Trial population representativeness
Who a trial enrols decides whom its results can be said to apply to. A group missing from a trial is a group about which the evidence says least, and each group is missing for its own scientific, ethical, practical or regulatory reasons. Regulators, sponsors and journals each act on who is enrolled and how the population is reported, and where trials run shapes who can take part. A difference between populations is a scientific finding to be shown rather than assumed, and it has to be weighed against what the statistics of a trial can carry.
Homogeneous trial populations and generalisability
The main concern about randomised trials as guidance for medical decisions is their applicability to daily clinical practice and their generalisability to the patient population at large. A randomised trial is built as a noise-free environment, almost a laboratory, with a homogeneous population of patients very much alike, so as to be as sure as possible that any difference seen at the end of the study is attributable only to the interventions applied in the study arms. Randomisation and specific, often long, inclusion and exclusion criteria make the study population homogeneous and the arms similar, which suits the trial to capturing the efficacy of the intervention. Once a drug is approved and used by ordinary physicians, its patients differ widely: some are obese and some very thin, some pregnant and some old, and often none of them resembles the patients in whom the trial was run. Trial populations differ from the ultimate target population, being usually free of factors such as multimedication and comorbidities that might blur the results, and this is perhaps one of the most important reasons for the gap between efficacy in a trial and effectiveness in clinical practice.
An analysis of the inclusion and exclusion criteria of 17 randomised trials of ulcer treatment, covering venous, diabetic foot and pressure ulcers, calculated the proportion of a typical wound-care patient population each trial would have excluded, as a surrogate for its applicability to daily clinical practice. It estimated that more than 50% of patients would have been excluded in 15 of the 17 trials, and concluded that their results are unlikely to effectively support the management of a typical wound-care population.
A 2019 meta-analysis compared randomised trials from Medline and the Cochrane databases published between January 2007 and December 2016, each with at least two sites and more than 100 adults on dialysis for end-stage kidney disease, with the 2011 United States Renal Data System cohort of more than 500,000 patients. The median trial had 211 participants from 15 sites in a single country, and 7 months from randomisation to final data collection. Across 189 trials and 80,104 participants, trial participants were significantly younger, more likely to be male and less likely to have diabetes or diabetic nephropathy than registry patients, and their mortality was substantially lower, less than half that of registry patients, both overall and in trials recruiting only in the United States. The authors concluded that caution should be exercised when generalising trial results to the broader dialysis population. An accompanying editorial called for older patients and those with serious comorbid illness to be included in randomised trials if their evidence is to inform decision-making for all patients.
Population differences in drug response
Inclusion and diversity are debated across society and politics, not only in clinical research, and for clinical trials an important question is what would be missing if they were not examined.
Some regard inclusion as optional, but in some fields it has a deeper scientific meaning. It has been known for decades that ethnic factors can affect the dosing, efficacy and safety of medicinal products. ICH E5, from the 1990s, introduced the concept that ethnic groups may differ in ways that bear on the validity of clinical trials, and brought in bridging trials to avoid unnecessary repetition of full trials. ICH E17, in 2017, went further by setting out principles for multiregional clinical trials, under which it may be better to design a trial from the beginning to involve different ethnic groups.
Some medicines approved in the West do not work at all in the neonates of a clinical trial in Congo, and can even harm them, according to a clinician treating them there, so the effect of a treatment can depend heavily on the ethnicity of the people treated.
In a kidney transplant study, the immunosuppressant was metabolised at almost twice the rate by people of African descent as by people not of African descent, so patients of African descent given an insufficient quantity would reject the organ. Unless that population is included in studies and the dosage increased from the start, how patients of African descent who need a kidney transplant respond to the drug is never known.
Physiological and pathological functions differ between the sexes, so those differences are examined in clinical research. Between 1997 and 2000 the FDA withdrew eight prescription drugs that all showed more adverse effects in women than in men. More frequent prescribing to women does not explain the difference, because for at least four of them prescribing was equal between men and women. In 2013 the FDA announced that women taking zolpidem for insomnia were at risk of excessive daytime sedation and impaired driving after bedtime doses, and lowered the dose for women, because the recommended dose had been based on male participants.
