Yeah it’s not a typo! You heard it
right, saturated fat and cholesterol have nothing to do with your risk of developing heart disease!
In this article I’ll go on to show you
examples of how people and even scientists will only believe what they want to believe,
no matter how much evidence to the contrary exists. Here is a take on the
popular notion that a low-fat diet low in saturated fats and cholesterol will
reduce the incidence of CHD and related deaths.
Going through our education we were
taught the basic science tenet that if a phenomenon occurs for a given set of
conditions and if those set of conditions were to be replicated, the same phenomenon would occur. That was enough to establish causality (cause
and effect). This basic tenet holds for any scientific investigation. And if
what you thought was the ‘condition’ that made a certain phenomenon happen in
certain cases and made the exact opposite phenomenon happen in other cases,
then you would abandon the ‘condition’ as being the cause of that phenomenon…unless
of course that ‘condition’ happens to be high cholesterol and the phenomenon
happens to be coronary heart disease. Then, for some reason, the rules of
science don’t apply!
So, any guesses as to what a CHD
preventive diet should be? If you guessed a low-fat vegetarian diet, then you
are in majority by a huge margin and you should be in power (after all we live
in a democracy, right?). But too bad, science and democracy (or majority
consensus) have nothing in common. If they did then the earth would still be
flat and the sun would still be revolving around the earth.
If anyone has popularized a low-fat
vegetarian diet like no one else, I would give the credit to Dr. Dean Ornish. A
quick search on Google will show you that he is one of the most influential
opinion leaders on dietary intervention for prevention of CHD. If you want to
read more about his accolades and recognition, I suggest going to
http://www.pmri.org/dean_ornish.html since his accolades are longer than what I
am prepared to write in this article.
Before we jump into all the lifestyle
intervention studies that have been done to prevent CHD lets have primer on the
statistics behind clinical trials. Actually these principles are used in every
realm of science like process control, manufacturing control and various
testing methodologies and to a much more stringent level than used in clinical
trials (thank the heavens!). Here are a few terms to get used to:
Sample size: This is the
number of people tested. Its best to have the largest possible sample size
since it randomizes for variation in lifestyle (smoking habits, alcohol
consumption etc.), regional/cultural differences or other such confounding
factors. Obviously it’s hard to study a large sample size due to cost/resource
constraints so the experimenter has to choose from a small sub-set of the
representative population. Furthermore a multi-center trail is better suited
for generalization to a large population.
Factorial Design: When two or
more different treatments are evaluated in single study this type of design is
used to evaluate the effect of each treatment, independent of the other
treatment. If two different factors are lumped into one treatment then no
conclusion can be made about the effect of any one factor. It’s easy to see why
this is the case because you can’t un-couple the effect that each factor has on
the treatment. So it’s impossible to establish causality to any one factor. The
best you can infer is that both factors have an effect on the treatment, but
what percentage can be assigned to which factor will be unclear.
Blinding: The best way to
remove bias from a trail is by making it blind. Blinding (masking) is a method
of eliminating conscious or unconscious bias in how the experiment is conducted
or how the data is analyzed or interpreted. It’s basically a way an experiment
is conducted where the subject is not aware of which group is receiving the
treatment (single blind) or in the best case even the test giver, investigator
or the sponsor staff is unaware of which group receives which treatment. This
includes anyone that is designing the test or is selecting the candidates for
the test. Until such time that all opportunities for bias to affect the
endpoint and data integrity is passed all involved parties are blind. This is
maintained throughout the testing and analysis phase and when the data is
cleaned of outliers to an acceptable level is when the appropriate authorities
are un-blinded. This is the best and most objective way to conduct a clinical
trail. Now realize that it’s not always possible to conduct a double blind
trail especially in the case of lifestyle intervention trails, which is what we
will be talking about.
