Dr Araz Rawshani

Atherosclerosis, cholesterol and statins — Part 3 — Studies against cholesterol and statins

Contents

Atherosclerosis, cholesterol and statins: theories that speak against LDL cholesterol is dangerous, as well as statins are effective

This is the third part of our series about atherosclerosis, statins and cholesterol. Part 1 discussed atherosclerosis (atherosclerosis fattening/atherosclerosis) and part 2 discussed the importance of LDL cholesterol for atherosclerosis. The two big questions are:

  1. Is LDL cholesterol a cause of atherosclerosis, coronary heart disease, heart attack etc?
  2. Can statins reduce the risk of heart attack, stroke and death?

In Parts 1 and 2, we discussed studies that suggest that LDL cholesterol causes atherosclerosis. These studies included nature’s own genetic experiments (in the form of hereditary diseases leading to high LDL cholesterol), epidemiological studies, natural genetic experiments (where randomly inherited genes lead variation in LDL in the blood; this is called Mendelsk randomisation) and randomized controlled clinical trials. The latter two (Mendelsk randomization and clinical trials) are extremely important because high and low LDL cholesterol in these studies is only the result of chance. In the previous article, we came to the conclusion that studies that suggest that LDL causes coronary heart disease involve over 2 million people who have been followed for about 20 million years (in total) and that LDL cholesterol causes atherosclerosis.

Have you already read Part 1 (atherosclerosis) and Part 2 (LDL cholesterol)?

The meaning of chance: why this talk about randomization?

Random is English for chance. Random is crucial for medical research. If we want to investigate how LDL affects health, we need to study two groups of people, one of which has low LDL and the other has high LDL. It is crucial that high and low LDL is only a result of chance, and that no other factors (such as medicine, disease, weight etc) have affected LDL cholesterol. So, in the ultimate study, the LDL cholesterol of participants is only a result of chance. But how is it done? There are two variants:

  • The first is nature’s own randomization and it is called Mendelian randomization (after Gregor Mendel, who is the father of genetics). Because we humans inherit our genes at random. A mother has two copies of each gene, and so does the father. But the father’s sperm, as well as the mother’s egg, can contain only one copy of each gene and what copy they contain is completely random! That means your genes are a random sample of the variants your mother and father had. Further, for each gene there are different variants. You probably noticed that some have dark hair while others have light hair, straight hair, curly hair. So there are many varieties of all genes. In the same way as one can inherit different hair color and hair types, one can also inherit genes that affect LDL cholesterol. So some may have genetically high LDL cholesterol or genetically low LDL cholesterol and this is completely random (randomized)! That means that some people have high LDL, and others have low LDL, completely random from birth. This gives a unique opportunity to study the importance of LDL cholesterol.
  • The second variant is the clinical trial. The principle of such a is simple. It identifies a number of individuals who are believed to benefit from treatment and then randomly each person to receive either genuine treatment (e.g. statins) or placebo (“sugar pill”). If each individual has the same chance of receiving placebo or drugs, we will get two equal groups that are exactly the same in all respects (hair color, weight, gender distribution, genetics, height, abdominal range, sexual orientation, socio-economic status, etc.) except for one thing: the drug. One group receives medicines and the other is placebo. This means that if there is then a difference in (for example) the risk of a heart attack, that difference can be attributed to the medicine.

Why are we discussing this? Well, only this type of study can tell us whether a factor (for example, LDL cholesterol) causes an event (for example, death or heart attack). If you want to express an opinion on cause and effect (which is called causality), you must have support from this type of study. In fact, randomization is so central to cause and effect that your everyday life is full of randomized trials. If you use Google, Hotmail, own an iPhone or just browse blogs, you’re guaranteed to be included in many randomized studies every day. Most sciences use randomization to draw clear conclusions about cause and effect.

Merging studies is even better

Randomized studies are the best we have because such studies allow us to express our opinion on cause and effect. In fact, randomized studies are happy to be merged because larger studies are even better studies (generally). A study that is a merger of other studies is called meta-analysis. So a meta-analysis is the ultimate, since all studies are included in that analysis.

Why do you need to know all this?

Most of the headlines you read in newspapers, blogs and social media are not based on randomized studies. These studies are therefore not reliable (although they are interesting!) and below you will see examples of this from the cholesterol and state scus.

