by Marion Nestle
Aug 6 2015

At last: two industry-funded studies with results that do NOT favor the sponsor’s interest

As regular readers know, I’ve been posting studies funded by food companies with results favorable to the companies’ interests whenever I run across five of them.   Since mid-March, I’ve posted 7 such collections for a total of 37 studies (two of the posts listed 6 studies).  These are all papers published since March.

With each set, I asked readers to send examples of studies that do not favor the sponsor’s interest.

They are rare, but do exist.  I’ve been sent two so far.  I’m guessing it will be a long time before I collect five, so have a look:

Butter increased total and LDL cholesterol compared with olive oil however resulted in higher HDL cholesterol than habitual diet. Sara Engel and Tine Tholstrup.  American Journal of Clinical Nutrition, July 1, 2015, doi: 10.3945/​ajcn.115.112227.

  • Conclusions: Moderate intake of butter resulted in increases in total cholesterol and LDL cholesterol compared with the effects of olive oil intake and a habitual diet (run-in period). Furthermore, moderate butter intake was also followed by an increase in HDL cholesterol compared with the habitual diet. We conclude that hypercholesterolemic people should keep their consumption of butter to a minimum, whereas moderate butter intake may be considered part of the diet in the normocholesterolemic population.
  • Sponsor: Danish Dairy Research Foundation
  • Comment: The data clearly show that butter raises blood cholesterol levels.  The authors spin it as positively as possible—higher HDL and it’s OK for people with normal cholesterol to eat moderate amounts of butter—but they make the downside quite clear.  In this study, “moderate” butter means 4.5% of calories or just 2/3 of a tablespoon for someone eating 2000 calories.  That’s not much, alas.

Influence of Pistachios on Performance and Exercise-Induced Inflammation, Oxidative Stress, Immune Dysfunction, and Metabolite Shifts in Cyclists: A Randomized, Crossover Trial.  David C. Nieman, Johannes Scherr, Beibei Luo, Mary Pat Meaney, Didier Dréau, Wei Sha, Dustin A. Dew, Dru A. Henson, Kirk L. Pappan.  PLoS One, November 19, 2014. DOI: 10.1371/journal.pone.0113725

  • Conclusion: In summary, 2-weeks pistachio nut ingestion was associated with reduced 75-km cycling time trial performance and increased post-exercise plasma levels of raffinose, sucrose, and metabolites related to leukotoxic effects and oxidative stress.
  • Funder: This work was supported by American Pistachio Growers. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
  • Comment:  It’s obvious that the funders had no role.  I’ll bet they are quite unhappy with the results.

This second study came out last year but I’ll take any of these I can get.  Please do send.

But if I’m just counting since March, the ratio is 37 studies favoring the sponsor’s interest, to 1 that doesn’t.  Coincidence?  I’m not convinced.

  • Food Scientist

    Coincidence? No, of course not, nor do I see anyone suggesting that it is.

    However, the results of your meta-analysis are not due to nefarious reasons, as you imply ad nauseam. Rather, simply realize that a company is unlikely to provide funding for academic research unless they ALREADY (key word!) have a very good idea that the results will be positive. They may have previously done in-house testing/experimentation that produced positive results (which you would surely eviscerate if published), or a company may form a hypothesis about an ingredient/product based on previously published literature. Why is this concept so hard for you to understand and/or accept? This is critical thinking 101.

    If you go into this process expecting to see a 50:50 split between favorable/unfavorable results for industry, you are guaranteed to be alarmed. Dr. Nestle, your analysis would be SO MUCH more valuable and worthwhile if you dug into these studies and picked apart the methodology or interpretation of results. It’s more work, but it is the right thing to do.

  • TedHutchinson

    I think everyone knows that Higher levels of Serum cholesterol concentrations lead to lower levels of all-cause mortality in older people. So as far as older people (over 65) are concerned is butter eating and it’s propensity to raise HDL necessarily a bad thing?
    The problem here isn’t the data the research provided but the interpretation placed on it by those with a vested interest in perpetuating the myth that higher cholesterol is always a bad idea.

  • Kateye70

    Just curious if the butter industry study looked at people who eat high carb-low fat diets only. It states “habitual diet” but doesn’t specify what that “habitual diet” consisted of.

