But now to talk statistics……run for cover, save the children.

‘Pitt concedes that the study was not adequately powered to observe differences in this end point (overall mortality).’

Now I am not a statistician – my brain fuses when confronted with mathematical formulae. But I do clearly understand the overall concepts. And I shall try to explain why the comment that the study was ‘not adequately powered to observe differences in this end-point,’ is bonkers. And can only be accepted if you are the sort of person who thinks that the Bill Clinton defence ‘But what is, is?’ sounds reasonable.

Almost all clinical trials need to enrol a sufficient number of people, over a sufficient time period, to show a significant difference between the ‘intervention’ group, and the ‘standard’ group. The greater the difference you expect to see, the less people you need to enrol. For example, if you expect that everyone who takes the drug will live, and everyone who doesn’t, will die, then it doesn’t take that many people to demonstrate a significant difference.

In my own mind I liken this to throwing a die to find out if it is ‘weighted’ or not. If the die is heavily weighted it will come up with a six almost every time, and it doesn’t take long to work out that you have a biased die on your hands. However, if the die is only very slightly weighted, it may take hundreds of throws before you can be sure that it is coming up six more times than is possible by chance.

Statisticians basically try to work out – before the trial starts – how many people you need to treat, for how long, to see if the die is weighted. Or if it is not. Or by how much the die is weighted. So you need to estimate, before the trial starts, what difference you think you are likely to find, then you can work backwards from this to establish how many patients you need for the trial to be properly ‘powered’ (with a bit of a buffer built in).

If the trial is insufficiently ‘powered’ i.e. doesn’t have enough people in it, you may see a difference, but it will not be statistically significant i.e. the difference could have been a ‘chance’ finding.

Still awake? Good.

TNT statin study designed to avoid the conclusion that statins do not lower total mortality or claim that it was due to random chance

Here is where we hit the ultimate Catch 22. The almost perfectly bonkers loop of self-referential logic in the TNT trial.

Step One – of mad logic:

Statins have been shown to reduce cardiovascular mortality in primary prevention studies (that is, in people who have not already got established heart disease). So a statistician, knowing this can say we need to look at (in the case of the TNT study) ten thousand people, with five thousand in one ‘arm’ of the study and five thousand in the other ‘arm’ over about five years or so. That is, if we expect to see around a 20% relative reduction in deaths from cardiovascular disease.

Step Two – of mad logic:

Because statins have not been shown to reduce overall mortality by much, if at all, we need a far greater number of patients in the trial to ‘power’ it, in order to see a significant reduction in overall mortality. In the case of TNT, at least 34,000 patients - apparently.

Step three – of mad logic:

When the study shows a reduction in cardiovascular mortality this is claimed to be statistically significant.

Step four – of mad logic:

When the study shows no difference in overall mortality, it can be confidently stated (without being struck down by a bolt of lightening from above) that the study was not powered to look at overall mortality, and therefore the finding could have been ‘chance’ and is not significant.

TNT statin study claims increase in total deaths is insignificant, but a decrease in cardiovascular deaths is significant

This is despite the fact that, in the TNT study, more people died of non-cardiovascular disease than cardiovascular disease, and that more people, total, died in the statin ‘treatment’ arm. This, we are told, is chance. Whereas the much smaller overall reduction in cardiovascular mortality is statistically significant.

And lo, the greater number of deaths is insignificant. Whilst the lesser number of deaths is significant. And the statistician looked upon his work and he was pleased. And because no one understands statistics, and they are terrified of making a statement that may make them seem foolish, we accept this complete rubbish without question.

Rubbish From Statin Supporters

Of course, accepting this rubbish does allow various doctors to make statements such as the ones that follow:

‘This study provides direct evidence of benefit of intensive treatment in stable coronary heart disease patients, a population that numbers about 30 million in the US. That’s huge.’ Dr Christopher Cannon (Brigham and Women’s Hospital, Boston, MA)

He went on to add: ‘To be able to prevent major cardiovascular events on this scale will have a huge public-health implication.’

Or how about a little more from John La Rosa (State University of New York Health Science Center, Brooklyn).

‘The results of the TNT study indicate that the linear relationship between reduced LDL and reduced CHD risk demonstrated in prior secondary-prevention trials holds true even at very low LDL cholesterol levels. These data extend the growing body of evidence that lower LDL cholesterol levels well below the currently recommended targets can further reduce the healthcare burden associated with cardiovascular and cerebrovascular disability.’

And they can still sleep in their beds at night. Cor blimey – as we say in the UK.

—————

This article was previously published on THINCS.org

This article was previously published on THINCS.org (The International Network of Cholesterol Skeptics).

I republished the article here with Dr. Kendrick’s permission.

Malcolm Kendrick’s Email Address—.(JavaScript must be enabled to view this email address)

Dr. Malcolm Kendrick can be reached at .(JavaScript must be enabled to view this email address).

Malcolm Kendrick, MD is the author of the wonderful, eye-opening, paradigm-shifting book book The Great Cholesterol Con: The Truth About What Really Causes Heart Disease and How to Avoid It .)

.)

Articles on the same subject can be found here:

• Cholesterol drugs

• Cholesterol Hypothesis Skeptics

• Doctor - Malcolm Kendrick, MD author of "The Great Cholesterol Con"

• Statins