Is the Use of Placebo Controls Ethically Permissible in Clinical Trials of Agents Intended to Reduce Fractures in Osteoporosis?


  • Baruch A Brody,

    1. Center for Medical Ethics and Health Policy, Baylor College of Medicine, Houston, Texas, USA, and Department of Philosophy, Rice University, Houston, Texas, USA
    Search for more papers by this author
  • Nancy Dickey,

    1. Office of the Vice Chancellor, Texas A&M University System Health Sciences Center, College Station, Texas, USA
    Search for more papers by this author
  • Susan S Ellenberg,

    1. Division of Biostatistics and Epidemiology, Center for Biologics Evaluation and Research, U.S. Food and Drug Administration, Rockville, Maryland, USA
    Search for more papers by this author
  • Robert P Heaney,

    1. Department of Medicine (RPH and RLO) and Center for Health Policy & Ethics (RLO and RBP), Creighton University, Omaha, Nebraska, USA
    Search for more papers by this author
    • Dr Heaney has research contracts from the following companies: Eli Lilly and Co., General Mills, GlaxoSmithKline, Lane Laboratories, Mead Johnson, National Dairy Council, Proctor and Gamble Pharmaceuticals, Rhodia, and Roots, Inc. He serves as a consultant for Bayer, General Mills, GlaxoSmithKline, and National Dairy Council. Dr Heaney is also a member of the Speaker's Bureau for the Eli Lilly and Co. and National Dairy Council. He is an emeritus member of the Board of Trustees for the National Osteoporosis Foundation. Please note that the research contract and consultant income is not retained by Dr Heaney but is paid to Creighton University

  • Robert J Levine,

    1. Interdisciplinary Program in Bioethics and Center for Interdisciplinary Research on AIDS, Yale University, New Haven, Connecticut, USA
    Search for more papers by this author
    • Dr Levine has served as a consultant for Eli Lilly and Merck. All other authors have no conflict of interest

  • Richard L O'brien MD,

    Corresponding author
    1. Department of Medicine (RPH and RLO) and Center for Health Policy & Ethics (RLO and RBP), Creighton University, Omaha, Nebraska, USA
    • Center for Health Policy and Ethics, Creighton University, Omaha, NE 68178, USA
    Search for more papers by this author
  • Ruth B Purtilo,

    1. Department of Medicine (RPH and RLO) and Center for Health Policy & Ethics (RLO and RBP), Creighton University, Omaha, Nebraska, USA
    Search for more papers by this author
  • Charles Weijer

    1. Department of Bioethics, Dalhousie University School of Medicine, Halifax, Nova Scotia, Canada
    Search for more papers by this author


Substantial progress has been made in developing treatments that reduce the risk of fractures in osteoporosis. However, available treatments are only partially effective, they are not widely used, and there is need to search for more effective means of fracture prevention. Currently known effective means of reducing fractures were found using randomized placebo-controlled trials. The use of placebo controls in clinical trials has been a subject of significant controversy in recent years. The Declaration of Helsinki revision of October 2000 caused great concern among clinical investigators about the future use of placebo controls if known effective therapeutic agents are available. A working group of ethicists, clinical trial design experts, and clinical investigators examined the current state of knowledge of osteoporosis treatment and trials. They concluded that if placebo controls put subjects at substantial risk of serious outcomes, they are not ethically permissible. Placebo controls in osteoporosis trials with fracture as the measured outcome are permissible only under narrowly defined conditions. Placebo controls may be used if competent, well-informed patients refuse approved therapies for sound reasons, there is a reasonable basis for substantial disagreement or lack of consensus among professionals about whether approved treatments are better than placebos, or subjects are refractory to known effective agents. Active control trials are permissible and desirable if they can be designed and conducted in ways that overcome the interpretive difficulties often associated with such trials.


IN OCTOBER 2000, THE World Medical Association issued a revised Declaration of Helsinki,(1) article 29 of which declares: “The benefits, risks, burdens and effectiveness of a new method should be tested against those of the best current prophylactic, diagnostic, and therapeutic methods. This does not exclude the use of placebo, or no treatment, in studies where no proven prophylactic, diagnostic or therapeutic method exists.”

This clause of the revised Declaration quickly raised serious concern among clinical investigators and Institutional Review Boards. Many interpreted it as a prohibition against placebo controls in therapeutic trials if any effective therapy is known and approved for use. Because there is a widely held perception that there is a regulatory requirement for placebo controls in antifracture efficacy studies in osteoporosis, many investigators have expressed concern that strict adherence to Helsinki would make it operationally impossible to introduce and test new agents to reduce fracture risk in osteoporosis.

