One of the most strongly held beliefs in medicine - that dummy pills or other sham treatments greatly help many patients - has been called into question by Danish researchers who found little or no "placebo effect" in dozens of studies.
That led the researchers and other doctors to recommend that for ethical reasons, placebos, or inactive substances, no longer be given to patients outside of controlled medical experiments.
"The shoe is on the other foot now. The people who claim there are placebo effects are going to have to show it," said Dr. John C. Bailar III, a just-retired professor of health studies at the University of Chicago who wrote an editorial accompanying the research in Thursday's New England Journal of Medicine.
Arthur Caplan, director of the Center for Bioethics at the University of Pennsylvania, said: "I was shocked by this study. This just goes completely against the grain."
In many medical studies, patients in one group receive an experimental new treatment, while a comparison group gets a lookalike dummy pill or other placebo. Neither the patients nor the doctors know who is getting what. The goal is to see if medicines being tested do more good than people's will to get better.
Studies occasionally include a third group not getting a placebo. Those patients receive nothing, or just get the standard treatment for a condition if the study is testing whether combining a new treatment with the current one is better.
The Danish researchers combined the findings of 114 such studies from around the world, involving dozens of conditions ranging from colds and seasickness to Alzheimer's disease and schizophrenia, to see how the sham treatment stacked up to no treatment.
In most of the studies, the placebo group fared about the same as the group getting no treatment. The exceptions were studies of pain treatments and some others with subjective results, meaning patients reported how much symptoms bothered them, rather than having an objective measure such as blood pressure.
Placebo recipients in the pain studies averaged a 15 percent reduction in pain, and patients in the other subjective studies had even smaller improvements.
Many past studies and textbooks suggest that about one-third of patients given placebos in medical experiments get better, presumably because they believe they are getting an effective treatment. But the new research casts doubt on this long-held belief.
"The high levels of placebo effect which have been repeatedly reported in many articles, in our mind are the result of flawed research methodology," said Dr. Asbjorn Hrobjartsson, a professor of medical philosophy and research methodology at University of Copenhagen who ran the study with colleagues at the Nordic Cochran Center there.
The researchers and other experts said the improvements in subjective measures might be explained by "reporting bias," where patients, thinking they are getting a powerful treatment, incorrectly judge their condition or overstate any improvement to please their doctor.
Dr. Stephen Schneider, a professor of medicine at University of Medicine and Dentistry of New Jersey in New Brunswick, has worked on many trials where patients on placebo improved. He said some of those patients would have gotten better with no treatment. In addition, he said, simply participating in a medical study induces patients to eat better, exercise more or otherwise pay more attention to their health.
Schneider and Caplan said many doctors still routinely prescribe placebos to reassure patients who want every possible treatment.
That includes giving antibiotics to people with colds, which cannot help, recommending heavy doses of Vitamin C for the flu, or prescribing a drug that is unlikely to work but might give hope to a patient with a terminal illness.
Richard Gracely, a scientist specializing in pain research at the National Institutes of Health, noted that Hrobjartsson's report did not state how much the active treatment in the 114 studies helped patients.
Gracely said that might be important because the magnitude of the placebo effect in past studies generally was proportional to the active treatment's effect.
On the Net: http://www.nejm.com