* undisputed notion that people with high cholesterol get heart disease? I thought there were people with low cholesterol also getting heart attacks? I know *you* didn't say that but I'm kind of going WTF here.

  26. The sad part is that cholesterol is not a lipid. It's technically a waxy alcohol. Lipoprotein (which is not cholesterol) isn't even a proper lipid–it's a lipid plus a protein. Just as a lectin isn't a pure protein, but is a combination of a protein and a sugar.

    I would expect the media to get all that stuff confused, they're just journalists and most of them do not have a science background. But I expect better from actual scientists.

  27. Ha … glad Melissa commented first! Don't disagree with you that it's important to differentiate the two, and I'll certainly defer to you on the proper use of the term "lipid hypothesis."

    But good luck with the rest of the (mainstream) world. Most seem to believe/accept "dietary fat -> blood lipids" so it's just an extension of the proper definition!

  28. I don't know if I agree with this since it seems that lipid hypothesis has come to mean the group of hypotheses that postulate that fat causes disease.

    " On the opposite side, Keys postulated the lipid hypothesis: that excessive dietary lipid intake caused systemic diseases. "

    "the context of a lipid hypothesis where alterations in lipid homeostasis might underlie the pathogenesis of schizophrenia."

    For example, "lipid hypothesis of osteoporosis." here and

Leave a Reply

Your email address will not be published.