Now that we got the basics down let’s
look at the popular studies that have been done to assess the effect of
lifestyle intervention on preventing CHD. Let’s look at the popular Ornish
study also known as lifestyle heart trail [1] in which he claims that his
particular diet regresses (meaning improves) the atherosclerosis in the blood
vessels. Below is the abstract right from the publication:
In a
prospective, randomised, controlled trial to determine whether comprehensive
lifestyle changes affect coronary atherosclerosis after 1 year, 28 patients
were assigned to an experimental group (low-fat vegetarian diet, stopping smoking,
stress management training, and moderate exercise) and 20 to a usual-care
control group. 195 coronary artery lesions were analysed by quantitative
coronary angiography
And the conclusions stated (again from the publication):
The
average percentage diameter stenosis regressed from 40.0 (SD 16.9)% to 37.8
(16.5)% in the experimental group yet progressed from 42.7 (15.5)% to 46.1
(18.5)% in the control group. When only lesions greater than 50% stenosed were
analysed, the average percentage diameter stenosis regressed from 61.1 (8.8)%
to 55.8 (11.0)% in the experimental group and progressed from 61.7 (9.5)% to
64.4 (16.3)% in the control group. Overall, 82% of experimental-group patients
had an average change towards regression. Comprehensive lifestyle changes may
be able to bring about regression of even severe coronary atherosclerosis after
only 1 year, without use of lipid-lowering drugs.
So the authors concluded that the artery
diameter stenosis (ADS) improved by 5.5% ([40-37.8]/40) in the experimental
group (intervention group) while it deteriorated by 8% in the control group (no
intervention). So apparently the intervention improved things by 13.5%. A
pretty big number if you ask me. So now
you tell me, OK its settled, so will you please give in, bow down, bow out,
take a hike, buzz off and shut up and agree that a vegetarian diet low in
cholesterol and saturated fat is the answer to preventing CHD. I’ll say wait a
minute, but let’s look a little closer at the results.
A few points to note about this trial:
The sample size is pretty small (48) so making a generalization based on this
study would be premature because it couldn’t have possibly accounted for
regional, cultural, dietary and other socio economic factors in the sample
size. How could you possibly make dietary recommendations for people world over
based on the 48 people you tested? You wouldn’t.
The second point I would like to bring
forth is the measurement error in coronary angiography. The numbers reported
are absolute, without any error bars. Any real instrument has repeatability and
reproducibility errors. Also take a closer look at the standard deviation in
the measurements (the % numbers in parenthesis). It’s in the range of 16%. A
quick search on google for measurement errors in coronary angiography shows the
error to be about 10% standard error (http://spo.escardio.org/eslides/view.aspx?eevtid=33&fp=3022).
In light of this fact do you think the numbers reported in the research have
any statistical significance? Absolutely not! The measurement variation
reported is in the signal noise.
The third point I would like to bring
out is that the study made an attempt to correlate the effect of 4 parameters (low-fat vegetarian diet, stopping smoking, stress
management training, and moderate exercise) on ADS. So this study cannot
isolate the individual contributions from any one parameter and its effect on ADS.
So based on this study can we recommend that a vegetarian diet will lower your
chances of developing arterial lesions? I bet you, even the most diehard vegan
wouldn’t agree if someone made a claim like that based on this data. And yet this
study is cited for recommending a vegetarian diet low in saturated fat as a
preventive step to lower your risk of CHD. Hilarious isn’t it?
Now let’s look at the other side of the coin and see how many studies
have failed to show a correlation between dietary lifestyle modification and
the development of CHD. The most popular one that is cited over and over again is
called MRFIT (Multiple Risk Factor Intervention Trial) [2]. This one was special
intervention (SI) program consisting of stepped-care treatment for hypertension,
counseling for cigarette smoking, and dietary advice for lowering blood
cholesterol levels for the treatment group and no intervention for the control
group. Below figure 1 shows the results of this study in a snap shot.
Figure 1. MRFIT study
parameters and results snapshot.
Quoting straight from the research
paper:
The Multiple Risk Factor Intervention Trial was a
randomized primary prevention trial to test the effect of a multifactor
intervention program on mortality from coronary heart disease (CHD) in 12,866
high-risk men aged 35 to 57 years. Men were randomly assigned either to a
special intervention (SI) program consisting of stepped-care treatment for
hypertension, counseling for cigarette smoking, and dietary advice for lowering
blood cholesterol levels, or to their usual sources of health care in the
community (UC).