Genetic variation, race and ethnicity
Genome research has confirmed overwhelming DNA similarity and negligible DNA differences between humans, although genetic variation cannot account entirely for the phenotypic differences among them. A DNA modification not linked to disease is called a genetic variant, and research has found more genetic variants than sociocultural categories such as race. Genome-wide association studies have stratified populations by clusters of gene expression rather than by geographic origin, and most genetic variation is found among people belonging to a single population. The questionable "race correction" of spirometric measurement, used in the US since the 19th century, has not yet been updated using modern methods. Given the difference in genetic variants, the warfarin dosing algorithm may differ between White and African-descendant patients.
Some genetic variants may be associated with the risk of disease. End-stage kidney disease is much more frequent in African Americans than in White people, and in non-diabetic people with West African ancestry it has been associated with polymorphisms at the APOL1 locus, a variant selected because it protected against sleeping sickness, but the polymorphisms do not account for the increased risk and no mechanism has been demonstrated. Phenotypes also vary for other reasons: the risk of disease may be linked to external factors acting on the epigenome, and the low mortality of SARS-CoV-2 infection in Africa compared with Europe, the US and Asia can be explained by differences in environment rather than race.
South Asian populations living in the US or EU lack well-characterised genetic and epigenetic profiles, while including people with African ancestry has identified new loci in obesity, metabolic syndrome and multiple sclerosis. Genetic and epigenetic profiles can help identify subpopulations at risk of syndromes and diseases, which may be fundamental for prevention strategies. Limiting the inclusion of subpopulations prevents the objective of precision medicine from being reached, whereas including diverse subpopulations in genome-wide association studies may strengthen the research, cast light on genotype–phenotype interactions and identify new drug targets.
Underrepresentation and its consequences
Population diversity in clinical trials is essential, yet many trials are not yet enrolling a diverse population that reflects the target population. Historically, trial populations have been dominated by White men, and marginalised racial and ethnic groups, women and other historically disenfranchised populations have been substantially underrepresented. This makes a comprehensive assessment across the whole affected population impossible, and leaves the clinical relevance of a drug to its target population a matter for post-marketing activities. Because most registrational trials include mostly White patients, drug approvals lack data on the different effects of products in various populations. Race and ethnicity, alongside other determinants, seem to account for differences in insulin regulation and glycaemic response to carbohydrates, yet given the scarcity of studies, recommendations on insulin dosing and formulations in more diverse populations were still lacking in 2022.
Writers can make clear the benefit of a more diverse research population, which matters even more than research dollars. In the view of a director at the National Institute on Minority Health and Health Disparities (NIMHD), more diverse participation in clinical trials is the only way to know whether a treatment or medical advice has broad applicability. A programme officer at the NIMHD holds that more work is needed to ensure that the evidence base for clinical guidelines and policy recommendations includes trials with sizeable numbers from the populations with the highest disease burden and mortality. Prostate-specific antigen screening guidelines, for example, rest mainly on large US and UK trials that enrolled very few African Americans, although African American men are twice as likely as men in the general US population to die of prostate cancer. Diabetes trials should include Mexican Americans and Puerto Ricans, populations with high rates of diabetes, and prostate cancer trials would be remiss to leave out African American men, who are twice as likely as White men to be affected by and die of the disease.
Disparities in healthcare access due to race and ethnicity are crucial confounding factors with severe consequences. Underrepresented subpopulations in clinical studies can mask the epidemiology of several diseases, and the missing data make it difficult to identify connections between socio-demographic and genetic determinants and clinical variables. The underrepresentation of minority populations limits the identification of risk profiles for diseases, fuels false beliefs and implicit bias in clinical decisions, hinders interpretation, generalisation of findings and prevention planning, and increases socioeconomic disparities in healthcare access. The persistence of false beliefs in race and ethnicity categories in randomised clinical trials and observational studies may affect clinical decisions. Limited studies on the health of ethnic minorities can have a negative impact on healthcare expenditure for diseases such as diabetes, mental health or infectious diseases.
