National Placebo Initiative
Summary of Conference Proceedings
This Summary reports only on the formal presentations of the Conference. Summaries of the discussion held at the end of each session and of the breakout sessions are in the full report.
KEYNOTE ADDRESS
David Sackett
Why are placebos sometimes used in Canadian research?
Placebo controls were included in the Canadian trial in the 1970s that established the net benefit to patients suffering from repeated minor strokes (transient ischemic attacks) of the use of aspirin to prevent major strokes. Patients were randomized among groups, one treated with two drugs thought to be of benefit, two treated with one of the drugs plus a placebo for the other, and one treated with placebos for both drugs. However, all patients also received best medical care, including any other drugs that they needed. Placebos were used to allow the trial to determine whether either drug (or both together) would benefit the patients, and to maximize the chances that the trial would yield a valid result; the trial result would be invalidated if the patients could know or guess what they were receiving and 'self-medicate', or if knowledge by patients or researchers of what individual patients were receiving might bias the assessment of the trial outcomes.
This example, and others outlined more briefly, provided the foundation for the conclusions that: placebos are not the same as no treatment; placebo controls provide the most powerful and reliable trial design for tests of the efficacy of drugs; randomization and blinding, where possible, are important techniques for ensuring the integrity of the performance of the trial; and patients treated with placebo inside a trial often do better than patients treated with known therapies outside trials.
What is the problem?
Kathleen Glass:
Ethical/legal concerns regarding withholding standard treatment
A number of principles and values in research ethics express the obligations of researchers and establish the REB's protective function: respect for persons and their human dignity; beneficence; and the physician's fiduciary responsibility to patients in clinical research. In seeking a favourable balance between benefits and risks, the major normative factor is the patient's right to medical treatment; the major scientific factors include integrity in conceiving and carrying out the trial, and a genuine uncertainty about the relative merits of the treatments being compared.
An argument is currently being put forward to justify a 'new' ethics of placebo- controlled trials. It proposes that physician/researchers seek answers to clinically relevant scientific questions and not optimal personalized care for individual patients.
Canadian and international documents (e.g., Helsinki, TCPS, CMA Code of ethics) and also the weight of legal commentary are unanimous in emphasizing the paramount responsibility of the physician to the patient.
Claims to the individual liberty of the patient to engage in actions not in their best interests must fail without an appropriate balance of risks and benefits. Also, consent, however perfect, cannot correct unethical research or incompetent medicine. Finally, considerations of ethics must include the question "Who benefits from placebo controlled trials?"
Robert Li
Scientific concerns regarding alternatives to placebos: active control trials.
Active control trials cannot be relied upon to establish efficacy of a test treatment. A non-inferiority active control trial, which can only demonstrate relative efficacy, does not answer the same question as, and cannot replace, a placebo controlled trial, which can establish absolute efficacy.
Randomized controlled trials, the gold standard for clinical trials, can be grouped into categories of superiority (test better than control), equivalence (test not much different from control) and non-inferiority (test not much worse than control). These categories involve very different null hypotheses and design features. Placebo-controlled trials always seek to show that the test is superior; active control trials can be in any of the three categories.
Non-inferiority trials can only assess efficacy relative to the (active) control, and must rely on evidence from outside the trial to reach a conclusion about the efficacy of the test drug. The quality of evidence that the active control is effective must therefore be assessed. Also, superiority and non-inferiority trials respond in opposite ways to poorly carried out research that gives noisy data, which increases the chance of not seeing a difference. Hence, observing that a drug is not inferior to an active control may be because both treatments are effective or ineffective or even harmful, or because the trial could not detect the difference that actually exists for technical, design or quality control reasons.
A recent trial of two drugs, one of which was regarded as standard therapy, against placebo failed to show that the test drug was inferior to the standard, but that both resulted in more deaths than placebo. Absence of a placebo control would have introduced yet another harmful drug.
Discussion of the ethics of placebo-controlled trials must include consideration of scientific and methodological realities. Can research that cannot give a meaningful result be ethical?
