by Kirsten Bibbins-Domingo, PhD, MD, MAS
Many of us do not routinely read the genetic literature, but for those of us interested in health disparities, a study in this week’s NEJM and its implications are critically important.
Here’s what the study found: Multiple patients in a leading genetic testing laboratory - all of whom were African Americans - tested positive for pathogenic mutations causally linked to hypertrophic cardiomyopathy (HCM) that were later determined to be benign variants. Since diagnosis of a causal risk variant prompts testing of first degree relatives, this means that others (e.g., brothers, sisters, parents, children of the affected individual) were also tested, given a diagnosis of causal genetic risk for HCM that later had to be (or should have been) amended to reflect that their genetic tests were benign.
How could this happen? Although African Americans are disproportionately affected by HCM, none of the studies that identified these variants now known to be benign included individuals of African descent. Although African Americans with HCM might have been included the original discovery phase, they were not explicitly included in the important secondary confirmatory studies. Simply put, by not including a diverse sample of patients, these studies got the wrong answer. And how much diversity would have been required to avoid this error? The NEJM authors estimate that just 10% diversity in the control sample would have given us a 50/50 chance of identifying the misclassification error.
Bummer! But let’s be clear what translating bad science into clinical practice means for African American patients. Although we don’t know the numbers, almost certainly some African Americans with these genetic variants but without signs or symptoms of HCM or with non-descript, atypical, or borderline signs or symptoms:
- Had to live with the anxiety of possible sudden death or potential heart failure in the future as a result of this diagnosis.
- Had to undergo serial cardiac testing and perhaps even treatment because of the high degree of concern for HCM.
- Had to forego good jobs that have physical requirements (e.g., with police or fire departments) because of this higher genetic risk.
- Had to cease high performance athletic activities, missing out on scholarships and other opportunities for elite athletes. But importantly, others likely altered even basic physical activity essential to good health.
- May have had difficulty with insurance coverage because of this pre-existing genetic condition (although there are protections under the law).
- Altered their plan for having children, fearful of passing on this genetic risk.
- Were falsely reassured about the absence of genetic risk, because the errors in this body of literature can go in both directions (a true genetic risk may have falsely been called benign).
The authors point out the need for more representation of diverse patient populations in clinical studies. They also highlight the need for open access to data, as it was only through subsequent analyses by other groups that the variants initially identified as causal were found to be benign. Finally, they note the critical need for agile responsive electronic health records that can be updated to reflect the latest scientific thinking so that mistakes in science that become part of clinical practice can be more quickly rectified.
I wonder how many of the studies of HCM used known racial disparities to justify the need for their work to the NIH or other funders. I wonder how often racial disparities were used in applications to the IRB to support enrollment of patients in genetic studies and even included in informed consent to highlight potential benefits of studies. We continue to let racial disparities be used to bolster the significance section of research grants, but we do not hold scientists and funders (e.g., the NIH) accountable when poor science is performed simply because study populations do not reflect the diversity of the US or, in this case, even the diversity of the patients affected by the disease.
Assuring diversity in our study populations is not just a noble goal - it is essential to good science. And the consequences of the poor science that results from these types of studies are not benign - when translated into clinical practice, they cause real harm for our patients.
For those of us interested in addressing disparities in health, advocating for representation of diverse participants in our studies is never more critical than it is today, lest the disparities we seek to address are in fact worsened by the poor science we – by our silence – support.
Thanks to Esteban Burchard, Sam Oh, and Josh Galanter for comments on an earlier draft of this post.