An article caught my eye in this month’s edition of the American Journal of Clinical Nutrition. Based on the title, “Separate effects of reduced carbohydrate intake and weight loss on atherogenic dyslipidemia,” I assumed this paper by Ron Krauss et al would address one of the age old problems with studying the effects of low-carb diets on blood lipids: how much of the correction is due to the weight that’s lost and how much is due to the metabolic effects of the carbohydrate reduction?

At first blush I thought the paper was pretty good; on another read through along with some back-of-the-envelope calculations I thought it was pretty awful. I then really got it down, rooted around in it and was able to snuffle up a few good insights to pass along.

One of the problems in dealing with dietary regimens–especially those that cause weight loss–is to determine what the heck it is that causes the changes found? Are the changes a consequence of the lost weight? Or of the change in macronutrient composition? Or both? Or something else altogether?

This problem arises because there are really four different variables involved in any diet–fat, protein, carbohydrate, and calories–all of which vary with respect to one another, and all of which have an influence on the outcome of any specific diet.

When I check blood values six weeks after I put patients on low-carb diets,  I expect to find lower triglyceride levels along with a fairly substantial weight loss. I can claim it is the lowered carbohydrate levels that bring about this triglyceride lowering. Others might reasonably conclude that since the patient lost weight, the triglyceride lowering is simply a manifestation of that lost weight. Others yet could claim that it was the restriction in calories causing both the weight loss and the reduction in triglycerides.  In fact, that is the primary argument of those who don’t believe that the restriction of carbohydrates does anything other than restrict calories.

The arguments of these non-believers are perfectly valid. If one compares two diets that contain an equal number of calories but varying amounts of carbohydrate–let’s say two 1500 kcal diets, one of which contains 50 grams of carbohydrate, 90 grams of protein and a little more than 100 grams of fat (about 63% of calories), and the other of which is made of 200 grams of carbohydrate, 90 grams of protein and a little less than 40 grams of fat (only 23% of calories)–and finds that subjects following the lower-carbohydrate, 1500 kcal diet lose more weight and lower their triglycerides more than those following the higher-carbohydrate, lower-fat diet (which is almost always the case when tested), can we say for sure it was the lower carbohydrates that did the trick?

Not necessarily.

Since the lower-carbohydrate diet had to contain the same number of calories as the higher-carbohydrate diet to remove any caloric difference from the equation, the calories lost by reducing the carbohydrate amount had to be replaced with fat because in this example the protein content was the same in both diets. So who’s to say that the changes didn’t come about because of the higher fat content? Or a combination of lower carbohydrates and higher fat?

So let’s check that out. We’ll design a diet that keeps the carbohydrate at 50 grams, keeps the protein at 90 grams and sets the fat at 40 grams just like the other diet, then we can say that any changes are surely due to the lower carbohydrate level in this diet. Not so fast. Now we’re dealing with a 920 kcal diet instead of a 1500 kcal one, and our naysayers are going to point out that all the weight loss and any other changes we might find can be attributed to the lower calories. See what I mean? It’s not an easy problem to deal with.

Another similar problem is trying to figure out whether some change that occurs with diet comes from the diet itself or simply from the weight lost while dieting. For example, let’s say I give you a pill that makes you so nauseated you lose your appetite, don’t want to eat, and drop 20 pounds over the next two months. I then check your blood pressure and find it’s a lot lower and tell you the pill has lowered your blood pressure. You might well look at me and think: what a moron! My blood pressure is lower because I lost 20 pounds. It’s the same with diet.

Go on a low-carb diet, lose 20 pounds, and your blood pressure falls. Is it from the lowered carbs, the lost weight, or both? That is the question the paper under discussion deals with, and is why I was so excited when I first read it.
The authors are firm believers in the lipid hypothesis and are interested in looking at a couple of specific lipid parameters they believe are the most important: triglyceride levels, HDL levels, and small, dense LDL particle levels. (For what it’s worth, if there is even a flicker of a lipid-hypothesis-believing flame barely alive anywhere within me, I would finger these exact parameters as the culprits.) These authors point out right off the bat that obesity causes elevations of these parameters and acknowledge that carbohydrates do so as well.

