How should trial teams make decisions about the proportions and diversity of the ethnic groups in their trial?

The need for trialists to include diverse ethnic groups in their trials is widely recognised [1, 14, 16, 18,19,20] and sometimes mandated [21, 22]. The start of designing an ethnically diverse trial is being clear about which ethnic groups need to be in the trial to answer its research question and at what proportion. As our panels found, these are not easy questions to answer.

Indeed, a limitation of our work is that we were generally unable to say in concrete terms which ethnic groups should have been involved in the trials we discussed or at what proportion. With two exceptions (the two maternal and infant health trials), our panels were not discussing trials they knew from personal experience. Instead, panel members depended on the summaries we provided and their own knowledge and experience. It is possible that working directly with trial teams may have led to different results. That said, in both cases where we had trial team members in our panel, the final target suggested was the census-based default that became part of our Recommendation 6. The full reports we used for the trials, mostly substantial NIHR reports, did not generally present detailed discussion of how ethnicity affected trial design, although PROPELS, one of the diabetes trials, was an exception [23]. This suggests that having more trial team members on our panels would have been unlikely to change our conclusions or recommendations.

Policy initiatives to support implementation

The eight recommendations shown in Table 2 will, if implemented, help to make trials more ethnically diverse. The rub is the qualifier—if implemented. Making recommendations is easy; implementation is where things get difficult. Trial teams, both in academia and industry, can start using the recommendations in Table 2 immediately. Some, such as not worrying about the practicalities of how the team will recruit and retain people until after you have identified the groups you need (Recommendation 5) or thinking about ethnicity when choosing trial sites (Recommendation 8), are relatively easy to implement. Others, such as the need for more diverse trial teams (Recommendation 1), are system-level challenges, although research teams need to have this in mind when training and hiring staff.

We, the wider Trial Forge team, and others are disseminating these ideas to trialists and others, most recently at the 2024 International Clinical Trials Methodology Conference (https://ictmc.org), including through KK’s keynote address. However, this type of dissemination is unlikely to be enough. Below, we outline five policy initiatives that we think would provide an environment in which our recommendations would have a greater chance of improving the ethnic diversity of trials. Without these initiatives, change will be slow and the ability of trial teams to make their trials ethnically diverse will be hampered. Some of the policy initiatives are already underway in whole or part in some jurisdictions, giving reason for optimism. However, implementation of these initiatives needs to be accelerated. It should also be noted that while we focus on ethnic diversity, we are confident these policy initiatives will support greater diversity in trials more generally.

Policy 1: Financial signalling by funders

To generalise, many ethnic groups have a low level of trust in trials and research, and trial teams have a low level of knowledge in how to engage effectively with people in these groups. Given this, everyone involved in the process of funding and delivering trials needs to accept that ethnically diverse trials will take longer and cost more than trials that are less inclusive. This may change as trust increases and evidence about how to engage effectively improves.

Funders need to clearly signal to trial teams that the time and costs needed for greater ethnic diversity are not only accepted but also expected. The expectation part is important: funders need to signal that greater ethnic diversity is not optional, but a requirement. Public acceptance by the funder of the greater time and cost this entails is a necessary consequence of the requirement.

An exemplar of the importance of signalling is support by NIHR, a UK funder, for Studies Within A Trial (SWATs) [24]. These evaluations of trial process alternatives were largely limited to a few enthusiasts in the UK until NIHR signalled the wider need for SWATs by offering specific funding of £10,000 and, more recently, £30,000 per SWAT. Other funders such as the Irish Health Research Board and the Canadian Accelerating Clinical Trials program signal their encouragement through similar schemes, the latter funder offering nearer to £57,000 for a SWAT. Explicit signalling is needed to give trial teams the confidence to take on the time and costs of making trials more ethnically diverse.

Policy 2: Health and social care providers need to routinely collect diversity-related data, and these data need to be made available to trial teams

The first question any trial team designing an ethnically inclusive trial asks is how is disease prevalence, severity, and progression spread across different ethnic groups.

Finding data to answer this question is hard. For example, our panel discussing PROSPER (https://www.isrctn.com/ISRCTN35358984), a UK trial of exercise to prevent shoulder problems in patients undergoing breast cancer treatment, wanted data on the ethnicity of people who get shoulder problems during breast cancer treatment. The PROSPER trial team did not mention ethnicity when discussing shoulder pain [25] and neither did a systematic review of shoulder and arm pain in patients after breast cancer treatment [26]. In a country as diverse as the UK, with around 20% of the population being part of an ethnic minority group, it seems unlikely that the 8% ethnic minority involvement in PROSPER reflects the clinical population of people with shoulder problems during breast cancer treatment. PROSPER is not a special case: our panels struggled with limited or lack of data for all the trials we discussed.

The UK Health Data Research Alliance Ethnicity Coding Standard Special Interest Group, which KK co-chairs has already called for ethnicity data and other determinants of health to be routinely and consistently collected across the health and social care sectors [27]. We agree—collecting these data needs to become standard policy. Anonymised versions of these data then need to be made easily available to those planning and commissioning trials; otherwise, trial teams will be designing in the dark.