The Great Pit: Epidemiological Studies (Observatory Studies)

Randomization is crucial for deciding on cause and effect (whether the study is about atherosclerosis, headache or joint pain). The opposite of a randomized study is an epidemiological study (also called an observational study). In such a study, you select a group of individuals, map them and then follow them to see who develops (for example) a heart attack. You can then investigate how LDL cholesterol is linked to the risk of heart attack but in these studies high and low LDL values are not randomly distributed to participants! This means that you can never express your opinion on cause and effect. We can take an example:

Imagine that we are investigating people with diabetes at a health center in Gothenburg. We find 100 people with diabetes and 50 of them use blood thinning, and the other 50 don’t use blood thinning. We decide to follow up on these people to see who is doing best. When the follow-up is over, we note that as many have died in both groups. What does this say about the effect of blood thinning treatment? Answer: absolutely nothing. It could be that blood-thinning treatment has no effect (therefore, as many die). But it could also be that there was a reason why some people had blood thinning – they might have a very high risk of dying, and then blood thinning must be very good because just as many died in both groups… Maybe you realize you can come up with more arguments and theories.

The bottom line is that an observational study can never express an opinion on cause and effect and it is not infrequently that surprising results can be found in such a study.

Doctors and biostatistics do not go together

Biostatistics form the basis for all clinical research. Doctors, who are mostly front figures in medical research, find biostatistics difficult. For the last 10 years, I have met no more than 5 doctors (from graduate students to professors) who can design and analyze a study on my own. We doctors simply do not have that competence and our ignorance is very great, which is disastrous because biostatistics are the basis of the research that we represent. Fortunately, epidemiologists and statisticians are in order to get this in order. If you ask a statistician/epidemiologist what it thinks about doctors’ ability to design, analyze and interpret studies, you will hear a good laugh, unfortunately. You will soon see many examples of how little we doctors really know about causality and interpretation of studies.

Doubtful results in doubtful journals: Problem with Open Access

The publishers that publish medical science have a turnover of billions. Publishing a study in a medical journal has always been exhausting. They submit their results and then wait up to 12 months of discussion with the journal and experts hired by the journal (the experts work free of charge) to review the study. A few years ago, several publishers started with a new concept called “Open Access” and that means that the authors simply pay (between 20 and 30 thousand SEK) to publish their article. This proved to be extremely lucrative and therefore several journals have switched to Open Access only and many researchers prefer this because it is quick and easy to publish. I have done this myself and it is a joy because the study does not encounter any patrol at all, or very fastidious one. I realize at the same time that this means that my articles are not scrutinized as hard (sometimes I wonder if they were reviewed at all). Just recently, I myself reviewed an article for an Open Access (August 2017) and then I was one of 4 experts who would judge a terrible bad job. None of the experts had any opinions about the work but thought it could be published. Nor did I have the power to make any significant contribution. I’m sorry. I’m sorry. Of course, more people have realised that the Open Access journals are deplorable in this respect. One demonstrating this was John Bannon from Harvard who conducted an experiment. He fabricated a whole scientific study. The study was about how to make cancer cells grow slower, and Bannon designed the study terribly poorly, which was deliberate. He made big and clear mistakes in all aspects of the study and he says that “a person with high school competence should have discovered that there were major errors in the study”. Bannon sent the study to 304 Open Access journals. How did it go? The article was accepted for publication in over half of the magazines! Why are we discussing this? The studies that speak against cholesterol and statins are mostly published in just Open Access journals. The question is whether it has affected the quality of the studies.

Now to statins and cholesterol

Let’s look at studies that dispute the prevailing opinion that high levels of LDL cholesterol are harmful, as well as that statins are effective in reducing the risk of heart attack and stroke. We will go through the studies that are often discussed in Sweden. The discussion mainly takes place online at various forums and blogs, although it is sometimes raised in research contexts. There are several cholesterol and statine critics in Sweden. It seems that Dr Uffe Ravnskog and Dr Rolf Sandberg, both docs and doctors, are the most prominent statin and cholesterol critics. They have researched and published articles on both cholesterol and statins and they question the importance of cholesterol (they mean that high cholesterol is healthy) and statins (they mean that statins are only good for those who have already had a heart attack and then the effect is negligible). Both Ravnskog and Sandberg are active in the debate; a quick search on Google provides, among other things, the following statements:

Statins and cholesterol: industry that is languishing?