In consequence, the American Society for Bone and Mineral Research sponsored a working group to explore this issue. Participants were chosen because of documented expertise in the conduct of clinical trials with osteoporosis, clinical trial design, human subject research ethics, and participation in the development of international guidelines addressing human subjects research. A conference was convened by the Creighton University Center for Health Policy and Ethics in Omaha, June 25, 2001 to examine the question: is it ethical to use placebo controls in clinical trials of new agents being tested for efficacy of reducing the incidence of fractures in osteoporosis? This report summarizes the outcome and rationales of that meeting.


During the last 10 years, substantial progress has been made in developing treatments that reduce the incidence of fractures in patients with osteoporosis. However, fracture-reducing measures are at best only partially effective.(2)

Calcium and vitamin D supplementation, when tested against placebo controls, have been shown to reduce fracture incidence by 22–70%, varying according to the population studied, whether the subjects met diagnostic criteria for osteoporosis, whether they entered with or without prevalent fractures, and the site and nature of incident fractures.(3–7)

In addition, two classes of pharmaceutic agents, bisphosphonates and selective estrogen receptor modulators (SERMs) have been reported to reduce fracture incidence by 10–49% in subjects with osteoporosis, again varying with the population studied, existence of prevalent fractures in subjects, and site of incident fractures.(8–14) Raloxifene (a SERM) produced no significant reduction in nonvertebral fractures.(8) Bisphosphonates reduced nonvertebral fractures, which generally are associated with greater morbidity and higher mortality than vertebral fractures, by 10–36%.(9–14)

From these studies, one may conclude that absence of therapy carries risk of six to eighty preventable vertebral and nonvertebral fractures per 1000 patient-years, depending on bone mineral density (BMD), age, gender, and the existence of one or more prevalent fractures. Many of these will be morphometrically diagnosed rather than presenting clinically. However, as will be referenced below, fractures of all sites may be associated with significant morbidity, and placebo-controlled trials require an excess of at least 24–30 (and frequently more) preventable fractures in the placebo arm to be statistically valid.

It has also been reported that calcitonin administered intranasally reduces vertebral and other fractures in osteoporosis, but that study is clouded by a high subject dropout rate and the lack of a dose response to calcitonin.(15)

All cited studies showing fracture reduction in patients with osteoporosis were conducted as randomized double-blind trials with placebo control arms. In the studies of pharmaceuticals and calcitonin, the placebo and treatment arms included calcium and/or vitamin D supplementation.

In addition to the fact that currently approved fracture-reducing therapies for osteoporosis are only partially effective, others factors contribute to the need to develop and test the efficacy of new agents for treating osteoporosis. Currently effective measures are not widely used for several reasons. Osteoporosis often is not diagnosed. Even when diagnosed, physicians frequently do not prescribe known effective agents.(16–19) Currently approved agents have side effects that many find intolerable.(8, 20)


There is wide agreement among ethicists that research involving human subjects must satisfy several essential standards. The research must have value for human knowledge, and in the case of biomedical research, health. It must be designed and conducted so that scientifically valid conclusions may be drawn from the results. Subjects or the populations from which they come should be able to realize the benefit of the research. No more subjects should be enrolled than the minimum number necessary for scientific validity. Voluntary informed consent of subjects or appropriate permission for those incapable of consent is required; subjects must know that the trial is a search for knowledge and whether or not it will entail any diagnostic or therapeutic measures that may be useful; they should know of available proven therapies and offered the opportunity to choose them instead of participating in the trial; they must know the risks they are assuming; subjects must be free to withdraw from a study at any time. It is not acceptable to allow subjects to accept unreasonable risk for any reasons. Informed consent alone cannot justify enrollment. Research subjects who are harmed as a result of the research should be cared for and compensated; although the ethical obligation to provide medical treatment and compensation is widely supported by ethicists and the reports of virtually all national deliberative bodies on the ethics of research involving human subjects, it is not required by federal regulations. Human research must be subject to independent review with the authority to approve, disapprove, modify, and terminate any study. There must be a reasonable relationship between risks and benefits, and risks must be minimized.(21–24)


The ethics of the use of placebo controls in any circumstance where effective therapeutic measures exist has been questioned.(25–27) Several recent works urge moderation and describe circumstances under which placebo controls may be considered ethically permissible. (28–32)

Most ethicists and investigators agree that even if approved therapies exist, placebo controls are permissible if there is a reasonable basis for substantial disagreement or lack of consensus among professionals about whether approved treatments are better than placebos. Many also hold that placebo controls are permissible if subjects are not put at greater risk of serious and/or irreversible outcomes than they would be in a study treatment arm. Placebo controls may not be used if subjects are put at increased risk of death or serious irreversible outcomes, relative to subjects in the treatment arm, as a result of participation in the trial. Many ethicists and clinical investigators hold that risks of certain reversible outcomes (e.g., severe pain, intense fear, great emotional or physical discomfort) also militate against the use of placebo controls.

