Ethics of Medication-Free Research in Schizophrenia

Bola's study highlights two important points about the ethics of medication-free research in early episode schizophrenia.¹ First, the question of whether “biological toxicity” results from short periods of medication-free research in early episode schizophrenia is an important, though very narrow, question in the debate over ethics of medication-free research in schizophrenia. Second, even on this narrow question, the paucity of good data is striking.

Based on a meta-analysis of 623 participants in 4 randomized controlled trials and 3 quasi-experimental studies otherwise meeting his selection criteria for study inclusion, Bola concludes that “in the absence of substantive evidence of long-term harm from short periods of medication-free research in schizophrenia, a categorical prohibition of medication-free research in early episode schizophrenia on ethical grounds of harm to human subjects should probably be reconsidered.” Instead, he suggests applying Emanuel and Miller's “middle ground” approach.² Following Bola's suggestion, we illustrate the important, though limited, role of the biological toxicity question in the debate over the ethical acceptability of medication-free research in schizophrenia and comment on the implications of his findings for the ethical debate.

Emanuel and Miller's “middle ground” approach² involves assessing individual research protocols for the ethical acceptability of brief medication-free periods, using three criteria: (1) a compelling scientific rationale justifies the study and its design; (2) research participants experiencing deliberate study-related medication-free periods “should not be substantially more likely than those in the active-treatment group to die; to have irreversible morbidity or disability or to suffer other harm; to suffer reversible but serious harm; or to experience severe discomfort”; and (3) trials that meet these methodological and ethical criteria must nevertheless institute safeguards to minimize risks of harm.

Compelling Scientific Rationale

The first criterion for the middle ground approach is that there must be scientifically compelling reasons for even short-term medication-free periods in order for the study to be ethical. At issue here is who decides what reasons are “compelling enough.” For example, there are many types of research questions that would require a medication-free arm to answer, such as the scientific and clinical question of what subgroup of individuals experiencing a first psychotic episode may experience remission without use of medications or may safely discontinue medications after only a brief exposure.³ How and who decides if these research questions are sufficiently valuable to ethically justify a study requiring a medication-free arm remains to be answered. Similarly, if a study is initially acceptable, what type of interim analyses and additional data should be required to justify continuing medication-free periods in ongoing studies or competitive renewal of grants?

Furthermore, some studies in which some participants are deliberately left medication-free do not directly test hypotheses related to the scientific or clinical utility of being off medication. Are studies aimed at getting FDA approval for marketing or for gaining market share sufficiently compelling scientifically to justify medication-free arms? In some cases (e.g., studies of novel anti-tumor agents or novel anti-epileptic agents), society generally says no, while for others (e.g., studies of a novel anti-nausea compound during cancer chemotherapy), society may say yes.

Additionally, some studies suggest that medication-free groups can exert significant “design effects” on research outcomes. For example, a recent meta-analysis suggests that the presence of a placebo group in a trial is associated with higher dropout rates in both arms, potentially affecting generalizability,⁴ while another finds that antipsychotic medications show higher rates of improvement in active-controlled trials than the same dosage shows in placebo-controlled trials, potentially affecting the ability to make accurate comparisons across studies.⁵ How, and if, such data ought to be taken into account in the ethical decision-making regarding research with medication-free arms remains to be answered.

“No Serious Harm or Significant Discomfort”

Bola's study most directly addresses this issue. Nevertheless, with regard to the “biologic toxicity” hypothesis, no evidence of harm does not necessarily constitute sufficient evidence of no harm. While it is interesting, and important, to illuminate the poor quality of data upon which the “biologic toxicity” hypothesis gained favor in schizophrenia, doing so does not invalidate the hypothesis. Kindling effects and possible neurotoxicity related to stress and distress remain a concern in many brain disorders, and while the available data do not support such effects in schizophrenia, neither are they of sufficient quality to exclude it as a possibility. Nevertheless, that these longer-term safety issues remain to be resolved should not distract investigators and IRBs from focusing on the importance of minimizing shorter-term risks, as discussed further below.

While the question of “biological toxicity” is important, it is only one of many potential harms or significant discomforts that are of ethical concern. The potential for “biological toxicity” was never the only ethical ground upon which medication-free periods in schizophrenia research stimulated concern. For example, some individuals argue strenuously that “dignitary harms” are as important as physical harms. While there is disagreement as to what constitutes “dignitary harm” and how much weight it should carry in ethical analyses, one thing is clear: data such as these do not speak to dignitary harms and would do nothing to assuage concerns over dignitary harms.

Another important ethical issue is that of adequate informed consent. Many studies across several types of illnesses and research designs show that we still are not good at ensuring adequate informed consent for all research participants.⁶ Research involving individuals with schizophrenia seems to do no better, but no worse.⁷ Finally, Bola's data do not speak to the issue of the potential suffering caused by medication-free periods in research. Some discount the importance of the suffering caused by such research as no different than that experienced by the vast majority of individuals with schizophrenia who stop medications for a variety of reasons, particularly early in the course of illness, and because medications themselves contribute to a certain amount of discomfort and suffering (and biological harm). However, this does not erase the fact that significant discomfort and suffering can result from medication-free periods in research, even brief ones. Allowing such discomfort and suffering requires honest recognition that it exists and must be mitigated to the extent possible. Furthermore, the potential of experiencing discomfort and suffering while participating in research must be discussed with potential research participants, including discussion of the interventions that can and cannot be used within the context of the study to alleviate discomfort, and reiteration that participants can exit the study whenever they want.

Provisions to Minimize Risks of Harm

While the second criterion allows for some harms and discomforts in research, the third criterion reminds researchers that, nonetheless, these must be minimized. However, there are no predetermined acceptable thresholds. Questions remain as to how many episodes of illness exacerbation should be allowed. How long an episode? Or is the better question, how much suffering should be allowed? Should a limit be imposed, like with radiation exposure, allowing so many numbers of episodes or quantum of suffering? Data allowing consideration of such thresholds do not currently exist. Nevertheless, grappling with the implications of Bola's study, along with other empirical research sparked by ethical questions, points to the types of studies and data that could be helpful in moving the ethical analyses forward.⁷

In light of this discussion, what is an ethically appropriate way to move forward in the ethical analysis of medication-free research? The middle ground criteria are a good start, recognizing that as we gain experience and collect data that address these three criteria, the thresholds for what is considered ethically acceptable within each will shift. In the meantime, we suggest that consideration be given to forming some type of data repository into which core data elements (which would need to be defined) can be deposited from all research participants who experience medication-free periods in schizophrenia research to develop the evidence base necessary to ascertain what the levels of risk for biological and non-biological harms are, and what, if any, potential benefits there might be (with full recognition of the practical and ethical challenges inherent in such an endeavor). Nevertheless, if such a repository is constructed in a way that periodic follow-up of participants could be done, empirical data of sufficient quality to better inform more aspects of the ethical analysis of medication-free research might ultimately be possible.