|
Animal Fats Don't Cause Breast Cancer
Did you notice how low-carb diets have hit the news in 2003? Wheat farmers
in the USA were finding they can't sell their crop as people were giving up
eating bread. Potato farmers faced a similar crisis in Britain. And I heard a spokeswoman from the Vegetarian Society admit that 600,000 people had given up on vegetarianism.
The next thing
to happen may well be all the people who have taken the 'healthy' advice and
put weight on or developed diabetes as a consquence may decide to sue those
nutritionists who gave them that bad advice.
During the third week of July, 2003, two studies were published which purported
to show that an increasing intake of animal fat increased the risk of breast
cancer. It looks like the conventional 'fat is bad for you' nutritionists are
stepping up their rearguard action in a vain attempt to prevent this inevitable
backlash.
The first was an American study published on 16 July in the
Journal of the National Cancer Institute
; the second out on 18 July in the British medical journal,
Lancet
. Below I have included abstracts from the first, plus the published figures. As the second can be read online, I have given the URL. Look carefully at them and you will see what the real truth is. I must confess,
when I read them both, I was decidedly underwhelmed by the arguments. As the
media — and the researchers — tend to hype up their results for
maximum impact, I have added comments to each to explain what the figures mean
in real terms.
The first study was reported in the American press thus:
Animal Fats Linked to Increased Breast Cancer Risk, Study Finds
July 15 (Bloomberg) — Eating high-fat red meats and dairy products such
as cream may increase the risk of breast cancer in pre-menopausal women,
according to a study published in tomorrow's issue of the Journal of the
National Cancer Institute.
A diet high in animal fat raised the risk by as much as 54 percent, said lead
author Eunyoung Cho, a nutrition researcher at Boston's Harvard Medical School
and Brigham and Women's Hospital. The eight-year study enrolled 90,000 women
aged 26 to 46.
The findings suggest that the Atkins diet and other regimens that encourage
people to eat meat to lose weight may harm younger women, Cho said. Her study
found no link between breast cancer, which kills about 40,000 people a year in
the U.S., and high levels of vegetable fat or animal fat from chicken, turkey
or fish, she said in an interview.
"I would not recommend that diet for pre-menopausal women, unless they replace
red meat with poultry and fish," Cho said.
Women at the high end of animal fat consumption got 23 percent of all their
calories from meats and dairy products, almost twice as much as those who ate
the least animal fat. Some researchers believe that a high-fat diet may
increase the risk of breast cancer by spurring the body to make estrogen, which
can contribute to tumor growth, Cho said.
Now let's look at what the study really revealed:
Eunyoung Cho, Donna Spiegelman, David J. Hunter, Wendy Y. Chen, Meir J.
Stampfer, Graham A. Colditz, Walter C. Willett Premenopausal Fat Intake and
Risk of Breast Cancer.
J Natl Cancer Inst
2003;95:1079?85
Background:
International comparisons and case?control studies have suggested a positive
relation between dietary fat intake and breast cancer risk, but prospective
studies, most of them involving postmenopausal women, have not supported this
association. We conducted a prospective analysis of the relation between
dietary fat intake and breast cancer risk among premenopausal women enrolled in
the Nurses' Health Study II.
Methods:
Dietary fat intake and breast cancer risk were assessed among 90 655
premenopausal women aged 26 to 46 years in 1991. Fat intake was assessed with a
food-frequency questionnaire at baseline in 1991 and again in 1995. Breast
cancers were self-reported and confirmed by review of pathology reports.
Multivariable relative risks (RRs) and 95% confidence intervals (CIs) were
calculated. All statistical tests were two-sided.
Results:
During 8 years of follow-up, 714 women developed incident invasive breast
cancer. Relative to women in the lowest quintile of fat intake, women in the
highest quintile of intake had a slight increased risk of breast cancer (RR =
1.25, 95% CI = 0.98 to 1.59; Ptrend = .06). The increase was associated with
intake of animal fat but not vegetable fat; RRs for the increasing quintiles of
animal fat intake were 1.00 (referent), 1.28, 1.37, 1.54, and 1.33 (95% CI =
1.02 to 1.73; Ptrend = .002). Intakes of both saturated and monounsaturated fat
were related to modestly elevated breast cancer risk. Among food groups
contributing to animal fat, red meat and high-fat dairy foods were each
associated with an increased risk of breast cancer.
