2A-02 - Racial-ethnic and social determinants of health variation in drug safety evaluations: limitations of spontaneous reports data and the importance of advancing risk stratification capabilities in real world data

Background: Spontaneous reports data are routinely used for safety surveillance and have known limitations, including missing measures for important patient characteristics, such as social determinants of health (SDoH), ethnicity, and race. Surveillance strategies that rely primarily on spontaneous reports data are therefore not well-suited to assess adverse event differences by these sociodemographic contextual factors.

Objectives: To evaluate an overall known drug-outcome association (i.e., Sodium-glucose cotransporter-2 inhibitor (SGLT2i) use and an increased risk of genital mycotic infections (GMIs)) and potential sociodemographic factors (SDoH, ethnicity, and race) using a multimodal surveillance approach.

Methods: Overall, associations between SGLT2i use and GMIs were assessed using the Food and Drug Administration’s Adverse Event Reporting System (FAERS), a spontaneous reports database, and separately, in 2014-2023 OneFlorida+ network, a state-based electronic health record (EHR) dataset. Disproportionality analyses with Multi-item Gamma-Poisson Shrinker (MGPS) model were used to examine associations in FAERS. A standard threshold for MGPS of the lower limit of the 90% confidence interval (CI) being higher than 2 was used. The OneFlorida+ analysis included adult patients diagnosed with type 2 diabetes who initiated a second-line glucose-lowering drug (GLD) during the study period, comparing the effect of SGLT2i versus another second-line GLD. Propensity score matching with a Cox regression model was used to estimate hazard ratios (HRs) and 95% CIs.

Results: In FAERS, our findings support a known significant association between SGLTi use and GMI (overall: MGPS=30.9, 90%CI(29.3-32.7); males: MGPS=75.3, 90%CI(67.4-83.9); females: MGPS: 32.2, 90%CI(30.1-34.3)). In OneFlorida+ EHRs, we confirmed the association (overall: HR=1.3, 95%CI (1,2-1.4)). Subgroup analyses suggest that Hispanic patients with lower income (HR=1.6 95%CI(1.1-2.2)) and educational attainment (HR=2.0, 95%CI(1.4-2.9)) had a higher risk of GMI associated with SGLT2 use; these subgroup analyses could not be conducted in FAERS.

Conclusions: Our analyses demonstrate the importance of developing a capacity to enable the examination of ethnic and racial variation in SDoH for safety outcomes using a multimodal surveillance approach. Assessment of these sociodemographic contextual factors using real world data, such as OneFlorida+ EHRs, may provide insights into risk subgroups for drug- or vaccine-outcome pairs that may otherwise be missed if analyses of spontaneous reports data are solely relied upon. Findings suggest that ethnic variation in SDoH may exist in the association between SGLT2i use and GMIs, which warrants further exploration.