Risk factors, causes, and the diet-lipid hypothesis

A conversation with a reader about medicine's Ptolemaic epicycles

Share with your friends


I shared with a reader an editorial I co-wrote in 2010 entitled “Risk-Factor Medicine: An Industry Out of Control?” Subsequently we had the following e-mail exchange, which I thought might be of interest to other readers of Alert and Oriented.  I was impressed by Robert’s comments and learned a few things from him and from the links he provided.

On April 29, 2016, Robert wrote:

Hi Michel

After some thought I have been able to formulate my view on risk factor medicine.

The main problem is that “risk factor” it is an illegitimate category from a standpoint of metaphysics and human action.

Risk factor is an illegitimate category because it groups together causes and effects.  Only causality is guide to human action.  But risk factors are acted on as if they were causes.

It may be useful to measure certain markers which are effects as an indicator of the progress or regress but it does not make any sense to treat effects directly.    If a patient reverse their disease, e.g. by stopping smoking or otherwise working in the direction of causality, and as a consequence of this, certain effects, e.g. lung function improve, then the effects can be used to measure the progress achieved by causality.

The problem with risk factor medicine is that they have created a fuzzy category not based on cause and effect, which deliberately obscures causality.  The next step was to reintroduce causality through the back door by by using the term “risk factor” as a guide to action as if it were a cause.  In terms of how medicine is practiced, you could substitute “cause” for “risk factor” and the actions of physicians would be explained by that.    The best example of this is LDL which is an effect of CVD, not a cause, yet there is a vast industry of people trying to lower their LDL directly, which would only make sense if LDL were a cause.

There is an even more cynical layer to this, where drug companies test compounds, searching for something measurable that is changed by the action of their compound on humans.  When they find this, they manipulate the process to have that measurable number declared to be a risk factor, which then must be treated as if it were a cause.


My reply

Hi Robert,

I generally agree with you, although in the specific example you mention I am not sure what you mean by LDL is the effect of CVD.

The devaluation of the concept of cause pervades all the modern errors!  And in biology (and even more so in medicine and human affairs), even strict attention to efficient causality is not sufficient.  Organisms are goal directed, therefore we need to resurrect the notions of final and formal causality…

Hope you’re well.  When I have time, I will try to organize a little group of like-minded people interested in health and healthcare.

Stay in touch and have a good weekend,


Robert replied

I am not sure where we are getting lost on LDL but so I will say something possibly related.

By CVD I mean a degeneration of the artery wall in which scar tissue and plaques build up.     The question is whether LDL particles in the blood cause the scarring process or whether LDL is part of the body’s response to the inflammation that is the actual cause.   If the latter, then removing LLD by preventing the body from synthesizing is not operating in the direction of causality.

High LDL is not a disease, it is a marker, though I wonder in medicine if the two have been conflated?   Is having a high risk factor itself now considered indistinguishable from a disease? I did have a doctor tell me that high cholesterol is a disease, so I suppose some doctors must think that.

This does make me think of another point.  The concept of “diagnosis” has been similarly divorced from causality.  A diagnosis should contain information that is not present in the description of the problem, even if it is a proximate cause rather than an ultimate cause.

I deal with this all the time in diagnosing operational problems with computer systems.  “The system is slow ” is not a diagnosis.  It is a description of the problem.    “My car won’t start” => diagnosis is “your car won’t start”.   You get the idea.

If the patient says, “Doctor I feel tired all the time”, and the diagnosis is “You are not getting enough sleep” that is a valid diagnosis because it moves one step up the chain of causality.  Perhaps in that case, the ultimate cause has not been reached, which might be too much stress at work, or drinking alcohol late at night.  But the diagnosis is still moving in the right direction, toward causation, which will guide human action.

A lot of medicine now involves measuring a risk factor X, and then providing a diagnosis is the latin translated version of “risk factor X is too high”.  E.g, “doctor my back hurts” => your diagnosis is “idiopathic lumbar-itis”.   Not a diagnosis, only a description of the problem itself translated into medical terminology.

If a risk factor is a cause and the diagnosis is the risk factor then the entire process of medicine has devolved into measuring and treating self-contained markers that have no relationship to anything other than themselves.    You can see how different this is from finding the cause of a problem, which is what I do at work with computer systems.


My response

I understand what you are saying, and yes, it is conceivable that there is one causal process that leads both to an elevation in LDL and to damage of the arteries, and therefore that LDL is simply a marker.

That may be true, but we must bear in mind that there are genetic conditions (defects in the liver LDL receptor) that give rise to extremely high levels of LDL from birth and can lead (in homozygous cases) to arterial disease in childhood.  Such conditions can be treated with liver transplant (and in future, possibly with gene correction).

