(A-200) Strengthening inferential studies in the U.S. FDA Sentinel initiative: Results from a methodological demonstration project

Background: Historically, Sentinel’s reliance on insurance claims data has led to insufficiency in addressing some emerging safety questions requiring more granular clinical information. The FDA Sentinel’s Real-World Evidence Data Enterprise (RWE-DE), a data infrastructure linking large volumes of electronic health records (EHRs) with claims data for >25 million individuals, was created to help the FDA address emerging safety questions for which claims data may be insufficient.

Objectives: To conduct a demonstration project that aimed to evaluate the applicability of the RWE-DE in a use case of the risk of acute pancreatitis following initiation of SGLT2 inhibitors compared to dipeptidyl peptidase-4 inhibitors (DPP-4i) in patients with type 2 diabetes mellitus.

Methods: We used data from the FDA Sentinel’s RWE-DE commercial network comprising two data partners — HealthVerity (2018-2020) and TriNetX (2013-2024). We leveraged the PRINCIPLED framework to conduct an inferential study. We deployed a previously validated computable phenotyping algorithm based on features derived from structured diagnosis codes, laboratory results, and natural language processing (NLP) of free-text EHRs. We employed propensity score fine stratification to adjust for 130 pre-exposure claims-based and 6 EHR-based covariates. Missing covariates in EHRs were multiply imputed. We used weighted Cox proportional hazards regression models to estimate hazard ratios (HR) and 95% confidence intervals (CIs) for both intent-to-treat (ITT) and per-protocol (PP) causal contrasts. Results from the two data sources were combined using inverse variance pooling.

Results: A total of 72,429 patients from HealthVerity (30,174 SGLT2i; 42,155 DPP-4i) and 24,690 patients from TriNetX (11,943 SGLT2i; 12,747 DPP-4i) were included. Incidence rates of 2-3 per 1,000 person-years were observed across both databases for both treatment groups. After adjustment, the risk for acute pancreatitis following SGLT2i initiation was no different compared to DPP-4i initiation [HR 0.85 (95% CI: 0.67-1.07) for ITT and 0.84 (0.58-1.22) for PP].

Conclusions: This study within FDA Sentinel’s RWE-DE serves as a proof-of-concept for future protocol-based assessments in Sentinel, highlighting the value of EHR data for the capture of clinical information not available in claims data.