Which data to collect is a non-trivial question, and groups including the UK Health Data Research Alliance are working to answer it. A reasonable starting point is the PRO EDI list of participant characteristics [28], a core information set developed by ST, DD, and others for anyone wanting to describe the characteristics of a population to support equity, diversity, and inclusion judgements.

Policy 3: Funders and other policymakers need to signal that ethnic diversity in trials leads to more useful science

Ethnic inclusion targets are essential; we suggest using ranges (e.g. Black British people: 5% to 8%) as a good way to present them. But targets generate a thorny question: what should we do if targets for some groups are being missed, while those for others are being exceeded?

This is a question about who brings most scientific value to the trial. Trial teams and funders need to accept that the answer in many cases will be to halt recruitment of those groups where targets are being exceeded to allow time for other groups to take their place in the trial.

There is precedent. ILANA, a UK human immunodeficiency virus study, had targets of 50% women, 50% ethnic minority, and 30% over the age of 50 and used recruitment caps to enforce these targets [29]. Moderna, a pharmaceutical company, deliberately slowed recruitment of white people during COVE, a US mRNA-1273 COVID-19 vaccine trial, to allow recruitment of under-served groups, especially Black and Hispanic/Latinx people [30]. It did so because ‘Moderna and the US [United States] government understood that achieving a diverse study population was essential to ensure a more robust and representative body of clinical knowledge’ [30].

The policy of financial signalling mentioned above should acknowledge that decisions to prioritise the involvement of some people over others is appropriate where greater ethnic diversity improves the scientific and decision-making value of the trial. Like many things in trials, this is a judgement, but current research policy from major funders and regulators explicitly highlights the importance of greater involvement of ethnic minority individuals in trials [21, 22, 31].

Moreover, as part of the approvals process, approval bodies such as ethics committees should review a plan submitted by the trial team that outlines the characteristics of people who will be in the trial and why, what the recruitment and retention targets are by participant group, how targets will be monitored, and what will be done if progress against targets is not as expected. This policy is already coming in the UK [21] (and to which ST contributed) and the US Food & Drug Administration, a regulator, started doing this in 2022 [22].

A table of who should be in the trial compared to who is being recruited and retained should be a mandatory part of progress reports to funders, trial steering committees, and data monitoring and ethics committees. In all cases, action from the trial team would be expected where important differences appear between who should be in the trial and who is in the trial.

Journals publishing trial protocols should mandate that the protocol include a table describing the expected characteristics of people that will be recruited for the trial and why. Journals publishing trial reports should ask for that table to be updated with who was recruited and retained. There should be an expectation that results will be interpreted by considering any differences. The New England Journal of Medicine started asking for something like this in 2021 [32].

Finally, to support interpretation across trials and to avoid research waste, the reporting policies above should aim for a high degree of consistency regarding the core set of characteristics that trial teams need to collect and report. The PRO EDI table is, we think, a reasonable starting point [28].

Policy 4: Funders need to put more money into evaluation

The ability of trial teams to use evidence-informed recruitment and retention strategies to support ethnically diverse trials is hamstrung by there being almost no robust evidence on which to base those strategies [33, 34]. The current evidence on how to do ethnically inclusive trials is woefully inadequate and this urgently needs to change.

Some of us have previously suggested that trial funders should put 10% of their funding into applied methodology research and supporting infrastructure [35]. Lack of funding for methodology research, dissemination, and infrastructure is leading to research waste [35,36,37,38], of which insufficiently diverse trials are part.

Trial funders need to commission and support the coordination of research that gives trial teams evidence with which to answer their questions about how to achieve an ethnically diverse trial. We would prioritise work evaluating trial processes (especially recruitment and retention [39, 40]) that aim to increase involvement of specific ethnic groups and which are designed around known barriers and facilitators to the involvement of these ethnic groups in trials [18, 19, 41]. Evaluation should also include testing the impact of policies and recommendations such as ours: what difference do they really make? Once high certainty evidence is available, funders and national research infrastructure should actively facilitate its dissemination and use.

Policy 5: Good Clinical Practice (GCP) training should cover ethnic diversity in trials

Sponsors require almost all those involved with trials, and all those with participant-facing roles, to have mandatory research training, especially GCP training. For trials of drugs, GCP training is not just mandatory but a legal requirement in most jurisdictions.

GCP training should be modified to include material on the scientific importance of ethnic diversity in research, what can be done to support it, and what might make it less likely. It should cover the need for cultural competency training for those designing research, and those with participant-facing roles, essential diversity training that others have also highlighted [15, 16]. Some of this may be signposting, for example to sources of evidence on how to design inclusive retention strategies. We have for example started to do this in Aberdeen through discussion with colleagues at the Grampian Research Office, which offers GCP training. The reach of GCP training is so wide that it should be a cornerstone of increasing knowledge and skills around ethnically diverse trials.

留言 (0)

沒有登入
gif