Underrepresented groups
The scientific, ethical and societal questions raised by inclusion can differ for each underrepresented group.
Women
Women have been historically underrepresented in clinical trials, and in animal studies too. Women were once excluded from clinical trials on the grounds that they were too delicate and fragile, so drugs were tested only in adult men up to 55 years of age. Excluding women takes half the population out of the assessment of drugs they may be given, which are then tested only in people who do not share a demographic characteristic that may be critical. An important reason always given for their underrepresentation is safety and ethics, particularly for women of childbearing potential, a concern that goes back to the thalidomide tragedy several decades ago, and more modern scientific methods could now mitigate at least part of that risk.
The US Congress, through the 1993 NIH Revitalization Act, made research funding dependent on appropriate inclusion of women in clinical trials. Female participation in trials has improved significantly over the years: across 231 novel drugs approved between 2015 and 2019, 51% of all trial participants were female. Inclusion of women in earlier clinical research remains a challenge.
Pregnant and breastfeeding people
Pregnant and breastfeeding people are often excluded from clinical trials, and those who become pregnant during a trial are frequently discontinued from it. In 2025, according to data from the Clinical Trials Information System, fewer than 0.4% of clinical trials submitted in the EU included pregnant people, and 0.1% lactating individuals. As a result, product leaflets usually lack details on the benefits and risks of a medicine in pregnancy and breastfeeding, which can lead to suboptimal treatment decisions and potential harm. Yet the vast majority of pregnant people take medications, for example for chronic diseases, infections or pregnancy complications, and the situation is similar in breastfeeding populations.
For pregnant and lactating women, ethical aspects are especially important to consider. The ICH E21 guideline, in development in 2025, sets the scene for their inclusion in certain clinical trials while requiring a careful look at the available science and the ethical aspects. It holds that these groups should be included only where clinical data already exist in other populations, which lowers the risk further.
On 4 June 2025 the EMA released for public consultation, until 15 September 2025, a guideline developed through the ICH on including and retaining pregnant and breastfeeding people in clinical trials, with the goal that developers generate robust clinical data in pregnant and breastfeeding people so that they and their healthcare providers can make informed, evidence-based decisions. The guideline marks a change in paradigm in the development of medicines in pregnancy and breastfeeding. It holds that including pregnant and breastfeeding people in clinical trials should in principle be considered for all medicines intended for people who can become pregnant, and lays out the principles and conditions for the safety of participants and their foetuses and babies. It sets out scientific and regulatory principles and ethical considerations for including pregnant and breastfeeding individuals in trials both before and after authorisation, and encourages proactive planning and early consultation of regulators.
Children and adolescents
Children and adolescents have historically been excluded from clinical trials on ethical grounds: a child cannot give informed consent as an adult can, sensitivity to adverse events may be greater, and there is a risk of damage to the developing organism. Knowledge of dosing and experience with relevant endpoints were also lacking, and recruiting from the small patient pool can be difficult, and all of these remain limitations today. Modern designs such as basket and umbrella trials can at least circumvent these issues a little. Extrapolation approaches, a big topic at the FDA and EMA, may not always be feasible and need a very good scientific rationale, but where it exists they can limit how much has to be studied in a paediatric population where study may be risky. The concept of assent, through which a child can give their own agreement to the trial, may also allow ethics committees to agree to studies in children more than they could in the past. Challenges remain in dose-finding, where a modelling approach can help.
Older people
Of the underrepresented groups, older people may be the one that has seen the least progress towards better representation, because there are scientific limitations to their inclusion. Dosing recommendations for younger adults may not be applicable to them, and inclusion may carry a significant safety risk. ICH E7 aims to foster the inclusion of older people, yet in practice partially arbitrary inclusion and exclusion criteria and ethical concerns remain. Many older patients use several medicines at the same time, which makes it difficult to design trials whose data will not be confounded by polypharmacy. Patients over 65 are often excluded from clinical trials, for example because they are multimedicated, although they form a sizeable portion of the users of many drugs, such as hypertension therapies. Ethical aspects must also be considered, including how best to obtain informed consent from people with dementia and how to test new medicines in frail individuals.
