Look AHEAD Scientists: Trying to Move the Deer Crossing

The Look AHEAD: Action for Health in Diabetes trial has been halted two years early. Here’s Gina Kolata in The New York Times:

The study randomly assigned 5,145 overweight or obese people with Type 2 diabetes to either a rigorous diet and exercise regimen or to sessions in which they got general health information. The diet involved 1,200 to 1,500 calories a day for those weighing less than 250 pounds and 1,500 to 1,800 calories a day for those weighing more. The exercise program was at least 175 minutes a week of moderate exercise.

But 11 years after the study began, researchers concluded it was futile to continue — the two groups had nearly identical rates of heart attacks, strokes and cardiovascular deaths.

It’s clearly a negative result for “eat less, move more” as a health strategy for obese diabetics.

Was “Eat Less Move More” Harmful?

A few Paleo bloggers are not surprised; indeed, Peter Dobromylskyj speculates that all-cause mortality – which Ms. Kolata and the NIH press release do not report – may have been higher in the “eat less, move more” intervention group:

It seems very likely to me that more people died in the intervention group than in the usual care group, but p was > 0.05.

Call me a cynic, but I think they stopped the trial because they could see where that p number was heading.

Peter may be a cynic but cynics are sometimes right, and I will bet that he’s right about this. In general, calorie restriction and exercise are better attested against cardiovascular disease than against other health conditions, so if death rates from CVD were identical in the two arms after 11 years, it’s quite likely death rates from other causes were higher in the intervention arm.

Our Theory

We discuss in our new Scribner edition two reasons why “eat less, move more” can backfire:

  • On a malnourishing diet, “eat less” means even greater malnourishment. Less of a bad diet is a worse diet.
  • Excessive exercise may over-stress the body and harm health. In diseased people, the volume at which exercise becomes excessive may not be that high.

On the other hand, ultimately some form of “eat less, move more” is needed if optimal health is to be attained:

  • An energy deficit – eating less than the body expends – is necessary to lose fat mass, and obesity is probably incompatible with optimal health.
  • About 20 to 30 minutes of exercise per day at the intensity of running or jogging is needed for optimal health, probably due to the role of daytime activity in entraining circadian rhythms (see “Physical Activity: Whence Its Healthfulness?”, October 11, 2012). Most people would need to “move more” to achieve this.

So the challenge in weight loss is two-fold: It’s necessary to adopt a healthy diet in which malnourishment doesn’t occur despite calorie restriction, and to find a healthy level of exercise that improves health without overstressing the body.

Look AHEAD: Bad Dietary Advice

The Look AHEAD Study Protocol tells us what the intervention group was told to do.

From page 29, here is the diet advice:

The recommended diet is based on guidelines of the ADA and National Cholesterol Education program [96,97] and includes a maximum of 30% of total calories from total fat, a maximum of 10% of total calories from saturated fat, and a minimum of 15% of total calories from protein.

This gives 55% carbs and probably 10% omega-6 fat. The omega-6 intake is far too high – for weight loss and good health, omega-6 intake should be less than 4% – and so is the carb intake – for diabetics, reducing carbs to 30% or less is highly desirable.

From page 30, here is the exercise advice:

The physical activity program of Look AHEAD relies heavily on unsupervised exercise, with gradual progression toward a goal of 175 minutes of moderate intensity physical activity per week by the end of the first six months. Exercise bouts of ten minutes and longer are counted toward this goal. Exercise is recommended to occur five days per week.

Moderate-intensity walking is encouraged as the primary type of physical activity.

I think this is reasonable advice. It translates to 35 minutes per day for 5 days. The intensity is quite low. This level of exercise is hardly likely to be excessive; indeed, it’s probably grossly insufficient for optimal health. It represents about a mile and a half of walking per day, five days per week. This may have been a homeopathic level of activity.

There is another reason the exercise may have produced no observable benefit. Since I believe the health benefits of exercise occur primarily through circadian rhythm entrainment, it’s likely that daytime exercise is much more beneficial than night-time exercise. Night-time exercise might be ineffective or even harmful to health if it disrupts circadian rhythms.

Unfortunately many people find it difficult to find time during the day for exercise. If the walking was performed at night, even the modest benefits of the activity may have been lost.

Weight and Health: What’s the Direction of Causation?

The one “success” of Look AHEAD was that it brought about some weight loss: the intervention group lost 5% of their original weight.

We know that obesity is associated with poor health. Since causation implies correlation, the existence of this correlation suggests that either (1) obesity causes poor health, (2) poor health causes obesity, or (3) some third factors cause both obesity and poor health.

The Look AHEAD study presumed (1) – that obesity causes poor health. The “eat less, move more” intervention was wholly directed at weight loss. If obesity is the cause of poor health, Look AHEAD should have improved health. It didn’t. This tells us that the direction of causality is either (2) or (3). Obesity doesn’t impair health; other factors that impair health cause obesity.

It’s easy to make faulty inferences about the direction of causation. The Look AHEAD scientists made the same mistake this woman did:


The basic flaw in the Look AHEAD study was that it was designed to bring about weight loss, and hoped that weight loss would improve health.

A better intervention would seek to improve health through a more PHD-like diet and through circadian rhythm therapies. Successful health improvement would, more than likely, lead to weight loss.

For the overweight and for diabetics, the focus should not be on weight, but on health. Improve health, and weight loss will follow. Focus on weight with a simple-minded “eat less, move more” intervention without tending to the quality of your diet and lifestyle, and you might be doing yourself more harm than good.

