Friday, November 10, 2006

Spinning Low-Carbohydrate Diet Study Data

The last two days have brought us headlines trumpting the long-term finding that low-carbohydrate diets do not increase the risk of heart disease:

These and other headlines were reporting the findings published yesterday in the New England Journal of Medicine in Low-Carbohydrate-Diet Score and the Risk of Coronary Heart Disease in Women. In this paper, researchers from Harvard School of Public Health and UCLA analyzed the Nurses' Health Study data, with more than 82,000 subjects over a period of twenty years with a low-carbohydrate diet score. The score was based on percentage of calories from carbohydrate, fat and protein.

As is almost always the case, what we're reading in the media reports isn't exactly what was published in the data.

For example, the Chicago Tribune reported "those who ate a low-carb diet but got more of their protein and fat from vegetables rather than animal sources cut their heart disease risk by 30 percent on average, compared with those who ate more animal fats."

Forbes reported "Heart risk was also 30 percent lower for participants who got their protein and fat from vegetables rather than from meat."

The Baltimore Sun reported "In a separate analysis, researchers divided the women based on their consumption of vegetable fat, such as olive oil. Women who derived the highest percentage of calories from vegetable fat had 30 percent lower risk of heart disease than those who ate a higher proportion of animal fats."

The various news accounts leave readers with the impression that the finding of reduced risk from vegetable fats and proteins was in the context of a low-carbohydrate diet. Surprise - it wasn't. Let's step back for a moment and take a look at the methods of analysis in the study so you can understand why the various media reports are disingenous and inaccurate.

To review the potential risk or benefit of a low-carbohydrate dietary pattern, the researchers devised a scoring system for determining whom amongst the Nurses' Health Study consumed a low-carbohydrate diet. Based on total carbohydrate, total fat and total protein, the reseachers found that 3,693 subjects consumed about 37% of their daily calories from carbohydrate. In absolute grams each day, this worked out to be 139g a day, with the mean intake of 116.7g. This was compared with the highest consumption of 267g each day, with a mean intake of 234.4g.

As the researchers noted in their full-text, this level was "similar to that consumed by participants in the clinical trials of low-carbohydrate diets." It was not very low-carbohydrate, but similar to something like Atkins maintenance or the Zone. After conducting a multivariate analysis (that is accounting for confounding variables like smoking, BMI, hormone replacement therapy, etc.) they found those consuming the low-carbohydrate dietary pattern had a reduced risk of cardiovascular disease, with a 6% reduction in risk. However, this was not statistically significant as the range of risk in the confidence interval crossed the "1" - it ranged from 0.76-1.18.

The finding is however significant in this sense - for years we've been warned that a low-carbohydrate diet might increase risk of cardiovascular disease. This study shows us that over a long period of time - twenty years - it does not increase risk, and even reduces risk slightly. A key piece of data not discussed in the media or even the paper - the number of individuals who followed such a dietary pattern was very small - 4.4% of all subjects from the Nurses' Health Study. I'll get to why this is important in a moment - let's look at what the researchers did next in their analysis.

Using the same scoring system, the researchers now wanted to know about risk based on consumption of animal fats and protein. Using the scoring system with these macronutrients as the primary focus, they again grouped subjects into ten deciles - this time based on those with the highest consumption of animal fats and protein examined. Interestingly, those with the highest consumption of animal fats and protein again consumed the least carbohydrate - this time an average 128g per day. This was compared with the highest carbohydrate consumption group eating an average of 264g each day. And, again, using the multivariate analysis, an identical reduction of risk was found - 6%; with the range being 0.74-1.19; statistically insignificant when compared with the lowest intake of animal fats and protein. Noteworthy here - even less subjects consumed this dietary pattern - just 3.5% of participants in the study.

Both of these analysis were low in carbohydrate, but the researchers had one more question to ask - what about vegetable fats and protein? Once again they scored subjects, this time based on consumption of plant based calories for each macronutrient. And, AHA!, they found statistical significance! Those who consumed higher levels of plant based fats and proteins had a statistically significant reduction in cardiovascular risk.

Too bad it wasn't a low-carbohydrate diet. In this analysis, the lowest group intake of carbohydrate was 202g each day, compared with 242g each day in the highest intake group. There are two problems with this - one, the difference between the highest intake of carbohydrate and the lowest intake is minor, and two, in absolute grams each day, both the highest intake and lowest intake consumed almost identical animal fat and protein. Add to this, the lowest carbohydrate group now accounted for 9.3% of the subjects and it's easier to understand why this particular finding has nothing to do with low-carbohydrate diets - both of the previous analysis found less than 5% consuming a low-carbohydrate dietary pattern and here, we now have a larger population, consuming a much greater intake of carbohydrate, being used to conclude vegetable fats and protein are protective on a low-carbohydrate diet.


The dietary pattern in the third analysis had one major difference in foods eaten...and it wasn't red meat or any animal food for that was nuts.

