Skip to Main Content


Dilemma 2

Yale Medicine Magazine, 2004 - Spring


Human subjects research: when is use of a placebo ethical?

There are several standard drug therapies that can help prevent vertebral fractures in some women with osteoporosis, a potentially serious problem associated with a condition common among aging women. A pharmaceutical company asked me to design and direct a study of a new medication that shows promise of preventing fractures in many more women. I designed the study to compare the new drug with placebo, in which neither the clinician nor the patient would know whether she was receiving the drug or a pill having no effect. Based on statistical analysis, the proposed trial could not show the new drug’s effectiveness until the number of new fractures in the placebo group exceeded by 150 the number in the group receiving the new drug.

Our hospital’s institutional review board, or IRB, rejected the study, claiming it would be unethical to treat patients with a placebo when there are standard therapies that are known to be at least partially effective. The review board suggested that, instead of placebo, we give the control group one of the standard therapies. I disagreed because most women who have osteoporosis do not receive such treatment, particularly when they have no symptoms. (The women to be recruited for this study will have no symptoms of osteoporosis.) Furthermore, most vertebral fractures complicating osteoporosis have no symptoms, and the women who have such fractures often don’t even know it; it’s not as if they have serious pain or disability. Moreover, it will still be necessary to continue the trial until the number of new fractures in the control group exceeds by 150 the number in the group receiving the new drug. The same number of women will experience this injury; it will just take a lot longer to reach that number. In other words, there would be a substantial loss in efficiency of the trial with no compensating benefit. Was the review board’s action right in this case?

As chair of the Human Investigation Committee at the School of Medicine for 30 years until 2000, Professor of Medicine Robert J. Levine, M.D., HS ’63, was called upon to review hundreds of applications for testing new therapies. He is the author of the widely used book Ethics and Regulation of Clinical Research and founding editor of the journal IRB: A Review of Human Subjects Research. He agrees with the review board’s decision. “When a therapy known to be effective already exists, even if only partially beneficial, and withholding the effective standard may result in serious complications, it is unethical to use placebo.”

Levine suggests that the physician consider an alternative study using what he terms an “add-on” design for the trial. This is possible when a new drug acts by a different mechanism than the standard drugs currently used in medical practice. In osteoporosis, one of the standard therapies, a combination of vitamin D and calcium, is partially effective in the prevention of new fractures, according to Levine. Since vitamin D/calcium works differently in the body than the new drug, he believes it would be appropriate to give this combination to all patients and also give half of the patients the new drug and the other half the placebo. “In this way,” he says, “the trial has the advantages of being a placebo-controlled study without depriving the women of a known, partially effective therapy.” Although it will still be necessary to continue the trial until the number of new fractures in the control group exceeds by 150 the number in the group receiving the new drug, Levine believes the loss of efficiency will be compensated by the therapeutic benefits to the participants getting the placebo. “If I were on that IRB, I would approve such a study.”