Leave a comment ?


  1. Do we know if the participants truly followed a “rigorous diet and exercise program”? Did the participants fill out a survey (lied)?

    • Well, they started the patients off with a portion-controlled diet in which they delivered the food to the patients. They also started with weekly meetings, tapering to biweekly and then monthly.

      The press releases seem to suggest that enough people followed the program to make a difference, eg in weight.

  2. Paul,

    Excellent summary

    As I commented on CarbSane (before realising you had said all this here), the results have disproven their primary hypothesis – will the study authors come out and openly say as much?

    No doubt we will get things like “the participants didn’t lose enough weight to make a difference”, which ignores the fact, as you highlighted, that losing weight, in and of it self, *doesn;t* make any difference.

    Philosopher Karl Popper wrote that progress in science occurs by old theories being disproven, and replaced with new theories, that better fit the observations. It would appear that we are at such a point here.

    Naturally, disruptive results like this can be very damaging to those who are highly invested into them, but they are certainly beneficial to the advancement of health of the people – which is the whole idea of this stuff.

    The Hippocratic oath starts out with “first, do no harm..” It will be interesting to see if this intervention did indeed cause more harm than doing nothing.

    If they release all their data, that is…

  3. Hey Paul, nice discussion on this study’s interesting findings. I do have some ponderings on your weight and health section perhaps you can address.

    I personally think obesity promotes poor health (to some degree) and I don’t fully reject option 1. This study doesn’t support the causative role of weight on poor health, but in this post you suggested a possible explanation for the lack of correlation this when you brought up that the experimental group was eating a less nourishing diet.

    I guess I’m saying maybe we shouldn’t so readily discard the idea that obesity itself promotes ill health, but accept that it is one of a number of factors affecting health, of which the nutrient content of the diet (or lack thereof) is another.

    Does this seem like a reasonable possibility?

    • Hi M,

      Yes, I agree in practice. I think many of the same factors cause obesity and poor health. To fix your health you have to remove these factors, and to normalize weight you have to remove these factors. Any effective program of weight normalization will improve health, and any lifestyle or diet that causes obesity will generally harm health.

      Obesity is a protective response of the body, so I wouldn’t say obesity per se promotes poor health, but everything that brings about obesity does harm health, so maybe the best phrasing is “obesogenesis promotes poor health.”

  4. I think Dr. Arya Sharma had the best commentary on this study. Basically it was underpowered to detect significant differences between groups because the total event rates were low. They discontinued it not because of adverse effects, but because they decided the study was hopelessly underpowered.


    “This reduction in body weight was accompanied by significant improvements in glycemic control and numerous other health benefits (e.g. decreased sleep apnea, improved mobility and quality of life).

    However, given a remarkably low incidence of ‘hard’ endpoints in both the interventions and control groups, it became evident the the study would stand little chance of demonstrating superiority of the lifestyle intervention in terms of preventing cardiovascular complications.”

    Peter D’s speculations are baseless and too convenient for my tastes.

    Multiple controlled trials in multiple racial/ethnic groups have shown that a very similar intervention (modest weight loss plus exercise) reduces the incidence of type 2 diabetes by up to 60 percent! I discuss these studies in the link below. This type of intervention is clearly effective at improving health under some circumstances. It just didn’t pass statistical muster in the Look AHEAD study.


    • Thanks, Stephan.

      If the incidence of hard endpoints was “remarkably low” in both arms then that would suggest they were both health improving. If so, a victory for standard guidance. Or, perhaps a case of mean reversion — people who had inferior health at the start of the study reverted to average health over its 11 year course.

      Since according to Dr Kahn diabetes increases heart disease risk 2.5-fold, mean reversion is all you would need to show a 60% reduction in heart disease risk.

      It could be that the dietary guidance was too similar between the intervention and control groups and after the short initial period everybody ate the same diet, so there was no difference in outcomes between the two arms. It will be interesting to see the data when the studies are published, and how the outcomes of the two arms would compare with outcomes in the general population.

      It seems like a genuinely health improving diet ought to produce results fairly quickly. So probably studies with more participants and shorter durations, and more clearly divergent intervention and control arms, would generate better tests of the interventions.

      • I think one issue may have been with recruitment. Study selection criteria often exclude the sickest people with the highest health risks. Also, visiting a doctor regularly probably improves outcomes. Their endpoint rate in Look AHEAD was 1/4 of what the investigators were expecting. That must have been extremely disappointing.

        The intervention itself has already been proven valuable for reducing health risk (DPP trial and similar), just not in the context of pre-existing diabetes. Look AHEAD was basically a follow-up from the extremely successful DPP trial.

        • Hi Stephan,

          If that’s the case I wonder why they stopped the study. The starting ages were 45 to 76, correct? So now they’re 56 to 87. You’re bound to start having events soon.

          If events in both arms were 75% below the event rate in the general population, then it’s extremely important to see how the arms play out. If both interventions were highly successful, you’d think they’d want to find out how successful!

          Would you say what they’re thinking is that neither intervention was successful and they just recruited an unusually healthy population, and that both arms have now reverted to identical standard diets, so there will be no effect to detect?

    • Maybe Dr. Sharma’s comments sound better in context, but they sound like weasel words to me. So, the experiment was underpowered? Really, it is the failure of the intervention. The intervention itself was underpowered, and unable to do what was promised. How many thousands of dollars and bottles of Glucerna, for about 5-10 pounds and no differences in the important outcomes?

