17th February 2019
[Adherence to placebo saves lives]
To an extent I am cursing myself for doing what I am about to do. I have been dragged, yet again, into reviewing a paper that has made headlines round the world which proved, yes proved, that adherence to statins saves lives. I am doing this review because a lot of people have asked for my opinion on the paper.
I do feel like saying. ‘Look, I wrote the book Doctoring Data so that you could read papers like this and work out why they are complete nonsense for yourselves’. Clearly, not enough people have read my book, and I would therefore heartily encourage another million or so people to do so. [Conflict of Interest statement – I will get lots of money if this happens, which I think of as “win, win”].
The paper, in this case was called ‘Association of statin adherence with mortality in patients with atherosclerotic cardiovascular disease.’ It was published in the New England Journal of Medicine (NEJM) a couple of days ago.
The main finding was:
‘Using a national sample of Veterans Affairs patients with ASCVD (atherosclerotic cardiovascular disease), we found that a low adherence to statin therapy was associated with a greater risk of dying. Women, minorities, younger adults, and older adults were less likely to adhere to statins. Our findings underscore the importance of finding methods to improve adherence.’ 1
First thing to say is that this was an observational study. So, it cannot be used to prove causality, especially as the improvement in outcomes that they observed was an increased mortality risk of 1.3 (HR) in those who were least adherent – compared to those who were most adherent.
As many people know… sorry I shall rephrase that… as many geeks like myself know, if the hazard ratio is less than two, in an observational study, the best thing to do with said paper is to crumple it up and throw it in the bin. Because it is almost certainly meaningless. To quote Sir Richard Doll and Richard Peto, two of the fathers of medical research and epidemiology:
“when relative risk lies between 1 and 2 … problems of interpretation may become acute, and it may be extremely difficult to disentangle the various contributions of biased information, confounding of two or more factors, and cause and effect.”2
Observational studies with relative risks between one and two, are the type of studies which find that drinking five cups of coffee protect against CVD – or would that be increase the risk of dying of CVD. Or maybe it is tea, not coffee? [I apologise for mixing up odds ratios, hazard ratios and relative risk. For ease of understanding, think of them as the same thing].
For example, I was looking at this paper:
‘Tea and coffee consumption and cardiovascular morbidity and mortality’.
Where they found that drinking between three and six cups of coffee reduced CV mortality by 45%:
‘A U-shaped association between tea and CHD mortality was observed, with an HR of 0.55 for 3.1 to 6.0 cups per day.’3
That is a far better result than adhering to statins. After all it is a 45% reduction vs. 30% reduction. My advice therefore would be to stop the statins and have nice cup of tea instead. Life would be so much better, and you would live longer as well. Sorry, but I don’t know what sort of tea. English breakfast, Earl Grey, Darjeeling… So many questions. So many stupid studies to read. So much crumpling. So many bins to empty.
Leaving behind the nonsenses they are – the observational studies with a minute difference in hazard ratio – let us move on to the major confounder of this latest crumple, bin, paper. Which is that people who adhere to medications do far better than those who do not – even if that medication is a placebo.
This was first noted, with regard to cholesterol lowering medications, nearly forty years ago in another paper, coincidentally published in the NEJM. It was called:
Influence of adherence to treatment and response of cholesterol on mortality in the coronary drug project.
I have copied the abstract in full. In part because it is written in something akin to understandable English. Most unusual in any medical journal. In this study the researchers were looking at drugs used to lower cholesterol levels, prior to the invasion of the statins.
‘The Coronary Drug Project was carried out to evaluate the efficacy and safety of several lipid-influencing drugs in the long-term treatment of coronary heart disease. Good adherers to clofibrate, i.e., patients who took 80 per cent or more of the protocol prescription during the five-year follow-up period, had a substantially lower five-year mortality than did poor adherers to clofibrate (15.0 vs. 24.6 per cent; P = 0.00011).
However, similar findings were noted in the placebo group, i.e., 15.1 per cent mortality for good adherers and 28.3 per cent for poor adherers (P = 4.7×10-16). These findings and various other analyses of mortality in the clofibrate and placebo groups of the project show the serious difficulty, if not impossibility, of evaluating treatment efficacy in subgroups determined by patient responses (e.g., adherence or cholesterol change) to the treatment protocol after randomization.’ 4
I think it is worth highlighting the main findings again.
Those who adhered to taking clofibrate = 15% mortality
Those who had poor adherence to clofibrate = 24.6% mortality
Those who adhered to taking placebo = 15.1% mortality
Those who had poor adherence to placebo = 28.3% mortality
From this is can be established that it was worse for you to not take placebo regularly than it was to not take clofibrate regularly.
If we move forward in time, others have looked at adherence to taking statins. The first thing they noted was people who take their medication regularly are different in many, many, ways to those who have poor adherence.
The paper is called: ‘Statin adherence and risk of accidents, a cautionary tale.’ Published in the American Heart Association journal Circulation.
As they say in the introduction:
‘Bias in studies of preventive medications can occur when healthier patients are more likely to initiate and adhere to therapy than less healthy patients. We sought evidence of this bias by examining associations between statin exposure and various outcomes that should not be causally affected by statin exposure, such as workplace and motor vehicle accidents.’
As they conclude:
‘Our study contributes compelling evidence that patients who adhere to statins are systematically more health seeking than comparable patients who do not remain adherent. Caution is warranted when interpreting analyses that attribute surprising protective effects to preventive medications.’ 5
This takes us back to Hill and Peto:
“when relative risk lies between 1 and 2 … problems of interpretation may become acute, and it may be extremely difficult to disentangle the various contributions of biased information, confounding of two or more factors, and cause and effect”
In the case of this latest ‘nonsense’ paper on statins, it is not actually difficult to disentangle the various contributions of biased information.
We already know that people who take tablets regularly, and placebo regularly, are more health seeking than those who do not. We already know that if you take a placebo regularly, this almost halves your (absolute) mortality rate. These are both enormous confounders in the latest NEJM study.
In fact, the confounder effect unearthed in previous studies is far larger than the effect they found. Which, if you are going to be ruthlessly logical, would suggest you would be far better off regularly taking a placebo than regularly taking a statin. If you choose to do so, you could entitle their paper “Proof that statins have no beneficial effect”.
You sure as hell cannot use such data to suggest that adhering to statins is beneficial. Yet, the authors of this study have done so. I give their paper a mark of D-Fail, please try again.
Or else, I would say, please inform yourselves of the previous research done in this area before writing a paper. This will avoid wasting everyone’s precious time.
2: Richard Doll & Richard Peto, The Causes of Cancer 1219 (Oxford Univ. Press 1981).