Health Hopes Spring Eternal

Michael Eades:

For those who don’t know the Framingham Heart Study was kicked off back in the late 1940s in Framingham, MA. under the direction of the National Heart Institute … "to identify the common factors or characteristics that contribute to CVD [heart disease] by following its development over a long period of time in a large group of participants who had not yet developed overt symptoms of CVD or suffered a heart attack or stroke." 

… this longitudinal study that is still going on today. … Researchers are monitoring diet, medication history, and a host of other parameters, and have generated God only knows how many papers using this data that has been collected over the past almost 60 years.

One of the first ideas that the researchers had was to look at how diet related to cholesterol levels and how diet related to the development of heart disease. I’ll give you the surprise ending right up front: there wasn’t any correlation.  What’s interesting about this part of the study, however, is that the researchers didn’t publish it at the conclusion of the study, so it never really saw the light of day. You’ve probably heard the old saying that doctors bury their mistakes. Well, researchers often bury outcomes they don’t expect and don’t welcome. …

These guys tried as hard as they could to show a correlation between diet and serum cholesterol and between diet and the incidence of coronary heart disease, but failed. The data conclusively demonstrated no such correlations.

Imagine my surprise then I discovered the yellowed news clipping … from the Framingham newspaper dated October 30, 1970, … Apparently, despite all the supporting evidence, Dr. Kannel, the director of the study and the guy listed as lead author, wasn’t buying into all this nonsense about there being no correlation. He felt the need to `clarify’ the already crystal clear findings.

The clipping begins:

Although there is no discernible relationship between reported diet intake and serum cholesterol levels in the Framingham Diet Study group, "it is incorrect to interpret this finding to mean that diet has no connection with blood cholesterol," Dr. William B. Kannel, director of the Framingham Heart Study has stated.

Hat tip to Seth Roberts, who sadly doesn’t seem to have learned the lesson:

"If it is impossible to find exposure-response between changes of blood cholesterol and atherosclerosis growth in 22 studies including almost 2500 individuals a relationship between the two, if any, must be trivial."  Which sounds reasonable. But an even larger number of clinical trials failed to find clear evidence that omega-3 supplementation reduces heart disease. Yet I am sure that, with a large enough dose, it does. …  I don’t know why the big clinical trials failed to point clearly in the right direction. I can think of several possibilities: …
GD Star Rating
Tagged as:
Trackback URL:
  • william

    I apologize for posting such a long comment. But it is important to understand that reduction in serum cholesterol and fish oil consumption are beneficial.

    Serum cholesterol levels and cardiovascular disease are linked for a significant percentage of the population (about 60% if I remember correctly). It is amazing the link is this strong.

    There are two important factors to consider:

    1/ For 90-95% of the population, reducing cholesterol dietary intake to zero produces a maximum reduction of only 20% in serum cholesterol levels. Life without cholesterol is impossible. The liver synthesizes 80-90% of the cholesterol in our bodies from animal or plant triglycerides we ingest.

    2/ Cardiovascular (CV) disease does not directly depend on serum cholesterol levels and it doesn’t directly depend on the HDL:LDL (good:bad) cholesterol ratio. Here’s why. High and low-density lipoproteins are large heterogeneous particles made of a complex mixture of lipid and proteins. The HDL:LDL ratio reported in serum cholesterol screens measures a weight-percent ratio which assumes the surface area of HDL = that of LDL. But studies show the relative average surface area of each type of lipid-protein complex varies throughout the population. The ability of HDL to remove cholesterol form the CV system and the ability of LDL to deliver (deposit) cholesterol to the CV system depends as much on the particles’ surface area as it does on their mass. So a person with a “good” HDL:LDL weight ratio can have a mediocre or poor HDL:LDL transport ratio, and vice-versa. HDL and LDL particle surface areas depend on their protein composition which in turn is determined by genetic factors.

    Unfortunately it is very expensive to measure LDL/HDL particle size, so the mass ratio is the only practical data available.