      • Yes, it’s a good question, they thought the study was adequately powered when it was proposed and funded, why were they so far off? Partly it was that their study population was healthier than they thought they would be, apparently due to exclusion of those with serious health impairment; but partly it must be that the intervention was less successful (and/or the control more successful) than they expected.

        I will be curious to see how the papers analyze the study when they come out.

        • I am looking forward to additional information as well. Part of the problem with studies such as this is that there really is no control. Folks in that group probably aren’t doing SAD anymore. Either they have been bombarded by the low-fat message or they have been following one or many of a number of alternative eating styles that we, at least, believe are better than sad.
          But then again, count me in the cynical camp. I just don’t see how they would stop a funded-study unless there was an ethical issue. Better to stop early and leave everything up in the air rather than to have the real hard data showing that their darling diet makes things worse.

        • …and I guess the picky statistician in me would still disagree that the study is underpowered. Based on the available data at the time (and I don’t know who was excluded in the earlier studies…), the study was powered enough to detect the differences that they expected to see. If there aren’t any measurable differences between the two options, or if the differences are running the other way, having a million people in the study can’t help them. The power discussion should be with an OC curve, clearly showing the ineffectiveness of a large sample size when the differences are tinier.

          • Hi EB,

            Yes, clearly the study was well powered enough to exclude effects above a certain size. But it seems the interventions may have been so poorly designed that they wouldn’t be expected to produce much. The intervention diet in many respects resembled SAD, the main difference was it had fewer calories. Presumably its primary effect was to make people hungry and cause them to quit complying. Insofar as the diet advice was different from SAD, the control group got similar advice.

            I would also be in the cynical camp if the study weren’t costing over $10 million a year. They probably had pressure from NIH to give up their funding if they couldn’t show they would generate results.

  5. It makes you wonder how much proof is it going to take that diabetics have business on a high carb, low fat, omega 6 diet before people wake up… and that you can’t exercise off your lunch time muffin on the treadmill.

  6. For healthy diet and exercise this study help improve the over stressing and protect from obesity…over training is another part for unhealthy in diet.
    Well this post suggested how improve health and lifestyle.

  7. I gotta say Paul, quoting Peter’s nonsense speculations here made me cringe. I’m honestly baffled that you would bet he’s right that there’s a body count being buried somewhere amongst the ELMMers.

    The more studies I read and reader feedback/input I get, the less I see carb restriction as required or even necessarily better for diabetes in the obese. Unless the person is more successful at weight loss and maintenance using LC, it’s a bandaid. I have a reader who has lost 150 lbs on the DASH diet (http://dashdiet.org/) and totally reversed T2 diabetes. There are many many others out there.

    The paleo boards are riddled with people who switched to a healthy diet that has not produced weight loss, and sometimes even induced weight gain.

    As to the PUFA, IF they are following the calorie guides and getting 10% of calories from any fat that’s 13g to 20g (for 1200-1800 cal/day) or roughly one tablespoon to five teaspoons. Even if that’s pure corn oil, we’re talking 7 to 11g O6. Is there any evidence that these sorts of intake are metabolically deleterious? I haven’t seen it. Also, dieters who adhere to low fat prescriptions rarely squander fat calories on the usual suspects for O6’s, they’re far more likely to “indulge” in low fat dairy. Before the paleo trend took hold, many many diabetics turned to Atkins/LC eating a lot of mayo, soybean oil-based dressings and nut flours and no doubt consumed far more O6’s than the ISI group in this study. This would include any diet trials where the LC leg was based on Atkins. Most do not supplement O3’s and pay little to no attention to O6’s.

    • Hi Evelyn,

      My thinking was that ELMM should be more beneficial against CVD than other leading causes of death (notably cancer), so if there was no reduction in death or other hard endpoints for CVD, then there would likely have been an increase for the others.

      However, based on what Stephan and Dr Sharma say about low event rates, it looks more like the intervention might have been beneficial but the effect was hidden because the control arm behaved the same; and that the low event rate was partly due to selection of participants that was unrepresentative of (and healthier than) the general population. In that case there’s really little we can say about this study, other than that it wasted a lot of money.

      Re diabetes, I think reducing total calories is most important, but (a) total calorie intake will generally be lower with 30% carb consumption than 50% carb consumption (this is discussed in our new book), and (b) apart from total calories, it’s still helpful to avoid an excess of individual macronutrients, and I don’t think a sedentary person needs more than 30% carbs.

      A discussion of evidence for keeping omega-6 below 4% of energy is in our book.

      Obesity is a complex multi-causal phenomenon and often diet is only one set of causes along with infections, lifestyle (eg circadian rhythm disruption, sedentariness), stress, etc. Since the key factors differ across persons, no one diet will have the same effect in everyone.

      I think Atkins/LC on omega-6 mayo would be an unhealthy diet. The ill effects would be attenuated under calorie restriction, but I think a high omega-6 maintenance diet would lead to problems in the long run.

      • Wayne D Johnson aka Prouddaddy

        Paul, any chance of providing one good study reference on the unhealthfulness of omega 6s here? My library will probably not be stocking a new edition of your book.

        I keep reading that O6s are inflammatory and thereby bad for you (primarily by paleo advocates), but, like Evelyn, I can’t find the proof.

        • Hi Wayne,

          Most of the evidence was also discussed in the first edition, if you have that.

          The mechanisms are a bit complicated — the harm is more due to lipid peroxidation than inflammation I think — but under the right circumstances omega-6 fats can be inflammatory. The inflammatory idea was William E. Lands’s work. You can search “Lands WE” in Pubmed, some of his papers are free access.

