Skip to main content
2024 ISPE Annual Meeting Main banner
Icon Legend
This session is not in your schedule.
This session is in your schedule. Click again to remove it.
Back
Evaluation Video

Oral Presentation

Session: PS – There Are Others

Comparing OMOP and ConcePTION common data models in CPRD GOLD: direct oral anticoagulants versus vitamin k antagonists on the risk of bleeding and cardiovascular outcomes

Tuesday, August 27, 2024
9:00 AM – 9:15 AM CEST
Location: Conv Hall I D
Background: Common data models (CDM) attempt to reduce heterogeneity and increase efficiency in federated multi-database real-world evidence (RWE) studies. They provide syntactic, and in addition some provide semantic, harmonisation of the data contained in electronic health databases. The impact of the harmonisation approaches on the results of a RWE study is important when considering the strategies for federated analyses.

Objectives: We aimed to determine potential differences in patient characteristics, incidence rates of outcomes and the effect estimates of a RWE case study when mapping data to a CDM filled with study-specific data (ConcePTION) compared with a general CDM (OMOP).

Methods: A case study was conducted assessing the risk of direct oral anticoagulants (DOACs) vs. vitamin K antagonists (VKAs) on bleeding and cardiovascular outcomes in patients with non-valvular atrial fibrillation. The study data from CPRD GOLD (UK) was mapped to the ConcePTION CDM and the entire CPRD GOLD data source was mapped to the OMOP CDM. Baseline characteristics, incidence rates of the outcomes and Cox proportional hazards ratios using the same model were calculated and compared between the CDM approaches.

Results: The OMOP mapped study population included 80,701 persons (43.7% DOAC users) and the ConcePTION CDM mapped 76,726 (43.5% DOAC users), with 93,350 and 100,135 person-years follow-up respectively. DOACs showed no increased risk of CVD events compared to VKAs in the ConcePTION mapped CDM approach (HR 0.99, 95% CI [0.91;1.08]), but protective effects for the OMOP CDM approach (HR 0.82, 95 CI [0.74;0.90]). DOACs had a slightly increased the risk of stroke when compared to VKAs for ConcePTION CDM approach (HR 1.19, 95%CI [1.02;1.37]) and for the OMOP approach (HR 1.10, 95% CI [0.96;1.26]). No risk of major bleeding was found for DOAC use for both the ConcePTION (HR 0.97, 95% CI (0.84;1.13]) and OMOP CDM approach (HR 0.90, 95% CI [0.78;1.04]).

Conclusions: Slight differences were found for our case study when comparing data mapped to the ConcePTION and OMOP CDMs, with a greater study population and lower effect estimates for cardiovascular risks associated with DOAC use when using OMOP CDM. These differences partly stem from the cohort and variable construction methods utilised in the data ecosystems, despite direction from the same common protocol.