Ron Heselgrave
REB concerns regarding conflicting placebo policies.
The central question concerns the circumstances under which it is acceptable to use placebo controls. Should a more protocol-specific perspective be adopted which includes additional criteria?
Clinical equipoise requires honest professional disagreement among expert clinicians about the preferred treatment; an ethical trial has the potential to resolve this uncertainty.
REB concerns were identified as: FDA emphasizes superiority to placebo in demonstrating efficacy; the evidence for the effectiveness of standard treatment is often flimsy; acceptable conditions for use of a placebo control are poorly defined; and the consequences to the ability to treat patients from rejecting placebo control trials are unclear.
Placebo-controlled trials are carried out in all medical disciplines. Should specific protocols involving placebo controls be assessed on their own merits according to the usual criteria, rather than on global judgements?
Laura McAuley
Research concerns: Access and availability of data
Despite the enormous numbers of papers each year, not everything of importance is published. Analyses of the weight of existing evidence may be biased by what is available to the analyst.
Even the major journal indexing services do not cover more than 35% of the estimated number of journals. Also, the chances of getting an article published in indexed journals may be as low as 10%. Journals tend to publish reports that show statistically significant differences, possibly because researchers are reluctant to submit articles where statistically significant results are not seen.
These facts matter because they bias the results of cumulative analyses (meta-analyses) in favour of the conclusion that a therapy is effective. Trial registries might be a means of helping ensure that the results of all trials are reported.
Is a "Made in Canada" solution possible?
Thérèse Leroux (Director, Ethics Office, CIHR)
Heather Sampson (Chair, National Placebo Initiative Working Group)
The Conference and the National Placebo Initiative
This conference was organized by Health Canada and the Canadian Institutes of Health Research because imprecision in and disagreement between their statements on the use of placebo controls are causing confusion in the design and ethics approval of clinical trials in Canada.
The two statements are the:
Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans (1998) (TCPS);
International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use: Choice of Control Group and Related Issues in Clinical Trials (2000) (ICH E-10).
The uncertainties in Canada reflect international controversies, of which the Declaration of Helsinki (2000 and its clarification in 2001), the CIOMS draft documents (2001), and American FDA and OHRP policies have the greatest influence on Canadian research.
The purpose of the Conference was to share knowledge and perspectives on the appropriate use of placebos in clinical trials. The Conference is an early step in the National Placebo Initiative which the two organizations have agreed to undertake to develop a coherent Canadian approach to this issue. The objectives of the National Placebo Initiative are, through a process of stakeholder and public consultations, to develop recommendations to Health Canada and the Canadian Institutes of Health Research on a common policy in this area. A multi-stakeholder Working Group that was established to lead this initiative held its first meeting in February 2002.
Howard Brunt (Chair, Interagency panel on research ethics)
The section of the TCPS on placebos is a prime candidate for reconsideration. The willingness of the Panel to change the TCPS will be based on factors such as: the expectation that the changes will add to protection of research subjects and building of public trust; and the openness, inclusiveness and transparency of consultations. The Panel will consult widely and seek appropriate harmonization of TCPS with other policies within and outside Canada.
Robert Peterson (Director general, Therapeutic Products Directorate at Health Canada)
Health Canada is willing to consider an addendum to ICH E-10 on placebo controls. Though the ICH arose in Europe and is now led by Europe, the USA and Japan, Canada is "at the Table" and is influential in committees despite our relatively small share of the market in pharmaceuticals. In addition, regulations under the Food And Drug Act will continue to evolve. Health Canada welcomes this historic opportunity to work with CIHR and all stakeholders to develop a "made in Canada" position and is willing to consider changes that are consistent with Health Canada's mandate and that make sense to Canadians.
Report on focus groups
Maureen Smith
Public attitudes on placebos: Focus groups in Canada
The National Placebo Initiative has already started public consultations. Seven focus groups were held in February 2002 in Montréal (3, all in French), Winnipeg (2, both including First Nations people) and Toronto (2). Three groups of people were involved in different focus groups: the general public; type 2 diabetics; and people with common mental health conditions (depression, anxiety). The formal report (revision of translation in progress) will be posted shortly on the official site for the National Placebo Initiative.
