I have a question about the relationship between a graph of mortality rate and years of life. Something like the MRFIT study has graphs that show total cholesterol on the horizontal axis, and deaths per year on the vertical axis. Low cholesterol has more deaths, and high cholesterol has more deaths, but 180-260 mg/dL total cholesterol has minimum death.

Assume that there is a direct relationship between total cholesterol level and mortality rate i.e. if I change my cholesterol level, then I also move to the corresponding death rate. In that case, how does one work out the benefit of changing cholesterol? i.e. can a certain cholesterol change, from total cholsterol y that is outside the area of minimum death, to x which is in the range 180-260 for minimum death, be converted into a number of extra years per life by using the mortality graph?

Let's take a deeper look at the MRFIT numbers. Of a total of 300,000 men investigated (only 12,000 were included in the study), only 2258 had died by the end of the study. Another way to think of this statistic is that 99.4% of the study subjects were still alive at the end of the study, regardless of their cholesterol levels. 98.7% of those with the highest levels of cholesterol were still alive at the end of the study. 99.7% of those with the lowest levels of cholesterol were still alive at the end of the study. The difference is 1%. Not such a dramatic result as the study authors trumpeted, but just as accurate.

Now, let's look at some of the details surrounding those who died. Of those included in the study, there were no reports on how the cause of death had been established. While it's possible that every one of the 2258 men were autopsied, it's more likely that the vast majority had their causes of death determined on death certificates written by general practitioners. Not only are such death certificates highly unreliable, but in many cases (between 6 and 20%, depending on the analysis of the MRFIT study), death certificates were actually missing. Yet some of the reports gave a detailed list of diagnoses for all deaths.

Further, the 10% of the study group with the greatest number of deaths included all the individuals in the study with the rare, inherited disease of familial hypercholesterolemia. This group made up 10% of the top 10%, or 0.1% of the total study group. In addition to lifelong high cholesterol no matter what the treatment, this group also has a history of atherosclerosis (cholesterol plaque buildup on the arteries) and other cardiovascular disease. So, the study confirmed what researchers had long known: patients with an inborn error of cholesterol metabolism have a greater risk of dying from heart disease — no shocker there. (1) If you omit this outlier of a group, the difference in death between the groups with the lowest and highest cholesterol levels goes from 1% to 0.9%. For such an extensive study, these results aren't really worth writing home about.

So, maybe the mortality statistics are a little fuzzier than it would seem, but this study is still statistically valid, right? Well, during the initial screening, it came to light that one of the participating centers had falsified its data to increase the number of participants in the trial, possibly to obtain more financial support from the National Institutes of Health. This embarrassing matter received little mention in the follow-up reports, nor did the study authors mention the possibility that data falsification could have occurred in other centers as well. Instead, all discussion of the issue of quality control was studiously avoided. Wrote Professor Werko, who analyzed the MRFIT study and many meta-analyses of it, "in the many publications regarding the MRFIT screenees, it is obvious that the authors are more interested in the mathematical treatment of large figures than in the quality of these figures or how they were obtained."(2)

You will find problems of similar magnitude in every major study that asserts a relationship between increased cholesterol levels and either mortality or heart disease. For in-depth reading, look at these books:

• Lundell, Dwight, MD. The Great Cholesterol Lie. 2011
• Kendrick, Malcolm, MD. The Great Cholesterol Con. John Blake Publishing, London, UK. 2007.
• Ravnskov, Uffe, MD, PhD. The Cholesterol Myths. New Trends Publishing, Washington, D.C., 2000.
• Ravnskov, Uffe, MD, PhD. Fat and Cholesterol are GOOD for You! GB Publishing, Sweden, 2009

(1)Ravnskov, Uffe, MD, PhD. The Cholesterol Myths. New Trends Publishing, Washington, D.C., 2000, pp. 50-52.
(2)Werko, L. Analaysis of the MRFIT screenees: a methodological study. Journal of Internal Medicine 237, 507-518, 1995.

There's also a new book out that critiques the cholesterol theory of heart disease: The Cholesterol Delusion, by Ernest Curtis, MD (a cardiologist). Dr. Mike Eades discusses this book in this blog post.