"Low carb, high protein diet found to increase mortality" By Jake Young

Not good news for people who think low carbs is the answer. A recently published study has shown an increase in mortality associated with low-carb/high-protein diets.

Studies looking at the comparative effectiveness of different diets have shown that basically all diets are equivalent. (Equivalently bad, that is. All the diets tested produced moderate declines in weight, but adherence was low.) However, no real studies have examined what the mortality benefits or hazards might be associated with the low-carb/high-protein diets that are currently de rigeur.

Trichopoulou et al., publishing in the European Journal of Clinical Nutrition, have produced one of the first studies to look at this issue. (Happily it is not behind a subscription wall.)

They followed roughly 22,000 adult Greek citizens as part of the European Prospective Investigation into Cancer and nutrition (EPIC) from 1993 to 2003. During that period, there were 455 deaths in the study. They used information about the diets of the participants to calculate relative mortality rates for particular diets.

Just to cross the t's and dot the i's, they also control sex, age, years of schooling, smoking, BMI, physical activity, ethanol intake, and (in the data I am going to talk about) energy intake. (Someone always asks about that in comments, so I thought I would put it at the top.)

Generally what they found is in line with other studies about diet and mortality:

As presented in previous publications (Trichopoulou et al., 2003), the results with respect to non-nutritional variables were mostly in line with expectations, in showing that mortality was higher among men than among women, increased sharply with age, declined with increasing years of schooling, higher physical activity and alcohol intake and increased with smoking and total energy intake. (Link in the original.)

To quantify how much carbs and protein a participant is had, they reference mortality to what is called a low-carb/high protein ratio (LC/HP). This ratio is defined using the decile of each participants carbohydrate and protein intake relative to the rest of the study population:

[F]or each participant, ascending decile of protein intake and descending decile of carbohydrate intake were added to create an additive LC/HP using, alternatively, absolute and energy-adjusted carbohydrate and protein values. Thus, a subject with LC/HP score 2 is one with very high consumption of carbohydrates and very low consumption of protein, whereas a subject with score 20 is one with very low consumption of carbohydrates and very high consumption of protein.

The reason they use this kind of obscure measure is because one of the problems you have in a study like this one is total energy intake. Any change in diet composition is likely to cause a change in total calories consumed, and total energy consumption is also correlated with mortality. They argue that by using this measure, they have found one that does not correlate with total energy intake and is therefore more valid. (You can choose to buy that or not depending on preference.)

What they found when they compare mortality to LC/HP -- controlling for the factors listed above -- is a substantial increase in mortality for a relatively small change in diet:

An increase in the LC/HP score (energy-adjusted components) by two units was associated with an increase in mortality by 8% (95% confidence interval (CI), 3-13%). Therefore, a realistic increase in the LC/HP score by five units (corresponding to, e.g., an increase of protein intake by about 15 g/day and a decrease of carbohydrate intake by about 50 g/day) was associated with a 22% increase in overall mortality (CI, 9-36%). (Emphasis mine.)

Just for reference, a McDonald's cheeseburger has about 15 g of protein, and 2.5 cups of Cheerios (about a bowl) has about 50 g of carbohydrates.

To get a sense of typical diets in Western countries, they also mention these comparisons:

Very-low-carbohydrate diets typically contain less than 10% carbohydrates, 25-35% proteins and 55-65% lipids. For comparison, the average American diet contains 35% lipids (85 g/day), 50% carbohydrates (275 g/day) and 15% protein (83 g/day) (CDC, 2004). In Great Britain, the mean intake, as a percentage of total energy, is about 35% lipids, of about 48% carbohydrates and about 17% protein (Swan, 2004).

It is also reasonable to ask in a mortality study, what kind of deaths increased? Well in this study, they showed increases in deaths from all causes, but the only statistically significant increase in deaths was from cardiovascular causes.

What to make of this study?

Well, first there are a some caveats.