Research on public attitudes is important because, for example: the public is the most important stakeholder; very little is known about public views; and public input is a pre-requisite for public trust.
The main findings were that the public has limited knowledge about placebos, but a desire for new treatments, and that the initial perception was "I think it is OK", which evolved after background information was provided to "It's ok as long as.". The focus group participants expressed many suggestions and concerns of which the major ones concerned risk communication and informed consent, the need to follow patients closely during trials, and trials in mental health.
This initial focus group experience indicated their value because they can address common misconceptions held by the public and by experts, and they encourage public accountability and transparency. The limitations identified so far include that they offer only insights, for instance on how Canadians think about issues. Further public consultations are being planned, probably involving specific groups.
Ethical issues:
Perspectives on withholding standard treatment
Bernard Keating
The placebo question: The structure of the moral argument.
Five theses form the foundations of the ethical arguments dealing with the prescription and proscription of placebos.
Thesis 1. The existence of clinical equipoise results in a situation where the physician-researcher does not jeopardize the duty to beneficence. This thesis reflects the Hippocratic oath's emphasis that the duties of the doctor to his patient must take precedence over all other considerations, and underpins both the TCPS and the Declaration of Helsinki (2000).
Thesis 2. The risks associated with the use of placebos directly determine their ethical acceptability. Adherents to this thesis deny Thesis 1 except for major risks. The observation of equal rates of suicide in schizophrenic patients receiving placebo or standard therapy is evidence of the ethical acceptability of placebo controls.
Thesis 3. The scientific need for the placebo legitimates its use. Some argue that the use of placebos prevents approval of ineffective medications, that preventing placebo use could compromise development of new treatments, and that use of placebos would reduce the numbers of people subjected to experimental risk. Is a scientific or methodological need a true moral justification?
Thesis 4. To limit the use of placebos is to compromise the autonomy of the patient. Banning the use of placebos when there is no risk of significant or long-term harm would be paternalistic. This viewpoint forces us to question the boundaries of the physician's fiduciary responsibility. Which responsibilities can doctors relinquish in order to increase the autonomy of the patient/subject?
Thesis 5. The codes, declarations, and policy statements define an ideal to which we must aspire but not rules of conduct to which we must concretely adhere. This thesis exhorts its adherents to do as well as possible while aiming for the ideal. Would a court recognize a defence based on this thesis?
Trudo Lemmens
Avoiding a Jekyll and Hyde approach to the ethics of research and clinical practice.
It is often asserted that the ethics of research are different from the ethics of clinical care because the inherent tension between their different goals and practices means that their ethical standards are different. Is this assertion valid?
Although some authors have argued that the concept of clinical equipoise ignores differences between the ethics of research and clinical care, clinical equipoise was developed precisely as a concept in response to the inherent tension between the two. Clinical equipoise is a research-friendly moral concept that helps to tackle a fundamental ethical dilemma: how can physicians include patients into clinical trials. In contrast, arguing that the ethics of clinical research and practice are different because of the inherent tension amounts to ignoring this fundamental ethical dilemma.
It is problematic to focus the placebo debate on informed consent and avoidance of exploitation. Informed consent is a necessary but not sufficient condition for ethical research. Mere reliance on informed consent ignores the real-life circumstances of clinical research. There are increasing pressures to enrol patients in clinical trials; indeed, clinical trials (including placebo trials) are sometimes even presented as a route to fast track treatment.
What are the implications of the recent evolution of the Declaration of Helsinki? Paragraph 29 of the 2000 revision has been heavily debated and criticised. The clarification published in 2001, which was approved by the Council and not the General Assembly of the World Health Association, may not improve clarity. What is a "compelling and scientifically sound methodological reason"? Or a "minor condition"? Or how should "no additional risk of serious or irreversible harm" be interpreted? Based on established theories of interpretation, the clarification should be seen in the context of the general text, which firmly establishes the duty of beneficence physicians have towards patients.
