In response to a placebo, a neuron in the subthalamic nucleus shows a dramatic reduction in firing rate and bursting activity. Credit: Courtesy: Fabrizio Benedetti, University of Turin

For a phenomenon so often disparaged as a trick of the mind, the placebo effect has a starring role in medical research. But spurred by renewed debate on how clinical trials are designed, a handful of researchers are beginning to rethink placebos' use in clinical trials.

“'Placebo effect' is a catch-all phrase that involves a lot of things,” says Kathleen Propert, a biostatistician at the University of Pennsylvania who studies clinical trial design. “I think if we understood the individual pieces, we could use it more effectively.”

Over the past five decades, placebos—sham treatments that mimic the therapeutic effects of real drugs or interventions—have become indispensable control measures in judging the efficacy of new therapies. Based on studies conducted in the 1950s, researchers estimate that one in three people responds to placebos. The effect may be much higher for conditions with a strong psychological component such as pain and depression.

Prescribing placebos is generally considered unethical, but a recent study found that many doctors in Israel still do so (Br. Med. J. 329, 944–946; 2004). More commonly, researchers rely on placebos to isolate the nonspecific effects of a drug—such as the feedback a patient gets from the physician, the patient's expectations about whether the condition will improve, or the natural course of the disease—from the physiological properties of the drug itself.

Although placebos are used in nearly every clinical trial, however, researchers do not agree on what the placebo effect is, or even whether it exists. An analysis of 114 published trials found that placebos are statistically no more effective than no treatment at all, except in studies where patients subjectively rate their level of pain (N. Engl. J. Med. 344, 1594–1602; 2001).

Some researchers say the placebo effect may simply be false reporting from patients who tell their doctors what they want to hear. “I'm actually dubious that it occurs,” says Robert Temple, director of the Office of Medical Policy at the US Food and Drug Administration.

I'm actually dubious that [the placebo effect] occurs. Robert Temple,, US Food and Drug Administration

Still, there is increasing evidence for the physiological basis of placebos. For instance, using functional magnetic resonance imaging, University of Michigan researchers have localized placebo response to pain medicine to a distinct area of the brain: patients whose brain scans show the strongest response also reported the greatest pain relief (Science 303, 1162–1167; 2004). Another team at the University of Turin in Italy demonstrated the effect of placebos on a single neuron in patients with Parkinson disease. The researchers showed that saline injections mimic the therapeutic effects of drugs such as L-Dopa (Nat. Neurosci. 7, 587–588; 2004).

Even among the skeptics, few researchers dispute that placebo controls are necessary for good science—but the controls do not behave as predictably as researchers would like. “More than any other phenomenon, the placebo effect is exquisitely sensitive to the context,” says Ted Kaptchuk, assistant professor of medicine at Harvard University. “But science is about isolating phenomena from their context.” Kaptchuk notes that in their last trial, he and his colleagues used two different informed consent forms—and patients' response to treatment differed depending on which form they used.

Some researchers suggest expanding trials to include a group of subjects who receive no treatment at all, as this group would provide a baseline with which to compare both the placebo and the treatment groups. Others propose using active placebos that mimic the side effects of real drugs, to better mask whether a patient is receiving a drug or a placebo. But withholding treatment and subjecting patients to the side effects of drugs without their benefits both raise ethical concerns.

Scientists may also mistakenly assume that the patients don't know whether they are receiving the drug or a dummy pill. “We discovered that most people were able to tell, the tip-off being side effects,” says Roger Greenberg, professor of psychiatry at the State University of New York Upstate Medical Center. Asking patients what they think they got might solve that problem, Greenberg says.

Tweaking standard clinical trial paradigms might allow researchers to unravel how patient expectations influence the outcome. For example, the balanced placebo design uses four groups of subjects. One group knows it received the drug and a second group receives the drug but is told that it is a placebo; two groups of patients receiving placebo are likewise either told the truth or told they are receiving the drug.

In one study using this paradigm, only subjects who were told they were getting caffeine and were given the substance showed an increase in blood pressure and irritability. Participants who were given caffeine but believed it was a placebo did not show those effects.

In another approach, dubbed the open-hidden paradigm, researchers administer the same drug to two groups of patients—openly to one group of patients and covertly to another. Patients in one study who did not know they were being given morphine only felt half of the drug's effect, when compared with patients who knew (Prevention and Treatment 6, 2003, http://journals.apa.org/prevention/volume6/pre0060001a.html).

Not taking into account how patients' expectations influence the results of clinical trials can lead to wrong conclusions about a drug's efficacy, says Fabrizio Benedetti, professor of physiology at the University of Turin, who developed the open-hidden paradigm. But a consensus on how clinical trials should change to accommodate the new insights into placebos is unlikely for the present.