In the CARE study, breast cancer was indeed more common among those who took the drug than in the control group. In the treatment group, 12 women developed breast cancer during the trial whereas there was only one case in the control group, a difference that is highly statistically significant.
The authors of the CARE report were eager to explain away the increased
frequency of breast cancer, terming it
an 'anomaly'. It is possible that they are right, since the expected number of breast cancer cases, as calculated from the frequency normally seen in the general population, should have been five cases. Nevertheless, 12 is more than twice as many as five.
The gain in the numbers of fatal heart attacks was 1.1 per cent whereas the loss in numbers of breast cancers was 4.2
per cent. Calculated in the way that trial directors usually do-as relative, rather than absolute, risk-the difference was even more striking, with 12 per cent fewer attacks, but 1500 per cent more breast cancer. (However, side-effects are never calculated in this way-only positive effects.)
The number of side-effects has tremendous importance when it comes to assessing preventative treatment as the number of patients experiencing side-effects easily exceeds the number of prevented heart attacks. Unfortunately, this fact is often ignored or-worse-hidden by using the concept of relative risk to make positive effects seem larger than they are while
citing side-effects in absolute numbers.
The cholesterol myth
and where it came from
In 1953, Ancel Keys, director of the Laboratory of Physiological Hygiene at the University of Minnesota, published a paper that served as an early kickoff for the cholesterol campaign. Accord-ing to Dr Keys, high-fat food was the culprit. His proof was a diagram, published in 1953, which showed a close correlation between the total intake of fat and the death rates from CHD in six countries. But why did Dr Keys limit his data to a mere six countries when, at that time, information was available from 22 countries?
The reason is that, if all of the countries were included, the association be-came rather weak. The death rate from CHD in Finland, for instance, was seven times that of Mexico, although fat consumption in the two nations was almost the same.
To prove his idea about dietary fat, Dr Keys organised a study of CHD
in seven countries (Circulation, 1970; suppl 1: 1-211). He selected 16 local populations in the Netherlands, Yugo-slavia, Finland, Japan, Greece, Italy
and the US. The conclusion drawn from this gigantic project was that the factor most likely to predict the number of heart attacks in a country was the amount of animal fat consumed by the people of that country. Yet, no association was found with the total fat consumption, in striking contrast to Keys' previous study.
But within the individual countries, the number of heart attacks showed no correlation with diet. For instance, two districts of Finland had strikingly different heart disease and death rates, but virtually identical diets.
Other population studies have the same flaws. In the US, CHD mortality increased about 10 times between 1930 and 1960, levelled off during the 1960s and has since decreased slowly. During the decline of CHD mortality, the consumption of animal fat declined also but, during the 30 years of sharply rising CHD mortality, the consumption of animal fat decreased (Am J Clin Nutr, 1964; 14: 169-78).