Honesty: a new policy for drug companies?
I was struck by two different news stories today – one in the New York Times, and the other in the Washington Post. Although the topics were different, the underlying theme was unified: drug companies (particularly Glaxo Smith Kline?) are coming clean with research data and marketing messages. Sure, it might have taken a law suit by Eliot Spitzer to shine a light on the common practice of witholding negative research information from scientists… and it also may have taken a costly failure of an over-hyped diet pill by a competitor drug company to cause GSK to take the honest marketing track with Alli… But I like this new honesty, however we got to it.
Did you know that drug companies spend billions of dollars to research the safety and effectiveness of their drugs, but then are under no obligation to share what they learn with the general scientific community? No, they share what they want to – generally the studies that show the largest effect or the greatest safety profile. But now, physicians have been given access to the raw data collected in all the trials (showing benefit, no benefit, or harm) conducted by GSK. And they’re having a field day! A new study published in the New England Journal of Medicine is based on an analysis of GSK’s research, where they have found that Avandia (a popular diabetes drug) may put people at higher risk for heart attacks and heart related death.
Now here’s the devil in the details (as Dr. Charles rightly points out): the potential harm has been blown way out of proportion – the media has been citing “a 43% increase in heart attacks/myocardial infarctions and a 64%
increase in death from cardiovascular causes” when another way of stating what the authors found is that 86/14,371 patients or 0.598% of the patients taking Avandia had a heart attack, while 72/11,634 or 0.619% of people
not taking Avandia also had a heart attack. Gee… which sound bite seems more scary?
Honestly, I feel worried for the general public who are now (with the new full disclosure of drug company data) sure to be victims of an onslaught of media hype around all sorts of small differences found in research studies. Believe me, it’s important to sift through all this data to look for early signs of potential drug related health risks – but I think we should be careful before we terrify our patients with scary statistics.
Maybe in the midst of all this new honesty – we can have medical bloggers like Dr. Charles and the Revolution Health team help patients get to the bottom of things without having to have a PhD in biostatistics. We need a voice of reason to translate research data for public consumption. I’ll do my part – but since there are ~6000 research studies published per day in this world… I need some back up. Any takers?This post originally appeared on Dr. Val’s blog at RevolutionHealth.com.
Thanks for your perspective Dr. Val. What you are describing is ubiquitous in media as well as peer-reviewed research articles– these terms are called ‘relative’ versus ‘absolute’ risk reduction. For example, beta-blockers provide a 30% reduction in death in the setting of acute heart attack which translates to a absolute reduction of 3% (10%—> 7% mortality rate). Put another way, the probablity of death under treatment A MINUS the probability of death under treatment B (placebo).
The best overall summary statistic is the number needed to treat (NNT) which is defined as the inverse of the absolute risk reduction. Wikipedia(http://en.wikipedia.org/wiki/Number_needed_to_treat ) provides a nice example: For example, consider a hypothetical drug which reduces the risk of colon cancer by 50%. Even without the drug, colon cancer is fairly rare, maybe 1 in 3,000 in every 5 year period. The NNT for a 5-year treatment with the drug is therefore 6,000: by treating 6,000 people with the drug, one can expect to reduce the number of colon cancer cases from 2 to 1.
Hope this helps….