On Thursday morning, a lecture theatre at Norfolk and Norwich University Hospital will fill for the second Black Health and Research Conference. Entry is free, the first session invites questions from the floor, and the day closes with a networking hour that organisers want to turn into working partnerships. Regional gatherings rarely reach a national audience. This one carries a message worth taking to Whitehall and to NHS boards, because its subject sits beneath nearly every reform the government has announced for the health service.

That subject is who appears in the evidence. Studies that recruit too few participants from Black communities can produce findings that fit those patients less well. A presentation on the NIHR BioResource and the Improving Black Health Outcomes initiative will set out how participation can be widened and what support exists for researchers who want to help. The topic can sound like a matter for research managers. It is better understood as a question of data quality, and data quality can be improved.

The case for acting is well established. An independent review of equity in medical devices found that pulse oximeters can overestimate blood oxygen levels in people with darker skin, a problem that traced back to the make-up of validation samples. Maternal outcomes carry a similar signal, with Black women in the UK dying in pregnancy and childbirth at roughly three times the rate of white women. Each finding points to evidence that can be strengthened, and the means to do so are within reach.

The 10 Year Health Plan gives that work a timely home. It promises a single patient record and a larger role for genomics, and it expects AI tools to become routine in diagnosis and administration. These ambitions depend on data, and models perform best when trained on large, rich datasets. That gives the health service a chance to build representation into national systems from the start. Procurement at scale rewards early decisions, because contracts, pathways and clinical habits form around the first generation of products. Tools validated across a diverse population will serve every patient better and will be easier to defend under scrutiny.

The operating climate is demanding, which is why this work needs sponsors. Integrated care boards are cutting their operating costs by half, while trusts manage waiting lists amid strict financial constraints. Community engagement is easy to classify as discretionary in that setting, yet it repays the investment. Recruiting from communities takes staff time and sustained relationships. The conference theme this year, Honouring Our Communities, names the people who hold that trust: local volunteers who run support networks, and the nursing and care staff who deliver frontline services. Trusts that back those relationships gain research participants, and they gain patients who are more willing to engage with services.

The practical steps are modest. NHS leaders can report research participation by ethnicity beside access and waiting time measures, which makes progress visible at board level. Policymakers can ask for demographic reporting in publicly funded studies and for subgroup performance evidence before algorithms enter NHS use. Life sciences and health-tech firms have a commercial reason to join in, since a diverse trial population strengthens regulatory submissions and supports the UK's pitch as a destination for clinical trials. Researchers need funding for community work, which grant applications seldom cost.

One conference will not shift national recruitment figures, and earlier commitments on health inequalities have sometimes faded into annual reports. The Norwich organisers have set a testable aim, which is collaborations that outlast the day. If they succeed, other regions gain a template they can copy.

The government wants a data-driven NHS. Where participation widens, the information the service holds will describe all of its patients, and the tools built on it will treat them more fairly. The work can begin in a lecture theatre in Norwich this week.

‍