Excess body weight can result in changes in plasma lipids and lipoproteins that increase the risk of atherosclerotic cardiovascular disease (CVD), ie, increases in triacylglycerols and small, dense LDL particles, with variable increases in total LDL cholesterol, and decreases in HDL cholesterol. Atherogenic dyslipidemia associated with excess adiposity is highly correlated with reduced insulin sensitivity and is a major feature of the metabolic syndrome.

Dietary carbohydrates, especially simple sugars, can also promote atherogenic dyslipidemia, in large part because of effects on the metabolism of plasma triacylglycerol-rich lipoproteins. High-carbohydrate, low-fat diets have been shown to induce increased concentrations of small, dense LDL and expression of the small, dense LDL particle phenotype (LDL subclass pattern B) in a high proportion of healthy men.

The authors then note that low-carbohydrate, weight-loss diets have been shown to correct these lipid abnormalities and set themselves the task of designing a study to tease out whether it’s the carb restriction that does all the heavy lifting or simply the weight loss or both.

Here is how the study was designed. 178 men went on what is referred to as a ‘basal diet’ containing enough calories to maintain starting weight for one week. This basal diet was made up of 54% carbohydrate, 16% protein, and 30% fat. After this first week the subjects were randomized into four groups of varying carbohydrate content: one group stayed on the 54% carb diet, one group dropped carbs to 39%, another group reduced their carbs to 26%, and the final group also got 26% carbs but had about 15 grams of their monounsaturated fat replaced with saturated fat. All groups got the same number of calories as they did the first week

After the end of this three week weight-stabilization period all the subjects in all the groups got their intake slashed by 1000 kcal per day while keeping the same carb to protein to fat ratios as before. This phase of reduced caloric intake lasted for five weeks. The subjects lost weight during this five week period of reduced caloric intake and were then given an amount of calories sufficient to maintain their lower weight for the final four weeks of the study during which the macronutrient ratios were maintained as before.

I have a real problem with the way the data were presented in this study, which is why I thought this study was lousy my second time through. Nowhere was there a listing of caloric intake or grams of carbohydrates, protein or fat. All data were presented as percentages. The only hard number I had to go on was the fact that the average BMI measurements of the subjects was 29.2. Knowing that and that the average age was about 50, I was able to reverse engineer using the standard equations for total energy requirements for weight maintenance for 50 year old males with BMIs of 29.2 and found it to be roughly 2800 kcal per day.

So, during the basal diet the first week all subjects would have received 2800 kcal per day made of 378 grams of carbohydrate, 112 grams of protein, and 93 grams of fat. During the next three weeks of weight stabilization the subjects (still consuming 2800 kcal per day) were randomized to the different levels of carb restriction, those on the first level of carb restriction at 39% would have eaten 273 grams of carbs daily while those on the most restricted 26% carbohydrate diets would have gotten 182 grams per day, not an insignificant dose. (Remember, a cup of sugar weighs about 200 grams so those subjects on the most carb-restricted diet got almost a full cup of sugar equivalent daily. Not what I would call a real low-carb diet.)

After the 1000 kcal per day reduction the 54% subjects were consuming 243 grams of carb per day, those at 39% restriction, 175 grams per day, and those at 26% restriction got 117 grams per day, still not a real low-carb diet.
What the researchers wanted to see was what happened after the first four weeks of weight stabilization as the carbohydrate levels in the diet were reduced. Since there was no weight loss during this four week period (actually there was, but more about that in a moment), any changes in the lipid parameters under evaluation could be attributed to the change in carbohydrate content of the diet.

What happened?

In the initial diet, stable-weight phase of the study, the 26%-carbohydrate, low-saturated-fat diet resulted in reductions from baseline in total cholesterol, triacylglycerol, apo B, and total:HDL cholesterol that were greater than the changes observed in the group remaining on the 54%-carbohydrate diet.

