12 Questions We Need to Answer to Achieve Diversity in Clinical Research

Norman M. Goldfarb

Diversity, equity, and inclusion (DEI) has, deservedly, become a hot topic in clinical research. Numerous organizations are developing methodologies, products and programs. If “let a thousand flowers bloom” is the best way to achieve diversity, we are certainly on track. However, we should be able to minimize duplicative efforts and accelerate learning through collaboration. The best place to start this collaboration is to develop consensus answers to the following 12 fundamental questions.

1. How do we define “diversity”?

The following definition of “diversity” attempts to give equal weight to the five good reasons for diversity addressed by the next question:

• A diverse study population accurately represents the population of therapeutic relevance, taking into consideration fair representation of underserved populations, the practicality of the study design, scientific validity, and commercial imperatives.

However, the following definition, with its focus on ethical concerns, may better serve the clinical research enterprise at the current time:

• A diverse study population gives fair representation to underserved populations, giving consideration to the population of therapeutic relevance, the practicality of the study design, scientific validity, and commercial imperatives.

2. Why do we want a diverse study population?

Ther are at least five good reasons to increase diversity in clinical studies:

• Commercial relevance
• Therapeutic relevance
• Sharing of burdens and benefits fairly
• Support underserved populations
• Access a larger population

Pharmaceutical and medical device companies are, naturally, most interested in commercial relevance. The label on an approved medication reflects commercial relevance. Traditionally, FDA has focused on whether the study populations in the Phase III pivotal studies and any supplemental studies are consistent the label on the marketed product.

Therapeutic relevance is broader than commercial relevance. In particular, it covers off-label use. For example, if a medication is likely to be popular amongst the elderly, therapeutic relevance dictates testing in that population, even if it is not the official target market.

The related reasons of sharing of burdens and benefits fairly and support of underserved populations are matters of ethics covered by the Belmont Report’s principles of justice and beneficence. Fairness means placing the likely burdens on a subpopulation in line with their likely benefits. For example, testing an expensive drug on people who will not be able to access or afford it is unfair, i.e., exploitative. Some ethicists would say that these principles apply only to underserved populations. In other words, imposing the risks and other burdens on a middle-class, Caucasian, male population is ethically acceptable because of their privileged position in society. In other words, the fairness bar is much higher than that for underserved populations since it is assumed they can look after themselves.

Increasing diversity may cost time and money, but it also expands the available population for a study.

3. Which dimensions of diversity should we address?

In recent years, the clinical research industry has primarily focused its diversity efforts on the compound dimension of race and ethnicity, with some attention to age and sex. The FDA CDER “Drug Trial Snapshots Summary Report 2021” reflects this perspective.¹

FDA’s April, 2022 draft guidance, “Diversity Plans to Improve Enrollment of Participants from Underrepresented Racial and Ethnic Populations in Clinical Trials,” reflects a broader perspective.² It provides “recommendations to sponsors developing medical products on the approach for developing a Race and Ethnicity Diversity Plan … to enroll representative numbers of participants from underrepresented racial and ethnic populations in the United States…” FDA also “advises sponsors to seek diversity in … other underrepresented populations defined by demographics such as sex, gender identity, age, socioeconomic status, disability, pregnancy status, lactation status, and co-morbidity.” In addition to scientific rationales for diverse study populations, the guidance discusses issues of equity for “all clinically relevant populations,” e.g., “health disparities and differential access to health care in certain racial and ethnic populations, many of whom are part of underserved communities.”

Diversity should not just be about race and ethnicity with a nod to age and sex. For one thing, if we want to increase diversity in race and ethnicity, we need to address socioeconomic diversity as well.

There may be a practical limit to the number of dimensions of diversity that a clinical study can handle. If so, lower-priority dimensions could be watched but not managed. Every study participant represents multiple dimensions of diversity, so targeting one under-represented population may also increase diversity in other dimensions.

It may not be practical to select clinical research sites and develop a patient recruitment program optimized for each dimension of diversity, but it may be possible to address multiple dimensions of diversity at a single site or in a single program. At minimum, we can say: “No matter your race, ethnicity, age, sex, sexual orientation, income, education, employment or location, you are welcome to join our study.”

Each dimension of diversity has its own peculiarities. For example, socioeconomics is a multi-factor measure of vague composition that overlaps with other dimensions of diversity, e.g., race and ethnicity. The following is a draft definition:

• Socioeconomics relates to a person’s access to healthcare, based on social factors (e.g., race and ethnicity), financial factors, (e.g., income and wealth), and other factors (e.g., education, employment, location and access to medical, government, and community services).

If we are to address diversity in socioeconomics in an effective manner, we need a measurable definition, which the above definition clearly is not. One of the factors—access to medical, government and community services—may take the other factors into account and serve as a proxy for socioeconomics in general.

