Clinical Medication Trials Fail to Capture the Diversity of Their Locations, Potentially Limiting Study Findings

Anchiy/E+ via Getty Images

Clinical medication trials are often conducted in counties with diverse populations. But research shows that white enrollees are overrepresented in clinical trials. This lack of diversity can affect how well a clinical finding applies to the entire U.S. population.

Take the popular medication warfarin, for example. Prior to 2013, clinical trials on the efficacy of warfarin were conducted primarily on people of European ancestry. After the inclusion of a more diverse population in studies, it was discovered that proper dosing for the medication can differ by race.

In 2023, Congress passed a law aimed at improving diversity in clinical trials. To understand the current level of diversity in clinical medication trials, GoodRx looked at where trials took place in 2022 and which populations lived in those counties.

Over 13,000 medication trials recruited in over 1,000 counties in 2022

GoodRx Research used ClinicalTrials.gov to find out where clinical trials for FDA-regulated medications actively recruited in 2022. These were specifically trials studying drugs that were currently approved or seeking approval to be sold in the U.S.

We found that an overwhelming majority of people lived in a county with a clinical trial. In 2022,13,070 distinct clinical trials for FDA-regulated medications recruited for participants in 1,063 counties. Over 280 million Americans lived in those counties (85% of the total population). Clinical trials can recruit in multiple locations, and on average, a single clinical trial in 2022 recruited in about eight different counties.

The map below displays the number of interventional FDA-regulated medication trials that recruited for patients at any point in 2022, by county.

The top 10 counties with the most number of clinical trials for FDA-regulated medications that recruited in 2022 were:

1. Los Angeles, California (2,805 trials)

2. Harris, Texas (2,616 trials)

3. New York, New York (2,363 trials)

4. Suffolk, Massachusetts (1,852 trials)

5. Cook, Illinois (1,643 trials)

6. Miami-Dade, Florida (1,453 trials)

7. Philadelphia, Pennsylvania (1,336 trials)

8. Dallas, Texas (1,322 trials)

9. Maricopa, Arizona (1,265 trials)

10. Orange, California (1,215 trials)

Metro areas had the most number of clinical medication trials recruiting in 2022

To make sense of where trials are happening, we categorized counties by geographic area. The U.S. Office of Management and Budget (OMB) defines areas as metropolitan (metro) if they have at least one urban cluster with a total population of 50,000 or more and as micropolitan (micro) if they have at least one urban cluster of between 10,000 and 50,000 people.

These metro and micro areas can be further divided into either central or outlying counties. Central counties have a significant portion of their population in the urban cluster. Outlying counties have significant commuting ties to the central counties.

All other counties are neither metro nor micro. And they don't have any commuting ties to any metro or micro areas.

This grouping is more informative than using a rural versus urban classification because a rural county with commuting ties to a metro or micro area may have access to hospital systems, road networks, and sources of employment that rural counties without commuting ties may not have.

The table below shows how many distinct clinical medication trials recruited participants in 2022, by geographic area. Keep in mind that a clinical trial can recruit patients in multiple locations and can be counted multiple times in the table below.

When classifying counties by geographic categories, we found that central metropolitan areas had the most number of clinical medication trials, followed by counties with commuting ties to central metro counties. The total number of clinical medication trials that recruited in 2022 was 13,070. Of these trials,13,019 clinical medication trials recruited in at least one central metro county.

Counties with commuting ties to central micro counties had the largest number of clinical medication trials available to them on a per person basis. This means that in those counties, the number of trials available was large compared to the relative total population living in these areas.

But keep in mind that there were only two counties in that group. And one county had a facility with many trials available. That's what led to these counties having such a high per person number of clinical medication trials.

Counties with clinical medication trials are more diverse

Overall, we found that counties with FDA-regulated clinical medication trial sites tended to be less white than counties without clinical trials. In fact, counties with clinical medication trials tended to have a higher share of Asian, Black, and Hispanic or Latino populations. However, these counties also tended to have a lower share of people ages 65 and older.

The following bar chart compares the demographics in counties with a trial versus those without a trial in 2022. We specifically looked at the percentage-point differences in race/ethnicity and age between these groups.

A green color bar means that counties with a trial had a higher share of that demographic group than counties without. An orange color bar means that counties with a trial had a lower share of that demographic group than counties without.

We found that the Asian population share was 5.5 percentage points higher in counties with clinical medication trials than in counties without trials. The same trend was true for the share of the Black population (+4.8 percentage points) and the share of Hispanic or Latino population (+4.4 percentage points).

Clinical drug trials don't capture the diversity of their locations

But despite clinical medication trials being conducted in counties that are generally more diverse, many research studies point to low diversity in trials.

Research looking into trials that led to FDA approvals of oncology medications from 2008 to 2018 showed that Black and Hispanic populations were underrepresented when comparing their participation rates relative to their proportion of the U.S. cancer population.