1) The people in this study were not explicitly on low-carb diets, so in order to apply this research to them we would have to extrapolate the findings. Further, most low-carb diets involve substantially less carbs than even the lowest decile of carb consumption in this study. The authors, however, argue that it is fair to make that extrapolation:

In our study population, consumption of carbohydrates, even at the low extreme of the distribution, was higher than that advocated by the prescribed low-carbohydrate diets and few individuals consumed more than 20% of their energy from proteins. Nevertheless, it is unlikely that at the extremes of the low-carbohydrate-high-protein intake distribution there would be a reversal of the trend evident in our study population. Indeed, many of contemporary public health policies rely on extrapolations, so that if something is detrimental at a certain exposure level, its effect is likely to be more detrimental at a more extreme level.

2) The Greek diet is very high in monounsaturated fats -- from the olive oil, and this makes their diet different than say American or British diets. Monounsaturated fats have been shown to have some positive health effects. However, the authors argue that differences in fat intake were controlled for in the analysis.

3) Finally, the study did not at all address the issue of supplements, and these could affect the outcome.

In spite of those caveats, however, I would use this research as evidence to caution anyone from embarking on a low-anything high-anything diet -- including low-carb high-protein.

Extreme diets in any form are unwise because too much of anything is not a good thing. Dieting is a tough business, but the evidence shows that good diet is still -- in spite all the lame hype associated with alternatives -- based on healthy food like fruits and vegetables and lots and lots of exercise.

Sorry. The world sucks that way.

Pretty dubious article.

For example:



The fact that the study he cites is not strictly applicable to his anti LC hypothesis doesn't seem to bother him too much.

It certainly doesn't shake his faith in his position.

When you "control" for ten different variables in order to try and isolate the effects of that eleventh variable, I think the data will show pretty much whatever you want based on how you "control" it.

Until I hear of an epidemic of low-carbers dropping dead of heart disease, I'm sticking with this way of eating. I'm not buying any BS statistical study conducted by someone who probably already "knows" LC is bad for you.

One interesting statement was that total energy consumption is associated with mortality. Since LC causes you to wind up eating fewer calories as judged by a net loss of weight, I suppose that factor alone will decrease our mortality. Combine that with the positive changes in blood lipids, and I don't see how anyone can conclude that LC is dangerous in any way shape or form.

Quote: "Studies looking at the comparative effectiveness of different diets have shown that basically all diets are equivalent. (Equivalently bad, that is. All the diets tested produced moderate declines in weight, but adherence was low.)"

So anyone eating lowcarb/hiprotein is in danger but since 'adherence was low', there's no real risk. Lucky that most of us don't have the will to stay on any sort of diet. :-)

The increase mortality was cardiovascular so I assume the lowcarb/hiprotein group was eating a lot of animal meat. Plus these weren't dieters. The researcher decided what was lowcarb, what was hiprotein and slotted individuals into the category. But he states there were no truly low carb dieters included.

Everything I read keeps reinforcing that moderation is the only healthy way.

Instead of the first caveat, the author could have just printed "April Fools!".

For that's where he says that not a single person- not one- in the study of 22,000 was on a low-carb diet. He might as well attempt to predict the weather by looking inside his refrigerator.

How about this line: "...Indeed, many of contemporary public health policies rely on extrapolations.." Yes, we know.

Here's the rub. Unless the low carb eaters they were identifying as such were below the ketogenic threshold, i.e., their insulin was low enough to permit the burning of fat for fuel instead of carbohydrates, they would be unlikely to see many of the benefits that accrue to low carb eating. The extrapolation is patently false.

First of all, this is utterly no surprise. There are plenty of indications that at least some low carb/high protein diets are dangerous. Let's start with the explorer who went on what he thought was an Inuit diet. He ate lots of meat. And he started to get sick. Then he realized that he wasn't eating *everything* that the Inuit eat, he was just eating muscle meat. The Inuit ate pretty much all parts of the animals and fish they caught, including much of what we consider offal. When he included the other elements, his health returned.

Atkins, of course, is *not* supposed to be an LC/HP diet. It is an LC/HF diet, and it's recommended to eat lots of vegetables, which, of course, includes some carbs but, in particular, plenty of fiber. You can really do LC wrong, probably, and doing it LC/HP could be expected to be a problem

Some LC dieters, of course, try to play both ends. They go for LC and they try to stay low-fat, to hedge their bets. After all, isn't saturated fat bad for you, and isn't all fat, uh, "fattening"? Wrong.



