3

Defining, Collecting, and Sharing Comprehensive and Consistent Data on Trial Diversity

Panelists discussed key components of national, interoperable, and accountable systems for collecting and sharing condition-specific demographic data, including how to share clinical trial data across organizations and sectors to enable continuous learning and improvement in trial diversity. The session was moderated by Carla Rodriquez-Watson, director of research at the Reagan-Udall Foundation for the U.S. Food and Drug Administration (FDA), an independent, nonprofit organization that establishes public–private partnerships with diverse constituents to help advance FDA’s mission. She mentioned an example of a foundation program relevant to this workshop, the Real-World Accelerator to Improve the Standard of Collection and Curation of Race and Ethnicity Data in Healthcare

(RAISE), which is an initiative to improve collection and curation of race and ethnicity data in real-world datasets.¹

ONC INITIATIVES TO SUPPORT DIVERSE CLINICAL TRIALS

Stephen Konya, senior advisor to the Deputy National Coordinator and Innovation Portfolio Lead at the Office of the National Coordinator for Health Information Technology (ONC), discussed several ongoing ONC activities that support collecting and sharing trial data from diverse populations.

ONC oversees the U.S. Core Data for Interoperability (USCDI), which is the standardized set of data elements that should be shared through all electronic health record (EHR) vendor companies, Konya explained.² The goal of this program is to better enable data sharing by setting a minimum set of data categories that are available for exchange across all EHR vendors. This is relevant because clinical trials often rely on data in the EHR for various purposes. USCDI is in its fifth version and continues to expand and revise with input from public comments. Categories of recently expanded data elements include sexual orientation and gender identity; race, ethnicity, and language; women’s health; and social determinants of health. In addition, ONC released a concept paper earlier this year for public comment that outlines the agency’s commitment to advancing health equity by design in health IT systems.³

ONC recently partnered with the White House Office of Science and Technology Policy (OSTP) on a series of requests for information (RFIs) on optimizing data capture for clinical trials. Konya said this effort stems from the challenges faced during the COVID-19 pandemic in quickly initiating clinical trials across multiple sites and recruiting diverse patient populations. ONC is involved in three initiatives that emerged from the RFIs. The first, the Advancing Clinical Trials Readiness Initiative, is a collaboration with the Advanced Research Projects Agency for Health (ARPA-H) to improve the nation’s ability to conduct clinical trials by enabling 90 percent of all eligible U.S. individuals to participate within a half hour of their home.⁴ The second is the USCDI+ Cancer Program, which is developing a new set of oncology-related real-world data (RWD) elements to further prevention,

___________________

¹ See https://reaganudall.org/programs/research/raise (accessed September 9, 2024).

² See https://www.healthit.gov/isp/united-states-core-data-interoperability-uscdi (accessed September 9, 2024).

³ Advancing Health Equity by Design and Health Information Technology: Proposed Approach, Invitation for Public Input, and Call to Action (ONC, 2024). See https://www.healthit.gov/sites/default/files/2024–04/ONC-HEBD-Concept-Paper_508.pdf (accessed September 9, 2024).

⁴ See https://arpa-h.gov/news-and-events/arpa-h-advances-initiative-improve-clinical-trials (accessed September 9, 2024).

diagnosis, research, and care.⁵ It could also serve as a use case for building complex health IT infrastructure. Third is the Interoperability Bridge project, launching in 2024, which will develop interoperable implementation guides to support a range of activities (e.g., registering multiple sites with diverse patient populations) in collaboration with accelerators associated with the H7 Fast Healthcare Interoperability Resources (FHIR) health care data exchange standards (e.g., Vulcan, CodeX, FAST).

CATEGORIZING RACE AND ETHNICITY FOR DATA COLLECTION

Despite a modest level of improvement in the collection of data on race and ethnicity in recent decades, there is still “a tremendous amount of variation in how data are collected,” said Sarah Hudson Scholle, principal at Leavitt Partners and formerly a vice president at the National Committee for Quality Assurance (NCQA). One major challenge organizations face, according to Scholle, is deciding which source is best for determining race or ethnicity. When using different data sources, it is often not clear the extent to which data on race and ethnicity are current or accurate. She said that while self-reported data are the “gold standard,” organizations often combine data from multiple data sources. For example, Medicaid or Medicare enrollment data, which can vary in quality and completeness, may be different than self-reported race and ethnicity data collected through a survey.

Scholle described the recently updated standards for the collection of race and ethnicity data from the Office of Management and Budget (OMB).⁶ It calls for combining race and ethnicity into one question and allowing respondents to select as many answers as apply. It also adds a racial/ethnic category for those who identify as Middle Eastern or North African. She emphasized the role of leadership in making collecting race and ethnicity data an organizational priority. Scholle provided examples of how organizations are doing this, including the Centers for Medicare & Medicaid Services (CMS) highlighting differences in achievement on quality metrics by race and ethnicity and NCQA requesting stratification of health plan quality reporting by race and ethnicity.