Here you can at least see that the
study had a large enough sample size and went on for long enough to make some
conclusions. Here again it was a multifactor intervention program where they
modulated three parameters (smoking, cholesterol lowering diet and
anti-hypertensive drugs) simultaneously and studied its effect on CHD. The
primary end point was any MI event. The first thing that should stand out is
that the control group had 132% more smokers than the treatment group. Now looking at the improvement area we can
see that it reduced the coronary death events in the treatment group by 9
people. Now let’s look at people that died of other causes. Wait, the
difference is not 5 from what it seems like from the first results line
(265-260). All the first line states that 5 more people died in the treatment
group than in the control group. So overall more people died in the treatment
group then in the control group. Doing the math it seems like 150 people in the
treatment group died from other causes and 136 people in the control group died
of other causes. So, overall 14 more people died in the treatment group from
other causes. Now would you call the intervention program successful? I
wouldn’t call it successful. So would I call it unsuccessful? I wouldn’t conclude
that either. Basically I couldn’t conclude anything from these results. I think
the results were statistically insignificant and study was inconclusive. But wait here is what the authors concluded (drum
roll please):
Over
an average follow-up period of seven years, risk factor levels declined in both
groups, but to a greater degree for the SI men. Mortality from CHD was 17.9
deaths per 1,000 in the SI group and 19.3 per 1,000 in the UC group, a statistically non-significant
difference of 7.1% (90% confidence interval, -15% to 25). Total mortality rates
were 41.2 per 1,000 (SI) and 40.4 per 1,000 (UC). Three possible explanations
for these findings are considered: (1) the overall intervention program, under
these circumstances, does not affect CHD mortality; (2) the intervention used
does affect CHD mortality, but the benefit was not observed in this trial of
seven years' average duration, with lower-than-expected mortality and with
considerable risk factor change in the UC group; and (3) measures to reduce cigarette smoking and to lower blood cholesterol
levels may have reduced CHD mortality within subgroups of the SI cohort, with a
possibly unfavorable response to antihypertensive drug therapy in certain but
not all hypertensive subjects. This
last possibility was considered most likely, needs further investigation,
and lends support to some preventive measures while requiring reassessment of
others.
Let me put it in simple terms for you.
The data screams that the intervention didn’t work! At least not to a statistical
significance. But wait there is a possible explanation of why it didn’t work.
The authors hypothesize (because data is
telling otherwise) that the intervention may have worked within sub-groups of
the intervention group but a possibly unfavorable response to antihypertensive
drug therapy confounded the results and that the last possibility is considered
most likely. And why is that? Let me tell you why? Because the authors don’t
want to give up the original hypothesis (that the intervention works). So they
already were expecting (AKA bias) which way the results should have gone even
before testing the hypothesis. Wow, this is new age statistics for all of us! Old
school taught us to start an experiment with the null hypothesis and if data
disproves the null hypothesis then you take the alternate hypothesis. This new
way of still holding on to the null hypothesis when experiment shows otherwise
is called ad-hoc hypothesis and you will see many instances of this with
respect to the cholesterol hypothesis. It’s a nifty way to explain away
unfavorable results.
Ok so moving on now lets look at
another one. This is the famous Helsinki Study [3] that was conducted in
Poland. The primary intervention in this one was not only diet, smoking and
exercise but also anti-hypertensive drugs and for all the cholesterol haters
they also gave the preventive treatment group cholesterol lowering drugs. So it
must have been a win-win for the treatment group and I feel sorry for the
control group that didn’t receive any of the life saving miracle drugs or
dietary advice or counseling. The primary end-point was mortality. Figure 2 is
the snap shot of the test methodology and results.
Figure 2. The Helsinki
heart study methodology and results
Looking at the snapshot of the results
it seems like there were 8.5% more smokers in the control group. And lo and
behold only one person from the control group died of a coronary even while 4
people in the treatment group died from coronary event. If it were the other
way around you would be hearing things like “miracle drugs save your chance of
heart attack by 75%” ([4-1]/4). But no such miracles happened and no wonder you
don’t hear much about this study. Also
note that 6 people in the treatment group died from other causes while only 4
in the control group died from other causes. Suddenly I don’t feel so bad for
the people in the control group. The treatment group however…that’s another
story.
