walter willet repeats his low carb message

[Wuzzy]

This month's (MAy's) Obesity Reviews has a massive review on
dietary fat on obesity including willett again (deja vu from
AJCN) repeating in his title:

Dietary fat plays a major role in obesity: no

I'll post comments later the entire journal is
interesting to read..

[Jay Tanzma]

wuzzy wrote:
>
> This month's (MAy's) Obesity Reviews has a massive review on
> dietary fat on obesity including willett again (deja vu from
> AJCN) repeating in his title:
>
> Dietary fat plays a major role in obesity: no

You call that a "low carb message"?

-Jay

[Michael Ro]

On 18 May 2002 08:48:20 -0700, mypcos@hotmail.com
(wuzzy) wrote:

|This month's (MAy's) Obesity Reviews has a massive review on
dietary |fat on obesity including willett again (deja vu from
AJCN) repeating |in his title:
|
|Dietary fat plays a major role in obesity: no
|
|I'll post comments later the entire journal is
interesting to read..

wuzzy, I for one appreciate your posts.

[Wuzzy]

Something cool to note is that the following table represents
the types of whole grains eaten in the 70,000 nurses: (1st
quintile to 5th, ex. for dark bread it goes from 0.05 servings
per day to 1.73)

Foods (servings/d)

Dark bread 0.05 0.24 0.44 0.64 1.73 Whole-grain breakfast
cereal 0.04 0.14 0.30 0.44 0.55 Brown rice 0.01 0.02 0.03 0.04
0.06 Wheat germ 0.00 0.00 0.01 0.02 0.09 Bran 0.00 0.01 0.02
0.07 0.30 Other grains 0.00 0.00 0.01 0.01 0.02 Cooked oatmeal
0.01 0.03 0.05 0.08 0.11 Popcorn 0.03 0.07 0.10 0.13 0.21

Note that dark bread is best thing that represents "whole
grain" eaters (they eat it twice a day! Whole grain cereal
and bran are the next major protective grains eaten once
every two days.

In case anyone wonders what kind of "whole grain" diet the
nurses are on.. Popcorn once a week?

[Wuzzy]

>Foods (servings/d)

>Dark bread 0.05 0.24 0.44 0.64 1.73 Whole-grain breakfast
>cereal 0.04 0.14 0.30 0.44 0.55 Brown rice 0.01 0.02 0.03
>0.04 0.06 Wheat germ 0.00 0.00 0.01 0.02 0.09 Bran 0.00
>0.01 0.02 0.07 0.30 Other grains 0.00 0.00 0.01 0.01 0.02
>Cooked oatmeal 0.01 0.03 0.05 0.08 0.11 Popcorn 0.03 0.07
>0.10 0.13 0.21

Interesting:

the attempt is to group people by "whole grains" and then
predict disease:

Ex. do this by defining popcorn as high grain: therefore the
average "high-grain" person eats alot of popcorn. The average
"high-darkbread", however, may not eat any popcorn.

This is a curious approach, as opposed to: Group by disease.
Find out what high disease and low disease are eating. This
approach is probably not so good since disease doesn't cause
eating, eating causes disease. If butter causes hypertension,
group by butter, not by hypertension.

[Wuzzy]

>This is a curious approach, as opposed to: Group by disease.
>Find out what high disease and low disease are eating. This
>approach is probably not so good since disease doesn't cause
>eating, eating
causes
>disease. If butter causes hypertension, group by butter, not
>by hypertension.

Unfortunately this abolishes the use of 24hour-recalls: the
24-hour has a wider and flatter distribution so that it's
useless to classify people into quintiles unless you exclude
the first and last 'tiles..

Nonetheless knowing 24-hour profiles of quintiles of diabetics
is probably not useful since diabetes does not cause changes
in eating behaviour - changes in eating behaviour, on the
other hand, might cause diabetes.

[Wuzzy]

Jay Tanzman <jtanzman@sph.llu.edu> wrote in message
news:<3CE7C730.40D80670@sph.llu.edu>...
> wuzzy wrote:
> >
> > This month's (MAy's) Obesity Reviews has a massive review
> > on dietary fat on obesity including willett again (deja vu
> > from AJCN) repeating in his title:
> >
> > Dietary fat plays a major role in obesity: no
>
> You call that a "low carb message"?
>
> -Jay

Thanks for the correction, I do tend to "embroider" my
subject-lines..

