(246) Comparative Postmarketing Safety Assessment of Baloxavir Marboxil vs Oseltamivir using TreeScan™

Background: Tree-based scan statistics detect safety signals by leveraging a hierarchical outcome tree and controlling for multiplicity inherent in evaluating overlapping outcome groups. Baloxavir marboxil (BXM), FDA approved in 2018, is a first-in-class influenza-specific endonuclease inhibitor approved for acute uncomplicated influenza or post-exposure prophylaxis in adult and pediatric patients.

Objectives: To compare incident outcomes in new users of BXM with oseltamivir (OTV), a first-line influenza treatment and prophylactic having a different mechanism of action and greater postmarketing exposure.

Methods: Community-dwelling, ambulatory, nonpregnant patients with continuous medical and drug coverage and incident use of BXM or OTV were included. Cohorts included “all ages” (≥12 years), pediatric (12-18), adult (≥19), and older adult (≥65). We specified conventional and high-dimensional propensity score ([hd]PS) models including pre-index chronic comorbidities (plus 100 covariates in hdPS) and recent COVID-19; and age, race, sex, flu season, and quartile of Social Deprivation Index on the index date. Current acute respiratory illness was also included to differentiate likely treatment or prophylaxis. We used 1:1 and 1:2 PS matching (PSM) to balance cohorts, then described baseline characteristics and PS distributions before and after matching. We followed matched pairs for 84 d and identified incident outcomes (400-d washout) with a hierarchical tree of non-pregnancy, non-cancer-related ICD-10-CM codes, grouping at the 3rd-6th level. We used TreeScan™ software and unconditional Bernoulli tree-based scan statistics to identify statistically significant (p < 0.05) imbalances. We performed safety assessments in the inpatient (I) or emergency (E) settings, and in the I, E, or outpatient (O) setting as a sensitivity analysis.

Results: We identified 77K BXM and 3.5M OTV new users (83% BXM and 79% OTV in adult cohort), most indexing in flu season 2019. Before (hd)PSM, BXM users were less likely to be female, non-white, reside in a socially deprived geographic area, and have chronic comorbidities, but cohorts were well-matched on all characteristics except indicators of prophylaxis after matching. In most analyses, OTV users had a slightly higher volume of outcomes overall, indicating the potential for residual confounding. Compared to OTV, no outcomes were significantly imbalanced except in the adult 1:1 PSM I/E/O assessment, BXM users were more likely to have the code Acute Bronchitis, Unspecified (RR 1.09; p=0.01).

Conclusions: Our study did not identify new safety issues with BXM as treatment or prophylaxis for influenza compared to OTV in a large sample of real-world adult and pediatric patients, adding to the larger body of safety evidence.