In the initial stable-weight phase, LDL peak particle diameter increased to a significantly greater extent with both the 39%- and 26%-carbohydrate, low-saturated-fat diets than with the 54%-carbohydrate diet.
[larger diameter LDL particles are assumed to be much less atherogenic than the smaller, denser ones.]

In other words, there was a substantial improvement in those lipid parameters in the subjects on the lowest carbohydrate diet compared to those in the highest. Since all groups maintained their same weight, these changes could be attributed only to the decrease in carbohydrate in the diet. (Those of us who have been on real low-carbohydrate diets can only imagine the changes the researchers would have seen had these subjects been on about 50 grams of carbohydrate per day instead of the 182 they actually ate.) It turns out there was a slight fly in the ointment in this study during this first stable-weight period of the study that all of us who have been on low-carb diets will no doubt find amusing. As the researchers reported:

Despite our effort to maintain constant weight, the 26%-carbohydrate, low-saturated-fat diet group lost more weight than did the 54%-carbohydrate group during the stable-weight period. There was also a trend for a greater reduction in percentage body fat with the lower-carbohydrate diets.

Do tell, heh, heh, heh.

After the next part of the study during which the subjects lost weight due to the 1000 kcal per day reduction in intake an interesting finding emerged. The researchers had determined that the carbohydrate restriction by itself had lowered triglycerides, Apo B, HDL/total cholesterol ratios and produced larger LDL particles so they wanted to see what weight loss added to the picture. They found mild further improvement in these already improved parameters in the lower carb group, but found huge positive changes with weight loss in the highest carb group.

Let’s think about this a little. In the first phase the lowest carb group dropped their carb intake from 378 grams per day on the basal diet to 182 grams per day on the 26% carb weight-stable diet, a reduction of a whopping 192 grams of carb–virtually a cup of sugar–per day. When subjects in this same low-carb group reduced their caloric intake by 1000 kcal per day, they lost another 65 grams per day, an improvement to be sure, but a long way from the 192 grams they ditched when they switched from the basal diet to the iso-caloric low-carb diet.

When the high-carb group consuming the 378 grams per day reduced their intake by 1000 kcal per day they reduced their carb intake by 135 grams per day (378 minus 243), which was over double the 65 grams the subjects in the lowest carb group reduced when they dropped their 1000 kcal. It would stand to reason, then, that if it’s really the carb reduction that counts and not the weight loss, the group dropping the most carbs would have the greatest changes, which is exactly what happened.

So, to summarize: Carb restriction rules.

The lipid parameters evaluated in this study improved as a function of carbohydrate restriction. In their discussion and conclusion section the authors were obviously loathe to make these findings clear and saw fit to obscure them in a hodgepodge of academic gobbledygook but they’re there for anyone who wants to root them out.

Oh, and about that saturated fat they added to one of the low-carbohydrate diets… Well, it turns out that it didn’t do much of anything except make the LDL particles even bigger, which is just how you want them.

Once again, despite having to ferret the truth out of an article that does its best to obscure it, the low-carb diet wins by a knockout.

Graphic at top: Scholar of Natural Sciences by Carl Spitzweg, 19th century


  1. Good one Mike! You sure can dig in and root out. Too bad they didn’t measure lean tissue gains/losses as well. Researchers often forget how very important this is! We should be doing some research on this.
    BTW, the June issue of Allure magazine (just out) page 139 has our book mentioned with before after pix of a gal we trained to lose fat off her legs in 8 weeks. We did good. But the brainiac editor botched the details however stating the our gal lost a total of 5 pounds in 8 weeks using Slow Burn and following the Slow Burn Power Eating Plan in the book.
    Not so. And I gave them the details in full!! It’s as if they could not believe it so the refused to print it.
    She lost 5 pounds alright – 5.45 to be exact, but gained 4.8 pounds of lean making a total fat loss of 9.8 pounds! But here’s the kicker – she started out lean already and moved in to the ideal range on my body fat analyzer gizmo. THAT is no easy feat! Low carb and Slow Burn rules big time! Everyone else exercised for 5 + hours a week. Our gal – 1 hour or less.
    Hi Fred–
    Thanks for the kind words. As to the change in lean mass, the authors of the study did note that
    “There was also a trend for a greater reduction in percentage body fat with the lower-carbohydrate diets.”

