Risk assessment in clinical trials: it don't mean an ethical thing if it ain't got that probability ring!

Abstract

Since clinical trials are conducted in a context of inherent uncertainty, the assessment of risk is central to their scientific design and ethical conduct. Substantial evidence points to a pervasive misunderstanding of the concept of risk within the clinical research community that results in risk assessment being one of the least satisfactory parts of trial protocols. At its core, the problem rests with the fact that medicine, unlike the social sciences, economics, and philosophy which now pursue their respective scientific methodologies grounded in a probabilistic paradigm, remains wedded to a deterministic paradigm. However, it is from within the probabilistic rather than the deterministic paradigm that justification for conducting clinical research is to be found. Since prospectively their outcomes, including benefit and harm, are unknowable, clinical trials take place under conditions of uncertainty. In recognizing this, the Belmont Report (Belmont) [1] called for the assessment of risk involved in participating in clinical trials with three goals in mind. The first, for clinical researchers, is to determine whether the research design poses minimal risk of harm to participants. The second goal, for prospective participants, is to help them to decide whether to participate in the trial. The third goal, for institutional review boards (IRB’s), is to enable members to determine whether the risk of harm relative to the likelihood of benefit favors benefit over harm for trial participants. By way of confirmation, the National Bioethics Advisory Commission (NBAC) declared that “risk is a central organizing principle, a filter through which protocols must pass” [2] if scientific validity, patient participation, and IRB approval are to be considered reasonable. But given the centrality of risk assessment to scientific validity, a patient’s informed consent to participation and IRB approval, why in a 2002 study would up to two-thirds of respondents find that trial protocols provide insufficient information regarding the likelihood of both risks and benefits [3]? Why is it that so few protocols pass through the NBAC filter and in the way envisaged by Belmont? Closer reading of Belmont also reveals that risk assessment requires a specific understanding of the concept of risk which: refers to a possibility that harm may occur. However, when expressions such as ‘small risk’ or ‘high risk’ are used, they usually refer (often ambiguously) both to the chance (probability) of experiencing harm and the severity (magnitude) of the envisioned harm [1, pg7]. (Cont.)