    It seems that most people who have switched to high-fat, low carb (ketogenic) diet find the opposite, lower cholesterol, because the fat is being burned for fuel instead of carbs and not being converted to cholesterol.

  • Andre

    Links please. There may be a u-shaped curve, but not a reduction in risk as cholesterol goes up as you are implying.

  • George

    The butter study you cite does not put a “positive spin” on the results.
    The HDL increase is dismissed, using the concept of Mendelian analysis, which is probably a perverse interpretation of the idea (butter is not a gene, but an upstream factor – HDL is still an important risk factor).
    The importance of the LDL increase is exaggerated, using evidence from drug trials (butter is not drugs either, and drugs have multiple effects and side-effects other than lowering LDL).
    This twisting of the evidence is the exact opposite of a positive spin.
    You can take your Truther mindset too far and get consumed by the idea of a grand conspiracy suppressing the facts about butter, but the fact that saturated fats raise LDL and HDL is common, not suppressed, knowledge.

  • George

    This blog pulls together all the studies.
    Some are U-shaped (3rd quartile lowest mortality) others are not (4th quartile lowest).
    But low cholesterol – you don’t want that, unless maybe heart disease is the only death you care about, and sometimes not even then.

  • TedHutchinson
  • Andre

    It does not pull “all the studies.” It cherry picks. Simple googling identifies others with different results.

    <193 cholesterol is not "low." Many people with cholesterol between 150 and 200 have atherosclerosis and suffer cardiac events. That the studies are even set up with 30% calories from fat.

    Low cholesterol can also be a marker/consequence of disease – shouldn’t be confused with a cause of it. I’d expect the long term chronically ill, especially those in institutional settings, to have low cholesterol levels as they would be fed very low cholesterolemic diets.

    Heart disease is a leading cause of death, and is positively associated with with cholesterol.

  • Pingback: Food-industry conflicts of interest: newspaper revelations and five more studies with expected results: the latest collection - Exploring the News()

  • George

    Rather than being cherry-picked, they appear to be the largest cohorts available for those populations. Further, if your hypothesis were correct, it would be impossible to find multiple large studies that contradict it – one man’s cherry is another man’s black swan.

    Prospective cohort studies are long-term studies designed to eliminate the reverse causation you describe, by excluding the already-ill or those who become ill or die within the first year or two.

    The cholesterol cut offs are set by quartiles (4 equal groups) not whatever the fashionable diagnostic cut-offs happen to be.

    Cholesterol is associated with heart disease because genes associated with high cholesterol are thought to be causative in heart disease. In hospital populations eating exactly the same food, people with heart disease still had higher cholesterol. This has nothing to do with butter, which has not been shown to be associated with heart disease. The countries that eat the most butter today, such as France and Switzerland, have some of the lowest rates of heart disease. Countries with very low dairy and saturated fat intakes, such as those of the former USSR and its satellites, have some of the highest rates of heart disease in the world.

    Replacing saturated fat with carbohydrate lowers cholesterol but no-one today thinks this lowers the risk of heart disease (it doesn’t in substitution analysis of prospective cohort studies, and it doesn’t in RCTs).
    Replacing SFA or carbohydrate with PUFA is associated with lower risk of heart disease, lower cholesterol, but not lower all-cause mortality. One might as well replace bread with nuts to get this result, and keep the butter where it is. In fact in substitution meta-analysis, replacing carbohydrate with PUFA is associated with more benefit than replacing SFA with PUFA.

    None of this evidence indicates that butter is “bad for you”.

  • Pingback: Food-industry conflicts of interest: newspaper revelations and five more studies with expected results: the latest collection | Dana Hender()

  • BluebirdofUnhappiness

    Thank you. Worth repeating: “Rather, simply realize that a company is unlikely to provide funding for academic research unless they ALREADY (key word!) have a very good idea that the results will be positive.” But you have to also understand that turning everything into a conspiracy makes for good press and can further one’s career.

  • Pingback: Industry-funded studies with results that do NOT favor the sponsor! The score since March: 50:3 | Dana Hender()

  • Pingback: Early-Autumn Notes | Eggplant & Olive()