From National Cancer Institute
Previous studies have suggested that placebo treatment can have positive effects on a variety of disorders and disease-related symptoms. However, the methodology used to collect and interpret the data may not have been ideal, because the studies were not double-blinded or the endpoints were not properly validated. The purpose of the present study was to determine the probability of improvement in symptoms or quality of life and tumor response in cancer patients treated with placebos in randomized controlled trials. We hypothesized that administration of placebos would improve symptom control and quality of life but would not lead to tumor response.
Methods: We reviewed reports of randomized controlled trials in which there was a placebo arm (37 trials) or a best supportive care (BSC) arm (10 trials).
Results: In trials that assessed average responses for patients in the placebo arm, improvements in average levels of pain were reported in two of six trials and in appetite, in one of seven trials. No improvements in average levels of weight gain (six trials), in quality of life (as assessed by patients; 10 trials), or in performance status (as assessed by physicians; nine trials) were reported. In trials that assessed response to a placebo in individual patients, 0 percent–21 percent of patients showed reduced pain or decreased analgesic intake, 8 percent–27 percent of patients showed appetite improvement, 7 percent–17 percent of patients showed weight gain, and 6 percent–14 percent of patients showed improvement in performance status. Quality of life for individual patients was not reported in any trial. Tumor response assessed by World Health Organization criteria was observed in 10 (2.7 percent) of 375 patients (seven trials total). Response as assessed by a serum marker was observed in 1 (1.7 percent) of 60 patients (two trials total). The probability of symptom improvement in patients receiving BSC was generally similar to that in patients receiving placebo, although no improvement in pain and only one tumor response among 191 patients (five trials) were reported. Conclusion: In randomized double-blinded, placebo-controlled trials, presumably with minimum sources of bias, placebos are sometimes associated with improved control of symptoms such as pain and appetite but rarely with positive tumor response. Substantial improvements in symptoms and quality of life are unlikely to be due to placebo effects.
The placebo effect was first described by Beecher, who suggested that about 35 percent of patients with a variety of conditions could be improved or cured by placebos. A placebo effect can be defined as “the effect seen in patients who have received an intervention which is believed to lack a specific action”. The observed effect of an active drug is a combination of three parameters: the natural course of the disease, the specific effect of the drug, and nonspecific effects including the placebo effect. Randomized clinical trials, ideally double-blinded with a placebo control, have become the gold standard for clinical research, because they can separate specific effects of an intervention from those due to variation in the natural course of disease and from placebo effects.
Since the publication of Beecher’s article, many authors appear to have accepted that up to 35 percent of patients with a wide variety of disorders respond to placebos. Placebo effects have been well documented for relief of postoperative or other types of pain in nonmalignant diseases, in psychiatric disorders such as anxiety and depression, and in cardiovascular disease. However, Beecher’s article has been criticized because of misinterpretation of data and lack of appropriate documentation in his analysis. It has been suggested that the rate of response to placebo was overestimated. Certainly, responses in patients receiving placebos seem to be more frequent when the effect is a change in subjective sensation or when patients are anxious or depressed. Responses in such patients also seem to depend on the doctor–patient relationship, on the expectations of the patients and their doctors, and on the characteristics of the placebo, such as color, name, size, and route of administration.
For cancer patients, placebo effects are recognized when treatment is given for relief of symptoms. Placebos are regarded as essential in trials of antiemetics, but effective control of chemotherapy-induced nausea and vomiting does not exceed 15 percent in placebo groups and cannot be attributed specifically to placebo. Moertel et al. performed a review of four double-blinded, placebo-controlled, randomized trials of analgesic medication for cancer pain. They concluded that 113 (39 percent) of the 288 patients who received placebo experienced 50 percent or greater relief of pain. However, in these studies, the evaluation of pain was not based on a prospective comparison of validated scales that assessed the level of pain before and after treatment but on patients’ estimates of percentage of pain relief. Such estimates depend on the memory of the previous state and might lead to an inflated estimate of benefit. In a more recent study, Boureau et al. used validated scales (the visual analog scale and the French version of the McGill Pain Questionnaire) to assess pain in a double-blinded, placebo-controlled, randomized trial for cancer patients with bone metastases. They reported pain relief (as judged by the patient) in 51 percent of 38 patients at the end of twice-daily intramuscular injections of placebo, which persisted 7 days later. In this study, the authors described the placebo effect but provided no details of the active treatment or of its efficacy.
Many treatments, some with substantial toxicity, are given to patients with cancer. For patients with metastatic disease, it is rare that such treatments lead to improved survival, but they may lead to tumor response and/or improve symptoms and quality of life. All or part of these effects might also be due to placebo effects. If this were the case, it would be prudent to select nontoxic alternative treatments. The purpose of this review is to determine, on the basis of a review of the literature, the probability that symptoms and/or quality of life may improve and that tumor response may occur following the administration of placebos to cancer patients. We hypothesized that a substantial improvement in symptom control and quality of life would follow administration of placebos but that tumor response would occur rarely. •