Risk, responsibility, and choice in research ethics

In health-care settings, routine enquiry about experiences of trauma is good practice. Yet in research studies, whether to address trauma and how is contested. Some studies prioritise inclusive samples and questions about people's lived experience. Others avoid potentially retraumatising topics and exclude people considered too vulnerable to participate.

Safeguarding is essential, but when does protection become paternalism, and who gets to decide this? These questions matter and contribute to why we know so little about the extent and nature of domestic violence and of self-harm during the COVID-19 pandemic.¹

In 2000, Jon Nicholl wrote that it was immoral for ethics committees to have become barriers to ethical research that could help to improve health care.² Although the ethical review process has since become more efficient, Nicholl highlights a still relevant tension: how can we balance maximising the benefit of research to society while minimising risk of harm to the individuals taking part?

With the onset of the COVID-19 pandemic, face-to-face survey fieldwork around the world largely ceased (and 18 months later, few surveys have fully resumed). As new data needs emerged, all parts of the research process moved faster.³ Research that was recognised as a priority for the COVID-19 evidence base was given streamlined permissions, including expediated ethical reviews. However, this change in pace also contributed to a fall in lived experience involvement. Before the pandemic, patients and the public were involved in 80% of the research reviewed by the UK Health Research Agency. In March, 2020, public involvement was 22%. A related effect was a pause on research on specific topics, such as research asking probability samples of the general population about experiences of domestic violence or self-harm, and research focused specifically on affected individuals.¹

Although university and other ethics committees convened more regularly and processed applications more swiftly than before the COVID-19 pandemic, this came with increased risk aversion. Researchers avoided asking for approaches that were anticipated to maybe cause delay. Committee members were understandably concerned about approving research on sensitive topics, given the reduced ability to direct participants to curtailed and remote services or support. Since the start of the pandemic, few general population surveys have been permitted to ask about violence, abuse, or self-harm. This omission continues to have substantial consequences for the evidence base in England, UK, and elsewhere, serving to hide the scale of harm, and preventing people in need, including victims and survivors, from being heard.

As harm to participants is unethical, might exclusion from research also be considered unethical and an epistemic harm? Some standard mechanisms for protection, such as requiring participants to sign quasi-legal documents stating that their consent is full and informed, might serve to protect researchers, data guardians, and institutions more than participants. People who prefer not to sign such declarations are often excluded from research, constituting a hermeneutic injustice in itself. Those participants deemed too vulnerable—or too difficult—to ask might also be excluded on the basis of what others consider to be in their best interests. Even if participants can be informed about a study, that questions can be skipped, and that they might withdraw at any time, decisions about their fitness to be asked are pre-emptively made by a remote external regulatory body.

What assumptions are made about competency and protection when research about domestic violence or self-harm is not approved? We know, for example, that assumptions about vulnerability and victimisation intersect with inequalities related to ethnicity, gender, sexual identity, and age. Women have historically been left out of research,⁴ as have children⁷ and older people. Bayer and Tadd's study of ageism in research found that “of the 155 studies that were of relevance to elderly people, over half had an upper age limit that was unjustified” with none of these needless exclusions challenged by the ethics committees.⁶ Protection against disclosure is often cited when information on ethnicity and sexual identity are removed from or aggregated in archived datasets, restricting analyses.

Potential harms identified by ethics committees are not fictional.⁷ There are very real, but often surmountable, challenges. Balance is needed.⁵ For example, even though surveying all household members about domestic violence could alert a potential offender that a disclosure has occurred, we reduce this risk by including such questions only on surveys asked of one household member.⁸ Additionally, evidence suggests that rather than increasing risk, acknowledging and talking about suicide in research might reduce suicidal ideation and lead to improvements in mental health in treatment-seeking populations.⁹ Further steps to mitigate risk include training interviewers to manage psychological distress, consistently monitoring participants' emotional reactions, providing frequent breaks, debriefing, and providing information on available psychological or social services.¹⁰

The topics and people excluded from research cannot be counted or represented, and information on their experiences are not available to inform policy or services. Who is protected when such research is not allowed?

AJ reports grants from Medical Research Council. KB reports grants from the National Institute for Health Research, the Barts Charity, and the McPin Foundation. All other authors declare no competing interests.