In the end, courts may be confronted with the issue of placebo controls. They will ask the question whether a patient was treated according to the standard of care. The courts will not accept a violation of the standard of care on the basis of a questionable interpretation of an ambiguous clarification added to a document of ambiguous legal status and after approval by a non-regulated REB.
The law requires the physician's fiduciary duty to the patient. Physician - researchers cannot compromise on their responsibilities to their patients. This duty is reflected in regulations, guidelines, policies, etc., for research.
Lisa Kelly
Withholding standard treatment: Myth vs fact from a regulator's perspective
Standard treatment is withheld not only in a placebo control arm, and also in the test arm for the new drug. At the start of a trial, potential benefits and harms are estimated on the basis of accumulated knowledge, which is usually much less extensive for the new drug than for placebo or the standard treatment. Do the most worrisome harms arise in the arm with placebo or with a new drug?
Uncertainty raises the issue of risk management. When treatment failure can have very serious consequences for the patient, risk management considerations may lead to refusal to allow a placebo because treatment failures in patients receiving placebo could be perceived as negligence, a charge that would be less likely for failures on the standard or trial drug. Such considerations are consistent with policies that prevent placebo when the available treatment is known to prevent serious harm.
The regulator seeks to optimize the trial conditions to maximize the potential for safety in all arms of a trial. The regulator will seek to ensure that accurate information is provided to REBs, clinicians and patients. The regulator may reject the protocol or terminate an ongoing trial because harms cannot be managed.
We should consider the extent to which individual patient choice, based on accurate and complete information, should be respected.
Patricia Huston
International research ethics trends in defining appropriate placebo use.
The international trend in policies for appropriate placebo use is based on two criteria, scientific necessity and no increased risk of serious harm. ICH E-10 is consistent with this view. No international guidelines use the concepts of standard care or clinical equipoise. The duty of care remains essential to the appropriate use of placebos.
In 2000, international policies of placebos were at an impasse. ICH E-10 regards placebo controls as the best way to establish efficacy and accepts them if there is no risk of serious harm, safety is maximized, and there is informed consent. The Declaration of Helsinki (WMA) (7th revision, September 2000) stated ".. A new method should be tested against .. the best current ... methods".
CIOMS and the Declaration of Helsinki have evolved since 2000. The CIOMS (Draft Guidelines, Spring 2001) condoned the use of placebo controls when there are sound scientific and ethical reasons to use a control other than the best current treatment, such as when there is only trivial risk, and a comparison of two treatments would not yield reliable results. This position appears to be more conservative than the Tri-Council Policy Statement. A proposed modification in March 2002 was either trivial risk or a compelling scientific reason and no additional risk of harm. The clarification of the Declaration of Helsinki (October 2001) stated that a placebo control may be ethically acceptable when there is a scientifically sound methodological reason or the study involves a minor condition with no additional risk of serious harm. This confusing clarification seems to be more liberal than ICH E-10.
The 2002 CIOMS proposal appears to combine both the spirit of Helsinki and the ICH E-10 positions. It suggests that active control trials are preferred unless risks are trivial or an active control is not feasible. The CIOMS proposal is consistent with ICH E-10 in using risk as a criterion and in acknowledging the reliability problems of non-inferiority trials. Adoption of the CIOMS position would result in fewer placebo-controlled trials because they would all need to be justified. If efficacy has already been established, an active control should be used unless a placebo control involves trivial risks.
Due to its absence from international guidelines, the concept of standard care seems neither useful nor necessary to determine what control should be used.
Scientific issues:
Establishing efficacy of new agents when standard treatment exists
Stan Shapiro
Design options to assess efficacy when standard treatment exists
The use of placebo controls is relatively uncontroversial when there is no standard therapy, standard therapy is thought to be no better than placebo, the placebo and test are added to standard therapy, or when patients with minor conditions refuse standard therapy. However, when placebos are used instead of proven effective treatment there is conflict between Canada's regulatory and research ethics policies. Examination of the literature reveals that placebo-controlled trials continue to be proposed and approved even when meta-analysis of accumulated evidence shows that a therapy is clearly superior to placebo.