The group that was found to have a reduced risk ate four times as many nuts each week - 2.8 servings a week compared with less than 1 (0.7 servings per week). This led to something else being different - magnesium intake. The group consuming more nuts also consumed much more magnesium - 320mg each day compared to just 284mg a day.

The group with a reduced risk also had 29% taking a multivitamin compared with 23% in the group at the extreme other end. One last interesting finding with this group - they ate less fruit and vegetable than the other groups, consumed more coffee and, gasp!, consumed more saturated fat!

So, we find the analysis that claims that vegetable fats and protein are protective in a low-carbohydrate diet are not based on the data here - the dietary pattern was not low-carbohydrate, the group was not shunning animal foods (as implied), and they were not eating more fruits and vegetables (as implied).

The researchers did indeed find there was no risk to following a low-carbohydrate diet in the long-term; they even found that high intakes of animal fats and protein wasn't going to increase your risk of cardiovascular disease. What they didn't find is that vegetable fats and protein are protective in the context of a low-carbohydrate diet.

Instead they found, in the context of an intake range of 202g to 240g carbohydrate, when subjects are consuming similar intake of red meat, chicken, fish - a higher consumption of nuts, coffee, saturated fat and whole grains with less fruits and vegetables may provide a benefit in the context of such a dietary pattern higher in carbohydrate. Just don't expect them to tell you that - instead they'll continue to perpetuate the myth that animal foods and saturated fat is detrimental to your health.

Take home message here - the data is clear that if you're following a low-carbohydrate dietary pattern, a diet that is in the range of 100-130g per day - even one rich with animal foods, you do not have an increased risk of cardiovascular disease.

Since it's November, and it's National Diabetes Month, I'm going to ask - when will the American Diabetes Association start to seriously consider drafting a clinical guideline for those at risk for or diagnosed with diabetes to have the option of trying a low-carbohydrate diet to control their blood sugars?


  1. Thanks for the further info. Based on yours and the summary in the NEJM it appears as more of a "win" for a zone type diet. I haven't yet seen a coment from Sears, but Ornish is not happy from what I've seen.

  2. Is there a commonly agreed upon definition of what constitutes a low-carb diet? Under 150g of carbohydrates a day? Under 200g? Does this total include both fiber and sugars? Or sugars only?

    Thanks for the very informative blogging!

  3. Prorata - there is no definition of a low-carbohydrate diet that is agreed upon. That said, most hold that less than 60g net (deducting only fiber) each day is a ketogenic low-carb diet; any dietary pattern providing less than 130g a day is a low-carb diet; and usually up to 40% of calories (Zone type) is low-moderate carbohydrate when supplying more than 130g per day.

  4. Thank you for the info!

  5. Anonymous11:18 AM

    WOW! I just found this blog and it is wonderful.

    I too was confused about the data on this report, especially the emphasis that less animal protein and more vegetable fat is protective in a low carb context. Thank you so much for clearing that up! Your analysis of scientific data is very helpful; are you a scientist or medical researcher? If not, you should certainly be one! Thank you again for this analysis.

    P.S. I have been on Charles Clark's New High Protein Diet (the European version of Atkins), and have never felt better in my life. I will NEVER stop low carbing. I also know which blog I'll be frequenting from now on!

  6. The medical fraternity (especially the diabetes people) admit that low carb mgith benefit diabetics? Hell will freeze over first and the devil take the patients!
    Great post, though. Thanks for the clear analysis.
    I've lost 29 kgs and my migraines and other health problems followuing low carb and I'll never go back either.

  7. Regina,

    Excellent job exposing this latest attempt to redefine 'low-carb' as being high in vegetable fats and oils. I was suspicious of those article from the very beginning. Thanks for cutting through the smoke and mirrors.

  8. Thank you Regina, I have been waiting for someone to tell us the truth on this!

  9. Thanks!!! I knew that there was more to the story!!! And I knew you'd be the one to present it!!!

  10. great writeup and analysis of the study. i can continue to rest easy.

    - Mr F

  11. Anonymous10:34 AM

    Regina, it might be useful to know what is the take home message be for those who eat a moderate carb diet. Eat more food rich in magnesium ?

  12. RE the magnesium - I have another article coming tomorrow! I'm not convinced it was higher magnesium intake (although I'm sure that contributed) but something else...

    How's that for a teaser?

  13. Anonymous10:45 PM

    Don't assume that the dietary data is the least bit accurate though! They just administer a long list of "food frequency" questions - how often do you drink 8 oz of skim milk, how often do you eat 1/2 cup of carrots etc. These are known to be pretty inaccurate, except for the foods that a person never eats or that they eat every day. They have to 'adjust' the dietary data just to make it even look reasonable. These data are just based on people's inaccurate reports of what they eat, and the numbers are highly questionable.

  14. One thing that struck me about this study is that it was NOT reported in the New York Times. I think someone at the Times has serious objections to lowcarb. When Atkins died, the obit they published was very lukewarm. Maybe my memory is selective, but I'm sure that every time a story like this comes up, the Times manages to twist it into supporting the AHA status quo position.