          • Wayne D Johnson aka Prouddaddy

            Of the 48 free text hits I got, I didn’t find any human RCTs related to O6s.

            Of the hundreds of RCTs by others involving PUFAs, most indicate a beneficial effect. What I have never found, however, is such a PUFA study that completely separated the effect of O6s from O3s. So, I’m still looking for good evidence, as opposed to theory, that O6 ingestion is unhealthy.

          • Hi Wayne,

            We have an analysis of 8 major omega-6 clinical trials in the book. They are fairly consistent in showing elevated mortality with more omega-6. Sometimes the death rates were dramatically higher, as in the Rose Corn Oil trial.

            Reduced mortality seen in omega-3 interventions, such as the Lyon Diet-Heart Study, is indirect evidence for an excess of omega-6 and supports the idea that people would benefit from reducing omega-6 intake.

          • Wayne D Johnson aka Prouddaddy

            I’ll check your book out of the library and take a closer look than I did the first time.

            The Rose trial was only 80 folks, they all had existing heart disease, and the results didn’t quite reach statistical significance.

            Perhaps more on point and more recent would be

            Johnson & Fritsche 2012 “Effect of Dietary Linoleic Acidon Markers of Inflammation in Healthy Persons: A Systemic Review of Randomized Controlled Trials”

            in which no effect was seen in markers like hsCRP.

            Don’t you just hate dueling trials?

          • I don’t think CRP is a clean or robust indicator of inflammation. Inflammation is a complex process.

          • Wayne D Johnson aka Prouddaddy

            Ah shucks, mine is 0.028 and I was hoping it meant good things. In any event, the Johnson review also checked fibrinogen, plasminogen activator inhibitor type 1, cytokines, soluble vascular adhesion molecules, and tumor necrosis factor – alpha.

            I’m certain you wouldn’t rely on a mechanism or process that couldn’t be measured, so what are the best measures of inflammation?

          • I’ll take a look at the paper. Thanks, Wayne. A panel of cytokines and other immune mediators would have it covered.

          • The omega-6 studies seem to be vegetable oil or worst-case sources. What if the dietary PUFA came from whole foods like avocado, nuts, chicken, pork that wasn’t exposed to harse processing or high heat … could the 4% PUFA limit be different?

          • Hi Mark,

            I think the 4% limit on omega-6 is pretty solid whatever the source, but whole foods are healthier than vegetable seed oils, and they are also much lower in omega-6. So such a diet would probably work out OK.

          • Wayne D Johnson aka Prouddaddy

            Paul, Carl Lanore today pointed me to the work of Joseph Hibbeln of the NIH. He believes that linoleic acid in particular may be resonsible for the obesity epidemic based not so much on inflammation as the production of endocannabinoids from arachidonic acid. He also has critiqued the “healthy PUFA” studies by noting that those that distinguished between n-6 and n-3 showed n-3 as protective for CVD and n-6 as actually harmful. He is the first to caution, however, that the linoleic acid theory of obesity has only been indicated in rat studies with human studies still needed.

          • Hi Wayne,

            Thanks, yes I’m familiar with Hibbeln’s work (we cite him) and with the omega-6 endocannibinoid idea (although I didn’t quite have time to put that into the new edition of the book; there are other mechanisms too by which omega-6 promotes obesity, chiefly endotoxemia and mitochondrial dysfunction to lipid peroxidation). I agree with him.

          • Chris Cortes-Rojo

            I agree with you about toxic mitochondrial effects by lipoperoxidation induced by n-6 PUFA: the enrichment of mitochondrial membranes with PUFA leads to increased sensitivy to oxidative strees and ROS overproduction. Look at this paper authored by me in this regard.

            Cortes-Rojo et al. 2009. Elucidation of the effects of lipoperoxidation
            on the mitochondrial electron transport chain using yeast
            mitochondria with manipulated fatty acid content. J Bioenerg Biomembr 41:15-28

          • Hi Chris,

            Thanks, it looks like a very interesting paper. For those who are interested the full text is free: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2922399/.

            I just glanced quickly at it. Can you tell me if these iron concentrations are super-physiological? Would iron-induced oxidation be a significant cause of mitochondrial dysfunction in someone with hemochromatosis/iron overload? And did manganese/MnSOD worsen the problem?

          • Paul,

            Since PUFA is both omega-3 and omega-6, should PUFA recommendation of 4% be adjusted higher on the diet webpage?

            I think the 4% limit on omega-6 is pretty solid whatever the source


            Polyunsaturated fats should be a small fraction of the diet (~4% of total calories).

          • Hi Mark,

            I want to communicate that there should be a hard upper bound on omega-6 at 4%, but 3% would be better; while omega-3 around 1% is optimal; and total PUFA around 4% is a good target (3% omega-6, 1% omega-3).

          • It’s interesting to look at example menus for what 4% PUFA looks like. >=4% PUFA is pretty easy to reach eating whole foods.

            My FitDay for one week (3 eggs each day):

            3% PUFA for a day with no avocado, nuts, salmon, or olive oil.
            4% PUFA adding any one of avocado, salmon, or olive oil (I didn’t eat any nuts that week).
            5% PUFA for a day with 1/2 avocado and 1T olive oil.

            Here’s a sample menu from Mark Sisson’s Primal Blueprint that is handy with FitDay info.