Cancer trials have historically not enrolled older participants at levels reflecting real-world demographics: in 2004, 36% of participants were 65 or older against 60% of those diagnosed, an improvement on 25% against 63% in 1999. Common health complications in older patients can be barriers to enrolment, and data from trial participants aged 65 or older may not mirror actual outcomes for the general population. Older adults differ widely in genetic predisposition, environmental conditions and lifestyle, and may present with coexisting diseases, concomitant medications or poorer performance status than younger people. By 2019 the FDA and ICH warned against unjustified exclusions, such as of people over 75, that were once commonplace. The American Society of Clinical Oncology and Friends of Cancer Research recommend loosening eligibility criteria to allow prior or concurrent malignancies as long as this would not compromise safety and efficacy endpoints, and the American Society of Clinical Oncology also advocates special subgroup analyses to better inform cancer management for older adults.
People with obesity
Clinical pharmacology aims to ensure that the right dose is given to the right patient at the right time, which requires understanding the variability present in any patient population and its effect on drug response. Obesity is a factor that may alter the pharmacokinetic and pharmacodynamic response and the therapeutic dose, and its prevalence makes it a further challenge for drug development. A review of drug product labels and approval packages found limited information on dosing guidance in obese patients, and recommended that the evaluation of pharmacokinetics and pharmacodynamics in obese populations be incorporated into the general drug development process. In 2012 the consequences of obesity for drug disposition in children were less well studied than in adults, and an investigation of drug-metabolising enzyme activity in obese children concluded that further studies were needed to determine the consequences for the dosage of drugs those enzymes metabolise.
Socioeconomically disadvantaged groups
The WHO's 2024 guidance on best practices for clinical trials covers trials in general, but diversity and inclusion run strongly through it, and it recommends ways to enhance the inclusion of children, pregnant women and older people. Beyond the groups long recognised, such as sex and ethnicity, it gives a concrete list of underrepresented groups, all of them socioeconomically or societally disadvantaged. Socioeconomically deprived groups within richer countries, such as the Black and Hispanic populations of the United States, are underrepresented, and many low- and middle-income countries and their ethnicities remain totally underrepresented in trials.
Prerequisites for inclusion
In one practitioner's view, inclusion in clinical trials rests on three prerequisites: scientific progress, ethical and societal progress, and policy and regulatory backing. Whatever is examined should be of scientific value, and methods have to exist to establish that value. Society has to be willing to accept a certain level of risk for a certain population, and the role of women and of children in society is different from what it was twenty or thirty years ago. Policy and regulatory backing alone is not sufficient: older people have had it without actual progress. Only societal, scientific and regulatory actions combined can really drive inclusion forward.
Increasing the inclusion of groups should always be based on science, and only then will it increase the value of the trial results. Where including more groups can be shown to add to the science, society's trust in the results of the trial should grow. Medical and scientific justification limits the number of groups that can be included. Including individuals from many groups has to be weighed against the statistical possibilities of a trial. Defining too many distinct groups reduces the scientific value of a trial, so a sound balance is needed. Giving groups a voice in a trial is good, but a scientific difference between them has to be shown, and otherwise the subdivision becomes a problem for the trial's statistics.
The same practitioner holds that inclusion is a holistic task for all stakeholders, that societal and ethical progress is a key prerequisite for its success, and that how well a society implements inclusion in clinical trials may be considered a mirror of that society's focus and status as a whole.