Those who are critical of the importance of LDL cholesterol and the effect of statins often argue that everything is a scam that pharmaceutical companies staged to make money, as well as that the Academy’s representatives (researchers) are purchased by the pharmaceutical industry. There is a certain truth in this. Statins were once a huge cash cow for the pharmaceutical industry. Today, however, all relevant patents on statins have expired and this means that statins today cost as little as $50 per day (SEK 15 per month). You can examine this yourself directly on the state page www.fass.se (example: search for atorvastatin, then click on “FASS for health professionals” and scroll at the bottom to “Packaging” to see prices). There are certainly pharmaceutical companies that still make money from statins but this market is not nearly as lucrative as before. All Swedish county councils recommend statins where the patent expired (so-called generic statins). These are as cheap as 50 penny per day. It should also be added that the whole cholesterol-lowering drug industry is in decline, which has led pharmaceutical companies to stop investing in cholesterol-lowering drugs. Merck, one of the world’s largest pharmaceutical companies, has recently developed a cholesterol-lowering drug (anacetrapib) that reduced the risk of cardiovascular disease, yet Merck chose not to market it (read more). Statins and cholesterol are not as lucrative today as it was before.

Cholesterol and statine critics are essential for medical development

Don’t misinterpret me. Dr. Ravnskog, Dr. Sandberg and anyone else who is critical and Celtic, are incredibly important. One must question and reconsider all knowledge in order not to commit any unnecessary mistakes. We must therefore have deep respect for all those who question the prevailing ‘self-evictions’. As an example, we have recently seen how studies have been presented over the last 5 years showing that the last 50 years of dietary advice have been wrong (especially how fat and carbohydrates affect blood lipids and blood pressure).

Studies that speak against the importance of cholesterol

Study 1: Ravnskov U, Diamond D, Hama R, et al. Lack of an association or an inverse association between low-density lipoprotein cholesterol and mortality in the elderly. A systematic review. BMJ Open.

Where is the published: BMJ Open, which is an Open Access.

Is a statistician among the authors? Unclear (not visible). The use of statistical terms is incorrect (e.g. the term “multivariate” is used, although there is no multivariate analysis [multivariate and multivariable are not the same).

What type of study is: Summary of several observational studies. This is not a meta-analysis because individual studies are told separately.

What is the objective of the study? The aim is to investigate whether LDL cholesterol affects the risk of death and cardiovascular death among the elderly. This is how to write (highlighted in yellow):

This means that LDL cholesterol must be a risk factor for death and cardiovascular death in the elderly, if the theory of cholesterol is to be correct. The aim is thus to investigate whether LDL cholesterol causes cardiovascular disease and death among the elderly. Unfortunately, this is impossible to do with the help of an observational study (it is very common to forget about this).

How did the study be conducted? It is written that the study is a systematic review of the literature. To find relevant articles, PubMed has been searched and identified cohort studies where participants are 60 years of age or older, and where data were available on LDL cholesterol, cardiovascular mortality, overall mortality. 19 studies were identified in total and 68094 were included.

What is the conclusion of the study? It is said that higher LDL cholesterol was associated with longer survival and lower risk of cardivovascular death. So this is the exact opposite of what the rest of science says. This finding finds the authors in a total of 14 out of 28 studies they included and in the remaining 14 studies there was no correlation between LDL cholesterol and death.

Problems/weaknesses in the study

1 – The most important weakness is that this is an observational study and is therefore not valid for pronouncing cause and effect. It can at best be regarded as an interesting study that invites further research. This has nothing to do with the authors, the subject, the area or the like — this is a purely consequence of the laws of biostatistics. When you read the authors conclusion of the study, you realize that they do not understand the principles of cause and effect (causality):

As shown here, the authors write that this study gives reason to question the cholesterol hypothesis and to reconsider guidelines that suggest lowering LDL in the elderly. Unfortunately, this is a cardinal error that actually disqualifies the entire study but we still keep going through the study.

2 – It is said that you want to test Goldstein’s hypothesis that LDL causes cardiovascular disease but to do this you choose older people. This is strange, because Goldstein’s hypothesis is not just about older people, but all people. It becomes even more strange for those who know that all risk factors (diabetes, blood pressure, smoking, cancer, cholesterol) become less important the older people become. It does not matter which risk factors are investigated, studies show that the effect of risk factors decreases with age (this is a phenomenon discussed under the term survival bias for those who are particularly interested). You wouldn’t think diabetes is healthy anyway, would you? Well, it’s actually if you look among the very oldest, as shown in several studies, for example in this Swedish study found in the New England Journal of Medicine. It is strange to test Goldstein’s hypothesis in a group of people who you know are very little affected by risk factors, no matter what risk factor it is. However, previous research has shown that cholesterol is a risk factor up to 90 years (but it is as expected weaker the older you get).