            8% PUFA for a day with 3 eggs, 6 oz canned salmon, 1/2 avocado, 2 T olive oil, 1 oz almonds. To make more PHD compliant, swapping in safe starch carbs for smaller portions of salmon, steak, olive oil would lower the PUFA some, but still well beyond 4% PUFA.

            3 eggs
            1 pat butter
            mushrooms, peppers, salsa
            1 oz cream
            1 tsp sugar

            6 oz salmon, canned
            1/2 avocado
            2 T olive oil
            salad greens, tomatoes, onions, broccoli, vinegar

            1 oz almonds

            8 oz steak
            1 T coconut oil
            brussels sprouts

            after dinner:

            Seems like a pretty healthy daily menu.

            I recognize that the nutrition databases don’t have what’s considered higher quality meat sources and maybe my specific extra virgin olive oil, but it seems like a good estimate and I’m not sure how else I could check it for myself.

          • Hi Mark,

            They are all healthy foods but there’s quite a bit of oil and the salmon is more than one would need every day. We recommend 3/4 to 1 lb salmon per week which is 2 ounces per day; re oils 4 tbsp is a bit much. Drop the salmon to 2 ounces and replace the 2 tbsp olive oil with safe starches and the PUFA level would be pretty good I think.

          • For modern thinking on linoleate, google “Israeli Paradox”.

            I found the 4% presented some challenges when I tried to blog about fish oil: http://hopefulgeranium.blogspot.co.nz/2012/10/fish-oil-krill-oil-and-hepatitis-c.html

            Not all the PUFA in fish oil is EPA and DHA, it took me a few corrections to get my maths right.

            Just as there are cellular and acellular carbohydrates, the same applies to lipids. Almonds, for example, may not supply quite as much PUFA as the books say: http://www.ncbi.nlm.nih.gov/pubmed/22760558
            I think we may have a little more flexibility around nuts and fish, compared to oils. But it’s the lipids that are already in cells when we start a diet that take time to change – linoleate deficiency in the diet supposedly takes a long to produce effects.

          • This study is a good one, showing high-linoleate fat is more oxidizing in humans than high-oleic, and that butter (baseline diet) is best:

          • Wayne D Johnson aka Prouddaddy

            Rechecked PHD out of the library. It’s depressing to learn how much a 71-year-old brain can fail to retain! You made an excellent case as to the unhealthfulness of O6.

            I am still dubious about the inflammatory argument as the mechanism. Perhaps it is something else or we haven’t found thr way to measure it. Thus, I am very interested in your take on the Johnson (no relation) study I referenced above.


          • Hi Wayne,

            Here’s a link to the study: http://www.ncbi.nlm.nih.gov/pubmed/22889633.

            The trouble is that nearly all of these studies looking at inflammatory effects are very short term. The negative effects of high omega-6 diets build up over a period of years. It takes typically 4-5 years for omega-6 levels to build up at the cellular level and cause trouble. The body can sequester them in adipose tissue or elsewhere at first.

            So it’s really the long-running clinical trials that we discuss in the book that give the best evidence. Eg the Rose Corn Oil trial, LA Veterans study, etc.

      • The ill effects of high 0-6 would be suppressed on a VLC diet (Volek et. al.), but any extra AA accumulated in membranes would be later available to drive undesirable processes (sweet tooth, adipocyte expansion, insulin sensitivity and inflammation) soon after carbs are re-introduced.
        On a mayo-Atkins diet this could account for both rebound weight gain and the eventual assumption that life-long starch restriction was the only way forward.

        Metabolic disease happens in an immunological context (as much as the genetic and circadian ones) and any treatment needs to address this in some way; the right fibre foods, fermented foods, and herbs and spices; the right 0-6:3 balance; optimal vitamin D levels; outdoor air and dirt exposure…

        • Why would accumulated AA only “come out” when carb consumption resumed?

          When I look at nuts seeds and fatty fruits — aka plant foods man was likely to have eaten — they are pretty high O6. We can add peanuts/peanut butter to the list of O6 rich faves (and Taubes’ Westman inspired diet touts peanut oil as especially healthy!) of Atkins folks for eons.

          • There is a Volek paper showing decreased AA turnover – and thus higher AA levels – in membranes during VLC dieting. COX2 downregulation, consistent with this diet controlling diseases like arthritis.
            If this is a temporary result of the VLC diet, and if the diet happens to be high-PUFA, then taking the VLC brakes off means the AA becomes more active.
            As for the nuts, I think you were right about cellular and acellular lipids. Look at the difference between fish oil and krill oil; a few phospholipids rather than triglycerides make a significant difference in the type of effect.
            In nuts, compared to their oils, this (and the rest of the nutty nutritional matrix) would mean less retention of PUFA in the liver and more exporting and burning.
            This is maybe something that has hardly been researched, but you were onto something with that possibly throwaway comment.

        • Wow, that is an interesting idea. I think that happened to me. I was doing LC/VLC for many years and my main source of protein/fat was fatty chicken and chicken skin. My O-6:3 balance was way out of wack and I really had a hard time w/ adding back safe starches and fruit. Had a hard time not eating too much for a while and my weight went up a bit. Finally, it’s been a lot easier lately now that I’m favoring beef (mostly) with some and fish/seafood and rarely, if ever, eat chicken or pork.

