Why Alzheimer’s clinical trials need better data on representativeness
As precision neurology advances, Alzheimer’s research still faces major gaps in race and ethnicity reporting, trial eligibility, prescreening data and enrollment infrastructure — limiting confidence that new therapies reflect the populations they are meant to serve.
A therapy is only as trustworthy as the population in which it was tested. That principle is under real strain in Alzheimer's disease research, where a landmark 2026 systematic review in JAMA Network Open examined 88 US-based Phase 3 drug trials conducted between 1997 and 2023 and found that nearly half of published trials, 35 of 71, or 49.3%, did not report any data on participants' race or ethnicity at all. Among trials that did report race, the median share of White participants was 91.3%, what the study's authors at Yale School of Public Health characterized as "more than 90% White in a typical trial". Perhaps most consequential for clinical practice: only 3 of 71 trials — 4.2% — conducted any subgroup analysis of treatment effects by race or ethnicity, and none reported detailed findings on whether safety or efficacy differed across groups. That is not a diversity statistic so much as a data gap with direct clinical consequences: it means the evidence base cannot currently show whether approved Alzheimer's therapies work equally well, or are equally safe, across the population that will ultimately receive them.
A measurement problem as much as an enrollment problem
The scale of underrepresentation compounds an already documented pattern of diagnostic delay. A peer-reviewed analysis of Health and Retirement Study data linked to Medicare and Medicaid claims found that, among patients with delayed dementia diagnosis, the delay was 11% longer for non-Hispanic Black Americans and 40% longer for Hispanic Americans compared with non-Hispanic White Americans, translating to mean delays of 34.6 and 43.8 months, respectively, versus 31.2 months. Overall, the same study found that 42% of older adults with probable dementia had a missed or delayed diagnosis in their Medicare claims records, rising to 46% among Black beneficiaries and 54% among Hispanic beneficiaries. Later diagnosis means a smaller window for enrollment in trials targeting early-stage disease — compounding, rather than merely paralleling, the enrollment disparity documented at the trial level.
Complementing the diagnostic-timing data, a cross-sectional analysis of 19 completed Phase 3 anti-amyloid monoclonal antibody trials covering nearly 19,600 patients found Caucasian participants modestly overrepresented relative to disease prevalence, while Black and Hispanic participants were substantially underrepresented, with an enrollment-to-incidence ratio for Black participants of just 0.10. University of Miami researchers examining the underlying drivers found that socioeconomic factors, income, insurance access, and lifelong exposure to vascular risk factors tied to unequal access to care, rather than race or ethnicity itself, are the primary explanation for elevated Alzheimer's risk and, by extension, disparities in trial access, a framing that treats the disparity as a structural and data problem rather than a biological one.
Where trial design itself creates exclusion
A systematic review of 101 Alzheimer's drug trials conducted between 2001 and 2019 found that common eligibility restrictions, excluding patients with comorbid psychiatric illness (78.2% of trials), cardiovascular disease (71.3%), cerebrovascular disease (68.3%), or requiring mandatory caregiver attendance (80.2%), disproportionately screen out ethnoracially diverse candidates, since these comorbidities are more prevalent in Black and Hispanic populations. Biomarker-based eligibility criteria compound the effect: screening data from lecanemab and elenbecestat Phase 2/3 trials covering more than 10,800 US participants found significantly higher odds of ineligibility on amyloid-biomarker thresholds for Hispanic Black, Hispanic White, non-Hispanic Asian, and non-Hispanic Black participants relative to non-Hispanic White participants. Language access adds a further, largely avoidable barrier: a review of trials registered on ClinicalTrials.gov found that roughly 19% explicitly required English proficiency, a criterion frequently unrelated to a trial's scientific objectives, and one that excludes a meaningful share of the estimated 25 million Americans with limited English proficiency.
Documented mistrust of research institutions is a further, measurable barrier rather than an assumption: a 2024 Pew Research Center survey found 55% of Black Americans believe nonconsensual medical experimentation on Black people is happening today, and 51% believe the health care system was designed to hold Black people back. Financial and logistical barriers matter too: one patient survey found that while 72.5% of respondents said they would be willing to participate in a clinical trial, only 23.9% actually had, with 27.8% citing potential travel costs as a specific barrier.
Data infrastructure as the fix
The most promising interventions treat this as an infrastructure and data-standardization challenge rather than a matter of outreach alone. The NIA's Alzheimer's Clinical Trials Consortium built a centralized system to collect standardized prescreening variables, age, self-reported race and ethnicity, education, zip code, recruitment source, and reason for ineligibility, across sites for the AHEAD 3-45 trial, successfully capturing structured data on more than 1,000 participants in a vanguard phase and demonstrating that centralized, interoperable prescreening data can identify selection bias early enough to correct for it.
Community-based recruitment models built on that same infrastructure have shown measurable results. A pilot embedded in the AHEAD 3-45 trial, funded through the Alzheimer's Association's Equity AHEAD initiative and partnering with community-based organizations to conduct culturally tailored outreach to Hispanic, Filipino, and Korean American adults, engaged 654 individuals across 21 community events and yielded 71 pre-screenings and 25 enrollments from underrepresented groups. A related pilot using a similarly multi-pronged, culturally informed engagement approach produced a 268% increase in the monthly enrollment rate of Black and Latinx older adults into an Alzheimer's cohort study. At the federal level, provisions originally proposed in the bipartisan Equity in Neuroscience and Alzheimer's Clinical Trials (ENACT) Act, expanding outreach and education, increasing diversity among trial investigators and staff, and funding new Alzheimer's Disease Research Centers in areas with higher concentrations of underrepresented populations, were folded into the FY2023 federal appropriations package rather than passed as standalone legislation, but remain part of the current policy landscape supporting trial-diversity infrastructure.
Regulatory data requirements are catching up
The FDA has moved toward making diversity planning a structured, data-driven regulatory requirement rather than a voluntary best practice. The Food and Drug Omnibus Reform Act, enacted in December 2022, statutorily requires a "Diversity Action Plan" for Phase 3 and other pivotal drug trials, specifying enrollment goals disaggregated by race, ethnicity, sex, and age relative to a disease's clinically relevant population, along with the operational strategy for meeting them. Even ahead of the mandate taking full legal effect, voluntary submissions have grown substantially, from a combined 139 diversity plans across FDA centers in FY2023 to 206 in FY2024, suggesting sponsors are already building the data infrastructure the mandate will require. Complementary FDA guidance on trial design recommends specific, data-supported approaches for meeting these goals: broadening eligibility criteria where scientifically appropriate, avoiding unnecessary exclusions, and using adaptive or enrichment trial designs.
The informatics opportunity
For health data and clinical research informatics leaders, the throughline across all of this evidence is that representativeness is fundamentally a data problem: inconsistent race and ethnicity reporting taxonomies across trials, eligibility criteria that are not benchmarked against real-world comorbidity prevalence, and prescreening data that is not centralized or interoperable across sites. Solving it does not require waiting for a single legislative or scientific breakthrough. It requires the same kind of structured-data discipline, interoperability standards, and centralized reporting infrastructure that health systems already apply to other clinical quality domains — applied deliberately to the population that clinical trials are supposed to represent.
Kenneth R. Deans, Jr., DHA, MBA is the president and CEO of Health Sciences South Carolina.