articles in this week's JAMA have been published about an 8-year long
lowfat trial in more than 40000 women. No effect - neither as regards
breast cancer, colon cancer, CHD or stroke.
Surprise - surprise!
Dag Viljen Poleszynski
studies prove nothing, but Norwegian "experts" are doing
their best to make it appear as a support to high-carb diet!! If
anything, it shows that even a 320 kcal/d energy reduction for 7½ yrs
does not lead to significant weight loss when the diet is dominated by
There is one interesting thing in the JAMA weight study. Mean weigth
change was -1 kg. The Standard deviation however was 10 kg. So the
mean weight change was -1(+ -10) kg. That means that a large minority
of the women actually increased in weight. About 17 % of them
increased by more than 9 kg, and about 2.5 % increaseed by more than
19 kg. This is certainly not a normal distribution. One expects to
find behind such a wide distribution 2 populations with different
means, but merged in one. Probably the ones who increased and the ones
seems to be a lot of post-hoc justification going on - on both sides
of The argument. Of course from the usual suspects we get: 'Well, the
low-fat intervention didn't Prevent [insert condition here] because it
wasn't stringent enough (20% reduction target not met) or targeted
enough ('good' fats were decreased as well as 'bad' fats, oh and we
didn't know about trans fats then!) or the participants were all
overweight anyway (!!!) [so this result doesn't count]...'
Or the best one I saw 'the diet advice didn't include salt reduction
so that's why it didn't improve heart health' reported on http://news.bbc.co.uk/1/hi/health/4690230.stm)
On the other hand the
fact that 7-8 years of a low-fat diet did not produce weight loss has
turned into an unexpected bonus allowing them to say that it 'does not
result in weight gain' and take a pot shot at the anti-carbers! I bet
they didn't set out to see whether a low-fat diet caused weight gain
back in 1993! Of course low-carb people might shoot back that the %
energy from carbohydrate only increased by a small amount in the
intervention group, but clearly that did not make the intervention
women put on weight. Actually there's no mystery as to why the
intervention women on average lost a teensy bit more weight - on
average they ate a teensy fewer kcal as well.
not so sure either that the S.D. on the weight loss figures conceals
anything. For the smaller ethnicity sub-groups it does - look at the
error bars on the all the plots in Fig. 2 where they plot weight loss
over time by ethnicity group They're overlapping all over the place
except for the group as a whole and the largest group - non-Hispanic
whites. But this is just the greater statistical variation you expect
in small groups.
more interest are the plots of weight loss over time separated into
age groups. As far as I can see they show a completely natural
progression in older women – some weight gain up to and around
menopause, then, eventually, with increasing age, weight loss. Thus,
the older women lost weight (both groups) and the younger women put it
on (both groups) i.e. contrary to the report of a weight loss by the
end of the study - that was actually due to the older women. By the
end of the study, the youngest women had putweight on - just that the
intervention group had put on less. Similarly, contrary
to the key claim of 'no weight gain' women of a 'normal' BMI did put
don't think you can discount some kind of 'placebo' effect or perhaps
better to call it a Hawthorne effect. As Lorgeril pointed out in his
Med. Diet presentation, you can't blind a nutritional study! And (and
I wonder about this for the Lyon heart study too) the people in the
intervention group get much more attention - they get 'intensive
behavioural modification sessions' and 4 meetings a year and so on.
The control group just get handed a brochure. There may well be a
contribution from this attention and the additional focus it creates
in the mind of the participants. To me, this may explain the
infuriating way these studies (while although not proving their point)
always seem to be able to say '.. a small but nonsignificant trend to
less [insert condition here] in the target group...' thus keeping
their hopes alive!
agree. The authors had a further comment: "The possibility that
the effect might have been greater if the diet had been initiated at
younger age cannot be ruled out." I think it can because there
was no difference in outcome between those who had a low intake of fat
or SFA at baseline and those who had
a high intake. It seems not too farfetched to assume that the diet
recorded at baseline reflects these people’s usual diet.
Jacqueline, I'm all with you -- until you mentioned Lyon. The n=300 control group did nothing and the n=300 intervention
group did nothing [human nature], EXCEPT receive free for the whole
family a non hydrogenated
canola oil based margarine --and indeed they replaced butter with it--
and they were supposed to try and use some olive oil [sure] andeat
more fish [they did not] and eat more 'mediterranean' [10 more lentils/beans
per day etc]. I.e.
dietary advice [sorry dietitians] never works but people WILL take a
gift, or buy canola instead of sunflower oil [same price].
