July 26, 2022
Thank you for the opportunity to speak today on how current FDA policies contribute to health inequities.
The National Center for Health Research is a nonprofit think tank founded in 1999 that conducts research and scrutinizes research conducted by others to evaluate medical products, procedures, and policies. We do not accept funding from companies whose products we evaluate.
There are many reasons for health inequities in the U.S., but today I will focus on one that usually gets no attention: Federal laws regarding diversity in clinical trials.
The U.S. Department of Health and Human Services requires research studies to include people representing diverse racial and ethnic backgrounds. This is not always enforced, but the requirement is in the law and NIH, CDC, SAMHSA and other agencies make an effort to abide by this law.
The one exception among federal public health agencies is the FDA, which encourages but does not require diversity in clinical trials. They justify this because the agency doesn’t pay for the studies – the companies that make the products pay for the studies. However, taxpayers do pay for FDA staff that regulate these products, and taxpayers also pay for the products themselves.
Our NCHR Study of Racial Diversity
To see the impact of the lack of a requirement for FDA, we examined 22 of the highest risk medical devices reviewed by the FDA Advisory Panels for 4 recent years. We found:
The number of nonwhite patients in key trials ranged from 4 (3%) to 6,788 (17%).
Of 19 treatment devices, only 7 had analyses for racial groups for effectiveness and only 3 for safety.
69% to 99% of the patients in the studies were White. The number of nonwhites was as low as 4; 1 for Hispanics.
There were too few patients in most racial or ethnic groups to draw meaningful conclusions.
Even when subgroup analyses were conducted, their conclusions were often discredited by the FDA, blamed on chance differences due to small sample size. If there was a lack of diversity or even when racial or ethnic differences were significant, that information was were often not included on the label, which is the main source of information for physicians and patients.
Recent Examples from the FDA
When the FDA reviewed the data on Aduhelm for Alzheimer’s Disease, they focused on 2 studies comparing placebo to high and low dosage:
Study 301: 8 Black patients (half of 1%) and 37 Hispanic patients (2%)
Study 302: 11 Black patients (less than 1%) and 67 Hispanic patients (4%)
When the FDA reviewed data on the Reducer circulatory system device for angina, they found that there were no Black or Hispanic patients, and more than 80% of the patients were White or male, or both. And yet, most patients with angina are not White males.
Why is Diversity Important in Clinical Trials?
Response to treatment can vary due to genetics, health habits, metabolism, body part size/shape and other factors. If you exclude certain groups, you don’t know what works best for them. Keep in mind that you need enough patients to study safety and effectiveness for patients in each group.
Example – Lutonix drug-coated balloon catheter to open blocked arteries
You can see on this graph that the device seemed effective compared to the control group when men and women were combined. However, you can also see that the device was only effective for men, not for women. In fact, women did better with placebo. This is an example of how evaluating safety and effectiveness for a specific demographic group can provide information that is completely different from an analysis of a diverse group as a whole.
Shortcomings of Very Small Samples
When the racial or ethnic group is very small, the results may not be generalizable to the larger population that those patients represent. A few outliers can have an outsized effect on the outcomes – resulting in significant differences where they do not exist. Or, real differences may be apparent but not be statistically significant because the sample lacks statistical power.
If an analysis is conducted on a small number of patients of a particular ethnic group, any differences could easily be due to chance. For example, the graphic below shows that the new drug seems to be more effective than an older drug (40% effective compared to 30% effective) , but that difference is not statistically significant. However, if the same difference was based on a much larger sample, it would be statistically significant.
In conclusion, when the FDA approves a medical product that has not been evaluated on a relatively large number of patients in a specific racial or ethnic group, it is not possible to conclude whether the product is safe or effective for members of that group. If the FDA does not require adequate diversity in clinical trials used as the basis for FDA approval, then the agency should only approve those products for the types of individuals studied. That precision in approval decisions would create the incentive needed for companies to improve diversity in their trials and conduct appropriate subgroup analyses.
1. Fox-Rawlings SR, Gottschalk LB, Doamekpor LA & Zuckerman DM. (2018) Diversity in Medical Device Clinical Trials: Do We Know What Works for Which Patients? Milbank Quarterly, Vol 96 (3) 499-529.