Diverse Groups and Clinical Trials
Strenuous efforts are being made in many countries to ensure, as far as possible, that clinical trials include all groups in society, roughly in proportion to their presence in the general population. So much so that the Economist recently featured an article on this very topic, that involving diverse research participants will lead to new medical insights.[1] As paid-up grant holders under the NIHR we have attended events around this very issue. The argument put forward by many speakers and the Economist is incomplete. But there is a further argument for entry of diverse groups that is strong. Against the backdrop of the USA, a world leader in enhancing clinical trial diversity and equity, visibly halting its efforts,[2] it is important to consolidate our rationale for including diverse groups in clinical trials. Before covering these technical arguments, we would like to address non-technical ones – ethical and political points.
Moral Argument
The ethical argument (often implicit rather than explicit) is a social justice argument for inclusion of diverse groups. This argument is predicated on the notion that it is unfair to ‘exclude’ certain groups from trials. We will tackle later the issue of ‘fairness’ in terms of future benefits to groups who may benefit from the findings of trials. Here we consider whether it is unfair from the point of view of the participants in these trials. This argument cannot be sustained unless there is some benefit in participating. However, clinical trials are not intended to benefit participants, since the moral basis for trials is equipoise between trial arms.[3,4] People may benefit from participating in a trial, but only if the intervention turns out superior or if the intervention group receives treatment guidelines unethically provided to controls.[5] Under equipoise there is no expectation of net benefit, and the argument that a group not included loses out does not hold. An exception might arise in ‘equipoise-plus’ trials. Here there is an expectation of benefit, but the experimental intervention is restricted to trial participants.[6] Here, a fairness argument arises and there is a strong case to ensure that groups are not systematically excluded.
Technical / Scientific Arguments
Two scientific arguments are advanced for including diverse groups. The first is based on the principle of generalisability – trial findings will be generalised to the population at large and should therefore be derived from trial populations that represent groups in proportion to the general population. This argument is not as strong as it first seems. This is because guidelines are not based on national populations – as large as Britain is, only about 6% of the world’s trials are based there. Most of our guidelines are currently based on populations different from the UK. Nevertheless, there is still an argument that, across countries, populations should be diverse so that, ultimately, a wide range of different characteristics have gone into the mix.
The second scientific argument is the converse – since there may be differences in treatment effect by group, outcomes should be hypothecated by group. This calls for particularisation rather than generalisation. The problem here is that precision falls off rapidly when sub-group interaction effects are tested.[7]
The article in the Economist does not mention this inconvenient truth. Yet, the article is correct in arguing that there may be some reason to suspect an interaction between says ethnic group or age, and (absolute) treatment effect.
Here we need to consider the evidence that informs clinical guidelines – seldom are guidelines based on but one trial. Rather, it is synthesis of all the relevant evidence in meta-analyses that drives changes in practice. The corollary is that efforts should indeed by made to include diverse groups. Then, in the early trials, the results will be suitable for generalisation. In the longer term, they will be available for particularisation. In essence, this is an extension of the arguments that small (so-called ‘underpowered’) trials are not unethical.[8] Making anonymised trial data publicly available will assist sub-group analyses as part of future meta-analyses.[9, 10]
— Richard Lilford, NIHR ARC West Midlands Director; NIHR Midlands PSRC Co-Director
— Ameeta Retzer, Equality, Diversity and Inclusion Lead
References:
- The Economist. How to Improve Clinical Trials. The Economist. 16 January 2025.
- Steenhuysen J. US FDA drops web pages on improving clinical trial diversity. Reuters. 24 January 2025.
- Lilford RJ, Jackson J. Equipoise and the ethics of randomization. J R Soc Med. 1995; 88(10): 552-9.
- Edwards SJ, Lilford RJ, Hewison J. The ethics of randomised controlled trials from the perspectives of patients, the public, and healthcare professionals. BMJ. 1998; 317(7167): 1209-12.
- Braunholtz DA, Edwards SJ, Lilford RJ. Are randomized clinical trials good for us (in the short term)? Evidence for a “trial effect”. J Clin Epidemiol. 2001; 54(3): 217-24.
- Edwards SJ, Lilford RJ, Braunholtz DA, et al. Ethical issues in the design and conduct of randomised controlled trials: A systematic review. Health Technol Assess. 1998; 2(15): 1-132.
- Brookes ST, Whitely E, Egger M, Smith GD, Mulheran PA, Peters TJ. Subgroup analyses in randomized trials: risks of subgroup-specific analyses; power and sample size for the interaction test. J Clin Epidemiol. 2004; 57(3): 229-36.
- Edwards SJ, Lilford RJ, Braunholtz D, Jackson J. Why “underpowered” trials are not necessarily unethical. Lancet. 1997; 350(9080): 804-7.
- Trial Forge. Pro EDI: Improving How Equity, Diversity and Inclusion are Handled in Evidence Synthesis. 2024.
- Cochrane Methods Equity. PROGRESS-PLUS. 2025.