It is also important to earn the trust of the people being asked to provide their data. They must have confidence that their information will be used for their benefit and not cause harm. In Scholle’s current work on

___________________

⁵ See https://www.healthit.gov/topic/interoperability/uscdi-plus (accessed September 9, 2024).

⁶ Statistical Policy Directive No. 15 (Directive No. 15): Standards for Maintaining, Collecting, and Presenting Federal Data on Race and Ethnicity (Federal Register, 2024). See https://www.federalregister.gov/documents/2024/03/29/2024-06469/revisions-to-ombs-statistical-policy-directive-no-15-standards-for-maintaining-collecting-and (accessed September 9, 2024).

patient-reported data, people want to understand why they are being asked to provide their data, how their data will be used, with whom it will be shared, if they will have access to it, and whether the organization collecting their data is worthy of their trust.

IMPROVING RACE AND ETHNICITY DATA COLLECTION AND USE

Vindell Washington, chief clinical officer and director of the Health Equity Center of Excellence at Verily (formerly Google Life Sciences), reflected on a recent partial review of Verily’s ongoing registries, which showed that each registry used a different definition for race and ethnicity. He commended the recent update of the OMB standards for collecting race and ethnicity data discussed by Scholle. While the new standards are not perfect, having consistency across the clinical trial space is crucial. Collecting diversity data is becoming a routine part of the clinical trial process. This might be similar to the data collection people routinely encounter in other settings, such as government forms like the U.S. Census. It is important that people see themselves reflected in the race and ethnicity data elements presented to them.

Screening for clinician bias in enrollment is another area of focus for Verily. Perceptions and biased choices on the part of those providing care and conducting research can impact patient outcomes. Washington encouraged screening for bias among all persons who are involved in recruitment and conduct of a trial, as well as the system at large.

Finally, Verily is leveraging data gathered in clinical trial settings “to power more advanced data science methodologies for ensuring equitable participation in trials,” Washington said. The availability of standardized data to power advanced algorithms can improve the ability to ensure that limited resources are effectively applied to achieve health equity goals. For example, Verily published a framework for applying a Restless Multi-Armed Bandit model (which collects personal data to make personalized movie or music recommendations) to improve equitable recruitment for trials and evaluate resource allocation (Killian et al., 2024). Washington added that the success of trial enrollment should be measured relative to not only cost and time but also equity.

FDA INITIATIVES TO SUPPORT DIVERSE CLINICAL TRIALS

Jamie Brewer, a medical oncologist and clinical team lead in the Office of Oncological Diseases at FDA, said that FDA regularly observes disparities by race and ethnicity in datasets submitted to the agency for regulatory review. The FDA Oncology Center of Excellence has a range of initiatives

focused on increasing representation of historically underrepresented patient populations in clinical trials. One example is Project Equity,⁷ launched in 2020 to provide a formal mechanism to encourage sponsors to proactively improve diversity among participants in their studies. Project Equity is also working to promote consistency and coordination of review practices, policies, research, and engagement internally across review divisions. Brewer said that FDA is “developing policies and guidances, fostering research and policy collaborations, incorporating diverse perspectives in regulatory activities across the Oncology Center of Excellence programmatic areas, and engaging with external stakeholders.”

ELEMENTS OF SUSTAINABLE CHANGE IN CLINICAL TRIAL DIVERSITY

Michael Currie, a health care consultant and formerly chief health equity officer at UnitedHealth Group, offered his perspective on four elements of promoting equitable representation in clinical trials.

Understand the Audience

Currie said that different researchers, funders, and partners can have differing levels of cultural awareness and sensitivity regarding issues of diversity and inclusion in clinical trials. Not everyone involved in a trial will have that awareness. Conversations can be tailored to help those involved foster the heightened level of appreciation, awareness, and understanding.

Make the Business Case for Diversity

“The moral imperative [for diverse clinical trials] only takes you but so far,” Currie said. It is also important to make the business case for trial diversity. Funders and others should be made aware of the potential business benefits to them of investing resources in improving trial diversity. These conversations may require those developing and conducting trials to learn new concepts and language related to resource investment.

Promote Meaningful and Sustainable Change

Currie detailed which partners to involve when developing policies to guide meaningful and sustainable change, including who will do the work, how they will do it, and the time frame for it. First, involve potential

___________________

⁷ Also known as the “OCE Equity Program.” For more information, see https://www.fda.gov/about-fda/oncology-center-excellence/oce-equity-program (accessed September 9, 2024).

clinical trial participants. Second, develop policies in collaboration with providers, including clinicians and others in the community who are trusted sources of information. Those individuals can be important partners in recruiting participants to clinical trials. Third, partners are essential because the work cannot be done alone.