The dimensions of comorbidities and body mass index can be addressed by loosening a protocol’s eligibility criteria.

Language diversity can be addressed with translation, interpretation and community outreach, keeping in mind the need for effective communications in the case of a serious adverse event.

Approximately 60% of US adults have a disability related to hearing, sight, physical or intellectual function (not adjusting for people with multiple disabilities).³ The source for this statistic does not discuss degree of disability, but the 20% with hearing loss may, for example, constitute a significant underutilized population.

Approximately 7% of US adults identify as lesbian, gay, bisexual, transgender, or something other than straight or heterosexual.⁴ The AIDS/HIV epidemic activated this population for clinical research.

4. What categories within dimensions should we use?

Categorization within some dimensions is problematic. The dimension of “ethnicity and race” is deeply flawed. Lumping 60% of the world’s population into the category “Asian” is absurd. Combining Europeans from Spain and the native peoples of South and Central America into the Hispanic category makes no sense at all. To compound the confusion, many people fit into multiple categories or inaccurately self-report. Nevertheless, it may be adequate to look at self-reported race and ethnicity—even with the current flawed categories—as a proxy marker for actual diversity.

In theory, the dimension of socioeconomics could be addressed with a multi-factorial points system, but that would rely on unreliable and intrusive self-reported data. A better approach may be to accept self-reporting, with a question like the following:

• Given your income, employment, what you own, what you owe to others, your education, and your access to health, government and community services, would you consider yourself better off, about the same as, or worse off than other people in the country?

The length of this question highlights the challenge of socioeconomic categorization. Keeping in mind that the goal is not to accurately measure socioeconomic level but to assure socioeconomic diversity, perhaps it can be simplified.

The following is a much simpler question:

• Compared to other people, would rate your access to health services better, about the same, or worse than other people in the country?

5. What demographic data should we use?

The US Census Bureau publishes a wide variety of demographic data. Other reliable sources must be found for data that it does not publish. Some dimensions of diversity, e.g., socioeconomics, may require composite measures. To the extent possible, we should agree on a common set of demographic data.

6. How do equity and inclusion relate to diversity?

Equity and inclusion appear to support diversity by removing obstacles that discourage or limit access to a diverse population, especially the underserved, from enrolling in a clinical study. The relationships between the concepts may be more complicated.

We should agree on common definitions. We should be clear that equity does not work to anyone’s disadvantage. The following are draft definitions of equity and inclusion:

• An equitable clinical study gives all patients and study participants fair access to study participation and benefits, regardless of their abilities, needs or other characteristics.
• An inclusive clinical study welcomes the diversity of patients and study participants and makes them feel accepted, respected and valued.

7. Are any reasons, dimensions or categories more important than others?

If a given dimension or category is more important and difficult to achieve than others, we can give it priority. There are two ways to give priority to a subpopulation: representation and effort:

• “Representation priority” occurs when a study seeks to enroll more members of a subpopulation than its share of the otherwise-eligible portion of the general population. For example, a study sponsor may believe that past wrongs against a subpopulation justify its overrepresentation in a life-saving clinical study.
• “Effort priority” occurs when the study sponsor believes proportional enrollment will require extra effort to enroll certain subpopulations. For example, home visits may be much more expensive for study participants in rural vs. suburban areas. Historically underserved populations may require relatively expensive and time-consuming recruitment programs.

Effort priority addresses issues of equity and inclusion. It is probably less controversial than representation priority, just as academic scholarships are less controversial than admission quotas.

8. To what extent should US study sponsors apply the diversity concept to clinical studies in other countries?

The extent to which US study sponsors should apply diversity programs to clinical studies in other countries depends on why we want diversity. For example, if we want diversity for reasons of justice, every country had its own perspectives on ethical concepts related to diversity. For example, race and ethnicity are very different issues in the United States and Mexico. In many countries, discrimination is based not on race and ethnicity at all, but, rather, on tribe, religion, national origin, citizenship, language, enslavement or some other attribute. At a more fundamental level, in some countries, privileged populations consider discrimination a Western concept without relevance to their country. Should US study sponsors apply US values, local values, or some mix of values to questions of justice and beneficence as they relate to diversity?

From a scientific perspective, a US,-based global clinical study could recruit patients in African countries to boost Black enrollment but there are two issues: From a scientific perspective, there are no African Americans in Africa—diets, dietary supplements, medical treatments, and other confounding factors are different in the US and Africa. From an ethical perspective, enrolling Blacks in Africa does not address issues of justice and beneficence in the United States.

