(41) Establishing a Standing Cohort Using Claims Data to Mimic the Population of a Clinical Trial in The Cardiovascular Therapeutic Area

Publication Number: 1086

Background: Standing cohorts of real-world patients reflecting participants of randomized clinical trials (RCT) can be used during clinical development to contextualize potential adverse events observed in the RCT population. It is critical to derive a cohort that is comparable to the RCT participants on key baseline characteristics. However, not all selection criteria applied in an RCT may be practically applicable in a real-world dataset.

Objectives: To create a standing cohort of patients with myocardial infarction (MI) by applying varying selection criteria to mimic an RCT population and describe relevant characteristics of such patients in the real-world.

Methods: As an example, inclusion criteria applied in an RCT (AEGIS-II, ClinicalTrials.gov Identifier: NCT03473223) were retrieved and used sequentially as well as in several iterative combinations to identify comparable patients in IQVIA PharMetrics Plus database. Patients 18 years of age and older with an MI that occurred between 01 January 2016 and 31 December 2020 and had at least one year of observation in the database prior to the MI were the source population. Given the inherent limitation of the RWD source, not all RCT criteria could be applied. Sub-cohorts were created applying the following filters of various complexities: 1) Age ≥65 years, 2) History of peripheral artery disease (PAD), 3) ≥65 years with history of PAD, 4) History of diabetes mellitus (DM), 5) ≥65 years with history of DM, 6) History of DM and PAD, and 7) ≥65 years with history of DM and PAD. Descriptive analyses of baseline characteristics were performed. The number and percentage of patients in each category were used to describe categorical variables and mean and standard deviation (SD) were used to describe continuous variables.

Results: There were 457,068 patients who met the inclusion criteria in the overall cohort. The average age was 62.9 years and 63.8% were male. When the additional selection criteria were applied, the cohorts ranged in size from 30,210 patients to 185,707 patients, the average age ranged from 65.4 to 77.3 and the percentage of male patients ranged from 55.9 to 62.9%. The percentage of patients with defined comorbidities and medications also varied greatly. For example, in the overall cohort 33.9% had diabetes and ranged from 41.0 to 100% in the sub-cohorts.

Conclusions: When establishing a standing cohort to reflect a clinical trial population, applying various inclusion and exclusion criteria must be considered to generate a cohort as similar to the clinical trial population as possible.