New GEMRA Method Reduces Bias in Genetic Studies

Researchers have introduced a statistical method designed to improve the accuracy of genome-wide association studies by accounting for medication use. The approach, known as the genetic empirical medication reduction adjustment or GEMRA, aims to prevent biases that can distort estimates of true genetic effects when participants are taking drugs relevant to the trait being studied. The study, published in PLOS Genetics, demonstrates that this heuristic search for empirical adjustment offers a more robust alternative to common strategies such as including medication as a covariate, restricting samples to unmedicated individuals, or analyzing all samples without adjustment.

The GEMRA method applies an empirical adjustment to the phenotypic values of participants who report medication use. By doing so, it seeks to model the effect of medication on the trait of interest more accurately. The authors, including Amanda H. W. Chong and Gibran Hemani, utilized simulations to evaluate the performance of GEMRA. These simulations showed that direct genetic effect estimates using the GEMRA approach exhibited less bias and greater statistical power compared to standard methods for handling medication status in genetic analyses.

To validate the method in real-world data, the team applied GEMRA to estimate medication adjustments for specific traits using multi-population data from the UK Biobank and the Uganda Genome Resource. They focused on statin medication adjustment for the analysis of low-density lipoprotein cholesterol levels. Additionally, they applied antihypertensive medication adjustments for analyses of systolic and diastolic blood pressure.

The results indicated that a relative adjustment, rather than an absolute one, better modeled the effect of medication on low-density lipoprotein cholesterol, systolic blood pressure, and diastolic blood pressure. The study found that a 40 percent reduction in low-density lipoprotein cholesterol appeared consistent across the populations studied. For blood pressure, a 10 percent reduction in both systolic and diastolic measures also appeared consistent across the different groups.

Furthermore, the researchers demonstrated that spurious differences in medication effects can arise across populations if the effect scale is misspecified.

Subject of Research: Statistical method for adjusting medication status in genetic studies

Article Title: Adjusting for medication status in genome-wide association studies

Article References: Chong, A. H. W., Kintu, C., Cho, Y., Fatumo, S., Torres, J., Davey Smith, G., Gaunt, T. R., & Hemani, G. (2026). Adjusting for medication status in genome-wide association studies. PLOS Genetics, 22(11), e1012366. https://doi.org/10.1371/journal.pgen.1012366

Image Credits: AI Generated

DOI: 10.1371/journal.pgen.1012366

Keywords: genetics, statistics, medication, bias, UK Biobank, LDL cholesterol, blood pressure, GEMRA, Adjusting, status, genome-wide, association