The working group agreed to the premise that it is unethical to deny research subjects in osteoporosis clinical trials, or ask them to delay, the best approved therapy if doing so puts them at increased risk of serious outcomes.

There are some glaring terms of relativity in these statements. The terms “reasonable basis,” “lack of consensus,” and “serious” may be interpreted differently by different investigators and ethicists in different circumstances. However, Omaha conferees examined these terms and arrived at considerable consensus on their application in the specific case of clinical trials of fracture reducing agents in osteoporosis.

There is no doubt that patients with osteoporosis are at risk of fractures, vertebral and nonvertebral. The absolute risk at different sites varies with age, BMD, and the prevalence and sites of previous fractures. However, in trials with fracture as the end-point(s), a certain excess number of fractures must occur in a control group to attain statistical validity.(33) The crucial issue is whether or not fractures are serious outcomes.

Some may argue that some fractures in osteoporosis are not serious. Only about one-third of vertebral fractures are clinically diagnosed.(34) However, vertebral fractures, diagnosed or not, may be associated with significant morbidity including back pain, days in bed, limitation of activity,(35) and higher risk of mortality,(36) although the reason for higher mortality is not perfectly clear. Vertebral fractures also are associated with increased risk of future fractures, both vertebral and nonvertebral.(37–40) Forearm fractures also carry significant morbidity(41) and are associated with increased risk of future osteoporotic fractures.(42) Thus, any fracture in osteoporosis may or may not be a serious outcome. In trials with fracture as the end-point, it is not possible to predict which fractures will result in morbidity and which won't. To accept the risk of any preventable fracture is to accept the risk of, even anticipate and expect, serious outcomes.

Given the facts that the any fracture may be a serious outcome, conferees reached consensus that placebo controls in trials with fracture as the end-point(s) are permissible under certain narrowly defined circumstances.

Placebo controls may be permissible if subjects provide informed refusal of known approved treatments for sound reasons. Many osteoporosis patients refuse currently approved pharmaceuticals because of inconvenience and side effects they find unacceptable. To use placebo controls under these circumstances imposes on investigators a significant burden to assure informed consent. Consent must be truly voluntary, and subjects must understand the risks to which they are exposing themselves. Subjects must not be induced to refuse known approved therapy by investigators who lead them to believe that the proposed new treatment will be superior. Nor should subjects be induced to participate by other considerations (e.g., anything that may exploit patients made vulnerable by age, poverty, or social condition). Such persons are eligible to participate in trials, but special care should be taken to avoid exploitation.

Placebo controls also may be permissible if subjects have been shown to be refractory to approved effective agents. However, in osteoporosis, it may be difficult or impossible to determine if individuals are refractory to approved therapies. For example, the occurrence of incident fractures while on therapy may leave the uncertainty of whether the person is refractory, may have had more fractures if untreated, or would be better protected by another agent.

There may be a reasonable basis for substantial disagreement or lack of consensus among professionals about whether currently approved treatments for osteoporosis are more effective for fracture reduction than calcium alone or calcium plus vitamin D. The levels of calcium and vitamin D used when studied alone have been higher (as high as 1200 mg of calcium and 800 IU vitamin D) than those used in control arms of studies of pharmaceuticals (500–1000 mg calcium and 250–600 IU vitamin D). Both the rates of fracture reduction and of fracture incidence on the treatment arms of both sets of studies overlap significantly. However, differences in populations, entry criteria, and different levels of calcium and vitamin D make reasonable comparison of these studies extremely difficult, if not impossible. Therefore, one could hypothesize that calcium and vitamin D at higher levels of intake are as or more effective than the pharmaceuticals plus the lower levels of calcium and vitamin D used in drug studies. This may justify a trial comparing high levels of calcium and vitamin D, usually considered a placebo control arm, with the same levels of calcium and vitamin D plus a pharmaceutical agent. (In reality, this is a superiority trial of combination therapy tested against the active control of calcium and vitamin D and not a placebo-controlled trial. It is included here because of the convention of referring to calcium and vitamin D supplementation as the placebo control arm in studies of single pharmaceutical agents.)

Furthermore, any uncertainty that may exist provides only a temporary solution to the problem of placebo-controlled trials of pharmaceuticals. As soon as the first combined treatment is shown to be better than vitamin D and calcium at optimal therapeutic levels, it will then become the new “… best current… therapeutic method.” If trials with an arm consisting only of calcium and vitamin D are to be conducted, all potential subjects must be informed about the risks of withholding or delaying drug treatment, offered approved treatment with the opportunity to choose not to participate in the trial, and informed of their prerogative to withdraw at any time.


Superiority or non-inferiority trials with active controls are alternatives to placebo-controlled trials. But active control trials may require very large numbers of enrolled subjects to reach confident conclusions. Non-inferiority trials in particular require great care in design and conduct to reach the ethical standard of scientific validity.

