For randomised controlled trials (RCTs) to yield valuable results, the choice of the control group is critical. Since the disease as well as the intervention effect measures of health outcome, researchers must compare the effects of an experimental intervention on participants assigned to the investigational arm (or arms) of a trial with the effects associated with a control arm. Randomisation is the preferred method for assigning participants to a treatment arm because it increases the likelihood that measured and unmeasured factors influencing the outcome are equally distributed between study arms. A control arm helps ensure that the results of the study reflect the effects of administered interventions and extraneous factors rather than the disease process itself [1].

The use of placebos as controls in randomised controlled trials is the subject of considerable ethical debate. The ethical dilemma is particularly apparent in resource-limited settings and infectious diseases where it hinges upon the issue of “double standards”; there is one moral standard governing research in the developed world and an inferior standard for research in developing countries. A series of placebo-controlled trials on mother-to-child transmission of Human Immunodeficiency Virus (HIV) in the late 1990s brought this to the fore [2]. These studies were conducted in Sub-Saharan Africa at a time when an effective, albeit expensive treatment to reduce this transmission was already available in the developed world. The studies caused public debate due to concerns that double standards of clinical care were being applied to individuals from the developed and developing world, in disregard of the right to equal access to optimal clinical care.

More than twenty years after the controversial HIV trials in Sub-Saharan Africa, researchers and research ethics committees still cannot agree on when placebos should or should not be used. This disagreement was illustrated recently by the debate over the design of Ebola trials in West Africa in which proponents of placebo argued that placebo or “standard of care” arms were permissible, and in fact imperative due to the unavailability of a proven treatment [3, 4]. It was argued that a placebo-controlled trial would produce the highest quality evidence in the most efficient manner, therefore maximizing the probability of saving future lives. However, opponents of placebo argued that standard care would offer little benefit as mortality was as high as 70%, the affected populations were frightened of the progress of the epidemic and it would be highly unlikely that a placebo arm would be acceptable. Another big challenge was that novel agents were administered to Western aid workers without randomisation when they were treated in Europe and the United States [5]. They proposed that instead of an RCT, suitable treatments should be trialled in parallel non-randomised studies, comparing mortality with historical data; an approach which is methodologically weaker and frequently unable to render valid evidence [5–7].

In this paper, we propose six considerations for evaluation of whether the use of placebo is ethical and illustrate their application in a primaquine antirelapse trial for Plasmodium vivax malaria. These considerations evolved from published literature and our experience developing, debating and fine-tuning the study design including having lengthy discussions with ethics committees. The trial, known as the Improving the Radical Cure of Vivax Malaria (IMPROV) study, has been approved by relevant ethics committees in the United Kingdom, Australia and each of the countries where the trial is underway. This double-blinded, randomised non-inferiority trial is currently recruiting patients with P. vivax in Indonesia, Vietnam, Afghanistan and Ethiopia [8]. All patients are treated with either chloroquine or an artemisinin combination therapy (ACT) for the acute blood stages of the infection according to national guidelines. Patients are randomised to three arms for antirelapse treatment (termed radical cure): 7 days 1 mg/kg/day primaquine (total 7 mg/kg), 14 days 0.5 mg/kg/day primaquine (total also 7 mg/kg), or 14 days placebo. The primary objective of the IMPROV study is to determine whether a 7-day high dose primaquine regimen is as safe as and non-inferior to a WHO-recommended 14-day regimen in preventing P. vivax relapse. The hypothesis is that a 7-day primaquine regimen is well-tolerated and non inferior to a 14-day primaquine regimen.

The following paragraphs address our proposed considerations which are: social value; need for a randomised controlled trial and placebo; standards of care; risks of harm due to administration of placebo and harm benefit balance; clinical equipoise; and double standards.

Based on the above considerations, the application of placebo in the IMPROV study is ethically justified on the grounds that there is a big P. vivax burden and a need for a placebo-controlled trial. Furthermore, the current standard of care in most endemic places does not include prescription of radical cure. There is clinical equipoise among the study arms, the risk of harm of being in the placebo arm is that arising from the risk of relapse, and there are no double standards of care between developed and developing countries.

International, regional and national ethics guidelines set out useful general conditions for use of placebo in order to prevent double standards and to guide good practice in clinical research. For example the Declaration of Helsinki and Council of International Organisations of Medical Sciences guidelines state that placebo may be used when there is “no proven intervention” [19], or “no established effective intervention” [1]. We think that the question of whether or not a placebo arm is ethically permissible for a specific study depends on the objectives, population, and contexts of the specific study at hand.

In this paper, a set of considerations were employed to judge whether the use of placebo is justified - in a specific study in a specific context. Although these considerations evolved from a specific trial, our larger objective was to provide a starting point for unpacking the specific arguments for justifying the use of placebo in clinical trials, rather than solely relying on guidelines.

This paper focuses on placebo but these considerations also apply in observational control arms where no treatment is offered. The first is regarding the disease burden, which is related to the social value of a trial [1, 20]. This is an important requirement for any clinical research, not only placebo controlled trials. The second is whether there is a need for a randomised controlled trial and if other study designs are more scientifically appropriate or more acceptable by the community. For instance in some cases, historical controls or observational studies such as those suggested for the Ebola trials could be adopted [6]. These study designs may not yield the most robust data but may be more practical and acceptable by the community in situations of epidemic and fear.

If randomised controlled trials are the right design, the next question relates to the control arm. The default should be the standard of care, about which important questions must be asked. These include the nature of the evidence for the standard of care and if the standard is reasonable in the context. Are there costs, availability, safety, geographical, genetic, or medical differences due to age, gender, pregnancy or comorbidities that render the apparent “standard of care” non-standard?

The most important condition is that patients should not be knowingly allowed to suffer from serious harm if they are in the placebo arm. This condition applies even if there is no proven treatment. For example, it has been argued that placebo should not be used if there is high mortality such as in the case of the Ebola trials [5]. This is a pre-condition to a favourable harm-benefit ratio that renders a study ethically acceptable.

While the above criteria apply to all placebo controlled trials wherever they are conducted, the issue of double standards is particular to low-income settings. As a default, double standards are not ethically permissible unless there are compelling reasons, which are specific to the trial and specific to the context. The threshold for the acceptability of double standards should be set high; double standards must not be a rule but an exception.

For any clinical trial to be ethical regardless of whether a placebo is used or not, it must first fulfil several universal requirements, for example it must be scientifically valid, be responsive to local healthcare needs, must minimise the risks to which the participants are exposed, and must not be exploitative [20]. For the purpose of focus, this paper does not address these or other equally important ethical issues specific to the trial such as the non-inferiority study design, the role of a Data and Safety Monitoring Board, procedures which are in place in case of haemolysis in the primaquine arms, the treatment regimen if patients have relapses, the treatment for G6PD deficient patients, compensation for any increased burdens of being in the trial, post-trial access, informed consent procedures, as well as the quality of the placebo and blinding. These have been partially addressed in the published protocol and are beyond the scope of this paper [8].

In this paper, we present a set of considerations to systematically evaluate the ethics of placebo controlled trials. We propose that similar considerations should be prospectively applied to other placebo controlled trials and observational control arms where no treatment is offered.