Commitment, Perseverance, and Resources (“CPR”)

Sustainable change requires unwavering commitment from organizational leadership Currie said. It also takes perseverance and resources, including money, time, and “sweat equity,” which circles back to the need for commitment from leadership.

TOOLS AND IMPLEMENTATION

Rodriguez-Watson asked panelists to reflect on whether demographic data are insufficient to support recruitment of representative trial populations or whether the data that are available are not being used efficiently. Konya said it is a bit of both, based on the responses to the OSTP RFI and the subsequent engagements. Some responses to the RFI discussed issues with registering new or different trial sites, even though establishing new sites in communities where target populations live can reduce barriers to participation. Other responses discussed the challenges stemming from the varying ways platforms and organizations capture and share data. It is harder to efficiently use data when they are not shared in similar formats. Konya said ONC is developing “plug-and-play” application programming interfaces (APIs) to enable data sharing between parties, standardized APIs, and implementation guides. One goal is a clinical trial infrastructure for data collection that can be rapidly used to respond to public health emergencies.

Panelists then discussed resources and strategies for promoting implementation and uptake of existing tools for data sharing. From an insurer’s perspective, Currie highlighted three types of partnerships that can be leveraged: partnerships with philanthropy for the sharing of resources; strategic partnerships in one’s marketplace or geographic footprint, including allies as well as changemakers; and partnerships with the provider community, including individual providers, hospitals, and health systems. From an insurance perspective, the latter are key partners in designing awareness strategies that lead to robust enrollment. Scholle noted that capacity varies across provider communities, and many do not have the infrastructure to support robust data collection. For example, mental health providers were excluded from the Health Information Technology for Economic and

Clinical Health Act,⁸ which offered incentives to increase the uptake of EHR systems in primary and specialty medical settings. Community-based organizations, which provide social services and have the trusted relationships needed to potentially gather sensitive information, may not have the infrastructure to easily collect and share data. Scholle cautioned that steps to improve data collection and sharing should not become “a mandate on an underresourced part of the health care system.”

Washington and Scholle discussed the important role of regulatory agencies in this space, such as the National Institutes of Health (NIH) and FDA policy changes requiring diversity in trials. Also important, Washington said, is how agencies handle the absence of information; for example, CMS denied coverage for an early Alzheimer’s treatment due to insufficient diversity in the trial population. He noted that “no one lever alone is sufficient to actually drive the change.” Brewer said issues of insufficient data will persist if study sponsors keep using the same clinical sites when they know those sites lack the populations they need for diverse enrollment, describing it as “an unforced error of insufficient data.”

DATA ON WORKFORCE AND PROVIDER DIVERSITY

Ann Taylor, former chief medical officer of AstraZeneca and cochair of the Forum on Drug Discovery, Development, and Translation, observed that having health care providers representative of the populations they serve can be a component of building trust, but no central or standardized processes exist for collecting, storing, or using these data. Panelists discussed that efforts to collect data on provider race and ethnicity have been hampered by a range of concerns. Scholle noted resistance from some providers to include this information in a health plan directory due to concerns about bias against clinicians of different racial or ethnic groups.

Currie, who studied challenges to collecting provider demographic data, shared two barriers he observed. One is the lack of assurance that the information provided will not be used for a different purpose. Another is that many clinicians who identify with a racially and ethnically minoritized population want to be seen as a provider of quality care for all patients and are concerned about being labeled as a provider for only a particular race or ethnicity. Currie suggested that a work-around is to partner with organizations that already gather demographic information on providers. For example, the Council for Affordable Quality Healthcare collects data on provider race and ethnicity from those willing to share it. The American

___________________

⁸ Enacted under Title XIII of the American Recovery and Reinvestment Act of 2009 (Pub. L. 111–5).

Medical Association and National Medical Association also collect such data.

Washington said Verily has approached this question from a community perspective, asking “where do people live and who are the providers who are practicing in those areas?” He said the Restless Multi-Armed Bandit framework can be used to assess resource allocation by identifying channels that are performing to deliver the desired outcome (e.g., a diverse clinical trial population), and then directing more resources to support recruitment to that channel rather than deciding site funding levels up front based on community or practice demographics.

Drawing again from the responses to the OSTP RFI, Konya added that diversity is also needed among professionals in the clinical trials enterprise who are not providers, such as clinical informaticists and those who build IT systems for health care. Some RFI responses suggested that grant funders include a requirement to collect data on the diversity of the trial investigators, while others suggested that FDA require it. These data could enable calculating national-level statistics on the diversity of trialists, informaticists, developers of certified health IT products, and others involved in clinical trials. With regard to workforce, Konya mentioned the Public Health Informatics Training Program, an ONC initiative to support training a more diverse informatics workforce and developing a more culturally relevant curriculum.⁹

___________________

⁹ See https://www.healthit.gov/topic/interoperability/investments/public-health-informatics-technology-phit-workforce-development (accessed September 9, 2024).

This page intentionally left blank.