  2. A few things. As usual the calorie theory does not add up. My own back of the envelope calculations, assuming your 2800 kcal basal metabolic rate, of the results for a 5 week 1000 kcal diet would give an 18.0 lb weight loss vs. the actual result of 5.12 kg (11.3 lb) which is only 63% of the expected result. I’ve found this to be true of all the weight loss studies.
    The other point is the final sentence of the Results
    “lipoprotein changes with the higher saturated fat intakes were not significantly different from those with the lower saturated fat intakes, except for LDL cholesterol, which decreased less with the higher saturated fat intake because of an increase in mass of large LDL.”
    Finally, a plausible explanation for one of the few confirmed items of the lipid theory, that consumption of saturated fat raises cholesterol levels and especially the LDL cholesterol. It does this by increasing their size and thus their mass. Continuing this train of thought, consumption of saturated would actually appear to be anti-atherogenic (i.e. protective against heart disease) by raising HDL (the other ‘good’ cholesterol) and making the LDL less dangerous.
    All of this may seem obvious, but until bariatrics (the branch of medicine that deals with the treatment of obesity) has a scientific basis, the obesity problem will worsen. Thanks for the time and space to comment.
    Hi Mark–
    Well argued. I’ve said the same thing about saturated fat for years. It raises HDL, increases LDL particle size, and is one of the only agents known that reducles Lp(a).
    Old habits die hard.

  3. It is nice to see that Physiology prevails despite badly designed studies!!
    I just wanted to comment about other studies where low carbohydrate diets have been really used as such, for example those from Jeff Volek in which he uses what he calls “Very Low Carbohydrate Diets”, or VLCD, which I just call ‘true low-carb diets’. His studies have clearly shown that VLCDs improve the lipoprotein profile independently of weight loss. Of course, some people still point out that even if carbohydrate restriction improves lipid profiles, is still not as effective at lowering LDL cholesterol (LDLC). Go figure… how can a lipid profile improve but one of the main players doesn’t… Amazingly, the naysayers do recognize that carbohydrate restriction consistently improves postabsorptive and postprandial triacylglycerols (TAGs), HDL cholesterol (HDL-C), and the distribution of LDL-C subfractions to a greater extent than low-fat diets. So… the lipid profile does improve, doesn’t it? You gotta love semantics sometimes!
    I haven’t read the article you discuss yet but I’m curious to know if the authors cited any of Volek’s work. It seems to me that their question has already been addressed.
    What does seem to be independent of the diet, however, is the improvement in some inflammatory markers as Volek and others have shown comparing VLCDs and low-fat diets. Apparently, the improvement in inflammatory markers is similar and related more to weight loss than the type of diet. Even if that is so, given the boat load of other benefits brought about by carbohydrate restriction, that is even a better way to improve both, lipid profile and inflammatory markers.
    Hi Gabe–
    Thanks for the commentary. Volek’s work does seem to show that improvement in inflammatory parameters is more a function of weight loss than diet composition, but the works of others has shown a relationship between the glycemic load, i.e., carb content, and an increase in C-reactive protein.

  4. Mike, you’re right about glycemic load and CRP. I suppose that some anti-carb restriction wave the flag of ‘no difference’ to keep recommending against a carbohydrate restriction lifestyle. Others, better educated, would in turn recommend it to get more birds with the same ‘low-carb’ stone!
    Regarding the relationship between glycemic load and CRP, though, the one detail that sort of smears the studies a little (granted, mainly Willet’s work) is the use of the food-frequency questionnaire. Do you know of other studies where the estimation of the glycemic load has been done based on a 24-hour food journal for example?
    One study comes to mind, where the study subjects were given the food ahead of time, every week for the duration of the study. They would eat one meal on Monday and then go home with the rest of the food for the week (Pereira et al. JAMA. 2004;292:2482-2490). Interestingly, in the low-glycemic load food, fructose was included. So, the inclusion of seemingly ‘low-glycemic’ foods but that affect the metabolism just as bad as other carbohydrates is another detail that makes me not put too much stock in the glycemic load and/or glycemic index, even though the trend is clear. Ah… maybe I’m being too picky!

