(142) A multi-country, observational study of high-risk pre-event cardiovascular patients: design and analytical framework

Background: Recent data support early initiation of evolocumab to maximize clinical benefit in patients with established atherosclerotic cardiovascular (CV) disease. However, patients without prior CV events, who have established risk factors and poor lipid control may also benefit from intensive lipid lowering therapy. The effect of evolocumab on CV events in this high-risk pre-event population is being investigated in a phase-3 clinical trial (NCT03872401). An observational study was designed to describe the prevalence, characteristics, and treatment patterns of this high-risk pre-event population and to evaluate the incidence of major adverse cardiac events in the real world (RW). However, trial eligibility criteria such as imaging and lab values are difficult to replicate in RW data due to differences in data formats and code vocabularies.

Objectives: To describe a unified protocol and analytical common data model (ACDM) used across multiple databases to overcome the challenges of translating trial criteria to a multi-country RW study and to generate comparable estimates.

Methods: Using electronic medical records and claims data sources from eight countries in North America, Europe and Asia between 2010 and 2022, a high-risk pre-event population was identified. Code vocabularies across databases were reviewed to translate trial criteria into RW criteria. ACDM was used to standardize a limited set of variables into a common data structure, allowing the use of common analytics, definitions, and calculations across multiple datasets.

Results: Trial eligibility criteria were translated into RW criteria using codes captured across all databases. For example, peripheral artery disease (PAD) defined in the trial by % of limb artery stenosis, abdominal aorta treatment or ankle brachial index was translated to PAD diagnosis codes; estimated glomerular filtration rate of 15 to < 45 mL/min/1.73m2 was translated to chronic kidney disease Stage 3 or 4. Global and database-specific analysis plans were developed and translated into the ACDM of relational tables into which the patient demographics, lab, prescriptions and event data of each database are converted. To transform lab values, a conversion tool for different units was developed. Global programming code was developed to analyze data for each study objective within the ACDM. Each data owner will convert their data into the ACDM and apply the programming code. Results will be shared at aggregate level by each dataset and summarized across all datasets.

Conclusions: This work shows the framework to develop a trial-like population in the RW, using a federated model to provide comparable estimates of disease burden and progression of high-risk pre-event CV patients across global data sources.