Tag Archives: Karl Popper

You are a very black swan indeed

Here is Wikipedia on the wisdom of Karl Popper:

“The classical view of the philosophy of science is that it is the goal of science to prove hypotheses like “All swans are white” or to induce them from observational data. Popper argued that this would require the inference of a general rule from a number of individual cases, which is inadmissible in deductive logic. However, if one finds one single black swan, deductive logic admits the conclusion that the statement that all swans are white is false. Falsificationism thus strives for questioning, for falsification, of hypotheses instead of proving them.”

Sorry, although I am a great fan of Popper, his language is a bit, well, pedantic.  What he is saying here is that science starts with a hypothesis e.g. ‘all swans are white.’ If you find white swan after white swan, then you will mildly strengthen the hypothesis. However, once you find one black swan, the hypothesis is dead.  (Unless you decree that, as all swans are white, a black swan cannot – by definition – be a swan).

One of the great white swans of cardiology is that Familial Hypercholesterolaemia (FH) causes heart disease. When I give talks to other doctors informing them that the cholesterol hypothesis is bunk, one of the ‘facts’ that is triumphantly used to knock me down is that ‘People with Familial Hypercholesterolaemia die very young from heart disease.’ Case proven, raised cholesterol causes heart disease, now move on.

A little bit of context is needed here. Familial Hypercholesterolaemia (FH) affects about one in five hundred people. It is a genetic condition where those who have it (I refuse to use the words suffer from it) have very high Low Density Lipoprotein (LDL) levels. As everyone knows LDL is known as ‘bad’ cholesterol, which is considered to be the number one risk factor for heart disease. (Of course LDL is not cholesterol at all but, hey, why let scientific accuracy get in the way of…. Well, science).

Goldstein and Brown established that the cause of FH is a lack of LDL receptors on cells. LDL receptors are the things that bind on to LDL molecules and then remove them from the bloodstream. Cells manufacture LDL receptors when they are low on cholesterol and need more.

Once the receptor is made it is pushed out through the cell membrane to attract an LDL molecule. When an LDL molecule has been caught, by binding on to the receptor, the LDL and the attached receptor are pulled back into the cell and broken down. Because they function this way, LDL receptors only work once. If a cell wants more cholesterol, then it needs to manufacture more LDL receptors.

Clearly, if there are not enough LDL receptors being manufactured, the entry of LDL into cells is restricted. This means that blood levels of LDL rise, and you will be diagnosed with Familial Hypercholesterolemia. If, that is, you have a blood test.

Moving sideways for a moment I need to mention that most people with FH are heterozygotic, by which I mean they carry one gene for FH. Their LDL levels are therefore about double that of the surrounding population. However, those with homozygotic FH (carrying both genes) have LDL levels that can be twenty times ‘normal’.  More on this group in a later blog.

Back to FH. If you have a heart attack when young, by which I mean under about 55, and you have FH, doctors will nod sagely and that that ‘it was the FH that did it.’ (If you don’t have FH, they will say it was something else that did it).

Sometimes the relatives of those dying young of CHD with FH, are contacted. It is often found that there is a higher rate of FH and premature CHD in relatives. This type of evidence has been used as proof that FH causes CHD. Maybe. Maybe not. If someone dies young from CHD, and has FH, and other relatives have a higher rate of both FH and CHD, what have we actually proved? We have proved nothing – for certain.

All we have done is to establish that relatives of people with FH and premature CHD also have FH (as they must, as it is a dominant gene) and also have a higher rate of CHD. Now, it could be that the FH is the reason for their CHD. Or it might not. It could be that something else, genetic of behavioural, is causing their high rate of CHD.

Given this problem of inherent bias, how could you tell if the FH is causing the CHD or not? Or, better, how could you falsify the hypothesis that FH causes CHD? Can you find a black swan?

Well the best way to find a black swan in this ara is to turn your study inside out. Instead of looking at people with FH and premature CHD, then looking at their relatives to find FH and CHD, you need look for premature CHD first (knowing nothing of FH status), then see if FH is more prevalent in first degree relatives.

How do you do this? Well, firstly you ask hundreds thousands of students if their father had a heart attack, or died of heart attack before the age of 55. Then you measure the LDL level of those students to see if they have FH. At the same time you find age and sex matched control students to see if they have FH.

Now, if FH really were a major cause of premature CHD you would expect to find that FH was far, far, more prevalent amongst those students whose fathers suffered CHD before the age of 55.

Such a study was done once, in the Netherlands. The results were as follows:

In the EARS (European Atherosclerosis Research Society) studies, University students whose fathers had proven CHD before the age of 55 years, were recruited …Age and sex-matched controls were recruited from the same populations for each case.

  • 2 of 1089 students with family history of CHD had FH
  • 4 of 1727 controls had FH

Thus, the prevalence of FH in both groups was not significantly different at approximately 1 in 500, which is the estimated prevalence of the condition in the general population. The evidence that heterozygote FH is, of itself, a cause of atherosclerosis is unsatisfactory.

http://www.bmj.com/content/322/7293/1019/reply

In short, when someone finally did a study on the association between premature CHD and FH, where selection bias was removed, they found that FH was no more common in those with, and without, a strong family history of CHD. This was the blackest of black swans.

However, there is a twist to this tale. Which is that this study was never published anywhere. The only reason that I know about it, is that I was reading the responses section in the British Medical Journal, and the lead investigator of the study wrote the above letter in reply to a discussion on FH. (Something more than easily missed).

I contacted him, and asked why the study had not been published. He did not provide any answer that made any sense to me. The end result of the lack of publication is that the blackest of black swans is not actually black. It is invisible. Until now, of course.

P.S. As for homozygotic FH, I shall deal with this later.