Document Type

Dissertation

Degree

Doctor of Philosophy

Major

Nursing

Date of Defense

11-10-2025

Graduate Advisor

Umit Tokac, Ph.D.

Committee

Kimberly Werner, Ph.D.

Fan Li, Ph.D.

Kyle E Hultgren, Pharm.D

Abstract

Background: Adverse drug reactions (ADRs) pose significant public health challenges yet remain substantially underreported in traditional pharmacovigilance systems. Social media platforms and patient-reported outcome forums have emerged as potential complementary sources for capturing real-time patient experiences with medications, but their integration with established surveillance systems remains underexplored.

Purpose: This study examined ADR reporting patterns across FDA Adverse Event Reporting System (FAERS), X (formerly Twitter), and Askapatient to evaluate whether patient-generated digital data can complement traditional pharmacovigilance systems for enhanced drug safety monitoring.

Methods: An explanatory sequential mixed-methods design analyzed 107,753 reports representing 289,740 individual ADR occurrences for six medications (adalimumab, cetirizine, divalproex, levothyroxine, pregabalin, risperidone) collected from January 2020 to December 2022. Quantitative analyses included chi-square tests, two-way ANOVA, and correlation analyses to compare ADR frequencies and types across platforms. Sentiment analysis assessed the emotional tone of patient posts and tweets. Qualitative content analysis examined patient-generated descriptions to contextualize quantitative findings. The Social Amplification of Risk Framework (SARF) provided the theoretical lens for interpreting how platforms shape risk communication.

Results: Statistical analyses confirmed that ADR distributions differed significantly across platforms (χ² = 74,687.45, p < .001, Cramer’s V = 0.36). FAERS emphasized regulatory-relevant events including medication errors and device issues, while patient platforms prioritized quality-of-life impacts such as fatigue, depression, and cognitive effects described using patient-coined terminology such as “brain fog.” Platform accounted for 75% of variance in reporting frequencies (η² = 0.75). Sentiment analysis revealed uniform negativity on Askapatient (99.6%) versus varied sentiment on X (84.5-97.6% negative). Correlation patterns showed that platform alignment varied by analytical level: X and Askapatient showed stronger alignment at the drug level (ρ = 0.83), while FAERS and X demonstrated strong correlation at the ADR level (ρ = 0.90). Patient narratives provided contextual richness missing from structured reports.

Conclusions: Findings demonstrate that FAERS, X, and Askapatient capture complementary rather than redundant safety information. Integration of these sources enhances comprehensive drug safety surveillance by bridging clinical significance with patient experiences. This study contributes methodological, empirical, and theoretical foundations for evolving pharmacovigilance toward patient-centered systems that genuinely incorporate lived experience in drug safety surveillance.

Download

Share

COinS