A study of nearly 100,000 participants in the NIH’s All of Us Research Program identified thousands of previously unreported variants in 16 genes that can affect responses to prescription drugs; the findings were published in Science Translational Medicine on Nov. 11, 2026.

Every participant carried at least one variant that could affect a medication’s effectiveness or toxicity, and 24.4% had been prescribed at least one drug that could interact with a variant they carried.

Such variants can change how people process medications, potentially altering the right dose, causing side effects or making a treatment ineffective.

Researchers found a median of nine relevant variants per participant, with potentially interacting medicines including ondansetron, ibuprofen, omeprazole, pantoprazole and atorvastatin; the broader list includes antidepressants, statins, chemotherapy treatments and Plavix.

The team also identified more than 17,000 previously unreported variants and estimated the prevalence of known medication-linked variants across ancestry groups, including populations that have been underrepresented in earlier genetic research.

Using All of Us data, designed to reflect the diversity of the U.S. population, helped researchers address gaps and potentially biased risk estimates in earlier genetic studies.

The findings strengthen the case for genetically guided prescribing: genetic information in electronic health records could prompt clinicians to consider a different drug or adjust a dose.

More than 300 medications have labels describing genetic interactions; one colon-cancer chemotherapy drug can cause serious toxicity in up to 8% of people with certain variants, prompting testing before treatment.

Senior author Philip E. Empey said broad screening may be especially useful for people taking multiple medications or starting clopidogrel after cardiac stenting, but policymakers and insurers must weigh the benefits against costs.

Empey also urged people who have already had pharmacogenomic testing to share their results with pharmacists and prescribers so they can be added to electronic health records.

Wider clinical use still faces hurdles, including limited genome sequencing, physicians’ limited awareness of genetic effects on medications and the complexity of variants that can influence treatment.

All of Us aims to collect genetic data from 1 million Americans, but its future work faces funding pressure: a funding source established by the 21st Century Cures Act ends in fiscal year 2027, and the program expects $122 million after Congress finalizes a budget, down 80% from its 2023 peak.