A thought-provoking paragraph I read last weekend makes me wonder whether medical investigators are sometimes fooling themselves when they conclude that an experimental drug works better than a placebo.
As many of my readers know, medical trials typically involve some people who get a real drug and some people who get a placebo. The idea is to compare the two groups at the end of the study so you can essentially subtract any placebo effect from the presumably real effect of the drug.
But if people who are taking the placebo know they are taking the placebo, then there may not be a placebo effect. And if people who are taking the real drug know they're not taking the placebo, they may get an extra psychosomatic boost—a sort of super-placebo effect—from the knowledge that doctors are expecting them to do better than the poor chumps getting the fake medicine. Either way, comparing the two groups at the end of the study could lead researchers to make over-optimistic conclusions about how well a drug works against an underlying physiological condition.
This might happen more often than medical investigators would like to think.
And because they rarely account for the possibility when they design studies, investigators would have no way of knowing which of their conclusions are fatally flawed.
Here's the paragraph (from page 12 of the Oct. 16 issue of Neurology Today, a publication mailed to members of the American Academy of Neurology) that prodded me to think harder about the subject:
"At six months, there was significant improvement in fatigue for the patients treated with antibiotics compared to the placebo. The study authors noted that enough patients receiving antibiotic treatment correctly guessed they were in that group to conclude that this finding might be due to 'unblinding' and a placebo effect."
Hmm. In other words, antibiotics may have produced no true, biological benefit in this trial (which was treating people who had a chronic condition known as post-Lyme syndrome) and only appeared to improve symptoms because patients figured out that they were getting real medicine and unknowingly fooled themselves into thinking they felt better. Sure enough, the medicated patients did no better than placebo-treated patients on two other study measures, including one that measured immunit-related molecules in cerebrospinal fluid.
What's interesting to me is that very few medical investigators do what the authors of this study did: ask their patients whether they think they're getting the real drug or the placebo. And even fewer do any sort of rigorous statistical analysis to rule out the possibility that patients who've figured out which group they're in either improve or don't, according to whether they "should" be improving. At the same time, I'm told, "blinded" investigators (those who during a trial aren't privy to which patients are getting placebo) often discuss their guesses with each other about which patients are getting placebo.
This isn't the first time I've encountered this thing I'm calling the super-placebo effect. In 2003, I wrote an article about a placebo-controlled trial that concluded that creatine supplements may improve performance on cognitive tests. But Ronald L. Terjung, a researcher I spoke to at the time, had a very good criticism of the study. As I wrote in Science News: 'Despite the study's design, volunteers may have known when they were getting the real supplements. Creatine can cause bad breath, flatulence, and weight gain from excess water retention—cues that might have encouraged volunteers to unconsciously push harder on the tests, he suggests.'
So here's my hypothesis: Some placebo-controlled trials that conclude that a drug is efficacious are, in fact, reaching that conclusion erroneously. They may actually be finding evidence that volunteers and/or investigators involved in the trial can guess who is taking the real drug and who is not.