Regulatory requirements for diversity
Over the past few decades, regulatory guidelines on clinical trials have specified that enrolled participants should be representative of the population most likely to use the drug, and the persisting lack of diversity prompted the EMA and FDA to develop dedicated guidance encouraging sponsors to design trials that facilitate enrolment of the target population. Health authorities have signalled the need for greater diversity through the ICH guidelines on multiregional trials and on ethnic factors in foreign data, the FDA's Project Equity, and the FDA's guidances on enhancing trial diversity (2020) and on diversity plans (2022). In 2022 health authorities did not require sponsors to enrol subjects belonging to minorities, nor in numbers that allow proper analyses and conclusions on effectiveness, which lowers the robustness of meta-analyses and stands in the way of complete pharmacovigilance of a drug's adverse events.
In 2013 an FDA report to Congress highlighted demographic data gaps, which gave rise to the Diversity Action Plan, with recommendations to standardise data collection, improve data quality and public availability, ensure demographic representation and consider the integration of diversity throughout a drug's lifecycle. The FDA's 2016 guidance on collecting race and ethnicity data gave an update on a standardised approach to their collection and reporting in submissions for trials of FDA-regulated products conducted in the United States and abroad, and recognised that race and ethnicity categories are not anthropologically or scientifically based designations but describe the sociocultural construct of society. Its June 2024 revision promoted the inclusion of race and ethnicity information in labelling, with the baseline demographics of the study population given in the Clinical Studies and Adverse Reactions sections. The FDA's 2024 guidance on enhancing the diversity of clinical trial populations focuses on increasing the enrolment of underrepresented populations, with attention to demographic characteristics such as sex, race, diet, ethnicity, age and location of residence and to non-demographic ones such as organ dysfunction, comorbid conditions, disabilities, extremes of the weight range and low-prevalence conditions. It primarily recommends broadening eligibility criteria and using designs that reduce participant burden, to create a study population that more accurately reflects the patients likely to take the drug and to allow assessment of the impact of those characteristics on its safety and effectiveness.
The FDA's 2022 draft guidance on diversity plans recommended that sponsors develop and submit a diversity plan to help ensure the adequate participation of relevant and underrepresented populations and analyses of data collected from clinically relevant populations. The FDA's diversity action plans were intended to increase enrolment of participants from populations defined in the United States as historically underrepresented. The FDA issued several draft guidances on them, the latest in 2024, describing a Diversity Action Plan that sponsors should submit to show how they would enrol a diverse population into certain late-stage trials. The plan was to set aggregate enrolment goals by sex, age group, race and ethnicity, the rationale for those goals and the measures to meet them, thought out before enrolment rather than evaluated afterwards. Several sponsors found it difficult to establish what enrolment goals followed from such a plan, and where to draw the line between groups relevant enough to include or differentiate and groups that were not.
In one practitioner's assessment, the diversity guidance was quite complex, because it expected a case-by-case decision with the sponsor for every product and the groups to be examined had to be agreed with the FDA. ICH E5, by contrast, states which parameters can increase the risk of ethnic sensitivity, so a sponsor can apply the criteria, design a programme and discuss it with the agency. Community engagement, working with patient advocacy groups and local healthcare providers, and cultural training for investigators are all relevant, but the same practitioner finds it questionable that the guidance put them on the sponsor alone rather than treating them as a multi-stakeholder matter, which is much more the approach Europe takes on a similar topic.
The EU Clinical Trials Regulation No 536/2014 reinforces the requirement that participants should represent the population the drug is intended for, with a clear emphasis on age and sex, and sponsors are expected to justify a trial that does not reflect the target population. It makes no reference to race, which keeps ICH E5 the primary guidance on the impact of ethnic factors, and does not expand on considering intrinsic and extrinsic factors when designing trials; the EMA has adopted ICH E17 on multiregional clinical trials. ICH E17 recognises that differences in medical practice, diet, environmental factors, cultural or socioeconomic factors such as contraceptive use or a preferred route of administration, geographic location and access to healthcare can affect trial results, and may also affect recruitment, compliance and retention. In one author's view, an amendment to the EU legislative framework should be considered essential, and pharmaceutical regulations in particular should require the equitable enrolment of patients in clinical studies.
