Defense Date
2026
Document Type
Dissertation
Degree Name
Doctor of Philosophy
Department
Pharmaceutical Sciences
First Advisor
Elvin T. Price
Abstract
Adverse drug events account for approximately 1.5 million emergency department visits and 500,000 hospitalizations each year, resulting in an estimated $30.1 billion in annual medical costs. Older adults exhibit a markedly increased susceptibility to ADEs, with a nearly sevenfold higher risk of hospitalization compared to younger individuals. In this population, ADEs commonly manifest as disturbances in mood, mentation, and mobility. Pharmacogenomics enables the optimization of pharmacotherapy by informing individualized medication selection and dosing strategies, yet older adults remain underrepresented in PGx research. The objective of this dissertation was to determine the prevalence, clinical significance, and population-level impact of actionable pharmacogenomic variation in community-dwelling older adults, and to identify pharmacogenomic profiles associated with adverse clinical outcomes.
This dissertation utilized a combination of community-based pharmacogenomic cohort analyses and statewide administrative claims data to address these objectives. Initially, pharmacogenomic implementation strategies and resources relevant to older adults were systematically reviewed to establish the clinical basis for integrating pharmacogenomics into geriatric pharmacotherapy. Subsequently, actionable pharmacogenomic variants and gene–drug mismatches involving CYP2C19, CYP2D6, and CYP2C9 were characterized in community-dwelling older adults participating in the Translational Approaches to Personalized Health collaborative. Associations between gene–drug mismatches and clinical outcomes related to mood and mentation were assessed using the Patient Health Questionnaire-4 and Mini-Mental State Examination. The burden of actionable CYP2C19 variation among older adults prescribed CYP2C19 substrate medications was estimated by integrating genotype frequencies from the TAPH cohort with medication exposure and falls data from the Mobile Health and Wellness Program cohort. Finally, explainable machine learning approaches were applied to the Virginia All-Payer Claims Database to identify pharmacogenomic patterns associated with medication burden, fall-related injury, and emergency department utilization in older adults.
In the TAPH cohort, 68% of participants possessed at least one actionable pharmacogenomic variant. Multiple gene–drug mismatches demonstrated associations with mood and mentation outcomes, including significant relationships between CYP2C19 substrate mismatch and PHQ-4 scores, as well as between CYP2C19 and CYP2C9 variants and MMSE scores. Projection analyses revealed considerable overlap between actionable CYP2C19 variation and exposure to CYP2C19 substrate medications. The CYP2C19*17 and CYP2C19*2 alleles accounted for the majority of the projected pharmacogenomic burden among community-dwelling older adults and those with a history of falls. Analysis of the Virginia All-Payer Claims Database, encompassing 22,783 older adults and more than 5.2 million healthcare encounters, identified pharmacogenomically relevant medication burden as an important feature associated with fall-related injury and emergency department utilization. Explainable machine learning approaches further identified biologically plausible medication and drug–gene patterns associated with these adverse outcomes.
These findings indicate that actionable pharmacogenomic variation is prevalent among community-dwelling older adults and frequently coincides with exposure to medications implicated in adverse outcomes. The results support the integration of pharmacogenomic data into medication optimization strategies for older adults and establish a foundation for advancing pharmacogenomic implementation in aging populations that have been historically underrepresented in research.
Rights
© The Author
Is Part Of
VCU University Archives
Is Part Of
VCU Theses and Dissertations
Date of Submission
7-30-2026