During the next 18 years (Between 1974 and 1989) post trail mortality in the
Helsinki Study was followed up and the results straight from the publication
are:
the
total number of deaths was 67 in the intervention group and 46 in the control
group (relative risk [RR], 1.45; 95% confidence interval [CI], 1.01 to 2.08; P
= .048); there were 34 and 14 cardiac deaths (RR, 2.42; 95% CI, 1.31 to 4.46; P
= .001), two and four deaths due to other CVD (not significant), 13 and 21
deaths due to cancer (RR, 0.62; 95% CI, 0.31 to 1.22; P = .15), and 13 and one
deaths due to violence (RR, 13.0; 95% CI, 1.70 to 98.7; P = .002),
respectively. Multiple logistic
regression analysis of treatments in the intervention group did not explain the
15-year excess cardiac mortality.
Well here again you see the researchers in denial
as to why they got unexpected results. Even though they aren’t able to explain
the high cardiac mortality in the treatment group they aren’t able to fathom
the possibility that cholesterol has nothing to do with heart disease.
Let move on to the next study. This
one was called the Finnish mental hospital study [4] which had 600 women aged
49 yrs av. The treatment intervention was cholesterol lowering diet low in
saturated fats and increased polyunsaturated vegetable oils. The control group
had no intervention. This was the only study that looked at dietary
intervention alone. So anyone that understands anything about statistics should
put the most faith in this study. And for all the saturated fat and cholesterol
haters this is the basket you should be putting all your eggs (no pun intended)
into.
Figure 3. The Finnish
Mental Hospital study methodology and results
Here is a description of the study
conditions and methodology straight from the research article:
A
controlled intervention trial, with the purpose of testing the hypothesis that
the incidence of coronary heart disease (CHD) could be decreased by the use of
a serum-cholesterol-lowering (SCL) diet, was carried out in two mental
hospitals near Helsinki in 1959-71. The subjects were hospitalized middle-aged
women. One of the hospitals received the SCL diet, ie a diet low in saturated
fats and cholesterol and relatively high in poly-unsaturated fats, while the
other served as the control with a normal hospital diet. Six years later the
diets were reversed, and the trial was continued another six years. The use of
the SCL diet was associated with markedly lowered serum cholesterol values. The
incidence of CHD, as measured by the appearance of certain electrocardiographic
patterns and by the occurrence of coronary deaths, was in both hospitals during
the SCL-diet periods lower than during the normal-diet periods.
Please note that the SCL diet resulted
in markedly lowered serum cholesterol values during both phases of the study,
so there is no ambiguity about the cholesterol lowering capability of the intervention
diet. So now you can’t wait to see the conclusions of this study. Well I won’t
keep you holding your breath too much longer, so here are the conclusions,
again, straight from the paper:
The
differences, however, failed to reach statistical significance. An examination
of a number of potential confounding variables indicated that the changes in
them were small and failed to account for the reduction in the incidence of
CHD. Although the results of this trial do not permit firm conclusions, they support
the idea that also among female populations the SCL diet exerts a preventive
effect on CHD.
Now wait a minute. I am jumping up and
down! How did we come to this conclusion? Where did I see that this diet exerts
a preventive effect on CHD when 50% ([10-5]/10) more people died in the intervention
group of all causes, and 75% ([4-1]/4) more died of cardiac deaths compared to
the untreated control group? Am I missing something here? Can someone please
explain? I am dumbfounded, or maybe I am
just plain dumb!
The next one was a better controlled
study because the total fat intake was kept constant between the two groups but
the composition of fat in terms of saturated, poly-unsaturated,
mono-unsaturated and total cholesterol intake percentages were varied. This was
the Minnesota Coronary study [5]. The test design and administration, straight
from the paper states:
The
Minnesota Coronary Survey was a 4.5-year, open enrollment, single end-time
double-blind, randomized clinical trial that was conducted in six Minnesota
state mental hospitals and one nursing home. It involved 4393 institutionalized
men and 4664 institutionalized women. The trial compared the effects of a 39%
fat control diet (18% saturated fat, 5% polyunsaturated fat, 16%
monounsaturated fat, 446 mg dietary cholesterol per day) with a 38% fat
treatment diet (9% saturated fat, 15% polyunsaturated fat, 14% monounsaturated
fat, 166 mg dietary cholesterol per day) on serum cholesterol levels and the
incidence of myocardial infarctions, sudden deaths, and all-cause mortality.