Placebo-controlled trials are by their nature superiority trials. Active control trials, that is trials in which the control arm is an active control, can be superiority, equivalence or non-inferiority trials
A pivotal factor in the debate about the use of placebo versus active controls centers on the notion of assay sensitivity. ICH E10 (Sec 1.5) defines assay sensitivity of a clinical trial as ".the ability to distinguish an effective treatment from a less effective or ineffective treatment." The absence of assay sensitivity increases the chance that equivalency or non-inferiority trials will give the hoped for answer, and decreases the chances of demonstrating superiority. This has been used to argue for the scientific necessity of placebo-controlled trials.
The claim of assay sensitivity for placebo-controlled trials depends on a crucial assumption that has largely escaped discussion: effective blinding of the placebo arm. If blinding of the placebo arm is not effective, protection against expectation effects, biased assessment, contamination and co-intervention is lost. Superiority of a new treatment to placebo could merely be a consequence of loss of this control. An ineffective new treatment would spuriously appear to be superior.
Effective blinding is often difficult to achieve and needs to be demonstrated, not assumed. The absence of evidence of the success of blinding serves to undercut rather than support the claim that a placebo-controlled trial is scientifically necessary because it inherently possesses assay sensitivity.
In an attempt to obtain direct evidence about investigators' reporting of the success of blinding, a random sample of 200 placebo-controlled trials recently published in five top general medical journals and 4 top psychiatric journals was reviewed. This review showed that over 90% provided no information on the success of blinding. Thus there appears to be no basis to grant the claim that a placebo controlled trial is scientifically necessary because it inherently possesses assay sensitivity.
Where to from here? Should policies differ for 'me too' drugs? Should Canada encourage head to head comparisons of competitive treatments to compare efficacy and side effects? The regulatory challenges posed by active control trials do not justify dismissing such trials, especially when the alternative is to ask prospective subjects to forego standard treatment in favor of placebo. We need to consider the balances between questions of marginal value for most of society, and questions that are more difficult but more broadly relevant.
Robert Li
Regulatory approach to assessing the validity of active control trials.
When can a non-inferiority active control trial be relied on to establish the efficacy of a new treatment? The following conditions need to be met:
Good historical evidence that the active control is always effective;
Historical evidence that clinical trial results are sensitive to the effects of the active control;
The proposed trial design is similar to previous well-conducted trials using the active control;
The quality of conduct of the trial is high so that the noise in the data is low;
A properly sized non-inferiority margin is identified before the trial starts (not after seeing the data);
The clinical trial effect (placebo effect) in previous trials is consistently low;
The sample size is adequate to provide the needed statistical power.
Active control trials may be appropriate when treatments give high response rates with low variability or when responses to treatment far exceed spontaneous rates. Active control trials cannot be relied upon to demonstrate efficacy when placebo groups show substantial improvement and variability (high clinical trial effect); treatment effects with the active control are small or variable; or superiority to placebo for drugs considered to be effective cannot regularly be shown.
A scientifically acceptable trial can be ethically inadequate. But a scientifically flawed trial is always unethical. Proper ethical deliberation requires an understanding of scientific and technical limitations.
Don Addington
Clinical research perspective on placebo controlled studies of new psychotropic medications.
Mental disorders are of variable severity and duration. Depression, which can be a chronic disabling condition, may result in hospitalization or suicide. Epidemiological studies in the community show that, on average, an episode lasts 12 weeks; The average patient is not usually seen by physicians, let alone by specialists; approximately 20% receive an appropriate diagnosis and treatment. The challenge for the researcher is to find patients who are safe enough to enter a clinical trial but not so mildly ill that they remit spontaneously or respond too well to placebo.
Complex syndromes such as schizophrenia have more than one set of symptoms. A treatment targeted to one syndrome may not be effective against another. For example, antipsychotics provide useful control for the positive syndrome characterized by delusions and hallucinations. The same antipsychotics have only a marginal impact on the negative syndrome.
