(025) Expanding Capabilities for Unapproved Cannabis-Derived Product Surveillance in the United States, TriNetX, July 1, 2018 – June 30, 2022

Publication Number: 1021

Background: The emerging cannabis-derived products (CDPs) market has generated unique regulatory and surveillance challenges.

Objectives: To explore the use of electronic health records (EHR) to identify exposures to unapproved CDPs, including cannabidiol (CBD), as part of active safety surveillance efforts in the United States.

Methods: We conducted descriptive analyses of CDPs exposures using the TriNetX Live™ data network within the FDA’s Sentinel System, which provided EHR data from 76 health care organizations (HCOs) across 31 states; 15 of these HCOs had the ability to extract information from clinical notes and map it to RxNorm terms using a proprietary non-CDP specific natural language processing (NLP)-based algorithm. We required patients to have at least one encounter in the two years prior to a CDP exposure (day 0) recorded between July 1, 2018, through June 30, 2022, to define three exposure-based cohorts identified using relevant RxNorm terms: (1) CDP [excluding Epidiolex (FDA-approved CBD) or CBD 100 mg/mL]; (2) CBD (same exclusion); and (3) Epidiolex or CBD 100 mg/mL. We assessed demographics and clinical characteristics (days -365, 0).

Results: We identified 34,130 individuals with CDP exposure, 31,610 with CBD exposure, and 6,910 with Epidiolex exposure. The CDP cohort had a mean age of 50 (SD:±22) years, 61% female, and 83% White. These individuals rarely (~4-5%) had a diagnosis recorded for conditions for which Epidiolex is indicated. Patients’ characteristics in the CBD cohort mirrored those in the CDP cohort as most (93%) of the exposures in the CDP cohort were CBD; the remainder (7%) included cannabinol, hemp, cannabis seed oil and whole extract, and cannabigerol. The Epidiolex cohort included 6,910 individuals with a mean age of 28 (SD: ±22) years, 51% female, and 78% White. Of them, 29% had Lennox-Gastaut syndrome, < 1% Dravet syndrome/tuberous sclerosis complex, and 50% other epilepsies.

Conclusions: Results suggest the ability to identify unapproved CDPs exposures from clinical notes. However, the current NLP-based algorithms mostly identify unapproved CBD use and do not fully distinguish from the FDA-approved CBD product. Future improvements could be made to develop and validate algorithms to identify other CDPs (besides CBD), and to restrict data to HCOs with both NLP capabilities and brand name CBD information, in order to expand FDA’s surveillance purposes.
Disclaimer: The contents are those of the authors and do not necessarily represent the official views of, nor an endorsement, by FDA/HHS, or the U.S. Government.