Policy equipoise and interventional superiority

ABSTRACT

According to the norm of policy equipoise, it is permissible to randomly assign participants to two or more interventions in a public policy randomised controlled trial (RCT) when there is meaningful uncertainty among the relevant expert community regarding which intervention is superior. While this norm is gaining traction in the research ethics literature, the idea of interventional superiority remains unclear. Is one intervention superior to another if it is reasonably expected to more effectively realise one outcome of interest, even though there is uncertainty regarding other outcomes of interest? Or, must an intervention be reasonably expected to more effectively realise all outcomes of interest? I address this question in this paper. My aim is to develop and defend an account of interventional superiority for policy RCTs that are authorised, funded, or conducted by government institutions. I defend the greatest value view, according to which one intervention is superior to another if and only if it is reasonably expected to more effectively realise a set of outcomes with greater value.

KEYWORDS:

• Research ethics
• equipoise
• randomisation
• policy equipoise

Acknowledgements

This paper has benefitted greatly from the comments and questions of many people. First, I am grateful to three anonymous reviewers for their insightful comments and criticism, and to Alexandra Avdeenko for her excellent editorial guidance. Second, I am grateful to participants at the 2022 New Scholarship in Bioethics Annual Symposium (Kirstin Borgerson, Michael Garnett, Joseph Millum, and Marika Warren) for their detailed, thoughtful feedback. Finally, I thank audiences at the 2022 Annual Meeting of the Philosophy, Politics, & Economics Society Meeting and the 2022 Annual Conference of the American Society of Bioethics and Humanities.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. For discussion of the obligations of different types of research funders, see (Pierson and Millum 2018) and (MacKay 2015).

2. In a paper on clinical equipoise, Fred Gifford (2000, 423) draws a similar distinction between ‘one-dimension-considered’ equipoise and ‘all-dimensions-considered’ equipoise.

3. Thanks to Joseph Millum for raising this point.

4. This analysis also applies in cases where researchers are confident in the direction of an intervention’s impact on an outcome, but uncertain regarding whether one intervention is superior to another. To consider the case in Table 3, suppose A is reasonably expected to have a positive impact on Z, but the extent of this impact is uncertain. This is surely relevant to the question of whether policymakers should implement A or B, but I think it is permissible to conduct an RCT comparing A and B since there is still uncertainty whether A or B will more effectively realise Z. It might turn out that A has a small positive impact on Z while B has a large positive impact. If policymakers permanently implement A, they may end up implementing an inferior intervention. Thanks to an anonymous reviewer for raising this possibility.

5. Thanks to an anonymous reviewer for raising this objection.

6. Thanks to Alexandra Avdeenko for raising this objection.

Additional information

Notes on contributors

Douglas MacKay

Douglas MacKay is an Associate Professor in the Department of Public Policy at the University of North Carolina at Chapel Hill. He is also a Core Faculty Member of the Philosophy, Politics, and Economics Program and the Center for Bioethics. His current research focuses on the ethical dimensions of public policy and public policy research ethics. He is the co-editor of the recently published Oxford Handbook of Research Ethics.