The placebo effect is an effect seen in patients given an intervention, such as a placebo, that has no pharmacologically-mediated action against the disease. Previous studies have suggested that some cancer patients who had received a placebo for pain reported a reduction in pain after the intervention.
To determine the probability that a placebo will lead to improvement of symptoms and tumor response, Gisèle Chvetzoff, M.D., of the Centre Léon Bérard in Lyon, France, and Ian F. Tannock, M.D., Ph.D., of the Princess Margaret Hospital in Toronto, reviewed reports of 37 randomized controlled trials that compared a group receiving active treatment with a group receiving a placebo. The authors also reviewed reports of 10 randomized controlled trials that compared a group receiving active treatment plus best supportive care with a group receiving best supportive care alone. Some of the trials looked at individual responses, and other trials looked at group responses.
The researchers found that, in trials that examined average responses for patients in the placebo group, two of six trials reported improvements in pain and one of seven trials reported improvements in appetite. No trials reported improvements in average levels of weight gain, quality of life, or performance status.
In trials that looked at individual response to placebos, 0% to 21% of patients given a placebo showed improvement in pain or decreased use of pain medications, 8% to 27% of patients showed an improvement in appetite, 7% to 17% of patients showed weight gain, and 6% to 14% of patients showed improvement in performance status. By contrast, objective tumor response was observed in 2% to 7% of patients in a placebo group.
In addition, the authors found that the likelihood of symptom improvement in patients receiving best supportive care was similar to that of patients receiving placebo.
"The present review suggests that substantial well-documented patient-reported improvement in symptom control or quality of life (particularly when supported by objective evidence of tumor response) are unlikely due to placebo effects," the authors conclude. They add that they strongly support the use of a double-blind placebo-controlled design for randomized trials whenever possible because such a design can ensure equal conditions in both arms and minimize various types of bias.
In an accompanying editorial, Robert J. Temple, M.D., of the Center for Drug Evaluation and Research at the U.S. Food and Drug Administration, points out that only two of 10 trials reported objective tumor responses of more than 2% in the placebo group (one trial reported 7%). He says that "tumor response rates of more than 10% are unlikely to be seen in an untreated group and can therefore be considered evidence of a drug effect… ."
Temple also notes that in trials looking at symptoms such as pain and appetite, individual patient improvements were seen among people in placebo groups, suggesting that single-arm trials (those that include only a treatment group) are a suboptimal way to assess the benefits of treatment.
Contact: Vince Rice, Princess Margaret Hospital, 416-946-4501 ext. 5771; fax: 416-946-4585, vince.rice@uhn.on.ca
Editorial: Christine Parker, Center for Drug Evaluation and Research, FDA, 301-827-7251, parkerc@cder.fda.gov, or Crystal Rice, 301-827-1673. ricec@cder.fda.gov
Chvetzoff G, Tannock I. Placebo effects in oncology. J Natl Cancer Inst 2003;95:19–29.
Editorial: Temple R. Implications of effects in placebo groups. J Natl Cancer Inst 2003;95:2–3.
Note: The Journal of the National Cancer Institute is published by Oxford University Press and is not affiliated with the National Cancer Institute. Attribution to the Journal of the National Cancer Institute is requested in all news coverage.
Journal
JNCI Journal of the National Cancer Institute