  5. I have a question about LDL particles. I know they come in two varieties, large and small. My question is: Does VLDL have anything to do with the large and small particles, or are the large and small particles measured by a totally different test?
    The large and small are measured by electrophoresis whereas VLDL is measured by ultracentrifugation. And, small and large aren’t really small and large–LDL particals are on a continuum from large to small. What you want to have is more of the larger and less of the smaller.

  6. Just a minor quibble in an otherwise very interesting post. You write that “there are really four different variables involved in any diet–fat, protein, carbohydrate, and calories.”
    Wouldn’t micronutrients be a fifth?
    No, micronutrients wouldn’t count as a fifth in these calculations because micronutrients contain no calores. Weight-loss studies usually only look at the caloric content or of the macronutrients that make up that caloric content.

  7. What an excellent analysis! It seems that the medical/nutritional/drug communities would rather watch people suffer from poor nutrition rather than just admit that they were wrong about low fat/high carbohydrate diets. I have been fortunate to find a family m.d. who has seen his patients improve drastically by switching to this WOE and encourages it.

  8. Mike, the subject of micro nutrients has always left me with some questions. As Regina Wilshire has pointed out in her blog ?Weight of the Evidence?, low fat diets tend to be nutrient deficient because fat restriction limits intake of fat soluble Vitamins A, D, E and many antioxidants (the ‘new’ Vitamins). Also, any diet increases the possibility of nutrient deficiency because you are eating less.
    The idea of studying diets with an exclusive focus on calories seems likes an exercise in ‘tunnel vision’, i.e. treating the body like it’s an engine, rather than as an extremely complex biochemical feedback loop. Intuitively, a nutrient deficiency would seem likely to be interpreted by the body as an indication of starvation and thus tend to promote fat retention. On the other hand, any deficiency that interferes with absorption of nutrients and other anabolic metabolism (processes that build and repair the body) would likely promote weight loss.
    Reviewing the NIH website on Vitamins and Supplements ( shows only Calcium deficiency may be associated with weight gain while Chromium deficiency may be associated with diabetes and initial weight loss as symptom of diabetes. Notwithstanding the above examples, no vitamin or mineral deficiency appears to be associated with weight loss as a primary symptom of the deficiency. Your diet book (I know you hate that term) and many others recommend various supplements, so that it is clear that you don’t share the calories only perspective. Incidentally, like the others I have to congratulate you on the penetrating insights into this and other scientific articles. Thanks for the time and space.
    Hi Mark–
    I agree that micronutrients are important and are a necessary component of any diet. But, they don’t provide calories. In evaluating a diet strictly in terms of the energy it provide, one can only look at the macronutrients and/or the total caloric intake.

  9. Dr. Mike,
    I just finished reading, to my dismay, an article by the resident dietitian for news, health section, who states in an article, I believe posted today, that saturated fat is even more dangerous than previously thought.
    She mentions a new study that shows that saturated fat plays havoc with insulin levels. It dismays me that she provided no link or reference to that study, so that I could read it myself. For instance, I wonder if in that particular study, the folks who ate a lot of saturated fat also ate a lot of carbohydrates along with it.
    Dr. Mike, can you please comment? Thanks!
    Hi Sheryl–
    I did indeed read the insipid comments of this woman. There is no arguing with such unshakable stupidity. Whenever anyone resorts to the old “many studies have shown” or “recent studies have shown” argument without listing said studies, you know that person doesn’t have a clue as to what any studies–recent or otherwise–actually show.

Leave a Reply

Your email address will not be published. Required fields are marked *