A comparison of 397 new drugs approved in the US and Europe found inconsistent designations of race and ethnicity in labels and summaries of product characteristics, with the FDA emphasising race and ethnicity subgroups in labels more than the EMA, although such inclusion and reported differences were less frequent in clinical trials. The racialisation of pharmaceutical regulation, the way race and ethnicity have become important to drug testing and evaluation, has been rising in the US and EU, as case studies such as that of the antihypertensive BiDil report.
In January 2025, days after a US presidential executive order terminating federal diversity, equity and inclusion programmes, the FDA stopped implementing the diversity action plan guidance and removed it from its website without announcement. The removal raised concerns about the commitment of the FDA and other authorities, about the applicability of the statutory obligation on sponsors to submit diversity action plans, and about the acknowledged importance of trial diversity and health equity. The guidance was later put back on the website under a court order, accompanied by a statement from the US Department of Health and Human Services disagreeing with its content, and by December 2025 it was no longer being applied.
Representation in the protocol
Representation in clinical trials needs to improve, and the differences that matter are not only racial: socioeconomic status, physiology and disease characteristics may differ too, and so may gender, for transgender and gender-diverse people.
A protocol whose eligibility criteria say "sex: male or female" leaves the eligibility of a transgender participant undetermined, because it does not say whether it means sex assigned at birth or current sex or gender.
The Multi-Regional Clinical Trials Center of Brigham and Women's Hospital and Harvard publishes an Equity by Design resource and recommendations for diversity, equity and inclusion content in protocol templates, which in 2023 TransCelerate was understood to be examining for the common protocol template. The DIMES resource on diversity, equity and inclusion and digitised clinical trials bears on much of a protocol, including the risk–benefit argument, eligibility, which assessments are done and when, and stratification.
The US DEPICT Act requires late-stage trials to report enrolment targets by demographic group, including age, race, ethnicity and sex, and to provide a rationale for those targets, which forces study teams to construct a rationale for the people they intend to enrol. Arriving at a phase 3 trial with no picture of the demographic subgroups is a gamble at a late stage, so diversity planning is better brought in earlier. The inclusion of disproportionately affected populations is important to think about from the earliest planning of a trial: African Americans, for example, are at raised risk of Alzheimer's disease, so a plan to enrol and retain them should be in place from early in a study's development.
Making trials accessible to the participant groups a study wants means making sure the eligibility criteria are not too restrictive and that the sites do not require many hours of travel.
A disease may vary in severity and outcome with patient characteristics such as gender, race and ethnicity, and Bristol Myers Squibb works with epidemiologists to understand the aggregate disease and patient outcomes. Its Patient Voice programme talks with patients about their disease experience and the barriers to taking part in trials, and turns the insights into recommendations to protocol development teams that can reduce the burden of trial participation, and into knowledge of how best to target outreach and engagement at the most favourable points along the patient journey to support trial awareness and participation in diverse populations. Changes made on patient feedback have included a reduced site visit schedule, fewer invasive biopsies, a medication format that reduces the number of times a day a pill must be remembered, and culturally sensitive recruitment messages and materials. To reduce barriers to enrolling diverse patients, the company also aims to attract diverse talent, has an internal advisory board drawn from its People and Business Resource Groups that gives feedback on patient-facing materials and programmes, and keeps relationships with community outreach groups and patient advocacy organisations. The company aims to remodel aspects of trial design and execution, with the expectation that doing so will positively impact a broader patient population, more reflective of the real world and aligned with the epidemiology of the disease studied. The company's key operational strategies include assessing protocol language to minimise barriers to enrolment, embedding study diversity plans as a natural step in clinical development planning, and engaging with diverse patient representative groups. It holds diversity in clinical trials to be both a social and a scientific imperative, and has study-specific and non-asset-specific conversations with the FDA on diverse enrolment and receives feedback on regulatory documents that adds to its knowledge of the challenges to diverse enrolment.
Widening enrolment
Population diversity in clinical trials is vital to ensure that results are safe and applicable to the general population. Broadening eligibility criteria is one of the key regulatory approaches to increasing the enrolment of a diverse population, but broadening them too far can increase variability and complicate the interpretation of results. Increasing the proportional enrolment of specific populations of interest is another recommended regulatory approach.