3 – There is no SAP (statistical analysis protocol). SAP is a protocol that is completed before work begins and it should be published (preferably signed with date). SAP shows which methods it is intended to use, which persons should be included in the study, and so on. The purpose of SAP is to prevent the study from taking shape as it continues (as this risks introducing complex systematic errors into the study). We can therefore never know whether the study is the result of what it was originally intended to investigate, or whether it was redirected as the articles were read.

4 – Only one database was searched, namely PubMed (usually searched in several databases). All the more strange is how to choose to search. This is how the authors write:

Thus, the search term “lipoprotein AND (old OR elderly) AND mortality NOT animals NOT clinical trial” has been searched. In other words, they have not been searched for “LDL” or “cholesterol”, which means that they missed all studies that did not use the term “lipoprotein”. This is messy because many studies use the terms LDL and LDL cholesterol without writing out the abbreviations.

5 – There is no list of which studies have been excluded, so we cannot see what is actually presented by the original sample of studies.

6 – There are no reasons given why individual studies have been excluded.

7 – You write that you included studies where “participants reflect the population, or where you could not see any difference between the participants and their population”, which is highly unclear what that means, but it leaves plenty of room for subjective interpretations of what should be included in this study.

8 – The authors write: “we included only studies where adjustments were made for the relationship melan LDL and mortality”. But the fact is that Table 2 shows that the study of Bathum and colleagues — who actually make up two-thirds of the 68085 people included in the study — did not provide any details about what and how to adjust the analyses.

9 – It writes that it excludes the study of Ptsay and colleagues (reference 6) because “it contained the same patients as another study”, which was Fried et al (reference 7). In fact, Ptsay et al was a re-analysis of data from Fried et al, and this re-analysis was done 6 years later and there was no correlation between LDL and mortality.

10 – Have studies been excluded that should have been included? Yes, examples of such are Postmus et al, published within the time period that the authors made their study. It was found that high LDL cholesterol increases the risk of death and low LDL cholesterol reduces the risk of death. Postmus et all is actually a Mendelsk randomization, which is why it is very strange that it was not included.

11 – See highlighted text in the following table from Ravnskog’s article:

It’s only 3 out of 19 studies that have taken altitude for statins! In an observational study (not randomized) investigating the importance of LDL cholesterol, it is crucial to take elevation for statins! A high school student should also realize that. This is not good. All the more strange is that Ravnskog and colleagues write: “It is worth noting that some with high LDL cholesterol may have received statins during the study. Such treatment may have increased the life expectancy of people with high cholesterol”. And in the same paragraph it says, “However, the beneficial effect of statins should be minimal, since statins have little effect at a maximum of 2% “. The next sentence says: “It is relevant that the 4-year mortality rate among those with the highest LDL cholesterol was up to 36% lower than those with the lowest LDL cholesterol”. This is not acceptable because it can deceive the inexperienced reader. The first figure (2%) is a figure indicating the absolute reduction in risk, while the second figure (36%) is the relative risk reduction. So when Ravnskog talks about the effect on statins, he uses absolute measures, but when he wants to highlight how healthy it is with high cholesterol, he uses relative measures. Relative dimensions are always more powerful!

12 — Those with the highest LDL cholesterol are most likely to receive statins during follow-up! Can this explain that they have the lowest mortality rate? Everything is possible in an observational study.

13 — Of course, we need to look more closely at the three studies that actually had information on statins.

  • In Jacobs et al, there is no correlation between LDL and death, but it notes that statins reduce the risk of death by 43%. If you look closer to the population that Jacob et al studies, you can see that those with high cholesterol are healthier individuals. Those with high cholesterol were more often women (women live longer than men), they estimated their health higher, they smoked less, they had 10% less coronary heart disease, lower incidence of heart failure, lower incidence of diabetes, lower incidence of cancer, lower incidence of stroke, lower incidence of morbidity in other things. Some of these differences were adjusted but that does not change the fact that (1) did not adjust for all differences, which one should definitely do and (2) there are strong indicative that these two groups were not comparable because the only group was healthier (that’s why we randomize. .. to avoid this…) .The next study, Fried et al. Ravnskog and colleagues retrieve their results from Fried’s statistical models that are built using the method “stepwise selection” and this is so substandard (poor), especially for these purposes, that it is not worth discussing this study (you who are interested in why refer to arbitrary modern book about prediction models). In addition, it was this study that was done six years later and the results changed so that no longer had a protective effect of high LDL cholesterol. It can also be added that the figure that Ravnskog refers to in this article (hazard ratio 0.66) is an unadjusted figure; Fried et al did not make adjusted calculations for LDL cholesterol. Thus, this study would not really have been included in the Ravnskog study.The third and final study is Blekkenhorst et al. There you can find no correlation between LDL and mortality or cardiovascular disease.