In schizophrenia, the typical patient - subject in a clinical trial has had 10 years of illness causing 5 hospital admissions and has already been treated with 4 antipsychotics. Standard antipsychotics typically show 50% improvement rates, but 75% of patients have side effects, some serious; 25% of patients on placebo show improvement, but without the side effects.
The selection of a reliable and valid primary outcome measure is critical to the success of any clinical trial. In clinical trials of psychotropic medications, observer rating scales are often the primary outcome measure. For example, the extent of depression is measured by assessing the patient's response, on a 4-point scale to a question such as "How would you describe your mood over the last 2 weeks? Do you keep reasonably cheerful or have you been very depressed or low-spirited recently?" Researchers need to train their raters and demonstrate that ratings are reliable between raters.
Many precautions are taken to assure safety in trials in mental health. Rates of suicide or attempted suicide in placebo-controlled trials in depression showed no differences between placebo, active control or new drug.
According to the Canadian Psychiatric Association Position Paper, placebo controls are indicated when placebo response rates are high, and/or variable, and/or close to response rates for effective therapies, or established interventions carry high risks of side effects and/or are effective only against certain symptoms. These recommendations are logically derived on the basis of the ethical principles in the Tri-Council Guidelines and the clinical nature of the disorders being studied.
Patients have a clear right to effective treatments. Placebo-controlled trials are the most powerful means of demonstrating safety and efficacy. Appropriate placebo-controlled trials must meet the principles of appropriate ethics guidelines and policies. Current guidelines can and should be improved.
Joe Pater
Placebo-controlled trials in cancer: emerging challenges
Trials in oncology have almost always used active treatment controls because: the trials have traditionally used hard endpoints like survival; established treatments could not ethically be withheld; and the trials have sought to improve on current treatments. Oncology trials have traditionally been impossible to blind because the procedures or drugs could not be provided in placebo form.
Newer oral drugs with few side effects are being developed. Also, endpoints such as quality of life or time to progression of the disease are increasingly being used for reasons of their importance to patients, and the increasing complexity of treatments. These new endpoints are susceptible to patient or observer bias. Use of placebos and blinding are therefore increasingly important in trials in oncology.
Placebo controls are much more complex to run than open label trials.
When should the blind be broken for individual patients? When the disease progresses? At completion of the therapy under the protocol? At the request of the patient? At completion of the study? The first three options could compromise the trial, and hence defeat the purpose of blinding which is to protect the integrity of the trial. There is no clear consensus on questions such as these, and guidelines are clearly needed.
Access to information:
Implications for placebo-controlled trials
Merry Bujaki
Access to information at Health Canada
The Access to Information gives Canadian citizens or permanent residents access to all records in the possession of the federal government, with some exemptions. These exemptions include trade secrets, confidential financial, commercial, scientific or technical information, or information that could adversely prejudice third parties in defined ways, mostly to do with financial or competitive considerations. Such exemptions would include clinical trial applications, pre-market adverse event reports, unapproved submissions, methods of manufacture of the dosage form, etc.
Health Canada has clearly defined processes and accountabilities to meet the requirements of the Act. The Office of the Information Commissioner investigates any complaints under the Act.
Abby Lippman
Access to data: Implications for patients and the public.
The major points of this presentation were: If patients or the public lack data, they cannot exercise an informed choice about trial participation; If REBs lack data, it is not ethical for them to approve a protocol; If the public lack data, they cannot determine if their health is protected by regulatory organisations; If researchers or regulators lack data obtained from direct input from the public or patients about policies and protocols, they cannot do solid research or establish sound policies.
Inaccessible data are a major cause of concern. The domination of private sector funding for health research sets a limited agenda and may foster practices that make data unavailable and/or inaccessible.
Evidence that patients see trial participation as a means to get faster and better care increases the need to be certain that potential research subjects have complete, detailed, unbiased information if they are to make informed choices. Patients need: comprehensible information on the standards of treatment and alternatives; details about all "harms," social as well as physical, of drug use; counselling and discussion with disinterested parties, which take time; follow-up information about the trial results; financial arrangements in the trial. However, an indigestible mass of detail can inhibit understanding.
