Real-World Data Estimates of Racial Fairness with Pharmacogenomics-Guided Drug Policy

点击翻译按钮获取中文摘要

Project Summary A pharmacogenomics-guided drug policy includes the genomic profile of an individual’s drug response with other clinical characteristics (age, body weight, etc.) and may improve the safety and effectiveness of drug therapy. Thus, in recent years several medical centers in the United States have implemented clinical pharmacogenomics services to support such policies. Among the services that can be supported, preemptive clinical genotyping services produce pharmacogenomic data before it is known that a particular drug may be needed by a patient. Preemptive clinical genotyping services that cover genetic markers primarily based on populations of European ancestry, however, can have reduced performance of a policy to identify well-tolerated medications in understudied groups. Worse performance in the understudied groups is, in part, due to being more likely to have an indeterminate drug response phenotype when compared to a European ancestry group. Having more indeterminate drug response statuses in some racial subgroups translates in to more occurrences of “missing data” in assessments of an individuals’ drug response, thus resulting in lower racial fairness. One possible solution to this challenge of knowing if low racial fairness is a problem, is to estimate the pharmacogenomic- guided drug policy performance and fairness for different racial subgroups a priori. The specific objective of this project is to use All of Us research program (AoU) data to derive evidence of the potential unintended consequence of low racial fairness that can exist with a new pharmacogenomic-guided drug policy. The AoU data is uniquely suited to generate such evidence given that it includes a diversity of racial subgroups and a variety of data types, including from electronic health records and clinical whole genome sequencing data. We will conduct an observational cohort study using the AoU data to assess the performance of pharmacogenomics- guided drug policies to identify well-tolerated medications (Aim 1), and quantify the potential impact of differential data access among patients on performance (Aim 2). We will also study the impact of differential data access on the racial fairness of pharmacogenomics-guided drug policy (Aim 3). Outcomes of this work will demonstrate one strategy to produce evidence from real-world data that can be expanded upon and studied further in future research. Presenting this type of evidence prior to approving pharmacogenomics-guided drug policy holds promise to inform Pharmacy & Therapeutics committee decision-making.