ETHICAL ISSUES IN THE SELECTION AND USE OF CONTROL GROUPS
There are special ethical concerns in behavioral intervention trials related to the use of control groups because treatments are determined by chance (Street & Luoma, 2002). The arms of the intervention trial must be in clinical equipoise, which is central to the protection of human participants. The equipoise principle states that participants should be assured of receiving the best available standard therapy in any therapeutic study (Schwartz et al., 1997). Over the past several decades, clinical research has documented reduced morbidity and, in some cases, increased survival among medically ill populations receiving a psychosocial intervention (Schwartz et al., 1997). Given these effects, it is increasingly difficult to justify relying solely on a no-treatment control in many behavioral intervention studies. Although no-treatment controls have an appealing simplicity, as discussed earlier, they also raise ethical concerns. The use of these controls may be ethically defensible when the experimental treatment targets a problem without a treatment indication, or when the trial focuses on a population with no immediate risks (e.g., a trial for the prevention of depression; Mohr et al., 2009). However, the use of no-treatment controls may be less acceptable when the trial targets severe disorders, for which effective treatment is both indicated and available (e.g., a trial for patients with major depressive disorder).
Ethical issues are also raised when other control conditions are employed such as placebo controls (Emanuel & Miller, 2001; Schwartz et al., 1997). Where no effective treatment is available, the use of a placebo can be ethically justified. Placebos are also ethical when there exists no permanent harm in delaying active treatment for the duration of the trial. However, in the context of a treatment to prevent or delay death, it is difficult to defend the use of a placebo control. By denying treatment to controls, an investigator might run the risk of “resentful demoralization” on the part of the control group, which may lead them to seek out treatment similar to the active intervention. It may be inappropriate or unethical to ask control participants to avoid using a resource that has been demonstrated to be health enhancing (Schwartz et al., 1997). Even when there are debates about the efficacy of an intervention, and the scientific community remains neutral, ethical issues may still exist. Participants and their care providers may perceive benefit of the treatment when none exists.
There are also ethical issues involved in economic considerations in the design of behavioral intervention trials. Economic factors can influence trial design and the choice of treatment and control conditions. As Schwartz et al. (1997) point out, if resources are scarce, then the trial should be designed for clinically significant and economically feasible outcomes rather than focusing on statistical significance per se. In this case, the investigator might use less expensive, but perhaps less effective, commonly used treatments for a given control condition rather than a no-treatment control. The no-treatment control might be less costly because of lower sample size requirements, but the comparison conditions should be designed according to the treatment and effect sizes that are economically feasible in the real-world setting (Schwartz et al., 1997).
Finally, so-called “cluster RCTs” of behavioral interventions pose special ethical challenges because the intervention is delivered at the level of the group or community rather than at the individual level (Christie, O’Halloran, & Stevenson, 2009). In this case, it may be difficult to ensure individual choice when the intervention is delivered to the group. For example, if a health promotion intervention is delivered to an entire continuing care retirement community, individuals within that community may be exposed to the intervention whether they choose to be or not. Before starting a cluster RCT, investigators should determine if the intervention will actually be delivered to the community as a cluster without the possibility of individual choice. It is especially important for investigators to safeguard the rights and interests of the individual when designing and implementing cluster RCTs.