Conclusions:
Intake of animal fat, mainly from red meat and highfat dairy foods, during
premenopausal years is associated with an increased risk of breast cancer.
COMMENT:
Now let's look at their figures for animal fats.
|
Animal fat intake quintile
|
1
|
2
|
3
|
4
|
5
|
P
trend
|
|
No. of women
|
17 994
|
18 150
|
18 188
|
18 216
|
18 106
|
-
|
|
Range of animal fat intake, % of energy
|
<14
|
14-16
|
16-18
|
18-21
|
21-46
|
-
|
|
No. of cases of cancer
|
123
|
145
|
151
|
161
|
134
|
-
|
|
Median intake animal fat, % of energy
|
12
|
15
|
17
|
20
|
23
|
-
|
|
Age-adjusted RR (95% CI)
|
1.00 (referent)
|
1.22 (0.96 to 1.55)
|
1.26 (1.00 to 1.60)
|
1.38 (1.09 to 1.74)
|
1.15 (0.90 to 1.47)
|
.04
|
|
COMMENT: Percentage of women who did NOT get breast cancer
|
99.32
|
99.20
|
99.17
|
99.12
|
99.25
|
.04
|
COMMENT:
If eating animal fat increased the risk of breast cancer, one would expect
that the more animal fat is eaten, the more breast cancer there will be. That
is clearly not the case. Those eating the most animal fat (5th quintile) have
less
breast cancer than those eating less (3rd and 4th quintiles). That is the first
sign that something is not quite what it should be.
The second is that the range of findings cuts across
1.0 (below 1.0 indicates benefit; above 1.0 indicates harm; and 1.0 means
neither benefit nor harm). This indicates that the figures may not be
significant.
This is then confirmed if one looks at the "P" figure.
This is a measure of the likelihood that the final figure HAS been arrived at
by accident. The higher the figure, the higher the likelihood that the
outcome is an artefact rather than a true finding. A 'P' of 1 is 100% — a
certainty that it is an accidental figure — and .01 means a 1% chance
that this is an accidental result. Thus a figure less than .01 is usually
required to indicate statistical significance; .001 would be highly
significant. The P figure for animal fats is higher at .04.
This study uses what is known as a 'data dredge' (see below). And it
contradicts all other studies that have gone before which all show that
increasing animal fats have a cancer protective effect. It also tries to
contradict our entire evolutionary history: cancer has really only 'taken off'
in the last couple of centuries, yet we have been eating animal fats for
millions of years.
Another thing that is strange about this study is that this group published a paper only 4 months ago in which they were unable to find a breast cancer risk for red meat intake (see abstract below). Interestingly in the new paper above, analysis was not even mentioned. That abstract is below. It refutes the above study.
Holmes MD, Colditz GA, Hunter DJ, Hankinson SE, Rosner B, Speizer FE, Willett WC. Meat, fish and egg intake and risk of breast cancer. Int J Cancer. 2003 Mar 20;104(2):221-7.
Intakes of animal protein, meat, and eggs have been associated with breast cancer incidence and mortality in ecological studies, but data from long-term prospective studies are limited. We therefore examined these relationships in the Nurses' Health Study. We followed 88,647 women for 18 years, with 5 assessments of diet by food frequency questionnaire, cumulatively averaged and updated over time. We calculated the relative risks (RR) and 95% confidence intervals (95% CI) for risk of developing invasive breast cancer, over categories of nutrient and food intake.
During follow-up, 4,107 women developed invasive breast cancer. Compared to the lowest quintile of intake, the RR and 95% CI for the highest quintile of intake were 1.02 (0.92-1.14) for animal protein, 0.93 (0.83-1.05) for red meat and 0.89 (0.79-1.00) for all meat. Results did not differ by menopausal status or family history of breast cancer. We found no evidence that intake of meat or fish during mid-life and later was associated with risk of breast cancer.
|
The second study was published in
Lancet.
Again it was hyped up by the media as 'conclusive proof that a high animal fat
diet causes breast cancer' but, again, it is nothing like conclusive! Using similar
methodology to that which I have applied above, you will see a similar pattern
here:
Are imprecise methods obscuring a relation between fat and breast cancer?