REBs need all the relevant information about a protocol, including a high quality literature review and evidence for gender-based analyses where appropriate. Public representatives need support and validation to participate fully. Cognitive diversity is essential. REBs need to ascertain the importance and relevance of the research questions to the interests of society: a basic question for all protocols is "who is likely to benefit, and who is likely to be harmed." They need evidence that a drug to be tested will meet a real unmet health need before approving a protocol involving volunteers under the rubric of research.
All citizens need full access to and participation in regulatory processes, evaluations and policy-making.
A central registry of all clinical trial protocols that includes all study outcomes should be established, and REBs should require registration as a condition of acceptance. An open, accountable, transparent and well-governed regulatory process that includes enhanced public participation must be established and maintained. Special processes and policies for drugs to be used in healthy populations are required. Risk/benefit assessments, even consideration about whether "equipoise" applies, must never over-ride the establishment and maintenance of safety in research, in policy-making, and in post-approval monitoring.
Laura McAuley
Access and availability: The implications.
Meta-analyses tend to overestimate treatment effects because, for example, they may not include the grey literature and they may not exclude duplicate reports of the same trial. Incomplete coverage for linguistic reasons reduces the precision of estimates of effectiveness but does not appear to bias conclusions.
Cumulative information is needed to decide when there is sufficient information that further trials of the efficacy or effectiveness of new treatment has been accumulated, and thus prevent further placebo-controlled trials, which may remove patients from proven reliable treatments.
All research should include a comprehensive review of relevant literature and current studies. Efforts to establish registries are starting to help; these are limited because registration is, for the most part, voluntary and the registries tend to focus only on protocols, and not to include outcomes.
Access and availability are probably not the only problems. Lack of confidence in the generalization and applicability of the results may demand further study.
William Pryse-Phillips
Full disclosure of data in industry-sponsored clinical research.
The recent editorial in the Canadian Medical Association Journal and other major journals about publication of industry-sponsored trials were discussed under a series of headings. Extensive reference was made to the Glaxo-Wellcome Policy Manual.
Data reporting. Is the objectivity of subject recruitment and data analysis and reporting threatened by the current intellectual environment for clinical research? Not generally. Types of subjects and controls are frequently dictated by regulatory agencies. Also, negative results are less interesting to authors and editors. Evidence that papers supporting safety of a medicine are more likely to have been industry sponsored might encourage journal editors not to discriminate against negative results. Creating an impartial www site for peer reviewed results would help.
Phase 4 trials. "The use of clinical trials mainly for marketing . is misuse of a powerful tool." Though clinical research lacking a valid and relevant outcome is unethical, such trials: familiarize physicians with the new drug under tightly controlled conditions; provide free treatment; and add further knowledge of drug performance.
Design and Conduct of Trials. Should all trial researchers have a substantial say in trial design? Not all investigators have the interest, time or ability to make significant changes to the protocol, but all can decide whether or not to participate. As a minor player, Canada can expect to have only limited influence in major international trials. Also, how can investigators take responsibility for any activities outside their own centres? What power does any investigator have who contributes a few percentage of patients?
Full Data Disclosure. Few investigators seek the full data; most prefer a summary. The complete raw data must be available to all investigators, but a complete dataset is huge. Also, no investigator has rights over results from another centre, and results from one centre are likely to be statistically insignificant and inconsistent with the overall results.
Authorship and publication. It is necessary to differentiate authors from all who participate as investigators. Most trial investigators might be better described as contractors rather than as cutting-edge researchers. The Vancouver Guidelines for Multi-Centre Studies set standards for selecting authors for the publications.
The industry needs to adopt self-regulating policies to prevent decisions not to publish the results of clinical trials.
Do the authors or the study sponsor own the intellectual property of a submitted manuscript? Most of the 'authors' did not write the paper which is about a molecule that they did not discover, but which they supervised within their own centres in a trial that they did not design.