        • Chris Cortes-Rojo

          Hi Paul

          The higher iron concentration used in that study (100µM), is for sure, out of the range of physiological concentrations, but levels of iron approaching the lower concentration used during the study (12.5µM), have been detected in fibroblast mitochondria from patients suffering friedreich’s ataxia, while in healthy rat liver, mitochondrial iron can reach concentrations up to 16µM. Thus, it could be expected that iron concentrations higher than the above mentioned may be reached in liver mitochondria from patients with diseases involving hepatic iron overload (e.g. hemochromatosis). Indeed, oxidative damage in mitochondrial respiratory chain by iron overload has been involved in the development of liver and neurological diseases. In this regard, you could e-mail me to christiancortesrojo@gmail.com to send you a copy of a minireview that I published last year about the role of mitochondrial oxidative damage in disease development and some issues about the use of mitochondria-targeted antioxidants.
          Respect to the role of MnSOD in oxidative damage, the data available is controversial due to overexpression of MnSOD decreases life span of some experimental models of aging, while in others had the oposite effect. i think that in the case of mitochondrial iron overload, the overexpression of MnSOD may be deleterious because the consequent rise in hydrogen peroxide could increase the formation of the dangerous hydroxyl radical through Fenton’s chemistry.

  8. There is something to be said for MarkES’ point, but we mostly got here by eating too much oil, and I wouldn’t even consider it till the effects of that have been reversed.
    PUFA in cellular lipids (phospholipids) probably do behave differently from acellular fats (triglycerides). We see this with krill oil compared to fish oil. So a case can be made for nuts being different from oils.

  9. Paul, I love your work, but I think you and others are too quick to dismiss the argument that the people in this study didn’t lose enough weight.

    As Evelyn points out, for someone weighing 250 lbs., a 5% drop will take them down to 237.5 lbs., or, in other words, from obese to slighly less obese. Would people really expect that to accomplish much?

    • Hi Jonas,

      Well, as Stephan has written, there is a line of thought that excess energy states are responsible for much of the negative health effects of obesity, and just being in normal or modestly negative energy balance for an extended period of time may restore normal health even if not much weight is lost. That’s a theory for why some studies found good health results from ELMM.

      In other words, if it’s not obesity per se, but being on an obesogenic diet after adipose cells have ceased storing fats, that causes problems, then just getting weight down a little, restoring insulin sensitivity and adipose function, might be enough to greatly reduce adverse health events.

      The other factor to consider is that there will be a dispersion of results and although the average was a 5% drop, some people will have lost 30%. So if larger weight loss would restore health, you might expect the 5% or 10% of people with the greatest weight loss to have stellar results, so the event rate should have dropped at least 5% or 10%. That didn’t happen (at least in intervention vs control group comparisons; as Evelyn said in her first blog post, data on the individual level could show that big losers had big benefits).

      Finally, if the way to get weight loss is ELMM, and this intervention produced too little weight loss on a 1200 to 1800 calorie diet, then a more aggressive intervention would presumably call for eating even less. As I’ve written, I don’t think people can be healthy (well-nourished) on less than about 1300 calories a day. So either “move more” would have to be the primary prescription, as on Biggest Loser, or ELMM would be a failure because it doesn’t effectively generate enough weight loss to matter.

      So, I would say that yes, Evelyn could be right that the problem is the intervention was ineffective; but though poor at bringing about sustained weight loss it could still have been effective against diabetes or CVD; and if it was ineffective against everything, that could be taken as evidence against ELMM.

      • Hey Paul, In my second post, I included the distribution of weight losses. Only 9% of IDI lost 15% or more, while the 4% of the controls did so I have my doubts we’re talking many in the 30% or more range. In my latest post I brought up Ebbeling — in that study weight losses at 12 weeks put the losses in this “intensive” study to shame, when in fact we were talking about a similarly restrictive caloric deficit. Either by one year weight was already on the rise or these people lost very little on average indicating they didn’t eat a whole lot less. When obese people are put on such caloric levels in controlled circumstances they always lose weight.

        I’m concerned that you don’t think less than 1300 cal/day can be healthy. Jaybird lost his weight eating 1200 cal/day of PHD as I recall.

        If someone doesn’t count calories I’d agree they are likely to eat less on a 30% carb diet than a 50% carb diet, but upping the protein caused weight loss and spontaneous reduction in intake in this study I blogged on: http://carbsanity.blogspot.com/2010/04/high-protein-diet-induces-sustained.html

        • Hi Evelyn,

          1200 and 1300 are about the same. Jaybird was pretty much 100% in accord with our recommendations.

          Thanks for the info about Ebbeling, I basically agree with your posts on Look Ahead. It looks like an ineffective intervention, behavior of the controls was pretty similar to the intervention group, plus they selected unusually healthy diabetics for the trial so they had low event rates and we can’t infer anything from the difference in event rates between study participants and the general population.

          Counting calories will probably bring everyone closer to a target, but I doubt it completely suppresses a tendency to eat more calories on a 50% carb diet than on a 30% carb diet. I agree that increased protein intake tends to suppress appetite and calorie intake.

  10. You might find the individuals in the study who lost the most weight did noticably better in terms of DM2.
    In which case the failure is not of the hypothesis, the proof of concept would be there. The failure would be the “intention to treat” type.
    Giving this kind of advice and guidance to patients fails overall as a therapy; but following it does not, for those who do actually follow it. Perhaps.
    We’ll see.

  11. Paul, any ETA on that Forum? 😀

  12. I just came across an interesting website/blog that talked about this sort of thing, and referenced some studies on the Prudent diet from back in the 60’s.