It has not occurred to this author that there might be a systemic flaw in the studies. In order to try to avoid participation bias, they attempt to *put* people on a diet. People tend to resist this. A recent study, for example, which found that Atkins was somewhat better than other diet from the point of view of weight loss and cardiac risk indicators, had low compliance. What does this mean about a diet *if you follow it*? Almost nothing. No diet is going to work if people don't follow it!

Instead of looking at the study design and why people did not adhere to the diet, this writer draws a conclusion which is not supported by the studies. Simply handing a person a copy of Atkins New Diet Revolution, and having them attend a lecture about it, quite possibly from someone who doesn't know beans about the diet or who actually opposes it, isn't enough to get a person over the hump of the huge body of erroneous assumptions and social difficulties that stand in the way of successful low carb dieting. The writer holds one of these erroneous assumptions himself, that an LC diet is necessarily an HP diet.:



I agree with part of this, with some caveats. First of all, there are some epidemiological studies which somewhat address the issue. Unfortunately, much of this work was not only poorly done, but it did not really study LC diets, except for some work with the Inuits. And an Inuit diet isn't what LC dieters are eating, though there are some resemblances. In addition, epidemiological studies can suffer from severe bias due to population differences. If a certain population eats diet A, and another population with different traits eats diet B, and there are differences in health, we don't know from the study whether or not the health difference are caused by diet, or by some other factor, such a genetics or social habits. Such studies can only suggest avenues for further research. Unfortunately, one major study was used, with political involvement, to set up the whole low-fat myth and pretend that it was scientifically based, together with a host of ready, knee-jerk assumptions, such as the idea that cholesterol in the diet will translate to cholesterol in the blood, though the two are only weakly linked. Or that fat, that icky, sticky stuff, if you eat it, will clog your arteries. It was an obvious assumption. And clearly wrong. Ultimately, we don't know the answers to these questions to any reasonable degree of certainty. I'm convinced about saturated fat, it is practically my staple now, but I can't claim that this has been clearly proven to be safe. It just probably is.

Secondly, of course, the low-carb diets that are "de rigeur" are not necessarily the kinds of diets that were studied. The writer does note this later, almost as a footnote. Don't you love it when a headline screams something and a footnote reverses the impression?



This could be an interesting study. However, what the writer puts in his footnote is actually pretty obvious from the start. This is not about "diets," i.e., specific eating plans followed for weight loss or gain or other health reasons, but about "diet," that is, various eating patterns followed by subgroups of people.



Good, again. Still studies like this are notoriously subject to various kinds of bias. It is extremely difficult to eliminate all possible confounding factors.



Again, not surprising, though the "total energy intake" would be expected to be complex, and it would interact with other factors, such as exercise. Someone who exercises is expected, wouldn't you think, to have, other things being equal, higher energy intake. But in a population, it would not be surprising that total energy intake rising would be correlated with obesity, a clear health risk, if it is extreme.



The more complicated the measure, the more likely that something will go wrong. The elephant in this living room: fat is not mentioned. Even if we assume that the metric they use is a good one, and that is a very complex question, I'm sure, is the difference in mortality associated with the reduction in carb, the increase in protein, or both?

We know that a diet that is totally carbs, no protein, is unhealthy. So this metric, if it depends on the carb/protein ration, would be expected to show high mortality when the metric is very low, lower mortality when the metric is at some middle level, and then higher mortality again as the metric increases further.

But what is utterly unmentioned is what happens if you replace carbs with fat instead of with protein? Most LC diets do just this, instead of stuffing you with protein. Protein consumption may be elevated, by my impression is that it is not to a level hat makes the diet "high protein." The dieter is not burning protein as fuel, rather, fat is being burned.

Now the study does report lipid consumption, saturated and unsaturated. I don't have time to go over it in detail. From the study discussion, there is this, though:


This claims directly that the Atkins diet is LC/HP. Let's look at the references:

Volek J, Westman E (2002). Very-low-carbohydrate weight-loss diets revisited. Cleve Cl J Med 69, 849–858. http://www.ccjm.org/pdffiles/Volek1102.pdf

"The studies presented in this review examined very-low-carbohydrate diets that contained less than 10% carbohydrate, 25 to 35% protein, and 55% to 65% fat." Atkins is not mentioned, but a typical set of meals in the study looked like Atkins to me. Bacon and a tubful of lard for every meal. Just kidding! :-) Lots of vegetables and greens....