Technically Willett's diet is mediterranean, his pyramid is
remarkably similar to:
http://www.uicc.org/nutrition/greeknutri.shtml

Except that he adds gollops of olive oil and fat to his foods,
that is, fat is at the base of his pyramid.

[Wuzzy]

How much olive oil is too much,
ie. how much can you capitalise on the fact that veggies have
almost no caloric density?

-Wuzzy
----------------------------------------------------------
There was an article published recently that suggested that
the protective effect of veggies was their "starvation"(low
calorie) effect:

In the early 90s a series of articles were published on
oligomenorrheic females who were consuming high fiber,
veggies, salad.. (they had low c-peptide) As they re-fed them
fat apparently they became regular again and c-peptide went up
again.. This is odd since it goes against the usual "high fat
lowers c-peptide" attitude of low-carbers.. obvious that
veggies have a "starvation-like" effect and is a large part of
its benefit..

[Wuzzy]

>Foods (servings/d)

>Dark bread 0.05 0.24 0.44 0.64 1.73 Whole-grain
>breakfast cereal 0.04 0.14 0.30 0.44 0.55 Brown rice
>0.01 0.02 0.03 0.04 0.06 Wheat germ 0.00 0.00 0.01 0.02
>0.09 Bran 0.00 0.01 0.02 0.07 0.30 Other grains 0.00
>0.00 0.01 0.01 0.02 Cooked oatmeal 0.01 0.03 0.05 0.08
>0.11 Popcorn 0.03 0.07 0.10 0.13 0.21

What I mean to say is that if they had grouped people by
"blood sugar levels", chances are the low blood sugar people
are not eating ANY whole grains..

Appologies for repeated posts to get single message
across, hehe..

[Jay Tanzma]

wuzzy wrote:
>
> >Foods (servings/d)
>
> >Dark bread 0.05 0.24 0.44 0.64 1.73 Whole-grain breakfast
> >cereal 0.04 0.14 0.30 0.44 0.55 Brown rice 0.01 0.02 0.03
> >0.04 0.06 Wheat germ 0.00 0.00 0.01 0.02 0.09 Bran 0.00
> >0.01 0.02 0.07 0.30 Other grains 0.00 0.00 0.01 0.01 0.02
> >Cooked oatmeal 0.01 0.03 0.05 0.08 0.11 Popcorn 0.03 0.07
> >0.10 0.13 0.21
>
> Interesting:
>
> the attempt is to group people by "whole grains" and then
> predict disease:
>
> Ex. do this by defining popcorn as high grain: therefore the
> average "high-grain" person eats alot of popcorn. The
> average "high-darkbread", however, may not eat any popcorn.
>
> This is a curious approach...

Curious? It's standard.

> ...as opposed to: Group by disease. Find out what high
> disease and low disease are eating. This approach is
> probably not so good since disease doesn't cause eating,
> eating causes disease.

You're grouping people both way simultaneously. It's a two-way
classification: level of exposure by disease status. The odds
ratios are the same either way you calculate them. However,
relative risk is by definition the relative risk of a possible
future event, namely developing a disease.

-Jay

[Dolores]

mypcos@hotmail.com (wuzzy) wrote in message
news:<d996c21a.0205201711.988f876@posting.google.com>...
> >Foods (servings/d)
>
> >Dark bread 0.05 0.24 0.44 0.64 1.73 Whole-grain
> >breakfast cereal 0.04 0.14 0.30 0.44 0.55 Brown rice
> >0.01 0.02 0.03 0.04 0.06 Wheat germ 0.00 0.00 0.01
> >0.02 0.09 Bran 0.00 0.01 0.02 0.07 0.30 Other grains
> >0.00 0.00 0.01 0.01 0.02 Cooked oatmeal 0.01 0.03 0.05
> >0.08 0.11 Popcorn 0.03 0.07 0.10 0.13 0.21
>
> What I mean to say is that if they had grouped people by
> "blood sugar levels", chances are the low blood sugar people
> are not eating ANY whole grains..
>
> Appologies for repeated posts to get single message
> across, hehe..

Why wouldn't the low blood sugar people be eating any whole
grains? What is low? I am T2 and eat whole grains at each of
my 4 meals per day---HbA1c is 5.1. Or do you mean really
really low?