    The idea of the Prudent diet was to replace sat fats with unsaturated (margarines, soy oils, etc), and that the amount of PUFA should be *double* the amount of SAFA. This was done partly be reducing red meat, eggs etc and by replacing butter and lard with margarine oils etc. sounds almost like the modern SAD, but without the HFCS…

    “The men on The Prudent Diet were mostly teachers in colleges and universities in the city. They were dedicated to following The Prudent Diet. The controls were men of wealth on Wall Street who were found to live on a high cholesterol diet with nearly all the fat in their diet being the BAD saturated fats. Dr. Joliffe had a group of men that should have proven the expected benefit of The Prudent Diet.

    The trial ran for six years and the results were reported in JAMA in the Nov. 7, 1966 issue pp 129-134. The results were said to have been a great success proving the value of The Prudent Diet as serum cholesterol in the subjects living on The Prudent Diet had dropped to 200 mg % from 225 mg %. For about a week this was headline news, the first trial of The Prudent Diet had proved that it would prevent heart attacks. What the public was not told was that there had been eight deaths from heart attacks while living on The Prudent Diet whereas there had been no deaths from heart attacks among the controls living on the high-cholesterol, high-saturated fat diet. This was followed by the death of Dr. Joliffe. It was said that the cause of his death was due to his diabetic condition.”

    So, another side by side dietary intervention to create a cholesterol lowering diet, that did not even come close to the expected results.

    There are some interesting comments in there about Ireland too (huge butter eaters back in the day).

    And, of course, some evidence to support the theory that lowering cholesterol increases cancer.
    That would be what the statin drugs do!

    • Hi Paul,

      Good stuff. The “Prudent Diet” trial, actually known as the Anti-Coronary Club study, is discussed in both editions of our book, it is one of the better pieces of evidence against omega-6 fats.

      • Hi Paul,

        This is the part where I get confused as this seems like good evidence against processed seed oils … how does it apply to omega-6 fats from whole foods like avocado, nuts, olives, salmon, pork, chicken that are not subjected to high heat?


  13. A biochemist might say “a molecule is a molecule”, lineolate is exactly the same from any food;
    an immunologist might say that context influences metabolism.
    I think it’s reasonable to assume that fats in phospholipids Behave differently from those in triglcerides – up to a point, where their “molecule” nature becomes overwhelming.
    And we may find that triglyceride structure (esterification) also creates some differences;
    Antioxidants in food influence the fate of fats; sesame seed lignans inhibit elongation of GLA when linoleate is metabolised, which would make whole sesame seeds less inflammatory than their fats alone. Wine polyphenols enhance elongation of omega 3s. Olive oil may work in a similar way to produce less inflammation from its linoleate.
    Lipid science is in its infancy compared to carbohydrate and protein: lipids are hard to separate, triglycerides come in fantastic variety, and the effects of fats take time to show. Omega 3 was pretty much the last nutrient to be discovered.

    • “In most vegetable dietary fats, palmitic (C16:0) and stearic acids (C18:0) mainly occupy the 1- and 3-positions of the triacylglycerol molecule, whereas an unsaturated fatty acid such as oleic acid or linoleic acid usually occupies the 2-position. In animal fats, this is not the case. Interesterification of vegetable oils will enhance the amount of saturated fatty acids at the 2-position. Fatty acids at the 2-position are biologically different from fatty acids at the 1 and 3 position because they are handled differently during digestion and metabolism, and a relevant scientific question is whether there are health effects following from this.”

      • Hi George,

        Great observations.

        In our book we mention a peanut oil study. The peanut oil is highly atherogenic in its natural state, but if you randomize the fatty acid and organic molecules in the phospholipids, it becomes much less toxic. Clearly there is some particular combination in a phospholipid which is much more harmful to humans/monkeys than it is to peanuts.

        The whole topic seems almost unexplored.

  14. And when an oil is processed, the phospholipids are the first part discarded in the degumming: http://healthydietsandscience.blogspot.co.nz/2012/10/how-soy-phytosterols-are-manufactured.html

    Another good reason to avoid peanuts!
    If you think of all the possible fats… then remember that triglycerides can have any combination of 3… that phospholipids can have any combination of 2… and that the choline moitety of phospholipids can be replaced by ethanolamine, serine, inositol and many other head groups…
    the mind boggles.

    • Hi George,

      You’re quite the researcher and I must admit there’s a lot of info here that is beyond me, but it sounds much like a food altering chemistry experiment. It may look good on paper, but it seems to be when we get into trouble. I’m all for innovation, but maybe food should be kept more simple.


      • Indeed. I think the only useful conclusion is that we cannot exactly compare PUFA from oils with PUFA from wholefoods, and that fats have more surprises in store.
        Carbsane made the useful observation that perhaps the “acellular carbohydrate” hypothesis also applies to lipids and proteins.
        In which case it makes even more sense to source nutrients from traditional foods in traditional patterns, and to pay attention to our bodies’ own likes and dislikes. If some nut makes one itchy, or some fruit gives one the heebie-jeebies, or one particular sort of meat sits on one’s stomach too long, one shouldn’t wait for a chemical explanation before substituting something friendlier.
        I find the 4% PUFA guidelines useful and workable, and they do relate well to the oxidation literature and essential requirements. But I don’t bother counting PUFAs in the fish I eat or worrying about a couple of nuts here and there; it is more about the fats and the oils, the chicken and the pork (where the fats, with their PUFAs, can easily run free like oil).