The mean duration of time on the diets was 384 days, with 1568 subjects
consuming the diet for over 2 years. The mean serum cholesterol level in the
pre-admission period was 207 mg/dl, falling to 175 mg/dl in the treatment group
and 203 mg/dl in the control group.
So this intervention dropped the total cholesterol level for the
intervention group to 175 mg/dl and didn’t change it for the control group (203
mg/dl). All good right? But when you look at the summary of results in figure 4
(below) it seems like coronary deaths were about 13% higher and deaths from
cancer were 25% higher in the intervention group. All cause deaths and coronary
events were insignificantly different. But the results quoted from the paper
state:
For
the entire study population, no differences between the treatment and control
groups were observed for cardiovascular events, cardiovascular deaths, or total
mortality. A favorable trend for all these end-points occurred in some younger age
groups.
Now what does the term “A favorable trend occurred in some younger age
groups” mean? Does it mean that younger people will benefit from the
cholesterol lowering diet? And does it mean that by the time they get old this
diet wouldn’t save them the first heart attack because they are not young
anymore? I don’t know what it means but the results speak for themselves and no
one in their right minds could argue with such glaring data.
Figure 4. The Minnesota
Coronary study methodology and results
Now the next one is a study with a dietary intervention of changing
total fat intake to below 20% of total calorie intake (from 32% or more before
the intervention) and to include at least 5 servings of vegetables and fruits
and 6 servings of grains. This would be an ideal textbook diet for the
healthiest of them all. I am mean, vegetables and fruits and grains and fat
less than 20% of total calories! Who can argue with the CHD preventive benefits
of such a diet? No changes were made in the control group diet. This was the
Women’s Health Initiative (WHI) low-fat study [6]. Figure 5 shows the test
summary snapshot below.
Here again the test description states that the objective of this study
was:
To test the hypothesis that a dietary intervention,
intended to be low in fat and high in vegetables, fruits, and grains to reduce
cancer, would reduce CVD risk.
And the test itself was:
Randomized controlled trial of 48,835
postmenopausal women aged 50 to 79 years, of diverse backgrounds and
ethnicities, who participated in the Women's Health Initiative Dietary
Modification Trial. Women were randomly assigned to an intervention (19,541
[40%]) or comparison group (29,294 [60%]) in a free-living setting. Study
enrollment occurred between 1993 and 1998 in 40 US clinical centers; mean
follow-up in this analysis was 8.1 years.
I love this study for three reasons. One that the
sample size was huge and the participants were from a diverse background so the
results would be applicable to the general population. Second, the diet was an
ideal text-book diet that we have grown-up believing was beneficial to heart
health and third that the mean follow-up period was pretty long (8.1 years).
Looking at the summary of results in figure 5 we can see that the risk
reduction for heart disease was insignificant even with this textbook perfect
diet. Quoting the authors straight from the paper:
Over a mean of 8.1
years, a dietary intervention that reduced total fat intake and increased intakes
of vegetables, fruits, and grains did not significantly reduce the risk of CHD,
stroke, or CVD in postmenopausal women and achieved only modest effects on CVD
risk factors, suggesting that more focused diet and lifestyle interventions may
be needed to improve risk factors and reduce CVD risk.
Figure 5. WHI low-fat study
test methodology and results.
So it
seems like the textbook perfect low-fat diet with fruits and vegetables and
grains didn’t do squat in reducing the risk for heart disease. The authors
still had a back-alleyway for explaining away the unfavorable results by saying
that the lifestyle intervention wasn’t focused enough and there was more room
for improvement. Seems like ad-hoc-hypothesis land to me!