Research has identified many barriers to the inclusion of diverse populations in clinical trials. Participants from marginalised communities often mistrust the pharmaceutical industry, fear exploitation, lack awareness of their disease or of available trials, and face language barriers or operational constraints. Sponsors may show limited commitment and effort, conduct centralised studies, lack culturally or racially diverse staff and community engagement, and hold negative attitudes about marginalised communities' willingness to take part. Willingness is not the obstacle: numerous studies have shown racial and ethnic groups to be as interested as anyone in taking part in clinical trials. The problem is the history of the Tuskegee study and other events. Minority groups want a full discussion of anything that might harm them or their families, and want to feel comfortable trusting the researcher or doctor. This highlights the importance of communication, and shows that securing a more diverse pool of participants is not an insurmountable task. To increase the enrolment of historically underrepresented populations, sponsors may strengthen community engagement by providing cultural competency training for investigators and site staff, streamlining informed consent where risks are low, providing patient leaflets in several languages, and giving language assistance to participants with limited English proficiency.
An adaptive design, which plans modifications prospectively on the basis of interim results, can allow the eligibility criteria to be modified following interim results, so that underrepresented groups initially excluded by narrow criteria could be included and the trial could better reflect the diverse population likely to take the drug.
The National Cancer Institute and the NIMHD were addressing the lack of information on diseases prevalent in particular populations through ongoing trials and reanalysis of recent ones, and the cancer institute had 12 Minority/Underserved Community Sites focused on research with racial and ethnic minorities or rural residents.
Analysing subpopulations
Individuals may respond differently to drugs because of a combination of intrinsic and extrinsic factors, so sponsors of clinical trials are required to assess unusually large or small responses in population subgroups, examining for example whether there are differences by age, sex and race compared with the overall population. Study designs and statistical protocols are essential to highlight, deepen or confirm clinical similarities or differences among subpopulations, and to evaluate the contribution of socio-demographic variables and comorbidities. Subgroup analyses of clinical studies can highlight risk factors or determinants of disease in diverse subpopulations, and can also increase equity in the access to and provision of healthcare. Because including various populations entails a standard designation of race and ethnicity, study designs and their reporting need different protocol measures and suitable methods for calculating sample sizes.
African American and White men with newly diagnosed non-metastatic prostate cancer treated with standard healthcare access had comparable stage-for-stage prostate-cancer mortality after adjustment for demographic, cancer and treatment-related baseline differences and inverse probability weighting. In the POINT trial of 4044 US participants, Black participants (918, 22.7%) had a higher cumulative risk of stroke than White participants, and adjustment for demographic data, comorbidities and adherence confirmed their higher risk of early stroke recurrence after minor ischaemic stroke or transient ischaemic attack. In the REGARDS study of 9416 Black and 13,091 White people without a history of cardiovascular disease, Black participants had a significantly higher risk of sudden cardiac death at 6.1 years, confirmed after adjustment, but the adjustment variables did not account for the higher incidence. In 2017 the NIH-supported OPPERA cohort study on orofacial pain found no racial differences in tissue characteristics or nociceptive sensitivity. A 2020 NIH study that analysed pain perception pathways in three ethnic groups by functional magnetic resonance imaging found similar nociceptive pain processing among the groups. The perception of pain is complex and may be influenced by cultural differences, yet implicit bias about race and ethnicity in pain can increase the burden of pain, blur its assessment and mislead recommendations.
Reporting race and ethnicity
Race and ethnicity data are underreported or inconsistently reported in orthopaedics, surgery, anaesthesia and real-world data on medical devices, which hinders the identification of differences in treatments and outcomes. In 2022 most US and UK healthcare systems usually collected data on race and ethnicity, but most EU countries did not. Race and ethnicity data in clinical databases and algorithms are often absent, inconsistent, incomplete or contradictory, which leads to unreliable interpretation of results. Standardised data in registries favour measuring disparities in healthcare access among subpopulations.