Dolores

[Wuzzy]

Jay Tanzman <jtanzman@sph.llu.edu> wrote in message
news:<3CE9C14F.6F462FB5@sph.llu.edu>...
> wuzzy wrote:
> >
> > >Foods (servings/d)
>
> > >Dark bread 0.05 0.24 0.44 0.64 1.73 Whole-grain breakfast
> > >cereal 0.04 0.14 0.30 0.44 0.55 Brown rice 0.01 0.02 0.03
> > >0.04 0.06 Wheat germ 0.00 0.00 0.01 0.02 0.09 Bran 0.00
> > >0.01 0.02 0.07 0.30 Other grains 0.00 0.00 0.01 0.01 0.02
> > >Cooked oatmeal 0.01 0.03 0.05 0.08 0.11 Popcorn 0.03 0.07
> > >0.10 0.13 0.21
> >
> > Interesting:
> >
> > the attempt is to group people by "whole grains" and then
> > predict disease:
> >
> > Ex. do this by defining popcorn as high grain: therefore
> > the average "high-grain" person eats alot of popcorn. The
> > average "high-darkbread", however, may not eat any
> > popcorn.

woops, I am innacurate: The "high-grain" does not eat alot of
popcorn - only once a week! - but they do eat a slight bit
more. The "high-popcorn", though eats "high grain".

I know there is a difference because when I plot "popcorn" (X)
against "grains" (Y)and do a smooth curve, there is a big
slope but when I plot "grains" against "popcorn", the slope is
almost flat..

--------------------------------------------------------------
---------------

> You're grouping people both way simultaneously. It's a
> two-way classification: level of exposure by disease status.
> The odds ratios are the same either way you calculate them.
> However, relative risk is by definition the relative risk of
> a possible future event, namely developing a disease.
>
> -Jay

If you plot this simultaneously, like in a contingency
table (like the harvard group usually does), then it
probably doesn't matter: ex. _________|low
disease____|__medium disease_|__high disease_ low grain| |
| medium gr| | | high gr | | |

here it doesn't matter which is the X and which is the Y

but when you are looking at measures of association, like
doing a smooth curve I think it does, doesn't it?:
Non-diabetics may eat the slightest bit more grains, but high
grain-eaters are *very* non-diabetic.

[Wuzzy]

> > What I mean to say is that if they had grouped people by
> > "blood sugar levels", chances are the low blood sugar
> > people are not eating ANY whole grains..
> >
> > Appologies for repeated posts to get single message
> > across, hehe..
>
> Why wouldn't the low blood sugar people be eating any whole
> grains? What is low? I am T2 and eat whole grains at each of
> my 4 meals per day---HbA1c is 5.1. Or do you mean really
> really low?
>
> Dolores

I'm just trying to understand the nature of complex data: I
was saying that most normal people have normal blood sugar and
do not eat whole grains at all. Those who do eat it every day
have less risk of diabetes, though.

[Jay Tanzma]

wuzzy wrote:
>
> Jay Tanzman <jtanzman@sph.llu.edu> wrote in message
> news:<3CE9C14F.6F462FB5@sph.llu.edu>...
> > wuzzy wrote:
> > >
> > > >Foods (servings/d)
> >
> > > >Dark bread 0.05 0.24 0.44 0.64 1.73 Whole-grain
> > > >breakfast cereal 0.04 0.14 0.30 0.44 0.55 Brown rice
> > > >0.01 0.02 0.03 0.04 0.06 Wheat germ 0.00 0.00 0.01 0.02
> > > >0.09 Bran 0.00 0.01 0.02 0.07 0.30 Other grains 0.00
> > > >0.00 0.01 0.01 0.02 Cooked oatmeal 0.01 0.03 0.05 0.08
> > > >0.11 Popcorn 0.03 0.07 0.10 0.13 0.21
> > >
> > > Interesting:
> > >
> > > the attempt is to group people by "whole grains" and
> > > then predict disease:
> > >
> > > Ex. do this by defining popcorn as high grain: therefore
> > > the average "high-grain" person eats alot of popcorn.
> > > The average "high-darkbread", however, may not eat any
> > > popcorn.
>
> woops, I am innacurate: The "high-grain" does not eat alot
> of popcorn - only once a week! - but they do eat a slight
> bit more. The "high-popcorn", though eats "high grain".
>
> I know there is a difference because when I plot "popcorn"
> (X) against "grains" (Y)and do a smooth curve, there is a
> big slope but when I plot "grains" against "popcorn", the
> slope is almost flat..
>
> ------------------------------------------------------------
> -----------------
>
> > You're grouping people both way simultaneously. It's a
> > two-way classification: level of exposure by disease
> > status. The odds ratios are the same either way you
> > calculate them. However, relative risk is by definition
> > the relative risk of a possible future event, namely
> > developing a disease.
> >
> > -Jay
>
> If you plot this simultaneously, like in a contingency table
> (like the harvard group usually does), then it probably
> doesn't matter: ex. _________|low disease____|__medium
> disease_|__high disease_ low grain| | | medium gr| | | high
> gr | | |
>
> here it doesn't matter which is the X and which is the Y
>
> but when you are looking at measures of association, like
> doing a smooth curve I think it does, doesn't it?:

Not really. Like I said the odds ratio (a measure of
association) is the same, whether calculated as an "exposure
odds ratio" or a "disease odds ratio."

> Non-diabetics may eat the slightest bit more grains, but
> high grain-eaters are *very* non-diabetic.

You're only looking at two cells of the contingency table. You
have to look at all four. Go read an intro epidemiology text.

-Jay

[Wuzzy]

Jay said
> >
> > but when you are looking at measures of association, like
> > doing a smooth curve I think it does, doesn't it?:
>
> Not really. Like I said the odds ratio (a measure of
> association) is the same, whether calculated as an "exposure
> odds ratio" or a "disease odds ratio."

Thanks, this is very complicated [especially for a newbie like
me] and I'm still thinking about it:

I now realize that regression of X=Y is exactly like Y=X

--------------------------------------------------------------
---------------
My confusion is from the fact that I sometimes plot a
"spline matrix" for all my variables. (To do the spline
correlation matrixes in SPSS use IGRAPH /MATRIX command) The
splines always look different for Y=X as it does for X=Y.
Ex. a U-shaped curve will have a U-shape, but the spline
will be flat when you switch the axes. this doesn't occur
with linear graphs..

[Wuzzy]

mypcos@hotmail.com (wuzzy) wrote in message
news:<d996c21a.0205180748.45e7964b@posting.google.com>...
> This month's (MAy's) Obesity Reviews

One of the problems not mentioned in this review is that
most studies define obesity by BMI which isn't a direct
measure of body fat.

As we've seen, this is especially true of short term clinical
trials where the storage of glycogen (carbs) is a less
efficient energy store than fat because it is stored with
water, fat is stored by itself - so there might be a
paradoxical increase in weight with a decrease in body
fat(energy deficiency).. I wonder if athletes who carb-load a
week prior to competition experience any body weight gain.

The within-day variation in body weight is probably not
related to glycogen-stored water, I believe this is only in
the 500grams or so.. compared to the 5lb diurnal variation we
see.. I'll try get exact numbers on what accounts for diurnal
variation in bodyweight..

[Wuzzy]

> I was saying that most normal people have normal blood sugar
> and do not eat whole grains at all.

Just to clarify, this is wrong, diabetics eat less whole
grains than normal people prior to being diagnosed.

I hope I'm the only one who was confused..

-wuzzy

[Wuzzy]

More noise to add to this thread, appologies:

The model I had developed ("normal people don't eat whole
grains") has an error called: "improperly specified":

If you do not include other predictor variables (like
bodyweight, age, genes), then the coefficient of your exposure
will have alot of error, you'll only see a "weak" effect of
the exposure. The exposure actually has a strong effect on the
outcome. But if you don't filter out body weight then it will
appear like whole grains have no effect at all (improper
specification error attenuation)

Similarly 24hour recalls have alot of measurement error (much
like improperly specified models) and so their coefficients
are "weak"/attenuated.. there are ways for deattenuating error
by using gold-standard validation.

A similar discussion happened on another ng regarding error in
the dependant vs. error in the independant.. error on the
dependant doesn't matter..

[Wuzzy]

a properly specified paragraph would read:

Normal people have fluctuations in weight and exercise so the
effect of whole grains is minimal. But if you keep weight and
exercise constant, grains are protective.

okay, now that I finished Chapter 1 of the book I'm reading on
this topic, I will move to chapter two.

[Jay Tanzma]

wuzzy wrote:

> More noise to add to this thread, appologies:

Noise is right.

[remainder of noise clipped]

Give it a rest already.

-Jay