Confident translation of research evidence into clinical practice relies on a shared understanding of who is most likely to benefit from a particular intervention, the acceptability of an intervention to patients and health care professionals, the feasibility of delivering the intervention in the health system and the availability of the intervention to all who need it. Underpinning this process of translation is the assumption that findings from trials of the intervention are applicable to the target population. However, this assumption is often not true, and the dangers of failing to ensure the enrolment of an inclusive and diverse population into trials have been played out for the world to see in the COVID-19 pandemic. The populations recruited to the trials of the SARS-CoV-2 vaccines did not appropriately represent under-served communities1. This lack of community engagement eroded trust in vaccine safety and efficacy and influenced vaccine uptake in key under-served populations, who arguably stood to gain the most. This example illustrates how a lack of trust in trial findings influences everyone, from the recipients to healthcare professionals and policy makers.

As in other specialties, populations recruited to clinical trials in neurology are often not diverse, so the safety and efficacy results are not generalizable, sometimes at high cost to particular groups. For example, the exclusion of people with cardiac comorbidities from clinical trials of lamotrigine have resulted in uncertainty about its safety in that substantial population2. The landmark optic neuritis treatment trial in the 1990s did not include patients with neuromyelitis optica, which is more common among Asian and African people and those of Afro-Caribbean descent, resulting in suboptimal treatment for these groups during the subsequent decades3. Lack of inclusivity could also risk failure to identify ethnically related pharmacogenomic variability, with clinical implications; one example of this variability is the contribution of HLA haplotype to risk of hypersensitivity to carbamazepine4.

When no clear evidence is available to guide therapeutic decision making for a particular population — such as women with Parkinson disease5 — patients in that population feel disenfranchised and lose trust in their healthcare providers. Such evidence gaps can be addressed to some degree by post-authorization surveillance, but this surveillance needs to be robust and comprehensive, with a clear pathway to changing clinical practice on the basis of the findings. For example, the pregnancy register data relating to the teratogenic effects of valproate6 was first published in 2004, but the lack of a clear pathway meant that the NICE quality standard relating to this was not published until 12 years later in 2016.

Lack of inclusivity is important in contexts other than therapeutic interventional studies. Observational and epidemiological studies that are not inclusive can also produce results that are not generalizable and limit our fundamental understanding of disease. As an illustration, a recent call to action to promote diversity, equity and inclusion in Parkinson disease research and care sets out the problem7. Knowledge of how ethnicity and socioeconomic status contribute to the incidence, morbidity and mortality of Parkinson disease is limited. This gap in knowledge results not only from limited reporting of these data but also from the very low diversity revealed by transparent reporting — for example, only 8% of participants in clinical trials are not white.

What can we do?

For clinical research to answer clinically relevant questions and inform care decisions, the populations it is intended to benefit must be involved in all stages of its design. Ensuring representative involvement requires an understanding of which populations are under-served, transparency about the barriers that impede their involvement and solutions to remove those barriers.

Identification of under-served populations might seem simple, but a multi-stakeholder survey has indicated otherwise8. This survey demonstrated that under-served populations can be grouped by a variety of factors: demographics (for example, age, sex, ethnicity and education); social and economic factors (for example, employment, caring responsibilities, digital exclusion, social marginalization and geography); health (for example, multiple long-term health conditions, cognitive impairment, learning disability, visual or hearing impairments); and disease-specific factors (for example, rare diseases). Consequently, inclusivity is highly context-specific and, by nature, highly intersectional. It depends on the population, the condition being studied, the research question, the intervention and the observations made. No single, simple definition can encompass all under-served groups.

Existing research practices can also lead to a lack of inclusivity in trials, as discussed in detail elsewhere8. In particular, recruitment strategies and trial logistics often act as barriers to entry and retention. For example, recruitment via tertiary care centres that can be remote from the patients’ homes or in urban environments far from rural populations can result in challenges related to transportation and connectivity.

Solutions to engage under-served populations include the development of community-partnered participatory research that involves under-served groups to build long-term relationships and opportunities to participate with equitable involvement in all phases of the research process; the development of and access to training resources for designing and delivering health and care research that is tailored to all stakeholders and the needs of specific groups; the development of infrastructure to reach, engage, recruit and retain the under-served; and leadership from funders, regulators and other stakeholders to remove barriers in funding, regulations and policy.

Transparency around inclusion should be required not only by funders and ethics committees but also in the reporting of study protocols and findings. A variety of national bodies and policy makers are driving programmes to improve this transparency. In particular, the UK National Institute for Health and Care Research (NIHR) INCLUDE guidelines9 set out helpful questions to guide investigators, research funders, reviewers and research delivery teams in considering how best to design, review and improve inclusion at key steps in the research process. The guidance encompasses all health and care research, so it can be broadly applied, and is supplemented by essential training on implementation. Other notable initiatives are being driven by organizations such as the FDA, which recently published guidelines to drive industry to make changes that will increase the diversity of clinical trial populations, and, increasingly, patient charities and advocacy groups. Initiatives to mitigate evidence gaps, such as the systematic collection of real-world data in the OPTIMISE: MS study10, will also help to address the issue of trial participation. Ultimately, a collective effort is needed to deliver better health care for all.

“we must minimize bias and inequality in the generation of […] evidence”

Thinking more broadly

Though important, none of the initiatives to improve research inclusivity address the fundamental issue: under-served communities lack access to healthcare, which limits diagnosis and treatment and reduces engagement at clinical touchpoints that provide entry to research. Perhaps even more fundamental are the societal inequalities that shape the broader research landscape. Can we really deliver inclusive and diverse research if we do not have inclusive and diverse research teams? There are global systemic barriers to diversity of researchers, which probably start with academic attainment in schools, followed by attainment in higher education, exacerbated by a lack of diversity among decision-making bodies that are critical to academic progression (such as funder panels and editorial boards) — these factors compound the inequality faced by aspiring investigators and researchers from under-served communities. Unless we tackle the root of the problem — the social inequality we live with and that is ingrained in almost all of our institutions — we will only be scratching at the surface of a solution. We and all of our organizations have a role to play.

Conclusions

The importance of diverse and inclusive trials to improve healthcare is now firmly on the agenda of multiple stakeholders. In order to effectively translate evidence into clinical practice, we must minimize bias and inequality in the generation of that evidence. If we do not, we risk perpetuating the health inequalities that have been brought into sharp relief by the COVID-19 pandemic. Research opportunities and the benefits they bring to patients, researchers and clinicians should be available to everyone, not just a select few. This is our collective, global, moral responsibility.