These cases argue that LDL could be a causative factor in garden variety CVD of adults, especially since there is a mild correlation between baseline LDL levels and risk of future CVD in the general population.

Your reflections on the loose concept of diagnosis are correct and generally well recognized.  There is a wide range of information that a diagnosis can convey, from shorthand triviality to causal explanation.

But remember that the inherent goal in medicine is primarily to help, not to explain, even though one can ultimately help better if one can explain causes.  Much of the language and classification schemes are derived on the basis of their utility in being clinically, rather than scientifically helpful.  The two don’t always overlap for a variety of reasons, not least of which is that the science is very complicated and almost always incomplete or insufficient to explain everything.


Robert’s response

Reflecting on this, I think the point I am trying to make is not that we must understand causality of the illness in order to treat, although in some cases yes.

It is more about acting as there were causality when causality is not clear.   The process could be called “causality laundering” in the same way that money laundering converts the proceeds of crime into seemingly legitimate funds or assets, a weak correlation is turned into a causal relationship.


Robert later on added

On the LDL issue I will send you another paper.

I believe that the case of people with a genetic defect in LDL receptors is a separate case and not applicable to our understanding of common arterial degeneration.

One of the ideas that is out there is that the body generates more LDL for people who already have arterial damage as part of the body’s response to the problem.  I’m not sure if that is correct but it does illustrate how causality is reversed.

In this lecture a researcher talks about the lack of association that disproves causality.  I have run across multiple studies in different countries that have reached the same conclusions: 1) LDL inversely correlated with CVD mortality 2) LDL inversely correlated with total mortality, the effect being stronger in women  and with increasing age.

The origin of the heart disease theory came through an anatomist named Virchnow who dissected bodies and found a fatty substance in arteries.  He theorized that people must eat fat, small particles of fat adhere to the arterial walls and build up.  While this theory is false in all of its particulars, it has survived through a long process of “rebuilding the plane while it is in flight”.  As the wing falls off they have added a new wing, the tailfin falls off they add a new tail fin, and now after 100 years we have an entirely new plane in which every piece of it has been replaced yet it still looks like a plane.   This is a prime example of Ptolemaic epicycles in practice.


And in a separate email that included links to some papers, he also wrote

Here are my comments on the papers after reading them:

  • Plaque buildup in arteries is on the far side of the arterial wall away from the area where the blood containing LDL is flowing – how does the LDL deposit itself on the far side of the arterial wall?  Hmmm….
  • Pockets of plaque build up until they burst through the interior of the arterial wall, that is how caps form – not from deposits on the interior
  • If plaques were due to LDL deposits on the arterial wall due to something in the blood, they would appear throughout the entire artery, however plaques mostly form on the central part of the coronary artery
  • Contrary to what most researchers think the aterial wall is not a single layer thick uniformly (a lot of the theories about how LDL causes plaques rely on the single-cell thickness)
  • Plaques only build up in areas where the arterial wall is thicker than a single cell – again contradicting the range of theories (including the LDL-P osmotic flux) suggesting that small particles cause tears in the single-celled wall



[That last paper] points out various logical problems such as

“Therefore, the fact that lowering LDL levels does not prevent cardiac events in 60-70% of  individuals at risk [23] contradicts the causative role of LDL.  It is now known that HMG-CoA reductase inhibitors, cholesterol-lowering drugs known as “statins”, are almost 100% effective in populations with high LDL-C levels, but normalizing LDL levels only reduces the risk of cardio-vascular diseases in this group by approximately 30-40% [23,35-38], and the total number of coronary interventions (bypass and stenting operations) has increased significantly [39]. However, individuals with normal LDL-C levels suffer from coronary atherosclerosis, and although at lower risk, this includes vegetarians [40]. Numerous studies have demonstrated that coronary atherosclerosis affects all eutherian animals with a body mass comparable to or larger than humans, regardless of diet specialization and LDL levels [41-45]. Surprisingly, in these mammals, lipid accumulations in arterial walls were more common in herbivores than carnivores [43,46]. The lack of association between total or LDL cholesterol and degree of atherosclerosis in unselected individuals was demonstrated by a study during the 1930s [47] and has since been noted by many others”

To which I replied

Hi Robert,

I think the points made here are all reasonable challenges to the orthodox view.  Atherosclerosis is an enigma.  We must also bear in mind that it is not necessarily just one entity.  For example, the plaques that develop in some heavy smokers are not necessarily of the same composition and may not behave in the same way as those who develop in people with very high LDL-C.  We lump a lot of people under the label atherosclerosis, when in fact we may be dealing with a variety of conditions.  Very complex.


Robert then replied

I think that I forgot to send the link to the lecture with the epidemiological data showing that higher LDL is associated with longevity.  I will find that link.