The increase in protein is on the same order as the increase in fat. Carbohydrates have been reduced, and replaced with fat and protein, in more or less equal amounts, in terms of weight. Fat, however, is more energy-dense, so the matter is more complex.

The study concludes that:

Astrup A, Meinert Larsen T, Harper A (2004). Atkins and other low-carbohydrate diets: hoax or an effective tool for weight loss? Lancet 364, 897–899.
http://www.ncbi.nlm.nih.gov/entrez/...8&dopt=Abstract

I could not read the text of the article. The title is provocative, in the abstract, we have, however, "Long-term studies are needed to measure changes in nutritional status and body composition during the low-carbohydrate diet, and to assess fasting and postprandial cardiovascular risk factors and adverse effects. Without that information, low-carbohydrate diets cannot be recommended." Of course, there has been subsequent research that counters this, such as the JAMA study, which concluded that recommending Atkins was quite reasonable. (And the effect of recommending it is what was studied, compared to recommending three other options: the low-fat Ornish diet, the LEARN diet based on the USDA food pyramid, and the Zone diet. The actual diets weren't clearly studied, since compliance was relatively low in all cases.)

Lara-Castro C, Garvey WT (2004). Diet, insulin resistance and obesity: zoning in on data for Atkins dieters living in South Beach. J Clin Endocrinol Metab 89, 4197–4205. http://jcem.endojournals.org/cgi/co...tract/89/9/4197

"The widespread acceptance of popular low-carbohydrate high-fat diets (e.g. Atkins Diet, Zone Diet, South Beach diet) further underscores the need to evaluate dietary interventions regarding their safety and metabolic effects. These high-fat diets have been shown to be safe in the short term; however, their long-term safety has not been established."

Now, back to the Greek study:



While this is reasonable, it can be drastically wrong as well. Human beings are designed, quite obviously, to survive under different conditions, where different kinds of foods are available. The author is assuming some kind of positive correlation between an alleged harmful substance and and heath hazard. But the correlation between elements of diet and health can be quite complex. Something which is healthy in one context, even beneficial, can be harmful in another. Carbohydrate digestion, for example, is suppressed in the presence of fiber, fat, and protein. So if you eat these things with carbohydrates, the harm that can come from the insulin spike, over time, is reduced. Further, to switch from glucose metabolism to fat metabolism does not occur immediately, so how a food affects you may be related to what you've been eating for the past few days.

The LC-HF diets are called ketogenic diets, because they shift the metabolism of the body from burning glucose to burning fat (ketone bodies). An LC-HF diet is really a radically different dietary context! For example, it may well be that increasing protein while maintaining lowering fat and maintaining carbs (as represented by caloric content) will be harmful, but the same level of protein in the context of a high-fat, ketogenic diet -- which requires low carb or the body does not run the ketone-burning program -- could be without any harm whatever.

The author has done a bit of sleight of hand, we must note. He is drawing conclusions about lowering carbohydrates, and he is considering a modest lowering of carbohydrates. He then refers to common assumptions about *increasing* the level of exposure to toxic substances. Removing an allegedly beneficial substance is not the same as increasing exposure to a harmful one. In the former case, there may be a reduction in benefit, or, if the benefit is replaced through other means, no reduction at all or even a net benefit, but in the latter, we would quite reasonably expect the harm to increase. It might still not be true, if other aspects of the situation also change, but it is indeed something we can expect to see.

Back to the original review of this article:



The matter of most concern is the possible increase in mortality associated with high protein consumption. If we look at LC/HF diets, there may be some increase in protein consumption, but we do not know what the critical level is at which such an increase might increase health risks. The dietary context may be very important. The study participants were not "dieting," they were eating what they habitually ate. I would have to pour over the data in the article for a lot long than I have to extract much from it. The authors themselves note:



The caution about high protein is probably in order. But until we have more specific studies, we would have to rely upon epidemiological ones, and the most relevant involve the Inuit. They had a VLC-HF diet, I'm not sure about the protein levels, but I expect that they were comparable to the mild elevation in most LC diets. And they were healthy, generally, until modern foods arrived and they started eating much more processed food and carbohydrates.