    The quote on omega-3 fatty acids, fish oil, is unfortunate. The evidence for benefit in primary heart disease is mixed. However the clinical evidence of benefit for lowering serum triglyceride levels and in preventing secondary heart disease is strong. It would be a mistake to assume fish oil supplements are just another snake-oil scam.

    Here are some selected quotes and references form the American Heart Association. (

    “There is strong scientific evidence from human trials that omega-3 fatty acids from fish or fish oil supplements (EPA + DHA) significantly reduce blood triglyceride levels (1; 2; 3; 4; 5; 6). Benefits appear to be dose-dependent, with effects at doses as low as 2 grams of omega-3 fatty acids per day. Higher doses have greater effects, and 4 grams per day can lower triglyceride levels by 25-40%. Effects appear to be additive with HMG-CoA reductase inhibitor (“statin”) drugs such as simvastatin (7), pravastatin (8; 9), and atorvastatin (10). ”

    “The American Heart Association, in its 2003 recommendations, reports that supplementation with 2-4 grams of EPA + DHA each day can lower triglycerides by 20-40% (11). Because of the risk of bleeding from omega-3 fatty acids (particularly at doses greater than 3 grams per day), a physician should be consulted prior to starting treatment with supplements.”

    “C-Reactive Protein (CRP) levels : The data on fish oils and CRP is mixed (12; 13). While omega-3 fatty acids from both plants (ALA) and fish (EPA+DHA) have been shown to reduce CRP in some studies, others have failed to show an effect. There is growing evidence that reducing CRP is beneficial towards favorable cardiovascular outcomes, although additional research is pending in this area. Although statin drugs, weight reduction, smoking cessation, and COX-2 inhibitors all appear to reduce CRP, the evidence regarding fish oil remains equivocal.”

    “Several well-conducted randomized controlled trials report that in people with a history of heart attack, regular consumption of oily fish (200-400 grams of fish each week equal to 500-800mg of daily omega-3 fatty acids) or fish oil/omega-3 supplements (containing 850-1800mg of EPA + DHA) reduces the risk of non-fatal heart attack, fatal heart attack, sudden death, and all-cause mortality (death due to any cause) (14; 15; 16; 17; 18; 19; 11). Most patients in these studies were also using conventional heart drugs, suggesting that the benefits of fish oils may add to the effects of other therapies. Benefits have been reported after 3 months of use, and after up to 3.5 years of follow-up. Benefits of supplements may not occur in populations that already consume large amounts of dietary fish (20).”

    “Experiments suggest that omega-3 fatty acids may reduce platelet derived growth factor (PDGF), decrease platelet aggregation, inhibit the expression of vascular adhesion molecules, and stimulate relaxation of endothelial cells in the walls of blood vessels (21).”

