About half of American doctors in a new study regularly give their patients placebo pills without telling them.

That contradicts advice from the American Medical Association, which recommends doctors only use treatments to which patients have given their informed consent.

"It seems like doctors are doing things they shouldn't be doing," said Irving Kirsch, a professor of psychology at the University of Hull, who has studied the use of placebos. Kirsch was not linked to the research, published Friday in the British Medical Journal.

"Doctors may be under a lot of pressure to help their patients, but this is not an acceptable shortcut," he said.

Placebos were defined in the study as treatments whose benefits derived from patients' beliefs they would work, not from the actual drug itself. They included painkillers, vitamins, antibiotics, sedatives and sugar pills.

Studies have shown that patients given a fake treatment can often improve, despite the pill having no known impact on their condition.

Researchers at the U.S. National Institutes of Health sent surveys to a random sample of internists and rheumatologists across the country. They received 679 responses, of which 62 percent believed that using placebo treatment was ethically acceptable.

Half of the doctors reported using placebos several times a month. Nearly 70 percent of those who did so described the treatment to their patients as "a potentially beneficial medicine not typically used for your condition." Only five percent of doctors explicitly called it a placebo treatment.

Jon Tilburt, the study's lead author, said he believed the doctors surveyed were representative of internists and rheumatologists across the U.S. No statistical work was carried out to establish whether the study results would apply to other medical specialists like pediatricians or surgeons.

In the survey, doctors were asked if they would recommend a sugar pill for patients with chronic pain if it had been shown to be more effective than no treatment. Nearly 60 percent of doctors said they would.

Smaller studies done elsewhere, including Britain, Denmark and Sweden have found similar results.

The U.S. research was paid for by the National Center for Complementary and Alternative Medicine and the Department of Bioethics at the National Institutes of Health.

"It's a disturbing finding," said Franklin G. Miller, director of the research ethics program at the National Institutes of Health and one of the paper's authors. "There is an element of deception here which is contrary to the principle of informed consent."

The authors said that most doctors probably reasoned that doing something was better than doing nothing. Other situations where placebos were used included doctors giving vitamins to patients with difficult conditions, such as chronic fatigue syndrome, or giving antibiotics to patients with bronchitis to make them feel better.

Scientists still don't understand how the placebo effect works — whether it is just psychological or whether there is also a physiological reaction to the treatment.

Irving said it might be possible to get the psychological impact without using a fake pill. "If doctors just spent more time with their patients so they felt more reassured, that might help," he said.

Experts also don't know if the placebo effect would be undermined if patients were explicitly told they were getting a dummy pill.

But some patients said the truth was paramount.

"I would feel very cheated if I was given a placebo," said Ruth Schachter, an 86-year-old London resident who has had skin cancer. "I like to have my eyes wide open, even if it's bad news," she said. "If I'm given something without being warned what it is, I certainly would not trust the doctor again."