Drugs, risks and bad counting

Last week saw a lot of articles, like this, on the 'wonder drug' rosuvastatin (Crestor), which apparently cuts heart disease risk by 40-50%. 

To emphasise the benefits, the drug companies like to talk about relative risk reduction, rather than the more helpful absolute risk. To illustrate the difference, anyone can improve their chances of winning the lottery by 100%. All they have to do is buy a second ticket. The absolute chance of winning is still rather too high to change your way of life in anticipation of a win - about 1 in 7million.  So to understand real risk to an individual, the figures need to be examined. Ben Goldacre in his Bad Medicine column in the Guardian talks about this.

The details of the research project (Jupiter) show in fact that for the 18000 people in the study, the absolute difference in death rates for ALL causes between those taking the drug and those on a placebo was 49 people (247 to 198). Only 400 heart disease events, such as heart attacks and strokes occurred in total for both groups, about 2.2%., but over half were medical interventions to deal with symptoms rather than sudden illnesses. Only 12 people in each group died from strokes. At the same time, about 100 extra people on the drug developed diabetes. About a quarter of the people dropped out, before the study was stopped after less than two years because of its startling 'success'. If the study had continued for five years there would have been no one left at the end. Coincidentally, although it is not often mentioned, about 25% of people on statins have side effects such as mental impairment, muscle pain, memory loss etc. Other studies (GISSI-HF  and CORONA. for example) have shown that rosuvastatin provides no benefits.

So why are AstraZeneca funding such studies and trumpeting trivial outcomes as huge percentages in relative risk reduction? Could it be that they would like as many people as possible taking an expensive drug every day for 25 years? The current preferred drug, Simvastatin, is now off patent, and therefore cheap. As an illustration, one US commentator pointed out that to save one life using Crestor would cost almost half a million dollars (400 person years of the drug to avoid one death). So, rosuvastatin is a me too drug designed to reinstate an expired patent, rather than produce any other result. 

This is clearly not getting it right. The purpose surely should be to look for cost effective ways of reducing ill health, not selling drug products. Of course, these research studies are expensive to run, and all of them rely on funding from drug companies. None of Big Pharma can or will spend money on comparing their drugs with other solutions. Thus we have extensive testing trying to prove for example that statins reduce coronary heart disease compared with a placebo, but none comparing the drug with normal health improvement measures such as reducing weight, changing diet, exercise. 

There is one interesting fact that has yet to be explained about heart disease. The incidence started to increase noticeably in the 1920s, and peaked at the beginning of the 1970s, since when it has decreased dramatically, by about two thirds in real numbers (from 714 deaths to 194 per 100,000 men aged 55-64 between 1968 and 2005, for example).  Perhaps the right thing to research is what brought about these changes, and how we might extend it.