Total Pageviews

Thursday 19 June 2014

Assault on Science

David H. Newman, M.D.
Physician-researcher, leading expert in health care evidence

"Will it help me live longer?" 

When patients ponder the lifetime commitment to a statin drug, this is the question they ask. But a very public controversy in the scientific community has recently diverted attention from this central question -- and that just might be on purpose.
In the October issue of the British Medical Journal, the rare scientific journal that routinely questions convention, a group led by Dr. John Abramson, a teacher of healthcare policy at Harvard Medical School, reanalyzed the largest-ever research report on the statin drugs. Abramson's group found evidence that for all but the very highest-risk people, statins did not save lives and did not reduce the frequency of serious illness. 
This was surprising not just because it challenged convention but because it contradicted the conclusions of the Cholesterol Treatment Trialists (CTT) collaboration, a group granted exclusive access to report on raw data compiled by the drug manufacturers (who still harbor most of the data). In their highly touted 2012 report the CTT group concluded that the pills could extend life for everyone who took them, including healthy people at low risk of heart problems. This finding was prominently cited by the American Heart Association in their recent guideline recommending statin use for patients at low risk.
Why do the Abramson and CTT groups disagree about the same data? When the CTT group, led by Dr. Rory Collins, an epidemiologist at Oxford, reported the numbers, they failed to separate people according to risk level. So when they reported that statins reduce deaths, they did so based on a calculation including both the highest- and lowest-risk participants. When the Abramson group examined death rates, they looked at the low-risk patients separately, and the mortality benefits disappeared; for these patients, taking statins was the same as taking a placebo. 
This is undisputed. Neither Dr. Collins nor the CTT group nor the AHA has challenged the finding, which suggests universal agreement that for the great majority (80 percent or more) of people currently on a statin, and for all those newly recommended, the drugs don't save lives and don't reduce serious illness. 
The shocker, however, is that this is not the "controversy." Dr. Abramson and colleagues also reported that in the largest study of statin adverse effects, nearly 18 percent of patients had side effects and stopped the drug. Dr. Collins disagrees, saying that only 9 percent stopped the drug because of these side effects. 
That's it. Seriously. And Abramson's group agrees (while 18 percent of patients had experienced side effects, only half of those had stopped taking the drug; Abramson's group read the study incorrectly), so the BMJ has already corrected the error.
But Dr. Collins is still wailing -- perhaps to drown out something else. It does create quite a noise, after all, when medical recommendations affecting hundreds of millions of people are found to be incontrovertibly wrong.
Dr. Collins is calling for a retraction of the Abramson paper, an action typically reserved for studies in which the primary finding is both fraudulent and wrong (such as the one scientific paper linking autism and the measles vaccine). Retraction of a scientific paper is never used when the central finding is accurate and a minor error is discovered in a secondary point.
There are many reasons that Dr. Collins' statements on this issue are surreal. They seem to have left science behind, or worse. For instance, Collins seems to believe that randomized trial data like the data that his group has exclusive access to, as opposed to observational studies like the one Dr. Abramson cited, should be used to estimate side-effect rates. The Food and Drug Administration, and most others, disagree. It often takes years of post-marketing surveillance (i.e., observational) studies and case reports for dangerous side effects to emerge, which trials typically miss. This is partly because trials are usually focused on benefits, not harms. This is most obvious when drug companies, who have a financial stake in finding benefits rather than harms, pay for the trial -- like virtually all the CTT data. 
A classic example of why trial data should never be used to estimate side effects is a statin trial for which Dr. Collins himself was a lead investigator. In the Heart Protection Study of 2002, Collins' team systematically removed patients who suffered significant side effects, never reporting on them in the final numbers. Despite identifying 32,000 patients who seemed right for statin drugs, they dumped 12,000 of them when they had difficulty tolerating or taking a statin. The final paper reported on only 20,000 people. In other words, those with significant side effects were weeded out before the trial started. 
This is a dubious practice, defended as a way of identifying patients who will benefit most. And while this may be true, the practice clarifies why data from randomized trials are often irrelevant to people concerned about their chance of experiencing side effects.
To be sure, it may be reasonable to worry less about side effects for patients who could live longer if they take a statin. After all, fatal side effects are rare, and living longer is paramount to most patients. But side effects become critical when a statin won't extend life, and when the benefits are less common or less important than the side effects. But when low-risk people take a statin, the chance that the drug will cause diabetes is roughly equal (at best) to the chance that it will prevent a nonfatal heart attack. And a person taking a statin is 25 times more likely to experience muscle damage than they are to avoid a stroke.
Bottom line: The headline should be that the bestselling pills of all time don't save lives or reduce major illness for most who take them. But that has been obscured by a war on science. 
The war includes many assaults: performing trials that weed out side effects, claiming they should be used to counsel patients, maintaining secret databases, and, worst of all, trash-targeting a journal for publishing a discovery that brings truth to millions.
Graham

1 comment:

Anonymous said...

"And a person taking a statin is 25 times more likely to experience muscle damage than they are to avoid a stroke."
Stay clear for me.

Roy