Because the relevance of variables such as age, sex or ethnicity is not always known when a study is designed, the 2019 ICMJE recommendations hold that researchers should aim for representative populations in all study types and at a minimum provide descriptive demographic data, and that authors should define how they determined race and ethnicity, justify their relevance and use neutral, precise and respectful language. As of 2022, EQUATOR addressed the terms race and ethnicity only partially, CONSORT 2010 mentioned ethnicity only in examples on generalisability and gave no suggestion on reporting race and ethnicity, STROBE did not mention them, the European Association of Science Editors mentioned race only as a variable to be disaggregated, and COPE gave no specific core practice for reporting them. The New England Journal of Medicine has stressed the value of including various subpopulations for the generalisability of findings, and from January 2022 it asked authors for supplementary tables on the representativeness of the enrolled populations. Such transparency favours the decisions of reviewers and publishers on publishing manuscripts. Publishers' editorial guidelines, built on the ICMJE recommendations, should focus on reporting race and ethnicity data with clear clinical motivations related to the research question, editorial teams should harmonise their recommendations on collecting and reporting the data, and statements and designations of race and ethnicity should be applied by publishers and reviewers as well as authors.
Questions about a study's population
A writer reporting a study should consider who it included: did ethnic and racial minorities take part, and if not, why not? A study population that is overwhelmingly White calls for scepticism. If the researchers found differences between populations, the writer should ask whether they considered all the reasons such differences can arise, or jumped to the conclusion that the races must be biologically different. Asking these questions can help improve clinical research and ultimately help end health disparities.
Where trials are conducted
Over half a million clinical studies were registered worldwide in 2021 on the WHO's International Clinical Trials Registry Platform, a number that gives the misleading impression that clinical research is spread evenly across the world. Most of it is conducted in the global north, because the cost and technical expertise trials require restrict them in the underdeveloped parts of the world. In the decade to 2023 several Asian countries were turning into clinical research hubs, and China, India and the Republic of Korea were among the top 15 countries by number of ongoing trials. This benefits clinical research by promoting international collaboration, increasing diversity in the studied population and speeding up the process, as the rapid investigation of COVID-19 vaccines showed. Where trials are conducted in low- and middle-income countries, they tend to study diseases prevalent in high-income countries rather than those relevant to the host communities, leaving tropical diseases neglected, and a drug developed with participants there may never be launched there because those countries lack commercial potential.
A review by the US Department of Health and Human Services Office of the Inspector General of the marketing applications for 106 drugs and 15 biologics that the FDA approved in fiscal year 2008, representing 193 completed clinical trials, found that 80% of the applications contained data from foreign trials and that more than half of trial participants and sites were outside the United States. The percentage of foreign investigators conducting trials under Investigational New Drug applications had more than doubled over the preceding decade, and the FDA considered the trend likely to continue, citing Western and Eastern Europe, Central and South America, and China and India as regions where sponsors were conducting more trials.
Clinical trial capacity in low- and middle-income countries
Low- and middle-income countries lack national funding for research, which stands in sharp contrast to their high rates of preventable disease and the large patient pool that could be reached if the infrastructure existed. The WHO recommends that national governments build local clinical trial infrastructure so that they can run trials on their own, and has built a WHO Clinical Trial Unit model for doing so.
An African clinical trial network aims to build clinical research infrastructure across Africa as a public–private initiative that includes the pharmaceutical company Janssen. The African Medicines Agency had been officially launched only weeks before December 2025, and since harmonised regulation usually makes a region attractive to sponsors, as in the United States, a fully functional agency could be a huge boost for African research.
Public–private partnership is a very important model for advancing diversity and inclusion in clinical trials. Engaging communities at national and regional level, so that their voice is included in the decisions of health authorities, is important, also because it gives them a voice in decisions on clinical trial funding. The Global Alliance for Chronic Diseases addresses the needs of patients with noncommunicable diseases in lower-income countries and also of socioeconomically deprived patients with the same diseases in richer countries, groups that are underrepresented in trials.