    1 Harris WS. n-3 fatty acids and serum lipoproteins: human studies. Am J Clin Nutr 1997;65(5 Suppl):1645S-1654S.
    2 Sanders TA, Oakley FR, Miller GJ, et al. Influence of n-6 versus n-3 polyunsaturated fatty acids in diets low in saturated fatty acids on plasma lipoproteins and hemostatic factors. Arterioscler Thromb Vasc Biol 1997;17(12):3449-3460.
    3 Harris WS, Dujovne CA, Zucker M, et al. Effects of a low saturated fat, low cholesterol fish oil supplement in hypertriglyceridemic patients. A placebo-controlled trial. Ann Intern Med 1988;109(6):465-470.
    4 Roche HM, Gibney MJ. Postprandial triacylglycerolaemia: the effect of low-fat dietary treatment with and without fish oil supplementation. Eur J Clin Nutr 1996;50(9):617-624.
    5 Montori VM, Farmer A, Wollan PC, et al. Fish oil supplementation in type 2 diabetes: a quantitative systematic review. Diabetes Care 2000;23(9):1407-1415.
    6 Grimsgaard S, Bonaa KH, Hansen JB, et al. Highly purified eicosapentaenoic acid and docosahexaenoic acid in humans have similar triacylglycerol-lowering effects but divergent effects on serum fatty acids. Am J Clin Nutr 1997;66(3):649-659.
    7 Nordoy A, Bonaa KH, Sandset PM, et al. Effect of omega-3 fatty acids and simvastatin on hemostatic risk factors and postprandial hyperlipemia in patients with combined hyperlipemia. Arterioscler Thromb Vasc Biol 2000;20(1):259-265.
    8 Contacos C, Barter PJ, Sullivan DR. Effect of pravastatin and omega-3 fatty acids on plasma lipids and lipoproteins in patients with combined hyperlipidemia. Arterioscler Thromb 1993;13(12):1755-1762.
    9 Nakamura N, Hamazaki T, Ohta M, et al. Joint effects of HMG-CoA reductase inhibitors and eicosapentaenoic acids on serum lipid profile and plasma fatty acid concentrations in patients with hyperlipidemia. Int J Clin Lab Res 1999;29(1):22-25.
    10 Nordoy A, Hansen JB, Brox J, et al. Effects of atorvastatin and omega-3 fatty acids on LDL subfractions and postprandial hyperlipemia in patients with combined hyperlipemia. Nutr Metab Cardiovasc Dis 2001;11(1):7-16.
    11 Kris-Etherton PM, Harris WS, Appel LJ. Fish consumption, fish oil, omega-3 fatty acids, and cardiovascular disease. Arterioscler Thromb Vasc Biol 2003;23(2):e20-e30.
    12 Chan DC, Watts GF, Barrett PH, et al. Effect of atorvastatin and fish oil on plasma high-sensitivity C- reactive protein concentrations in individuals with visceral obesity. Clin Chem 2002;48(6 Pt 1):877-883.
    13 Hong H, Xu ZM, Pang BS, et al. Effects of simvastatin combined with omega-3 fatty acids on high sensitive C-reactive protein, lipidemia, and fibrinolysis in patients with mixed dyslipidemia. Chin Med Sci J 2004;19(2):145-149.
    14 Bucher HC, Hengstler P, Schindler C, et al. N-3 polyunsaturated fatty acids in coronary heart disease: a meta- analysis of randomized controlled trials. Am J Med 2002;112(4):298-304.
    15 Anonymous. Dietary supplementation with n-3 polyunsaturated fatty acids and vitamin E after myocardial infarction: results of the GISSI-Prevenzione trial. Gruppo Italiano per lo Studio della Sopravvivenza nell’Infarto miocardico. Lancet 1999;354(9177):447-455.
    16 Burr ML, Fehily AM, Gilbert JF, et al. Effects of changes in fat, fish, and fibre intakes on death and myocardial reinfarction: diet and reinfarction trial (DART). Lancet 1989;2(8666):757-761.
    17 Burr ML, Sweetham PM, Fehily AM. Diet and reinfarction. Eur Heart J 1994;15(8):1152-1153.
    18 Singh RB, Niaz MA, Sharma JP, et al. Randomized, double-blind, placebo-controlled trial of fish oil and mustard oil in patients with suspected acute myocardial infarction: the Indian experiment of infarct survival–4. Cardiovasc Drugs Ther 1997;11(3):485-491.
    19 Marchioli R, Barzi F, Bomba E, et al. Early protection against sudden death by n-3 polyunsaturated fatty acids after myocardial infarction. Time-course analysis of the results of the gruppo italiano per lo studio della sopravvivenza nell’infarto miocardico (GISSI)-prevenzione. Circulation 2002;105(16):1897-1903.
    20 Nilsen DW, Albrektsen G, Landmark K, et al. Effects of a high-dose concentrate of n-3 fatty acids or corn oil introduced early after an acute myocardial infarction on serum triacylglycerol and HDL cholesterol. Am J Clin Nutr 2001;74(1):50-56.
    21 Connor WE. Importance of n-3 fatty acids in health and disease. Am J Clin Nutr 2000;71(1 Suppl):171S-175S.

