Abstract TH928: Sequence Symmetry Analysis Shows Cardiovascular Safety Signals Associated with Non-Steroidal Anti-Inflammatory Drugs

Background: Non-steroidal anti-inflammatory drugs (NSAIDs) are commonly prescribed for pain, inflammation, and fever; however, real-world evidence on the cardiovascular risks of individual NSAIDs remains limited. Sequence symmetry analysis (SSA) is a signal detection method for identifying adverse drug events (ADEs) in large electronic health record datasets. This study applied SSA to identify cardiovascular ADE signals associated with oral NSAIDs. Methods: This cohort study used primary care records from the UK Clinical Practice Research Datalink (CPRD GOLD). Adults ≥18 years with ≥1 year of prior observation and a new NSAID prescription between 2013-2023 were included. Incidence of seven cardiovascular events (acute myocardial infarction MI, arrhythmia, deep vein thrombosis DVT, heart failure, hemorrhagic stroke, ischemic stroke, pulmonary embolism PE) was assessed within 180 days before versus after NSAID initiation. Adjusted sequence ratios (aSR) with 95% confidence intervals were calculated for each NSAID-event pair, with aSR>1 indicating a positive ADE signal and aSR<1 indicating protective effect. Positive (NSAID-edema) and negative (NSAID-cataract) controls assessed validity. Sensitivity analyses included stratification by age and sex, alternate windows (90 and 365 days), and stratification by prior proton pump inhibitor (PPI) use. Results: 77,570 patients met inclusion criteria (median age 66 IQR 54-76; 53.9% male). Positive and negative controls aligned with expectations. Naproxen showed positive signals across all events, with the highest for PE (3.03 2.63-3.51). Ibuprofen showed six positive signals, with PE being highest (2.2 1.88-2.59). Diclofenac and etoricoxib each had five positive signals, with MI (3.30 2.42-4.57) and ischemic stroke (3.84 2.11-7.43) as the highest, respectively. Celecoxib and meloxicam showed four positive signals, with the highest being heart failure (2.15 1.17-4.11) and PE (2.66 1.30-5.82), respectively. MI and arrhythmia were common across all aforementioned medications. Aspirin showed negative signals across all events except PE (1.08 0.96-1.22). Age, sex, and window sensitivity analyses were consistent, while prior PPI stratification showed aSR attenuation in prior users. Conclusions: Individual NSAIDs exhibit varied cardiovascular ADE signals, underscoring the need for personalized risk evaluation. SSA in real-world data enhances pharmacovigilance by identifying rare ADE trends.