I have read a ton about this issue.  I don’t know if you have read Gary Taubes’ book ?  The politics of it are quite interesting.  It is a prime case of the politicization of science, big government science, and research fraud.  The establishment of statin drugs as the primary means of treating this is afflicted by the same issues.

I don’t know if you follow this but the dietary fat axis of the diet-lipid-heart hypothesis has been rapidly collapsing over the past few years as well.  Saturated fats are increasingly reported as being beneficial, not causing elevated lipid levels in the blood.  The national health body of Sweden now recommends a high-fat diet, the popularity of butter and lard has been increasing, and at least a number of doctors now advocate high fat diets featuring saturated animal fats.   I also pay some attention to the the adoption of high fat diets among endurance athletes, which can be described as a small trend but arising out of nothing even five years ago.  Marathoners, ironman, and ultra runners are in some small numbers switching away from carb fueling to fat fueling strategies.

Anyway, after reading maybe two dozen books I have reached my own conclusion that the cause of the process of arterial degeneration is at present unknown, and I would agree with you that we may not be talking about all of one thing.


He then added

Here is the lecture citing the positive correlation between LDL and longevity.
There is not only one study showing this – I have run across multiple studies
This is the first I have heard of an inverse correlation between LDL and cardio mortality

As the speaker notes, association does not prove causality but lack of association disproves causality

My response

I know Virchow well.  He is a very well-known founder of cell biology and pathology and, incidentally, not a friend of liberty.  He was a staunch progressivist who had a famous quip: “Politics is medicine writ large”


And separately I commented

Taubes’ book is great.  Especially the first part that outlines the history of dietary recommendations.  I thought he was a little too keen about promoting his own hypothesis in the second part of the book, but interesting nevertheless.

Fifteen years ago, I was doing research on atherosclerosis in a basic science lab (my research was published in prestigious journals, but I knew at the time that it contributed nothing to our understanding…)  I had the chance to meet a very influential proponent of the diet-lipid hypothesis, Scott Grundy (I think he is mentioned in Taubes’ book).

At the time I was already very skeptical of the [lipid-diet] theory and I had an alternative theory that certain dietary fats are in fact protective.  I asked him what his explanation for the French paradox was (the observation that the French have very low rates of CAD despite a fairly fat rich diet) and he dismissed it, essentially saying that it did not exist…

I had a hypothesis that it was the ferments and microorganisms in the cheeses the French eat abundantly that were protective.  I also thought that the hypothesis would also apply to the Kenyan Masai paradox (these nomadic tribesmen were famously described in the 1960’s and 1970’s for having very low rates of atherosclerosis despite a very high fat diet—6% fat raw milk).  I even got in touch with George V. Mann, who had been one of the main researchers investigating this but has since passed away.  He was groping toward a similar hypothesis but could never prove it.  It got me into looking at the “clinical nutrition” literature and found it to be one of the most acrimonious in medicine!


Robert replied

I believe the opposite is true – medicine is politics writ large

Gary Taubes book confirms this – how much of widely medical science was established by politics


As many people have noted on this topic, when your data is full of “paradoxes” at some point you have to question the theory itself.

The explanations that are proposed for each of these paradoxes are Ptolemaic epicycles.

I don’t know if you follow astrophysics – they have introduced the concept of “dark matter” which is a plug factor to account for a discrepancy  between theory and experiment of 10-fold.

PS:  My editorial on risk-factor medicine is behind a paywall but I may give here a rendition of its main point in the next few days.

[Related post: A coming epidemic of normality]

Leave a Reply

Your email address will not be published. Required fields are marked *

4 thoughts on “Risk factors, causes, and the diet-lipid hypothesis

  1. Great discussion. I have heard of these things, although in somewhat broader strokes, having done a relatively small about of studying about it in nursing school and on my own. I decided I agree with the healthy fats approach. Sometimes it seems people make basic things about eating far more complicated than they need to be!

  2. For those interested in getting a copy of Dr. Accad’s editorial in the Journal of Cardiology, just put the title into Google Scholar and retrieve the PDF from academia.edu.

  3. Interesting discussion that brings up more basic questions like:

    1. whats the value of the ‘diagnosis’ in which situation? I find it often not too helpful and for lay people a source of unnecessary worry or identification. a diagnosis in oncology has a totally different use than for example in psychiatry.
    The regulators have used this label as a discrete value for controlling and reimbursement whis is a disaster, even worse for disability insurance.

    2. causality: isn’t it the core of the ‘machine-concept’? causality in chronic disorders is usually multifactorial, incl. psychological and lifestyle factors. the more chronic, the more complex. who is supposed to figure all that out, your MD?

    3. nutrition: how far away are we actually from a natural diet/lifestyle as we are made for?

    4. isn’t it the oxidized ldl – as a result of inflammation- that is harmful? and thereby a good marker for inflammation, as a result of the western style diet?