Minimal Risk in Pediatric Research: A Philosophical Review and Reconsideration

ABSTRACT

Despite more than thirty years of debate, disagreement persists among research ethicists about the most appropriate way to interpret the U.S. regulations on pediatric research, specifically the categories of “minimal risk” and a “minor increase over minimal risk.” Focusing primarily on the definition of “minimal risk,” we argue in this article that the continued debate about the pediatric risk categories is at least partly because their conceptual status is seldom considered directly. Once this is done, it becomes clear that the most popular strategy for interpreting “minimal risk”—defining it as a specific set of risks—is indefensible and, from a pragmatic perspective, unlikely to resolve disagreement. Primarily this is because judgments about minimal risk are both normative and heavily intuitive in nature and thus cannot easily be captured by reductions to a given set of risks. We suggest instead that a more defensible approach to evaluating risk should incorporate room for reflection and deliberation. This dispositional, deliberative framework can nonetheless accommodate a number of intellectual resources for reducing reliance on sheer intuition and improving the quality of risk evaluations.

KEYWORDS:

• Ethics and public policy
• history of research ethics
• human subjects ethics
• human subjects regulation and oversight
• pediatric research ethics
• research ethics

Disclaimer

The views expressed in this article are solely those of the author (RMN) and do not represent the views of the Food and Drug Administration.

Notes

¹ While the term “research” appears in the title of the U.S. federal regulations pertaining to human-subjects research (45 CFR 46 and 21 CFR 50), and while the category of “minimal risk” refers to “research,” other categories of the pediatric regulations use the term “intervention.” The use of the term “intervention” highlights that the unit of analysis for ethical and regulatory purposes is not the research project as a whole, but rather each intervention that is to be employed in the course of the research.

² Though one might try to distinguish “low risk” from “minimal risk,” there is no indication in Research Involving Children that the National Commission thought of the two terms as distinct risk levels. Instead, they seemed to use “low” and “minimal” risk interchangeably and more-or-less synonymously—though “minimal risk” is the term that was chosen and defined in the regulations. For the purposes of this article, we treat the terms synonymously.

³ To clarify, our goal is not to argue whether the current regulatory structure is ethically defensible. Rather, our goal is to ask how the regulations should be interpreted and applied, consistent with how they are currently written, and what the risk categories contained in them are supposed to represent. For example, it might be argued that the category “a minor increase over minimal risk” should be eliminated because it is morally indefensible, since there is no good reason why we should expose children with a “condition or disorder” to greater risk than healthy children in research (see Ross 2003; Kopelman 2004b; for discussion). This question does not relate to the interpretation of “a minor increase over minimal risk” but instead to whether we should have the category in the first place.

⁴ DeGrazia (2003) defines a considered judgment as follows: “I suggest that considered judgments, which have an initial credibility (though not infallibility), are moral judgments that are adequately informed in terms of both factual information and relevant moral alternatives, stable even when one is not under pressure to accept them, free from personal conflict of interest, and the like” (p. 221).

⁵ To wit: “the scope of parental authority routinely covers a child’s participation in many activities in which risk is more than minimal, and yet benefit is questionable” (Research Involving Children 1978, p. 2110). And: “The permission that parents give for children’s participation in research can be accepted as an exercise of their general role, as caretakers, to guide decisions affecting their children’s lives and activities” (Research Involving Children 1978, p. 2110).

⁶ Said another way, we should distinguish between a definition that is normatively inadequate and a definition that fails entirely to recognize the normative character of a term. The Commission’s discussion of minimal risk gives no indication they thought that just because something is the case (children are exposed to some risks in daily life) it therefore ought to be the case (they should be exposed to similar risks in research); that is not the logic of their discussion.

⁷ One minor complication to this point is that the “relative versus absolute” debate seems to hinge primarily on considerations of justice, rather than considerations of risk magnitude. However, we believe that this is best viewed as a second-order issue in the interpretation of “minimal risk,” since this issue arose only because of ambiguity in the wording of the risk categories and does not concern their basic meaning. Further, the issue here is not simply the abstract one that different children would be exposed to different levels of risk absent a good reason. Rather, a significant part of the concern at issue is that some children might be exposed to risks that are too high to rightfully be considered “minimal” or “low”—that without additional qualification, “the risks of daily life” are not a good substitute for risks that are minimal in magnitude.

⁸ These latter features are already supposed to be taken into account when IRBs consider whether a non-minimal-risk intervention might qualify as a minor increase over minimal risk. For minimal risk studies, IRBs might think about whether research risks are “reasonably commensurate” (45 CFR 46.406) with a child’s prior experiences, at least if a relative interpretation of minimal risk is ever appropriate. As well, once a study receives IRB approval and is recruiting participants, parents would need to think about how their individual child would fare in the research when making an enrollment decision.

⁹ The model we propose shows some broad similarities to the coherence model of ethical justification, often described as “reflective equilibrium.” Classically, the equilibrium is between our judgments about the ethical principles that should guide conduct and our judgments about what to do in specific cases: each informs and can revise the other. Our proposed approach is not principle-driven: we do not think that defensible judgments about minimal risk can be modeled as the output of specific ethical principles. Thus, this type of equilibrium would not apply. However, coherence can be thought of more broadly as involving ethical judgments that are stable under critical reflection, well-informed by empirical data, free of bias, and consistent with other ethical beliefs. These are also features of the dispositional model, and as already discussed the ideal conditions identified by the dispositional model might be justified only by appeal to reflective judgment. Further, when thinking about minimal risk as a judgment of acceptably low risk, the normative component of such a judgment (i.e., the “acceptable” part) would seem to rest principally upon reflective intuition, since as argued earlier principle-based justification does not seem possible at this level of specificity.

¹⁰ To wit: “Deliberation must end in a decision, but deliberative democracy does not itself specify a single procedure for reaching a final decision. It must rely on other procedures, most notably voting, which in themselves are not deliberative” (Gutmann and Thompson 2004, p. 18).

¹¹ One notable exception is Kimmelman (2004), who recommends a renewed focus on IRB composition and procedure so as to make IRBs more “nonarbitrary and rigorous, but also inclusive and comprehensive of different views of risk” (Kimmelman 2004, p. 387). This suggestion is entirely in line with our philosophical arguments here.

¹² The increasing reliance on regional IRBs makes it all the more important that the deliberative process be robust and transparent, insofar as a greater number of parties will be affected by the decisions of a single committee.