To get Lyon into Lancet [after NEJM refused it since the cholesterol
lipids did not change one iota], they negotiated with Lancet to make
it politically correct and they put in the word 'Mediterranean' [and
the ignored phrase: alpha-linolenic enriched]. Now, everybody promotes olive, Mediterranean et al while the
blood fatty acid demonstrated difference in the n-3's [ALA and derived
EPA] were the only significant changes that actually took place.
I have been concerned for some time about the thinking (or lack of it)
that surrounds so many
reports in the press, both medical and lay. The paper that I have put
on the group site attempts to rank evidence and I would appreciate any
comments. The paper has appeared in the BC
Medical Journal. It's fascinating to read the rationalisation that is
being written in the Canadian press (Globe and Mail) to attempt to
paper over the lack of effect of diet as reported in the JAMA. I have decided that the cholesterol myth is a
meme aided and abetted by vested interests, both academic and
The academicians and members of the profession in general are so
invested in the "fat" story that it would be professional
and academic suicide to accept the truth. Drowning people are said to
grasp at straws, but these straws have somehow kept them afloat,
largely because no one really wants to confront the reality of a 50
year old bit of nonsense that began with what was long ago shown to be
Well, the sun did ultimately stop orbiting the earth,................
The newspapers are putting their spin on the recent Women's Health
study results, saying that although the study showed that lowfat was
worthless, the results indicated that we should avoid trans and
SATURATED fats. Has anyone taken a close look at the data.
I would like to know whether the study found negative evidence about
See this link for a media portrayal of the Nurses Health Study: http://tinyurl.com/lapdb
study showed that the women on the low fat diet had significantly
lower LDL levels but no significant difference in cardiovascular
disease. In other words, it showed that if there is any relationship
between LDL level and cardiovascular disease it is too small to be
detected in a study on a very large group of middle-aged women. Why
have I seen no comment pointing it out? Am I missing something here?
Sally and sceptics, I don't know, I despair
sometimes I really do - when I see the findings of those
recent studies misrepresented to that extent... especially in the
popular press which is all most people read...
And surely Willett saying:
' What the findings do show, yet was often lost in
news reports, is that eating more fruit and vegetables as well as less
saturated and trans fats cuts the risk of heart disease and cancer.'
That evidence is 'quite strong' is an outright porky! It doesn't show anything of the kind! Even the
' weight loss' results (weight gain more like) as I
and others have already pointed out are mis-explained. And
as for 'even bad carbs aren't so bad' - that was not being tested guys
! - so you can't really
validly comment about what effect it did or didn't have.
(Footnote: strictly speaking - I
know I made this same error when commenting the longitudinal weight
changes between age subgroups in my previous email on this topic.
However, it was interesting to look at, added an extra perspective on
what was going on and I did acknowledge that the low fat group did end
up less heavy than the control group (because they ate fewer calories
Would they publish letters to the editor or even a rebuttal article to
correct this woman's nonsense?
As regards Sallys question. In a word,
NO. The study found no significant increase in CVD outcomes in
the low-fat group EXCEPT, they found the hazard ratio for women in the
low-fat intervention group who already had CVD to increase 1.26 (CVI
1.03-1.54) and this was significant (P=0.006). Interestingly they put
this bit in at the end of a section in a sentence which completely
omits to mention the direction of the 'interaction between the
intervention and the [baseline] disease'). However, they did do some
fancy calculations on CHD risk and found some groups where this fell
where saturated fat and trans fat consumption was lowered. Now this
can usually safely be ignored because 'risk' usually just means the
application of a model incorporating LDL levels, triglycerides,
diabetes, BP etc.
However, it's not clear how this was calculated. The paper says:
"Trends for changes .... were examined by evaluating CHD risk for
individuals stratified by quartiles of achieved levels of key
nutrients in year 1, using the rate in the comparison group as
reference. Analyses were adjusted for age, baseline CHD, ...
My question is - is this using actual CHD outcomes for the
intervention individuals or not? It's not clear because then they talk
about a 'CVD risk model' which includes the usual suspects (BP,
cholesterol level etc.). And then they compare each quartile
intervention group with the WHOLE comparison group!
If anyone can help me out here - it's in 'Additional Analyses' in the
Jacqueline is certainly correct that
the researchers in past work have used corrections for TC, LDL, etc.,
in the wrong direction for the age group. I found that by using their
raw data for events, opposite conclusions would emerge.
What is particularly shocking is that the
effect of the low-fat diet on clinical events was never tested before
it was recommended to the public. To get a drug into clinical practice
demands that it has been shown to be harmless or at least without
serious side effects, first in various laboratory animals, then in
healthy voluntary test subjects, and finally its therapeutic value
must have been proven in a number of clinical trials. But dietary
advices are given to millions of people based on speculation only.