  • Robin Hanson

    William, yes your comment is excessively long. And for such a long comment it is surprising that you don’t make clear whether you disagree or not with the claims quoted. The claims say no correlation between diet and heart diseases or cholesterol, between cholesterol and atherosclerosis growth, and between heart disease and omega-3 supplementation. Do you agree or disagree?

  • Jor

    One of the claims I think is very different from the others. I’m not a cardiologist or a nutritionist but my own understanding is that the evidence for serum cholesterol being correlated to and causing athersclerosis growth as well as increasing the risk of cardiac events is VERY STRONG. We have randomized trails with medicines that reduce cholesterol, shown to reduce cardiac risk.

    Whether you can alter serum cholesterol with diet, or effect your cardiac risk with diet, I’m not as familiar with the data on that, but I am under the impression that its not very good quality.

  • Paul Sas

    A fascinating phenomenon, described by Gina Kolata in the NYT, was articulated by the very acute statistician at Berkeley: [begin quote from 2/12/2006 NYT]
    “Scientists and the public, said David Freedman, a statistician at the University of California at Berkeley, “have this wonderful capacity for ignoring negative evidence.”
    Dr. Freedman, who has written books on the science and history of clinical trials, says he is reminded of a story about a pioneer in the medical application of statistics, Pierre-Charles-Alexandre Louis.
    In the 1830′s, Dr. Louis studied the effect of bloodletting, or bleeding — the standard treatment of the time — on pneumonia.
    “The data showed that bleeding didn’t work,” Dr. Freedman said. But, he said, “Dr. Louis rejected this as terrifying and absurd.”
    So, he made a recommendation: bleed earlier and bleed harder.

  • Floccina

    From reading a lot of this diet stuff, as far as I can tell as long as you have no vitamin, mineral or amino acid deficiencies and you are not eating a lot of poisons your diet (poison mushrooms etc.) what you eat has next to nothing to do with your health. I think that so many people cling diet as a very important factor in health shows a bias. Maybe that bias comes from a time when people suffered due to a lack of certain vitamins (vitamin c in sailors) and minerals (iodine in inland dwellers). Or it could be the “Listerine affect”, that is if it taste good it cannot be good for you.

  • Doug S.

    The first rule of bad management:

    “If something doesn’t work, do more of it.”

  • Constant

    I think that so many people cling diet as a very important factor in health [...]

    One explanation is that diet is something that we do have some influence over, so we can do something about it. As a point of contrast, we can’t do much about our genetic makeup.

  • ClarkW

    …so how do we overcome such “Bias” in professional medical research ??
    ” Facts are stubborn things, but statistics are more compliant. ”
    {– Laurence J. Peter}

  • Robin Hanson

    Jor, can you cite those randomized trials?

  • Mike Kenny

    I think Robin Hanson suggested that much health care is a way of signaling to others that we care about them and are trustworthy (why invest in someone if you’re going to abandon them?). Maybe the same is the case for researching health care.

    Also, if health care research, along with certain health care practices, is lacking in positive results, a researcher or provider or funder of research of health care, might feel not only that they lose their credibility of caring for others (“You tricked us! You pretended to care!”) but that even if no one questioned their motives, they’d be seen as incompetent and would lose some status in their group (“You wasted all this money and time on nothing?!).

    Some possible answers: try to make it easier for people to admit error. Don’t invest so much of yourself in one practice that might not end up doing any good.

  • Jor


    There is soo much data on lipid lowering drugs and heart disease, that I’m not even sure I can pick out representative studies. If you have access to UpToDate, they appear to have a good review of the major trials:

    If you dont’ have access to uptodate, here are what might be some good starting points (taken from uptodate review), but as stated before, not sure if these are the best studies.