This is unlikely to be a result of LC or of HP. The conventional wisdom would pin this on HF, but the short-term studies show no increased risk from HF diets, and there is no particular reason to believe that this would reverse in the long term. So I suspect that the effect they are seeing is indirect. For example, if reducing carbs modestly, which is what they were able to study, and replacing them with protein, leads to, say, weight gain, then we would expect to see a correlation with heart disease. But, really, this study does not answer the question. Like many such studies, it is a basis for further research.



It is indeed good advice in general to avoid extremes. The problem is that modern diets may be extreme in processed carb content, so our perception of what is extreme may be warped. The very dependence upon grains as a foundation of the diet is modern for humans, we were not particularly designed for this diet.

An Atkins diet is heavily dependent upon "healthy food like fruits and vegetables," which may come as a surprise to this writer. He really ought to read Atkins!

First of all, Atkins is not a long-list-of-forbidden-foods diet. You can, particularly on maintenance -- i.e, long term, which is where the concerns are -- eat pretty much whatever you want. But with some foods, typically high-glycemic index foods or other foods that will shift your metabolism, you need to be quite careful about quantity and frequency. That's all. Atkins was known to enjoy a baked potato from time to time, not that his personal diet is all that relevant. However, I'm going to guess that he put lots of butter and sour cream on it. That's what I do on the rare occasion I eat a potato! Yum!

You can be a vegetarian and follow a low-carb diet. Most of us eat meat, to be sure. But the foundation of my personal diet would be dairy, especially cream and butter. I like my vegetables with butter on them, and I apply it liberally. I drink heavy cream in my coffee, and I put it on blueberries.

Yes, blueberries. High fiber, not very high carb, plenty of antioxidants. And delicious. And I don't eat a bowl a day.

During induction on Atkins, the carb levels are reduced to quite low, below 20 grams per day net impact carbs (carbs minus fiber and other non impact carbs like sugar alcohols), but this is not intended to be long term; many think of Atkins as being the much more restricted induction diet, which is intended to kick-start benign dietary ketosis, and is recommended for two weeks.

Atkins is sustainable as a diet because the staple foods are easily obtained and they are, in fact, for me, the most delicious foods. They always said to me, long ago, "Have some bread with your butter!" When my friends and I went up and bought raw milk in 5 gallon jugs, I drank the cream off the top. I was raised on steaks and potato, with butter and sour cream. Now, I don't eat the potatoes, that's all. And I eat a lot more vegetables, Atkins has really taught me to appreciate them. And salads. A current favorite is to cut up good tomatoes, add mozarella balls, pour some extra virgin olive oil over it, and cut up some fresh basil.

And some of these papers we see speculate that Atkins works because the diet is so boring! It's amazing how out-of-touch some "experts" can be!

what annoys me about studies and statements like this is that there's no clarification of these figures. you're talking about 2% of the entire study group, with no explanation as to how many of those deaths were caused by external factors (mortal injuries), how many were due to medical conditions not associated with diet like old age or genetic conditions. even without that, a 2% mortality rate of any observed population is well within normal.

even with the supposed increase in cardiovascular deaths, how many of those were simply age-related? a 10-year study certainly qualifies for "long-term", but you still have to take into account the increasing age of the subjects.

and i agree, publishing findings claiming that low-carb diets increase mortality rates and then admitting that the subjects weren't REALLY on low-carb diets not only defeats the purpose, it produces fraudulent results.

That's not extrapolation, that's making up a pretense of knowing the relative results along an unplottable curve. There's no basis to make a calcuable extrapolation because there isn't any data that's relevant.

These people know what they are doing; extrapolation is essentially a linear guess through a series of calculations where there is a linear relationship between the controlling factors. But here there is no basis for justifying the linearity.

As we here know, once you lower carb intake below a certain point, the whole things shifts dramatically, then yet again becomes non-linear in a different way. [In essence, if we stay above the carb threshold, you can draw a vaguely linear inference as to what happens, but below things "flip" and you get a different set of results that might themselves be vaguely linear with respect to each other, etc. But there is no way to get one set to predict the other.]

