Monday, May 23, 2011

The China Study II: Wheat may not be so bad if you eat 221 g or more of animal food daily

In previous posts on this blog covering the China Study II data we’ve looked at the competing effects of various foods, including wheat and animal foods. Unfortunately we have had to stick to the broad group categories available from the specific data subset used; e.g., animal foods, instead of categories of animal foods such as dairy, seafood, and beef. This is still a problem, until I can find the time to get more of the China Study II data in a format that can be reliably used for multivariate analyses.

What we haven’t done yet, however, is to look at moderating effects. And that is something we can do now.  A moderating effect is the effect of a variable on the effect of another variable on a third. Sounds complicated, but WarpPLS makes it very easy to test moderating effects. All you have to do is to make a variable (e.g., animal food intake) point at a direct link (e.g., between wheat flour intake and mortality). The moderating effect is shown on the graph as a dashed arrow going from a variable to a link between two variables.

The graph below shows the results of an analysis where animal food intake (Afoods) is hypothesized to moderate the effects of wheat flour intake (Wheat) on mortality in the 35 to 69 age range (Mor35_69) and mortality in the 70 to 79 age range (Mor70_79). A basic linear algorithm was used, whereby standardized partial regression coefficients, both moderating and direct, are calculated based on the equations of best-fitting lines.

From the graph above we can tell that wheat flour intake increases mortality significantly in both age ranges; in the 35 to 69 age range (beta=0.17, P=0.05), and in the 70 to 79 age range (beta=0.24, P=0.01). This is a finding that we have seen before on previous posts, and that has been one of the main findings of Denise Minger’s analysis of the China Study data. Denise and I used different data subsets and analysis methods, and reached essentially the same results.

But here is what is interesting about the moderating effects analysis results summarized on the graph above. They suggest that animal food intake significantly reduces the negative effect of wheat flour consumption on mortality in the 70 to 79 age range (beta=-0.22, P<0.01). This is a relatively strong moderating effect. The moderating effect of animal food intake is not significant for the 35 to 69 age range (beta=-0.00, P=0.50); the beta here is negative but very low, suggesting a very weak protective effect.

Below are two standardized plots showing the relationships between wheat flour intake and mortality in the 70 to 79 age range when animal food intake is low (left plot) and high (right plot). As you can see, the best-fitting line is flat on the right plot, meaning that wheat flour intake has no effect on mortality in the 70 to 79 age range when animal food intake is high. When animal food intake is low (left plot), the effect of wheat flour intake on mortality in this range is significant; its strength is indicated by the upward slope of the best-fitting line.

What these results seem to be telling us is that wheat flour consumption contributes to early death for several people, perhaps those who are most sensitive or intolerant to wheat. These people are represented in the variable measuring mortality in the 35 to 69 age range, and not in the 70 to 79 age range, since they died before reaching the age of 70.

Those in the 70 to 79 age range may be the least sensitive ones, and for whom animal food intake seems to be protective. But only if animal food intake is above a certain level. This is not a ringing endorsement of wheat, but certainly helps explain wheat consumption in long-living groups around the world, including the French.

How much animal food does it take for the protective effect to be observed? In the China Study II sample, it is about 221 g/day or more. That is approximately the intake level above which the relationship between wheat flour intake and mortality in the 70 to 79 age range becomes statistically indistinguishable from zero. That is a little less than ½ lb, or 7.9 oz, of animal food intake per day.


Peter said...

Then you have Richard Feinman's view: “The deleterious effects of fat have been measured in the presence of high carbohydrate. A high fat diet in the presence of high carbohydrate is different than a high fat diet in the presence of low carbohydrate.”                                                              

John said...

So we're looking at a diet composed of pretty much just wheat and "animal foods"? The animal food isn't replacing the wheat, correct?

Gretchen said...

Couldn't these results be interpreted to mean that the more wheat people eat, the less of some nutrient that is not in wheat they get? The meat could provide that nutrient.

IOW, it's not that the wheat is causing increased mortality except in that it's substituting for healthier foods.

Is there any way you could analyze for that?

I'm not saying wheat isn't bad in some people.

cpsdcarroll said...