    Trials showing lipid lowering meds improve survival.
    1.)… The Long-Term Intervention with Pravastatin in Ischaemic Disease (LIPID) Study Group. NEJM.
    2.)West of Scotland Coronary Prevention Study Group. NEJM

    Trials showing Lipid Lowering medications reduce athersclerosis progression
    1.) Progression of coronary artery disease predicts clinical coronary events… Circulation.
    Use of intravascular ultrasound to compare effects of different strategies of lipid-lowering therapy… Circulation

  • nick

    You are only seen to care for others if you can plausibly claim that your knowledge is helping them. A discovery in genetics doesn’t directly do that, because there is nothing one can directly do to change their genetics. Indeed, telling someone they have a genetic defect signals meanness, not caring. Moralizing about diet, on the other hand, is a plausible signal of caring.

    Sometimes one diet recomendation can send multiple signals. For example vegetarianism signals caring mainly through the implication of caring for animals, but also through the implication of caring by practicing and urging a healthier diet.

    Health care is not the only part of life where signalling dominates. Look at energy conservation: the focus is on the most visible, for example light (look at all the hoopla over switching to lower power light bulbs). Cars with the “Prius” label sell far better than the equivalent hybrid under a traditional car name (“Accord”, “Civic”, etc.). Meanwhile we go on spewing out more greenhouse gases than ever, but we feel good about it.

    Politics and voting have no personal repercussions so they almost entirely a matter of signalling. Democrats signal that they care for people, even if just with other people’s money. Republicans signal that they care about our tribe, even if just with other people’s lives.

    Actual cause and effect — that is, whether the act taken or recommended is actually produces the claimed helpful consequence — is irrelevant to signaling unless it is obvious. For a signal to succeed it need only be plausible that a person could believe that their act or other’s acting on their proclaimed beliefs will help. Whether it will really help is otherwise irrelevant. It is intent, not reality, that counts.

  • nick

    Jor, the studies you cite in conjunction with the many studies showing no correlation between lipid levels and heart disease are consistent with the theory that the cholesterol-lowering drugs help lower heart disease not by lowering cholesterol, but by one of their many “side effects.”

  • nick

    Bottom line: the vast majority of human conversation and activity, including almost all of what passes for academic debate, is about signalling — mainly signalling altruism — not about truth or reality. Robin’s observations about health care are just the tip of the iceberg.

  • Robin Hanson

    Nick, yes, signaling is behind a lot more human behavior than we realize. This is an easier to truth to accept about other than about ourselves.

  • nick

    Robin: “This is an easier to truth to accept about others than about ourselves.”

    I don’t know about myself, but some people have Asperger’s or other syndromes that disrupt their signalling in such a way that they don’t know or care much about social consequences to themselves. They either don’t recognize the social consequences of poor signalling or have given up trying to influence them. The more severe their social handicaps, the more likely they are to be thinking and telling the narrow truth about what they see as purely rational issues (e.g. whether turning off indoor lights when it’s 30 degrees outside and the thermostat is set to 70 really saves much energy. The Asperger’s explicit answer: of course not, because most of the light energy is recaptured as indoor heat and makes the heater work that much less. Social signaller’s implicit answer: it does seem to save energy and is a very visible sacrifice, so it’s a great idea).

    Academia is the art of social signalling while pretending that one is debating reality and truth.

  • Seth Roberts

    Robin, clinical trials of whether omega-3 is good for your gums have failed to show clear effects. Tyler’s experience shows that they reached the wrong answer.

  • Robin Hanson

    Seth, I don’t follow you; are you saying that a single person’s experience is evidence for omega-3 effectiveness, overturning the many clinical trials?

  • Seth Roberts

    Yes, Robin, that is pretty much what I’m saying. When one person’s experience is that clear, it shows that the clinical trials were wrong. There weren’t “many” in this case.

  • Robin Hanson

    Seth, one person’s experience can’t tell you that people on average benefit from something; it could help some people and hurt other people. Is there a future clinical trial you are willing to bet on?