There
are many more studies that failed to show a direct correlation between fat or
cholesterol intake and the risk of developing CHD no matter how hard the
cholesterol haters have tried. I have yet to find a single one that succeeded to
show a correlation. And if someone can find that one elusive study that we all
have chased for the last half a century then please let me know. I am sure the
researchers would also love to hear about it. By now it should be fairly clear
that diet has nothing to do with your risk for developing CHD. So eat what you
want and you will not develop heart disease just because of a certain type of
food in your diet, be it saturated fat or cholesterol. And in fact I have reasons
to believe that a low-fat high carb diet will pre-dispose you to developing CHD
not directly but by impairing your body’s insulin response and making you type
II diabetic, because the correlation between a high carb diet and developing type
II diabetes has been shown in studies after studies. And studies have also shown
that having type II diabetes predisposes you to developing CHD. Now that data
is un-ambiguous and incontrovertible. Smoking and stress are the two other
significant risk factors for developing CHD.
So
eat all the saturated fat and cholesterol you like and you won’t develop heart
disease just because of that. There are of course merits to eating a diet rich
in vegetables and fruits (for preventing cancer, type II diabetes etc) but not
for reasons of preventing heart disease. Heart disease is a completely different
animal that is caused due to reasons other than diet. In fact in the coming
series of articles I will even show that a lower lipid cholesterol increases
your risk of dying of other causes after a certain age (47 yrs to be precise)
than having a higher level of cholesterol in your blood.
Why, then you ask, is cholesterol and
fat made out to be such a toxin if there are no studies that show a correlation
between lipid cholesterol levels and CHD? For that you have to understand the
relation between the expert opinion leaders in the medical community and the
pharmaceutical industry that sell the cholesterol lowering drugs and the
American Heart Association (AHA) that makes hundreds of millions of dollars
endorsing low-fat products with the famous AHA logo shown below.
But this is a topic for another
article. For now I will just say that cholesterol hypothesis is based on pseudoscience
or consensus science or what I like to call as ‘puppet science’. Just as
puppets can seem to defy the laws of physics because of the attached strings, cholesterol
science can be made to defy the laws of real science because of the attached ‘strings’
that make us ‘see’ something that doesn’t exist. Noble laureate Richard Feynmann
had an interesting take on pseudoscience that he called Cargo Cult science (http://www.physics.brocku.ca/etc/cargo_cult_science.php).
More to come, but in the mean time
enjoy that piece of steak that you have been avoiding because your doctor told
you so. And if your doctor has a problem with it please direct him to this blogJ
References:
1)
Ornish D, Brown SE, Scherwitz LW, Billings JH, Armstrong WT, Ports TA,
McLanahan SM, Kirkeeide RL, Brand RJ, Gould KL. Can lifestyle changes reverse coronary heart disease? The Lifestyle
Heart Trial. Lancet 1990;336:129-33
2)
Multiple risk factor intervention trial. Risk
factor changes and mortality results. Multiple Risk Factor Intervention Trial Research
Group. JAMA 1982;248:1465-77
3)
Strandberg TE, Salomaa VV, Naukkarinen VA, Vanhanen HT, Sarna SJ, Miettinen
TA. Long-term mortality after 5-year
multifactorial primary prevention of cardiovascular diseases in middle-aged men.
JAMA 1991;266:1225-9
4)
Miettinen M, Turpeinen O, Karvonen MJ, Pekkarinen M, Paavilainen E,
Elosuo R. Dietary prevention of coronary
heart disease in women: the Finnish mental hospital study. Int J Epidemiol
1983;12:17-25
5)
Frantz ID Jr, Dawson EA, Ashman PL, Gatewood LC, Bartsch GE, Kuba K,
Brewer ER. Test of effect of lipid
lowering by diet on cardiovascular risk. The Minnesota Coronary Survey.
Arteriosclerosis 1989;9:129-35
6) Howard BV, Van Horn L, Hsia J, Manson JE, Stefanick
ML, Wassertheil-Smoller S, Kuller LH, LaCroix AZ, Langer RD, Lasser NL, Lewis
CE, Limacher MC, Margolis KL, Mysiw WJ, Ockene JK, Parker LM, Perri MG,
Phillips L, Prentice RL, Robbins J, Rossouw JE, Sarto. Low-fat dietary pattern and risk of cardiovascular disease: the Women's
Health Initiative Randomized Controlled Dietary Modification Trial. JAMA
2006 Feb 8;295:655-66