The placebo effect — in which a false or 'dummy' treatment for a medical condition elicits a perceived or real improvement in the patient, simply because the patient believes the treatment will help — is one of the best-known phenomena in medical science. In recent years, the placebo effect has become so well-known among both patients and doctors that placebo treatments have become common components of medical practice, with various studies finding that large percentages of American and British doctors have made use of placebos in their everyday care. Perplexingly, some research suggests that the placebo effect is growing stronger, if only in the United States. But how strong is it, really? This 2010 meta-analysis of studies involving the placebo effect suggests that its perceived power may really be mostly in our heads. In fact, it may not exist at all in the physical sense that many believe it does.
Conducted by the Cochrane Collaboration, a think tank that studies health intervention effectiveness by processing large quantities of research, this paper analyzed hundreds of medical studies focusing on all types of conditions. Crucially, its authors didn't simply focus on studies that involved groups of patients receiving placebos, but on studies that involved a group of patients receiving placebo treatments and a group of patients receiving no treatment whatsoever, so that the impact on the two groups could be compared. This is crucial — a group of patients receiving placebo treatment could improve for many reasons besides the placebo effect itself, such as due to the fact that many ailments get better on their own over time, or due to regression to the mean. Without a no-treatment group to compare the placebo group to, it's impossible to tell whether improvements in a patient group's health derive from the placebo effect or from these other factors.
With this consideration taken into account, the Cochrane authors crunched the numbers on some 200 studies and made some surprising discoveries:
Overall, the researchers found that the placebo effect does not have the "large effects across many clinical conditions and outcomes", as it has lately been reputed to do.
When observers — such as doctors, researchers, research assistants, etc. — measured the strength of placebo treatments by physically evaluating the patient, they found virtually no impact.
The researchers only could detect meaningful placebo effect in studies where patients self-reported their condition (by reporting their pain level according to a scale, for example) while being treated for pain, asthma, and phobias.
This self-reporting placebo effect was small but consistent for nausea treatment, but varied widely from nonexistent to clinically significant for pain treatments. The Cochrane researchers essentially dismissed the phobia results on the grounds that the experiments they examined were of insufficient quality.
In short, this paper suggests that placebo treatments don't affect people's physical state in any measurable way, though they can improve patients' perception of their own condition. That's an important distinction — it suggests that while placebos may improve patients' mindsets, the popular idea that placebos can trick people's bodies into spontaneously healing themselves may be nothing more than smoke and mirrors.
Measuring the placebo effect is a tricky epistemological task, and as the Cochrane researchers acknowledge, their paper features a few shortcomings of its own. For example, it's impossible to sort experimental patients into a no-treatment group without their knowledge, and these patients' awareness that they're not receiving any treatment (as opposed to placebo patients, who think they're being treated but aren't) may affect their disposition. Further, the Cochrane paper relies heavily on self-reported data from patients, and patients may bias their own responses in an effort to please their questioners. Nonetheless, this meta-analysis casts serious doubt on some aggressive claims about the power of placebo.