They either know or don't care about the flip-over effect, and in any case produced results consistent with their expectations, just like any other aspect of junk science.

A general caveat that applies to all researched endeavors, far beyond "merely" our set of topics:

Research science has become a big business unto itself [I worked in the field for 20 years nearly 20 years ago; the seeds of the problem were already appearing even then, and it's gotten a whole lot worse!].

In essence, you have to show you are doing research that aids and abets the current fad of opinion about your field, whatever it is [including climate science], or else you don't get funded. Any research endeavoring to actually find something new that MIGHT disagree with prevailing "conventional wisdom" is likely to NOT get funded.

Grant advisors will tell you to carefully word your grant proposals in light of the prevailing "wisdom" so they don't see you as an enemy of the current opinion, but rather as someone just wanting to confirm it even more, but perhaps in a slightly different manner that will just strengthen their collective resolve.

Thus, it's no wonder it's hard to get anyone to believe other than the current propaganda about carbs, among a host of subjects held hostage to this unsavory business side of funded research, etc.

A usual cohort in this mess is that you should always go out of your way to include "computer modeling" of what you want to propose, especially if the only way to prove anything useful involves animal trials lest we might kill human subjects by accident due to any number of steps along the way we blunder at. That the computer modeling might be totally useless and inappropriate doesn't matter. Tell them you are researching a cure for cancer and need animal subjects to prevent human deaths during the experiments, and the end result could save large numbers of human lives, and you won't get a dime for research. But tell them that you intend to do worthless computer modeling that merely graphically shows the results of unprovable assumptions and you get way up the list for grant money, etc.

And, as we all know, all computers are PERFECT for everything! If a computer showed us something we made up the inputted parameters to the program, it must be true because we can see it with our own eyes, etc.!

The value of a lot of research has dropped significently over the years as professional researchers search for something to research that will get funded as opposed to research geared to actually accomplish something. It's turned into somewhat of a growth industry.

Given all of that pressure, it's no wonder we see clap-trap like this one, waving nonsensical propositions in our face as "scientific" etc.

cjl (remember, 4 out of 5 doctors .... is 80%)

I think this says all that needs to be said about how wrong the study is.
http://www.proteinpower.com/drmike/...ugar_hypot.html
And what makes the blood sugar rise, CARBOHYDRATES, of course.

Looking at it in another way.
It is generally agreed that high triglycerides and high insulin levels in the blood are bad. And what makes those levels high? Not protein! Or fat!

WOW! I just noticed three posts since I started my reply. I may need to edit this after I have a chance to read them.

Excellent analysis, Abd! Congratulations.

CLASYS, your comments on non-linearity were a good thing to point out. Many things in nature are non-linear, even to the point of being discontinuous. Electronics is a field filled with those things.

Abd: Excellent point by point refutation of the linked article.

and

CLASYS: Fine examination of the deceptive framing of the research cited in the piece.

The answer is in the "corrections" (an oxymoron int his context) for BMI and energy intake.

I believe I have good grounds to assume the low carbers generally have lower BMI. Then when you "correct" their death rates to what they "would be" if they had high BMI, you in effect multiply them by the improvement in death rate seen between the the low carbers and the high carbers. QED.

But don't stop there, you next "correct" for the lower energy intake of the low carbers and multiply by the ratio of improvement in death rate between low energy and high energy intake.

These type of repeated corrections are only valid to remove statistical noise. They are not valid if the various items being "corrected" for show correlation with each other, or with the subject variable of the study. IN this case lower carb versus higher carb.

If you really wanted to you could carry on and "correct" yet again for lower triglycerides in the low carbers.
Then you could "correct" for poor LDL/HDL ratio etc.
The longer you go on the more deadly the lower carb diet would become.

The single "correction" for BMI is a very common ruse in studies "discrediting" low carb. Seeing it used multiply here sort of gives the game away.

I just want to say, after reading the responses...... we have some tremendously bright people on this site!

I'm adding my two cents. HIGH FAT/low carb. Not only have I lost 50 lbs, I no longer need BP meds. My cholesterol has never been better and I have found that this is a great WOL. Not a DIET. I am MUCH healthier eating this way so I will use my own success as a rebuttal against "the study". I will be having my one year atkinsversary tomorrow.

Good job, Hellion