Another report found similar results: Black/African American and Hispanic patient inclusion in clinical medication trials failed to reach U.S. demographic levels on average across interventional trials in the past decade.

Yet another study showed that the Black or African American share of participation in clinical medication trials was representative of their share of the U.S. population, but not for other racial or ethnic minority groups.

It is important to note that race and ethnicity data has been reported inconsistently over time. This makes it difficult to truly capture the scope of diversity in clinical trials.

Boosting diversity in clinical medication trials

GoodRx research shows that clinical medication trials take place in counties with more diverse populations. But this diversity is not reflected in trial participation. Barriers to diversity in clinical medication trials are multifaceted and include willingness to participate, trust in the medical system, socioeconomic factors, and trial site selection.

In 2023, Congress passed a law directing drug companies to submit an action plan to enroll diverse populations for FDA medication trials. Sponsors can use the FDA's diversity plan guidance as a framework for their plan.

The guidance encourages setting enrollment goals for underrepresented racial and ethnic populations. It also recommends creating an action plan for operational measures and retention strategies to enroll and retain diverse participants, such as providing transportation or offering language assistance for better accessibility.

While the effects of the new law remain to be seen, federal policy supporting diversity in clinical medication trials is a step in the right direction.

Co-contributors: Tori Marsh, MPH, Jeroen van Meijgaard, PhD

Methodology

Clinical trials facilities data: We extracted data from ClinicalTrials.gov via the Aggregate Analysis of ClinicalTrials.gov (AACT) Database. This source contains static archived copies of the ClinicalTrials.gov database that are saved at the first of each month. For this analysis, we downloaded all static archived pipe-delimited flat files in 2022 (11 files).

Census demographic data: We used the U.S. Census Bureau's Hispanic or Latino Origin by Race (Table B03002) and Sex by Age (Table B01001) to obtain estimates of race, age, and sex.

Selection of clinical trials sites: The purpose of our study was to determine where drug interventional clinical trials for FDA-regulated drugs in the U.S. actively recruited/enrolled patients in 2022. To achieve this, we filtered to relevant facilities using two files from each of the static downloads from the AACT database: (1) studies.txt and (2) facilities.txt. We filtered studies using the studies.txt by selecting NCT IDs that had an overall status of "recruiting" or "enrolling by invitation," a study type of "interventional" and a flag of "t" in the FDA regulated drug flag. An NCT ID is a unique identification code given to each clinical study upon registration at ClinicalTrials.gov. We used the facilities.txt and filtered only to the facilities of our filtered list of NCT IDs from the studies.txt mentioned prior. We further limited facility country to "United States'' and focused on those that had a status of "recruiting" or "enrolling by invitation." Our final dataset is a list of facilities from each of the 11 static copies of the AACT database.

Assigning clinical trial sites to a county: Using the list of facilities from the 11 static copies of the AACT database, we filtered to a list of distinct facilities (name, city, state, zip code), and fed the facility information into the Google Geocoding API to get information on latitude/longitude of the facility. The latitude and longitude was then used to geolocate the facility to a county using Census 2020 TIGER/Line ShapeFiles. We linked a county to a core-based statistical area (a.k.a. geographic areas) using Census March 2020 Delineation files. The geolocated facility data was merged back onto the NCT ID data to assign an NCT ID to a geolocated facility. In total, we excluded 74 NCT IDs and facility combinations due to data quality issues (<1% of the total number of observations).

Analyses: To get the number of distinct clinical trials that recruited in 2022 in each county, we counted the distinct number of NCT IDs by county. An NCT ID is mapped to a county using the geolocated facility information as described above.

To get the number of distinct clinical trials that recruited in 2022 by geographic areas, we grouped counties by their mapped geographic areas and counted the distinct number of NCT IDs by group.To get the clinical trials per 100,000 person by geographic areas, we divided the distinct number of NCT ID by geographic area by the total population of the counties in that geographic area that had a clinical trial.

To get the percentage point difference in population characteristics between counties with a clinical trial versus counties without, we first grouped counties based on whether they had at least one clinical trial that recruited in 2022. We then found the proportion of each of the following population groups as a proportion of the total population for counties with a clinical trials and counties without:

(1) Female
(2) Ages 65+
(3) Hispanic or Latino
(4) White - not Hispanic or Latino
(5) Black - not Hispanic or Latino
(6) Asian - not Hispanic or Latino
(7) American Indian and Alaskan Native (AI/AN) - not Hispanic or Latino
(8) Native Hawaiian and Other Pacific Islander (NHPI) - not Hispanic or Latino
(9) Two or more race - not Hispanic or Latino
(10) Other race - not Hispanic or Latino
(11) No health insurance coverage

We then subtracted the proportion of the population group in counties without clinical trials from the proportions of the same population group in counties with clinical trials.

Public link

Please use the above public link if you want to share this noodl on another website.