Just and Ethical Study Designs in the Opioid Epidemic

As intervention researchers, what do we owe our most vulnerable, most socially marginalized study subjects? These men and women confront the ravages of the opioid epidemic, HIV, and other threats to their lives and their well-being. Facing these profound threats, our study subjects share their lives and their data with us, often in the most intimate and painful domains. What do we owe them in return?

Such chastening questions face many of us in medicine, social work, and public health. Of course, we owe all study subjects protections required by our university Institutional Review Boards, and by the legal and regulatory obligations that govern NIH and other scientific grants. We also have myriad pragmatic reasons—both scientific and self-interested—to provide value to all study subjects when follow-up is crucial, such as when we fear threats to study validity arising from differential follow-up across treatment arms in complex studies.

Is that enough? Do we owe people more than that? ^1,2 These questions are particularly pressing given the lethality of the opioid overdose epidemic, where many question the very idea that one can conduct an ethical randomized trial.

Randomized trials that involve a placebo/usual care arm understandably induce moral injury among study staff—some of whom might be tempted to defeat randomization to assist subjects assigned to a treatment arm likely to produce less immediate benefit. The same structure fosters reluctance among social service agencies and other partners, who are understandably ambivalent about collaborating in randomized trials that provide greater benefit to researchers, and to the society at-large, than they provide to many study subjects. Doolittle and Traeger (1990) surveyed service providers to understand widespread refusal to participate in a national job training trial. Representatives of more than 60% of contacted job training centers cited ethical and public relations challenges of randomized designs.³ (See also Heckman and Smith, table 2.¹)

We witness such reluctance among our colleagues. A police leader noted that his department was going to roll out an officer-initiated diversion trial.  They sought a research evaluation partner, but chose not to engage this partner when they learned that the researcher sought to conduct a usual-care trial. That leader was prepared to offer a valuable intervention to everyone, and thus concluded that a usual-care trial would cause harm.

As the social and political legitimacy of randomized trials are questioned, it’s important to recognize both the limitations and the necessity of randomized trials. In studying policy changes and other structural questions, nonexperimental designs are sometimes more reliable and generalizable than narrow experiments.^4–6 Nonexperimental designs are sometimes the only feasible or ethical path to study fraught questions. At the same time, nonexperimental designs often lack the academic and political credibility,  simplicity, or internal validity that accompany a solid prospective randomized design.⁷

Recognizing these challenges, an ethically informed trial design might address such questions with greater intentionality and greater humanity. Within this framework, the design of any study involving vulnerable subjects should include attention to one simple but difficult question: What can we do that brings genuine benefit to every study subject?

By genuine benefit, we mean something more than a McDonald’s meal or even a generously valued gift card in return for completing a survey and providing permission to access personal data. Placebo trials violate core principles of beneficence and social justice when they intentionally leave predictable and severe threats to individual well-being unaddressed. Placebo trials are particularly problematic when researchers fail to provide feasible assistance that could readily be incorporated into a study design.

Our team has tried to institute policies to address these questions. When we design a new trial or submit new grants, we ask: What do we owe every study subject? How can we improve or safeguard their lives in ways that are consistent with our scientific aims and constraints? Would available options, such as crossover designs or other strategies, allow us to achieve our scientific goals while providing greater participant benefit?

As colleagues in an NIH-funded multi-site randomized trial that explores the impact of interventions that link people leaving jails and prisons with treatment and harm reduction interventions,⁸ we sought to address the above ethical questions by avoiding a control or “usual care” arm. We have an overdose education/ naloxone distribution (OEND) arm, and a case management/peer recovery coach (CM/PRC) arm that receives OEND resources plus these additional resources. Given evidence of high mortality immediately upon discharge from carceral settings, we provide naloxone education,⁹ local service resource materials, and a naloxone packet to everyone. We provide incarcerated participants with OEND resources with their personal possessions when they leave jail or prison. Our primary outcomes concern linkage to treatment resources. We hypothesize that members of our CM/PRC arm will be more likely to engage evidence-based treatment. We do not expect (and do not wish) to see treatment-arm differences in access to basic harm reduction resources.

This approach brings genuine challenges, costs, and benefits. Narrowing contrasts between treatment arms for ethical reasons may weaken statistical power to detect small treatment effects. We may not be able to study some questions—e.g., the value of CM/PRC supports vs. usual (inadequate) care. Such constraints complicate study design and sometimes raise study or intervention costs.

However, this approach brings pragmatic benefits, too. We hope that people who use drugs will be more willing to participate in our studies, knowing that they will benefit whichever way the treatment-assignment coin is flipped. Ethically informed designs improve morale and pride among our staff and partners, who may lend more enthusiastic support to randomized designs that attend to such concerns beyond what’s required for human subject approval. Particularly in the post-George Floyd era, our staff members are more willing to ask the uncomfortable questions—and won’t stick around if we lack good answers.

As it happens, this approach may have rescued our study. A key partner wrote us a support letter. Their research arm then withdrew support because—like many other clinicians and social service professionals³–they didn’t believe placebo/usual care trials were ethical for individuals in their care. After clarifying that both study arms included OEND, i.e., that there was no “placebo” arm, the partner rejoined our study.

The opioid epidemic is an extreme case that underscores more general dilemmas. It’s a valuable lens that illuminates some of our own self-serving oversights as researchers. These same oversights arise in other studies that involve vulnerable subjects who often provide us with critical data, and who may need our help even more deeply.

We owe our subjects the values expressed in the Belmont declaration and other obligations covered in our required Collaborative Institutional Training Initiative (CITI) human subjects training courses: Respect for Persons, Beneficence, and Justice. A fourth principle also deserves greater attention in its practical execution: Gratitude. Expressing this gratitude—even more importantly, treating our study subjects with genuine beneficence and social justice–requires more than placebo—with or without an accompanying gift card.

References

[1] Heckman JJ, Smith JA. Assessing the Case for Social Experiments. Journal of Economic Perspectives. 1995;9(2):85-110.

[2] Caplan A, Annas G. Study of needle exchanges puts politics ahead of health. Washington Post. February 11, 1997.

[3] Doolittle F, Traeger L. Implementing  the National JTPA Study. New York, NY: Manpower Demonstration Research Corporation;1990.

[4] Deaton A, Cartwright N. Understanding and misunderstanding randomized controlled trials. Soc Sci Med. 2018;210:2-21.

[5] Imbens G. Understanding and misunderstanding randomized controlled trials: A commentary on Deaton and Cartwright. Soc Sci Med. 2018;210:50-52.

[6] Muller SM. Masks, mechanisms and Covid-19: the limitations of randomized trials in pandemic policymaking. Hist Philos Life Sci. 2021;43(2):43.

[7] Bannerjee A, Duflo E, Kremer M. The Influence of Randomized Controlled Trials on Development Economics Research and on Development Policy In: Basu K, Rosenblatt D, Sepulveda C, eds. The State of Economics, the State of the World   Cambridge, MA: MIT Press; 2020.

[8] Pho M, Erzouki F, Boodram B, et al. Reducing Opioid Mortality in Illinois (ROMI): A case management/peer recovery coaching critical time intervention clinical trial protocol. J Subst Abuse Treat. 2021;128:108348.

[9] Binswanger IA, Blatchford PJ, Mueller SR, Stern MF. Mortality after prison release: opioid overdose and other causes of death, risk factors, and time trends from 1999 to 2009. Ann Intern Med. 2013;159(9):592-600.

About the Authors