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

Background: The benefits of randomised trials are not shared equally, and people from ethnic minority groups are a key constituency under-served by clinical research and clinical care. The STRIDE project aimed to give trialists practical information about how to decide which ethnic groups should be in their trials, and at what proportion.

Methods: We considered trials in six clinical areas: cancer, cardiovascular, diabetes, maternal health, mental health, and smoking cessation. We created a summary for each, including participants-intervention-comparators-outcomes, and data on disease prevalence by ethnicity. These were discussed with panels with clinical expertise, trial and methodology expertise, lived experience, funding, and experience of working with and on behalf of ethnic communities. For each trial, we asked panel members to decide which ethnic groups should have been involved and at what proportion.

Results: We discussed 23 trials with 40 individual panel members. Panels found our questions difficult to answer. The lack of publicly available data on prevalence by ethnicity was central to this. Where data were available, decision-making was easier but not simple. The discussions led to eight STRIDE recommendations. We recommend that discussions involve diverse teams and that discussions need time, with access to the best available data. In the absence of data or consensus, we recommend the adoption of 'default' minimum rates of inclusion, with oversampling considered. These discussions should inform site selection, and the practical challenges of recruitment and retention should not determine which groups are to be included. We also suggest five policy initiatives to support implementation of the recommendations. Broadly, these are (1) funders need to signal that ethnic diversity is expected, (2) trial teams need access to better data, (3) funders and others need to signal that ethnic diversity means better science, (4) more funding is needed for evaluation, and (5) Good Clinical Practice training should cover ethnic diversity.

Conclusions: Agreeing targets for which ethnic groups to involve in a trial is essential but difficult. Our eight recommendations could help to make trials more ethnically diverse if followed, and we suggest five policy initiatives that would create a supportive environment for their implementation.