9. How transparent should diversity plans and results be?

The general public would like to see more diversity in clinical studies. The reputation of the pharmaceutical industry could certainly benefit from study population diversity. However, even the “best” diversity plan or results will be criticized by some part of the population. Sponsors of studies lacking diversity—even for good reason—may want to keep that information private.

10. Must every study have a diverse population or can diversity be measured across a clinical development program?

Increasing diversity in a clinical study generally increases variability, thus reducing the statistical power of a given sample size. (However, depending on the type of study and the type of diversity (e.g., gender or disability), increasing diversity may have no impact on statistical power.) Restoring statistical power means increasing the size of the study population, thereby increasing study duration and cost, leaving aside any additional enrollment challenges. Unless a study is very large, subpopulation analysis for smaller groups is impossible. Random serious adverse events may raise a false alarm for a small subpopulation. For these reasons, study sponsors commonly conduct their main studies with relatively homogeneous populations and then follow up with smaller studies on excluded (e.g., children) or underrepresented populations.

11. How do we measure results and progress?

Three initiatives have broken new ground on measuring diversity in clinical studies:

• FDA CDER’s “Drug Trial Snapshots Summary Report 2021,” mentioned above, provides diversity data across six broad therapeutic areas for female, White, Black, Asian, and age of 65 or older, based on 50 approved new drug applications (NDAs) and new biologics license applications (BLAs).²
• The 2022 IQVIA report, “Advancing Diversity in Clinical Development through Cross-Stakeholder Commitment and Action,” calculated an “Inclusivity Quotient (IQ)” of race and ethnicity diversity, based on ClinicalTrials.gov results data.⁵ The report found wide variations in diversity across eight therapeutic areas.
• The Clinical Study Diversity Score Project (CSDS) aims to calculate a single quantitative diversity score for each study.^6,7 CSDS scores can be calculated, tracked and managed from a study’s initial planning stage, through patient recruitment and study execution. The scores of different studies can be aggregated and compared to identify best practices. Clinical research sites can be selected based on granular diversity profiles across multiple dimensions.

12. What is a “satisfactory” or “excellent” level of diversity?

It is generally agreed that most past and current US clinical studies have subpar diversity. On the other hand, it is unreasonable to expect perfect diversity. How much diversity is enough? Will it always be a moving target or can we draw some lines? Instead of an absolute measure of diversity, should we focus on rate of improvement?

Final thoughts

Every major clinical research professional and trade association, every major pharmaceutical company and CRO, and numerous academic medical centers and health systems (notably the MRCT⁸) have launched significant programs to advance diversity in clinical research. Lewis Carroll, the author of “Alice’s Adventures in Wonderland,” said: When you don’t know where you are going, any road will take you there.” Answering the 12 questions above will not only help us decide where the clinical research industry is going on diversity but also help us get there.

Diversity—and DEI, more broadly—is the bright and shiny new priority in clinical research. However, if history is any indication, the clinical research enterprise’s focus will inevitably move on to the next bright and shiny new priority. The big question is thus: How can we best to establish robust diversity programs before everyone just says, “That will have to do”?

Norman M. Goldfarb is managing director of Elimar Systems, executive director of the Clinical Study Diversity Score Project (CSDS), executive director of the Site Council, and executive director of the Clinical Research Interoperability Standards Initiative (CRISI)

References

1. The FDA CDER “Drug Trial Snapshots Summary Report 2021” is available at https://www.fda.gov/drugs/drug-approvals-and-databases/drug-trials-snapshots
2. “Diversity Plans to Improve Enrollment of Participants from Underrepresented Racial and Ethnic Populations in Clinical Trials” is available at https://www.fda.gov/regulatory-information/search-fda-guidance-documents/diversity-plans-improve-enrollment-participants-underrepresented-racial-and-ethnic-populations
3. “Disability Statistics in the United States,” Bureau of Internet Accessibility, Nov 29, 2018, https://www.boia.org/blog/disability-statistics-in-the-united-states
4. “What Percentage of Americans Are LGBT?,” Jeffrey M. Jones, Gallup, Mar 3, 2022, https://news.gallup.com/poll/332522/percentage-americans-lgbt.aspx
5. “Advancing Diversity in Clinical Development through Cross-Stakeholder Commitment and Action,” IQVIA Institute for Human Data Science, Nov 2022, https://secure.constellation.iqvia.com/clinical-trial-diversity-report
6. “Calculating Diversity in Clinical Research Studies,” Norman M. Goldfarb, Feb 28, 2023, Applied Clinical Trials, https://www.appliedclinicaltrialsonline.com/view/calculating-diversity-in-clinical-research-studies
7. The Clinical Study Diversity Score Project website is at https://www.portolo.com/Diversity
8. The Multi-Regional Clinical Trials Center of Brigham and Women’s Hospital and Harvard website is at https://mrctcenter.org/