Sheila A Bingham, Robert Luben, Ailsa Welch, Nicholas Wareham, Kay-Tee Khaw,
Nicholas Day
Lancet
2003; 362: 212-14
Abstract
Pooled analyses of cohort studies show no relation between fat intake and
breast-cancer risk. However, food-frequency questionnaire (FFQ) methods used in
these studies are prone to measurement error. We assessed diet with an FFQ and
a detailed 7-day food diary in 13 070 women between 1993 and 1997. We compared
168 breast-cancer cases incident by 2000 with four matched controls. Risk of
breast cancer was associated with saturated-fat intake measured with the food
diary (hazard ratio 1·22 [95% CI 1·06-1·40], p=0·005, per quintile increase in
energy-adjusted fat intake), but not with saturated fat measured with the FFQ
(1·10 [0·94-1·29], p=0·23). Dietary measurement error might explain the absence
of a significant association between dietary fat and breast-cancer risk in
cohort studies.
COMMENT:
As its title suggests, this study is really looking at the way data are
gathered. What it is really saying is "We 'know' that saturated animal fat
causes breast cancer, but we can't prove it in trials." (This is, of course,
because all trials published so far — and there have been a lot —
have found that animal fats do NOT cause breast cancer, only vegetable fats do
— but that finding isn't politically correct). So now, this team looked
at the way data are gathered to see if they could spin their findings to show
what they want to see.
Note that they show a P value of .005 for a food diary, whereas the P for a
food frequency questionnaire, the usual way to gather information, is .23. This
means that the food diary is very much more reliable.
Now, using this method they studied 25630 men and women to look for effects of
fat intake and breast cancer.
They, too, split fat intakes into quintiles. There were
168 cases of breast cancer between January, 1993 and September, 2002, in
participants who had completed both dietary methods. Unfortunately, this study
doesn't break them down into how many ate how much fat, as the first study did.
However, 168 out of over 25000 in 10 years is not very many on which to base
reliable findings. And, again the data show clearly that the ones who ate the
most fat had less breast cancer than those who ate less. The numbers with
cancer in the two highest intakes (4th and 5th quintiles) was less than the
numbers in third.
So another waste of time!
|
Here is an example of a Data Dredge from
Number Watch
, an extremely interesting website if you want to know how scientists use
figures to mislead us lesser mortals. It is based on an Italian study that found that if women lived a more 'healthy' life, they would live longer.
Number Watch
decided to perform its own study, but not having generous
donors to fund a jaunt to Italy, it has to rely on the random number generator
in Mathcad. It was easy enough to use the same Trojan numbers, but necessary to
make a guess at the number of lifestyle habits and the proportion of women
adopting each one. The numbers chosen were fifty habits and a one in ten
adoption of each. A set of 50 binomial random numbers was generated (for the
cognoscenti by
rbinom(50,2800,0.1)
) and another set for the 4000 controls
with exactly the same number of habits and probability. The percentage
difference was then recorded and ranked, the five at each extreme being used to
form the table. The habits were named from the standard SIF hit list. The
results were as follows.
|
Habit
|
% change in risk
|
|
Tomatoes
|
-11.355
|
|
Jogging
|
-11.354
|
|
Green vegetables
|
-10.891
|
|
Aubergines
|
-10.223
|
|
Olive oil
|
-8.807
|
|
Insecticides
|
+8.753
|
|
Passive smoking
|
+10.883
|
|
Saturated fats
|
+12.128
|
|
Alcohol
|
+16.305
|
|
Smoking
|
+16.456
|
Not bad, when you consider that
there is no difference in the probabilities in
the two populations
! Combining the last three observations, we can now announce
to the women of the world that by giving up smoking, alcohol and saturated fats
they can cut their risk of breast cancer by one third — a similar result
to the good professor's at a tiny fraction of the cost.
As you can see, although the parameters were exactly the same for both the 'treatment group' and the 'control group', merely the fact that there were not exactly the same number of people in each group, skewed the figures. And that may be exactly what happened in the two studies above.
Last updated 19 July 2003
|
"NH&WL may be the best non-technical book on diet ever written"
Joel Kauffman, PhD, Professor Emeritus, University of the Sciences, Philadelphia, PA
The Perfect Weight Plan: Be Slim without Dieting
— a completely new kind of video and DVD.
|