Non-inferiority trials rely on historical data to support the efficacy of the active control group; if historical data are highly variable and not consistently positive, it may not be possible to show equivalence or non-inferiority of new agents tested against active controls.(43)

These challenges raises serious questions about whether placebo controls should be prohibited, given that the active control alternative is available.

Conferees also reached consensus on the use of active control trials. It is appropriate and permissible to conduct a superiority trial to test a new therapy against an approved therapy if credible preclinical data show that the new therapy may be more effective than the approved therapy. Animal data showing significant increases in mechanical strength of bone and BMD or significantly reduced toxicity may constitute such preclinical data.

It is also permissible, indeed desirable, to conduct superiority trials in which a new agent combined with an approved one with a different mode of action is tested against an active control arm consisting of the approved agent.

Conferees did not reach consensus on all topics discussed. Some participants hold that, if a combined therapy is shown in a superiority trial to be more effective than an approved therapy, it is permissible to test the new agent alone against the old approved therapy alone. Others concluded that such a trial would deny subjects the benefit of best proven therapy. They would support a trial of the new agent against the combined treatment if combined therapy were more effective than the old single therapy.

Most but not all conference participants agreed that a trial of a new agent versus an approved agent versus placebo would be ethically permissible if optimal levels of calcium and vitamin D were included in all three arms, only subjects with no prevalent fractures were enrolled, routine screening is conducted during the trial, and therapy is initiated in the event of incident fracture or bone loss. Those who support this design do so with the rationale that current treatments have not been shown to be superior to optimal doses of calcium and vitamin D, or even if it has been shown that current approved drugs used with optimal calcium and vitamin D are superior to calcium and vitamin D alone, the risk of symptomatic fracture is very low in this population. Those who do not support it argue that, if it is known that a drug plus optimal levels of calcium and vitamin D provide greater benefit than optimal calcium and vitamin D alone, those on the placebo arm (optimal calcium and vitamin D) are placed at greater risk of incurring fractures than those on the treatment arms. If it is not known whether any drug plus optimal calcium and vitamin D is superior to optimal calcium and vitamin D alone, this trial becomes an active control (calcium plus vitamin D) superiority trial similar to the one described above in the case of uncertainty about the relative effectiveness of the different treatment arms.

The working group considered reducing risks to research subjects by designing placebo-controlled trials that enroll only subjects with low bone mass who have never had a fracture or patients with a single radiographic nonclinical fracture in time-limited trials. All subjects who experience significant bone loss during trials and all subjects who experience fractures would be treated with approved therapy. Ethicists who find this impermissible point out that with fracture as the end-point(s), the trial is valid only if subjects experience the serious outcomes of fractures. Therefore, subjects are not only being exposed to risk but the serious outcomes are expected.


The possibility of reducing risk by using BMD as a surrogate end-point(s) was also discussed. No conclusion was reached; the value of BMD as a surrogate is uncertain. BMD correlates negatively with risk of incident fracture in untreated subjects.(44) Changes of BMD in treated individuals correlate positively but weakly with reduction in relative risk of fracture; the relationship is nonlinear.(45) It is also not known whether it would reduce the time to reach an end-point(s) sufficiently to reduce the risk to a level proportionate to benefit.

Since the conference was held in Omaha June 25, 2001, the World Medical Association has issued a “clarification” of Paragraph 29 of the Declaration of Helsinki. It reads “The WMA is concerned that paragraph 29 of the revised Declaration of Helsinki (October 2000) has led to diverse interpretations and possible confusion. It hereby affirms its position that extreme care must be taken in making use of a placebo-controlled trial and that, in general, this methodology should only be used in the absence of existing proven therapy. However, a placebo-controlled trial may be ethically acceptable, even if proven therapy is available, under the following circumstances:

  • 1.Where for compelling and scientifically sound methodological reasons its use is necessary to determine the efficacy or safety of a prophylactic, diagnostic or therapeutic method, or
  • 2.Where a prophylactic, diagnostic, or therapeutic method is being investigated for a minor condition and the patients who receive placebo will not be subject to any additional risk of serious or irreversible harm.

All other provisions of the Declaration of Helsinki must be adhered to, especially the need for appropriate ethical and scientific review.”

This clarification, although relevant to the issue of placebo controls when effective therapy exists, does not affect the considerations or change the conclusions of the working group, which considered the use of placebos in the specific instance of fracture-reducing agents in osteoporosis.


This study was supported by a grant from the American Society for Bone and Mineral Research.

The contents of this report do not represent the official position of any of the institutions with which its authors are affiliated or of the American Society for Bone and Mineral Research. The findings reported here were submitted to the Society and accepted as a report.