I'm not sure of when the data set was taken or the diets of the subjects but I often wonder if the older population has been "protected" so to speak, from an intake of more traditional foods early in life and for a longer time span than the younger generation. It seems that the problems of wheat consumption (and sugar and high n-6 vegetable oils) express themselves earlier for each new generation of industrialized peoples. It's as if we've depleted the diet from high nutrition foods like liver and then replaced the bulk of those calories with wheat, sugar, and vegetable oils. There is some good indication that the liver and eggs can be protective against the kinds of problems you see developing from the high intake of the frankenfoods. I remember listening to Sally Fallon talking on Super Human Radio about vitamins A and K and how important it is to establish them in the diet as soon as possible for children, and it just seems to me that the people who ate organ meats and real foods at an early age are at least protected from metabolic syndrome proportionally to their intake of these foods.

Ned Kock said...

Hi Peter. Fat is very calorie-dense, and carbs raise insulin, which promotes fat deposition. So eating those two in large amounts seems like a good idea if you are a Sumo wrestler trying to stick to the ground more when others attempt to throw you around.

Having said that, I do eat a reasonable amount of carbs and fat on days I exercise, and I have not been putting on any body fat lately. I have actually been losing body fat, but not weight.

That’s from someone who puts on fat very easily; I was once obese.

The key, in my case, is not to do that every day, and keep overall calories down.

Btw, the best low carb diet is a low calorie diet, in my opinion.

Ned Kock said...

Hi john. A. foods replace wheat somewhat, as we’ve seen before on other posts covering this same dataset, but I don’t think this fully explains the lack of effect of wheat on mortality in the 70 to 79 age range at the high a. foods intake range.

On this dataset we can actually tell that there is a displacement effect based on the right plot, the plot for the high intake of a. foods. The data points are somewhat concentrated on the left area of the plot, which is the area of low wheat consumption.

Still, we can also see data points on the right plot associated with high wheat flour intake (right area of the plot) and both low and high mortality (top and bottom areas of the plot).

The effect that seems to be the most salient in this analysis is the moderating effect of a. foods on the relationship between wheat flour intake and mortality in the 70 to 79 age range.

Ned Kock said...

Hi Gretchen. The only way to answer that question is to segment the dataset into high a low a. foods and analyze the subsets separately, which creates some problems of its own because of the small resulting sample size for both subsets.

Ned Kock said...

Hi cpsdcarroll. Interesting hypothesis.

As you can see on the right plot, there are data points associated with high wheat consumption and low mortality. We also find the opposite on that plot – low wheat consumption and high mortality.

Each data point is a county in China.

Ned Kock said...
This comment has been removed by the author.
Ned Kock said...

I should note that this is not the type of effect that one often sees being tested (i.e., a moderating effect), because most stats software packages make this type of test quite laborious.

Still, it allows me to address an issue that has been in the back of my mind for quite some time, and keeps coming up.

For example, I was recently in The Netherlands, having a meal with local employees of a company at their cafeteria. Everybody was eating wheat-based products, and those folks seemed to be generally healthy and relatively thin.

Paying a little more attention, I could see that those folks were eating butter and minimally processed dairy products. They served slice cheese, but it was not the kind of processed cheese that we call American cheese. It was aged cheese. One of the types of sliced cheese was actually aged raw milk cheese.

The Dutch seem to exercise more, biking everywhere, but I also think that their high intake of minimally processed dairy may have something to do with their good health.

Roberto said...


In the Netherlands, were they eating bleached wheat or unbleached/whole wheat?

I strongly suspect that if wheat is toxic it becomes so through the bleaching process. I've noticed that I can eat whole wheat and unbleached wheat with no apparent ill-effects, but bleached wheat gives me terrible acid reflex.

It frustrates me when people indict wheat altogether without considering the various levels of processing it undergoes.

Roberto said...

*acid reflux

anonymoose said...

Of course the catch is, eating half a pound of meat a day increases your risk of colon/rectal cancer.

Ned Kock said...

Hi Roberto. They were eating both types.

Ned Kock said...

Hi anonymoose. I’ve been suspicious of the alleged connection between red meat and colorectal cancer for a while. The reason is that it nearly always appears together with a connection between obesity/overweight and colorectal cancer, as long as a suitable measure is available (e.g., percentage of obese individuals in a group, average BMI in group etc).

Obesity appears to be a major causative factor in cancer:

Also, in at least one multivariate study I’ve seen (sorry, I can’t find the reference now), the effect of red meat on cancer “disappeared” when obesity was included as a competing independent variable, suggesting that obesity is a confounder that must always be controlled for in this type of study.

iolite said...

This diagram is showing and clearing concept about the wheat flour intake. But, may be it's not that the wheat is causing increased mortality except in that it's substituting for healthier foods.

Ned Kock said...

Hi iolite. I looked into the issue of wheat displacing other foods in this sample. It happens, but is not enough to fully explain the effect on mortality: