Stopping JUPITER: The Statistics Behind Astra Zeneca’s Crestor Cholesterol Medication Study
The decision by Astra Zeneca to stop the so-called JUPITER trial of its Crestor cholesterol medication last year garnered a ton of press attention. The New York Times captured the general tone of the coverage with this lead on a front-page story.
"A large new study suggests that millions more people could benefit from taking the cholesterol-lowering drugs known as statins, even if they have low cholesterol, because the drugs can significantly lower their risk of heart attacks, strokes and death."
What Abramson and other critics don't like about most drug study coverage is that reporters often start with the assumption that drugs are the best possible solution and outweigh any diet or lifestyle changes a patient might make. They also think reporters don't explain the possible ethical conflicts in these studies, as most of them, like the Crestor study, are funded by drug makers.
Abramson cites a report from Massachusetts General Hospital's Department of Psychiatry, which found that when scientists studying a drug have financial ties to the drug maker, the results are five times more likely to favor the drug over a placebo. A study in the Journal of General Internal Medicine found a similar result.
Abramson offers four things to keep in mind when writing about drug studies:
- Understand that the fundamental mission of commercially sponsored clinical trials is commercial.
- Search for the ways in which the commercial objective might have trumped dispassionate science-in the study design, analysis and presentation.
- Realize that identifying the financial ties isn't enough. Readers won't understand the extent to which the fundamental nature of the results can be spun.
- Recognize that presenting the comments of one or two experts who don't have financial ties to interested drug makers isn't enough. The majority of the reflection on the import of the findings should be from nonfinancially conflicted experts.
I would add one more thing, based on his commentary.
- Peel back the layers of data to understand how the authors drew their conclusions.
One of the few instances that I found of putting the numbers in perspective was in a story in my paper, the Los Angeles Times. The story pointed out that when the authors reported a dramatic 44% reduction in cardiovascular events among people who were taking Crestor they were actually talking about very small percentages. The real numbers are much less dramatic. The drugs dropped the heart attack risk from 2.8% to 1.6%.
Abramson makes much of the fact that there was "no difference in the overall rate of serious illnesses" between the Crestor and the placebo groups and that more people taking Crestor developed diabetes. He also thinks that, to truly measure the effectiveness of the drug, researchers should have included counseling on diet, exercise and other lifestyle issues in the study. By doing that, the authors could have reported whether Crestor improved the outcomes of people who were given lifestyle counseling - on how to eat better, exercise and give up smoking - and, one would hope, made an even bigger difference to people who were not counseled.
Maryann Napoli at the tiny nonprofit Center for Medical Consumers in New York underscores the importance of doing your own math with her breakdown of the numbers.
"For every 120 people who take Crestor for nearly two years, one cardiac event will be avoided. Or, put another way: Out of every 120 people who take Crestor, 119 will receive no benefit."