14 — The authors of this article (Uffe Ravnskog, Ralf Sundberg, David Diamond, Malcolm Kendrick etc) often criticize the use of “relative risk measures”, and consider it ugly tricks that scientists use to exaggerate the effect of statins and cholesterol. It is therefore very strange that this whole artery uses relative risk measures to highlight how healthy high cholesterol is.

15 – Has the food industry affected the results? One of the authors of the study has failed to report a bastard relationship. This applies to Alseem Malhotra, who is listed as advisor to the UK National Obesity Forum, which receives money from several industries.

16 — What did the experts responsible for reviewing the work of the BMJ Open magazine think? The experts’ comments can be read here. The first (David H. Newman) pointed out several of the above shortcomings and he did so several times. He also pointed out that the article could ‘be interpreted as being driven by an agenda’. He also pointed out basic methodological shortcomings that made the article not follow recommendations for how to do this type of study. On top of the barley, the expert himself is not an expert in the field; we searched PubMed for his previous work and cannot find any significant connection to cardiovascular research. The second expert (Flávio Danni Fuchs) is blood pressure scientist and writes as assessment:

Unfortunately, this review fails in two major points, the description and use of methods to perform a systematic review and the overall presentation of the script. The English is grammatically correct but the scientific style is far from the usual. The authors should discuss their own findings and leave aside at least the detailed discussion about several explanations for the non-association. The tone is excessively opinionated.

Conclusion

We have deep respect for colleagues and the researchers who wrote this article, but unfortunately it cannot be used for its purpose. It can also be added that paragraphs 2 to 16 are actually superfluous: the crucial point is paragraph 1, i.e. that this is an observation study. Cause and effect cannot be assessed in an observational study. In addition, there are well-made studies (with 900,000 people) that show absolutely opposite findings (i.e. that high LDL cholesterol is harmful at all ages). Let’s look at this study.

The connection between cholesterol and the risk of dying from a heart attack in all age groups

In a so-called meta-analysis (which means merging several smaller studies to gain more power in the study) that included 900,000 people, cholesterol was strongly associated with the risk of death from coronary heart disease at all ages (people up to 90 years). 2 The link Between total cholesterol and the risk of dying from cardiovascular diseases (coronary heart disease) is seen in this picture:

Sambandet mellan kolesterol och risken för att dö i kranskärlssjukdom
The relationship between cholesterol and the risk of dying from coronary heart disease

As shown in the picture, we see that for all age groups the risk rises when cholesterol rises. But apparently the ascent is greatest for the youngest and then gradually decreases over the years. This is a common phenomenon in medical research where survival is studied. The explanation is that those people who manage to grow old have qualities that protect them from the impact of risk factors. Anyone who succeeds in becoming 80 has proven that it has the conditions (it is strong enough) to resist the effects of various risk factors. This applies absolutely not only to cholesterol, but also to high blood pressure, smoking, diabetes and much more. The 80-year-olds who are alive when this study is conducted constitute the strongest individuals from their birth years.

Study 2. The effect of statins on average survival in randomised trials

This Danish study is often used in cholesterol and statine critical circles to claim that statins are ineffective. The purpose of the study is to calculate how much life expectancy is affected by eating statins. To make that calculation, randomized clinical trials have been identified that tested the efficacy of statins (both among people with and without cardiovascular disease). The study concludes that statins extend life 3 to 19 days. A drug that prolongs life 3 to 19 days, of course, is completely useless.

This study does not need many lines of discussion. The type of calculation that is intended to be made is called Life Years Lost Analysis, and such studies are fairly easy to do if you have data from clinical trials at hand. How did the authors do the analysis? Let’s see what they write in the method:

Yes, you read correctly, you’ve done the study in Microsoft Paint, the program that you draw stick figures in, but who was the statistician who was responsible for this? None – there is no statistician on board this study. Malene Lopez Kristensen is a pharmacist. Palle Mark Christensen is a doctor. Jesper Hallas is a pharmacologist. Thus, no statistician. What does one of Denmark’s most prominent epidemiologists/statisticians think about this study? He, Bendix Carstensen, uses the study to demonstrate how not to do.

We are going through another 10 cholesterol and statine critical studies in Part 4.

0/5 (0 Reviews)