  • Mike Kenny

    I was pondering this regarding self experimentation: how often is it likely a personal experience would reflect an average experience among people? If I pinch myself and see that as a result I feel pain, and assume others would report the same cause and effect relationship, I’d be right, but if I read Thomas Pynchon and liked it, and assumed everyone else did, I’d be wrong.

    This connects with what I said in the October Open Thread section. It’d be useful to know how often certain things seem to be true, in more than just an intuitive way. I have an intuitive sense of how often people are right when they tell me something, but I don’t have an actual record. It’d be great to have a table based on sampling that suggests the probability of certain common experiences being true. One useful probability figure to know would be the answer to the question “How often is my personal experience generalizable?” It may be very low (though would it? Wouldn’t the likelihood of your experience being average be the most reasonable assumption?), but even a low probability might be useful if the potential benefits are great of generalizing based on personal experience, and the costs are very likely low.

  • Richard Hollerith

    What nick said.

  • Robin Hanson

    Nick, is there any data to support your hypothesis?

  • Doug S.

    Seth, just because a person gets better after seeing a quack doesn’t mean that the quack cured them.

  • Seth Roberts


    My belief that in case of omega-3 and gum disease, the clinical trials — which failed to make the connection — were wrong and Tyler’s experience — which showed a strong connection — was right is supported by lots of facts: 1. Thousands or more cases where supplying a missing nutrient quickly cured a problem. E.g., lime juice and scurvy. 2. Evidence that gum disease is due to inflammation. 3. Evidence that omega-3 reduces inflammation. 4. My self-experimental results that show that flaxseed oil in certain doses (which Tyler used) had whopping effects on brain function. 5. The effects of flaxseed oil on my gums.

  • Seth Roberts

    I’d bet on a clinical trial about omega-3 and gum disease that

    1. used flaxseed oil.

    2. got the dose right (at least 1 Tbsp/50 pounds body weight)

    3. measured compliance so that non-compliant subjects were excluded.

  • Douglas Knight

    3. measured compliance so that non-compliant subjects were excluded.

    translation: you’re not willing to bet.

    Compliance is a big deal and it cuts both ways: it means that the effect measured in the study is less than the potential effect, but it also means that the effect measured in the study is greater than the effect will be in practice. I suspect this (and other ways that studies aren’t representative) is an important part of the discrepancy between studies and practice.

  • g

    Since different people will be capable of different levels of compliance, it would be good to decouple compliance from efficacy-given-compliance. If some dietary change is beneficial when actually followed but (for whatever reason) not usually followed by participants in studies, someone with above-average motivation or different tastes might be able to get more benefit from the change than non-decoupled studies would suggest. If they knew about it, which they never would if non-decoupled studies were the only ones.

    (Note that decoupling compliance from efficacy-given-compliance doesn’t have to mean just excluding non-compliant subjects. You could, obviously, publish the numbers both ways.)

    This all seems very obvious, but it also seems to make it silly to be dismissive about someone who says he thinks a good study should enable this sort of decoupling, so maybe it’s less obvious than it seems.)

  • nick

    Robin, my observations were based on no experimental data that I know of, just my own personal experience and some combination of reasoning and gut instinct. Anyway, if somebody wishes to test my theory here is a more precise statement: normalizing for IQ and education, an Asperger’s person will have a more accurate opinion of a physical fact (such as the energy savings from turning a light off inside when it’s cold outside and one is trying to heat the house) than a a non-Asperger’s person, where opinion of that fact can be biased by social signalling considerations (such as the benefit from being seen as a person who conserves energy). There may be other syndromes besides Asperger’s involving a deficit of social signalling that might also work here.

  • A broader view

    Does Diet AffectCholesterol?

    The usually insightful Overcoming Bias had a post about the Framingham study, a logitudinal survey which found no correlation between diet and cholesterol. Robin quotes an article from the Framingham newspaper; his take is that Kannel is trying to be c…