Last week I was going through some old files, and presentations, in a vague effort to clean up my computer. Whilst looking a one of many thousands of studies I had filed away I came across this paper: ‘Clarifying the direct relation between total cholesterol levels and death from coronary heart disease in older persons1.’
I read it, and immediately recalled why I kept it. For it came to the following, final, conclusion:
‘Elevated total cholesterol level is a risk factor for death from coronary heart disease in older adults.’
I remember when I first read this paper a few years ago. My initial thought was to doubt that it could be true. Most of the evidence I had seen strongly suggested that, in the elderly, a high cholesterol level was actually protective against Coronary Heart Disease (CHD).
However, when a bunch of investigators state unequivocally that elevated cholesterol is a risk factor for heart disease, I try to give them the benefit of the doubt. So I read the damned thing. Always a potentially dangerous waste of precious brainpower.
Now, I am not going to dissect all the data in detail here, but one sentence that jumped out of the paper was the following:
‘Persons (Over 65) with the lowest total cholesterol levels ≤4.15 mmol/L had the highest rate of death from coronary heart disease, whereas those with elevated total cholesterol levels ≥ or = 6.20 mmol/L seemed to have a lower risk for death from coronary heart disease. ‘
Now, I can hardly blame you if you struggled to fit those two quotes together. On one hand, the conclusion of the paper was that .. ‘Elevated total cholesterol level is a risk factor for death from coronary heart disease in older adults.’ On the other hand, the authors reported that those with the lowest total cholesterol levels had the highest rate of CHD; whilst those with the highest cholesterol levels had the lowest rate of CHD.
Taken at face value, this paper seems to be contradicting itself…. utterly. However, the key word here, as you may have already noted, is seemed. As in… those with elevated total cholesterol levels ≥ or = 6.20 mmol/L seemed to have a lower risk for death from coronary heart disease. ‘
Now you may think that this is a strange word to use in a scientific paper. Surely those with elevated total cholesterol levels either did, or did not, have a lower risk of death from CHD? Dying is not really something you can fake, and once a cause of death has been recorded it cannot be changed at a later date. So how can someone seem to die of something – yet not die of it?
The answer is that you take the bare statistics, then you stretch them and bend them until you get the answer you want. Firstly, you adjust your figures for established risk factors for coronary heart disease – which may be justified (or may not be). Then you adjust for markers of poor health – which most certainly is not justified – as you have no idea if you are looking at cause, effect, or association.
Then, when this doesn’t provide the answer you want, you exclude a whole bunch of deaths, for reasons that are complete nonsense. I quote:
‘After adjustment for established risk factors for coronary heart disease and markers of poor health and exclusion of 44 deaths from coronary heart disease that occurred within the first year, (my bold text) elevated total cholesterol levels predicted increased risk for death from coronary heart disease, and the risk for death from coronary heart disease decreased as cholesterol levels decreased.’
Why did they exclude 44 deaths within the first year? Well, they decided that having a low cholesterol levels was a marker for poor health, and so it was the poor health that killed them within the first year.
The reason why they believed they could do this is that, a number of years ago, a man called Iribarren decreed that the raised mortality always seen in those with low cholesterol levels is because people with low cholesterol have underlying diseases. And it is these underlying diseases that kill them. (What, even dying from CHD. And how, exactly does CHD cause a low cholesterol levels….one might ask).
In truth, there has never been a scrap of evidence to support Iribarren’s made-up ad-hoc hypothesis. [A bottle of champagne for anyone who can find any evidence]. However, it is now so widely believed to be true, that no-one questions it.
Anyway, without chasing down too many completely made-up ad-hoc hypotheses, the bottom line is that this paper stands a perfect example of how you can take a result you don’t like and turn it through one hundred and eighty degrees. At which point you have a conclusion that you do like.
Young researcher: (Bright and innocent) ‘Look, this is really interesting, elderly people with low cholesterol levels are at greater risk of dying of heart disease.’
Professor: (Smoothly threatening) ‘I think you will find…. if you were to look more carefully, that this is not what you actually found….. Is it? By the way, how is your latest grant application going?’
Young researcher: (Flushing red at realising his blunder) ‘Yes, by golly, how silly of me. I think I really found that elderly people with high cholesterol levels are at a greater risk of dying of heart disease.’
Professor: ‘Yes, excellent. Be a good lad, find a good statistician to make sure the figures make sense, and write it up.’
For those who wonder at my almost absolute cynicism with regard to the current state of Evidence Based Medicine, I offer this paper as a further example of the way that facts are beaten into submission until they fit with current medical scientific dogma.
As a final sign off I would advise that any paper that has the word ‘clarifying’ in its title, should be treated with the utmost suspicion. I think George Orwell would know exactly what the word clarifying means in this context. Facts do not need clarification.
1: Corti MC et al: Clarifying the direct relation between total cholesterol levels and death from coronary heart disease in older persons. Ann Intern Med. 1997 May 15;126(10):753-60