        • Thanks for summarizing, George, makes sense. 🙂

          • I try to eat one fish meal for each meal of chicken or pork, and eat greens with those; and I never eat nuts or seeds as a major fuel source, just as a condiment or ingredient or part of a snack. So I keep a sort of eye on the 3:6 ratio from whole foods. But as far as actual label-reading, google-based calculations go, that’s applied to fats and oils.
            It’s just not practical to weigh small amounts of nuts, or cooked fish… and then there’s this: much of the lipid from almonds is not actually digested:

        • it is more about the fats and the oils, the chicken and the pork (where the fats, with their PUFAs, can easily run free like oil).

          Does it matter if the free fats are produced under say sizzling hot or more gentle heat conditions, or is it simply the separation from the whole food making the difference. And if the separation is considered okay and essentially preserved, is it better to still eat those free fats with the meat as the parts equal the whole. That the meat without those fats would be something less than it’s natural whole, somewhat like low-fat dairy.

          And while PUFA is considered more chemically susceptible, would the “acellular” idea apply in some similar way for lower-PUFA free fats like coconut oil, clarified butter, beef tallow, MCT oil?


  15. There is a table of over 60 glycerophospholipids here: http://en.wikipedia.org/wiki/Phospholipid

    I am sure it is only the start…

  16. We’re really just guessing here so far, but my best guess is that the more saturated a fat is, the less it matters that it’s in triglyceride form rather than phospholipid, because it’s less bioactive anyway. That’s why I single out pork and chicken.
    Even a cellular fat is going to be a mix of triglycerides and phospholipids.
    The triglycerides (apart from maybe MCTs) are taken apart and rearranged before they enter circulation, so their esterification structure is maybe not important anyway.

  17. When I was a student I had the opportunity to visit one of the look AHEAD sites. This was years ago and I remember then that they were frustrated with the lack of difference between the two groups. Some of their intervention subjects had started to regain weight.

    I got to observe a cooking demo session and we all went on a group walk. They received an incentive for attending, little freebies.

  18. Great review of the study! I actually listened to a few of the investigators this week present at Harvard rounds and brought us through the results with an emphasis on all the successes from it. He even said they encouraged replacement of saturated fats with poly and monounsaturated fats. Maybe this should be a clue that even with very intensive conventional therapy, weight loss and biomarkers are not going to improve that much, and a look at the actual research to develop a new approach should be done!

  19. Hi –

    I’ve just finished reading the PHD book and found it fascinating.

    I’ve been following a general low carb Paleo Diet for about 2 years but have not lost the weight that I need to lose, and my blood sugar and cholesterol is terrible.

    My Background:

    40 year old East Indian male, 5’11” in height, 202 lbs, 38.5 inch waist

    History of Type 2 Diabetes and metabolic syndrome — my father and his two brothers. Father died of hear attack at age 67.

    I’ve been on a low carb diet for approximately two years.

    Carb intake approximately 100 grams per day.

    Recent Glucose Serum Fasting = 7.6, no A1C test because the lap forgot to take it, Total Cholesterol = 8.22, LDL 6.08, HDL = 1.36, Triglycerides = 1.71, TC/HDL-C Ration = 6, blood pressure 122/86 but am on blood pressure meds.

    I have been exercising regularly and lifting weights using a SuperSlow method (Dr. McGuff’s method) of 5 weight lifting exercises once every 7 days. I do 45 minutes of walking every day and occasional running sprints.

    My doctor agreed to give me 6 months to try one more time to bring my numbers in line through diet and exercise. He wants to put me on statins and probably glucose control meds.

    In the PHD book it states in the section on Key Steps to Weight Loss that a diet of 500 carb calories, 300 protein calories and 500 fat calories is the minimum caloric intake consistent with proper nourishment.

    This translates to 38% carb, 23% protein and 38% fat. Is that correct? It seemed odd to me that these percentages deviate so much from the normal PHD breakdown of 20 – 30% carb, 15% protein and 50 – 60% fat.

    I need to lose weight and get my blood sugars and cholesterol levels down — what is the recommended % breakdown of PHD I should follow?


    • Any thoughts or suggestions?

      • Well, I tried the PHD diet for 7 days and didn’t like how I was feeling and looking (more belly fat). I am going back to the Dr. Bernstein low carb approach of 30 grams of carbs per day max.

        • I would still be interested in any thoughts or suggestions on my question above. I have to go back to a Dr. Bernstein 30 gram / carbs diet for now as that is the only way I know right now to keep my blood sugars under control and to also not put on weight around my belly.

          • Hi,
            I don’t think exact percentages are the problem. You probably have a nutrient deficiency.

            If you read the Perfect Health Diet book, it says chromium is useful for diebetics in moving the glucose out of the blood and into the cell. So you may want to increase your intake of foods that are rich in chromium. Try not to rely on supplements; some supplements are made from things not even found in nature and never meant for the human body (ex: cyanocobalamin, B12, contains a cyanide molecule. the real B12 does not contain cyanide). Search for “chromium rich foods” on a search engine and eat them.

            Also, there are some studies in PubMed indicating that diabetes could be a Vitamin A deficiency. The theory is that the lack of Vitamin A makes the pancreas unable to create beta cells, which are the cells that make insulin. Without Vitamin A, cells cannot differentiate (they remain in a flux, I suppose). So, eat liver to increase your intake of vitamin A.

            Not sure what to say about cholesterol. So, you have cholesterol buildup that is not being converted into the cholesterol-dependent hormones, in my opinion. There is some post on this website of a person who had high cholesterol which was healed with copper and choline. You can search those on this website and read that post. I think it will help you.

            Good luck.

Leave a Comment

NOTE - You can use these HTML tags and attributes:
<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Trackbacks and Pingbacks: