MAGELLAN HEALTH DATA WAREHOUSE: A METHODOLOGICAL FRAMEWORK FOR REAL-WORLD PATIENT POPULATION IDENTIFICATION ACROSS THERAPEUTIC AREAS
Author(s)
Caroline Eteve-Pitsaer, MSc1, Cheikh Tamberou, Sr., MSc2, Ivan Palmer, MD3, Arnaud Fabre, MSc3, XAVIER ANSOLABEHERE, MSc4.
1European RWD-E Analytics Director, Cegedim Health data - Clinityx by Gers Data, Boulogne-Billancourt, France, 2GERSDATA, Boulogne-Billancourt, France, 3Clinityx by Gers Data, Boulogne-Billancourt, France, 4Clinityx by Gers Data, PARIS, France.
1European RWD-E Analytics Director, Cegedim Health data - Clinityx by Gers Data, Boulogne-Billancourt, France, 2GERSDATA, Boulogne-Billancourt, France, 3Clinityx by Gers Data, Boulogne-Billancourt, France, 4Clinityx by Gers Data, PARIS, France.
OBJECTIVES: The Magellan Health Data Warehouse (EDS) is a SNDS-derived database covering 65 million+ French beneficiaries (~99% of the population) with 10 years of longitudinal data from outpatient and inpatient settings, approved by the French data protection agency. This project aims to demonstrate the feasibility of identifying target patient populations across diverse therapeutic areas through three concrete examples.
METHODS: Three feasibility studies were conducted: (1) locally advanced/metastatic breast cancer (HR+/HER2-) after CDK4/6 inhibitor progression (ICD-10 C50, 2019-2022); (2) diabetes subpopulations stratified by treatment line (type 1, type 2 first-line/second-line, gliflozines); and (3) CAR-T therapy (Yescarta) recipients (2019-2023). Algorithms combined ICD-10 codes, long-term disease status, drug dispensing (CIP/ATC), and medical act codes (CCAM), with exclusion criteria to remove prevalent cases and surgical confounders.
RESULTS: All three studies demonstrated satisfactory patient volumes and data depth. For breast cancer, 33,140 patients were identified at index, with 16,500 in early progression post-CDK4/6 inhibitor. Diabetes populations ranged from 343,200 (type 1) to 1,413,820 (type 2 second-line). For CAR-T/Yescarta, 1,210 cumulative patients were identified (340 prevalent in 2023), with consistent growth trends (mean age 60.9 years; 66/34 M/F ratio). Breast cancer estimates were externally benchmarked against published epidemiological data, confirming the validity of the identification approach and supporting the launch of a full ad hoc RWE study on disease burden in this population.
CONCLUSIONS: Across three therapeutic areas, oncology, metabolic diseases, and innovative therapies, Magellan demonstrated its ability to reliably identify clinically meaningful patient populations. External benchmarking in the breast cancer case supported the validity of the algorithm-based approach. The consistent application of a standardized coding framework (ICD-10, CIP/ATC, CCAM) across all three cases further illustrates the reproducibility of this methodology. These findings position Magellan as a credible and multipurpose data source for population identification and for the full spectrum of RWE research in France.
METHODS: Three feasibility studies were conducted: (1) locally advanced/metastatic breast cancer (HR+/HER2-) after CDK4/6 inhibitor progression (ICD-10 C50, 2019-2022); (2) diabetes subpopulations stratified by treatment line (type 1, type 2 first-line/second-line, gliflozines); and (3) CAR-T therapy (Yescarta) recipients (2019-2023). Algorithms combined ICD-10 codes, long-term disease status, drug dispensing (CIP/ATC), and medical act codes (CCAM), with exclusion criteria to remove prevalent cases and surgical confounders.
RESULTS: All three studies demonstrated satisfactory patient volumes and data depth. For breast cancer, 33,140 patients were identified at index, with 16,500 in early progression post-CDK4/6 inhibitor. Diabetes populations ranged from 343,200 (type 1) to 1,413,820 (type 2 second-line). For CAR-T/Yescarta, 1,210 cumulative patients were identified (340 prevalent in 2023), with consistent growth trends (mean age 60.9 years; 66/34 M/F ratio). Breast cancer estimates were externally benchmarked against published epidemiological data, confirming the validity of the identification approach and supporting the launch of a full ad hoc RWE study on disease burden in this population.
CONCLUSIONS: Across three therapeutic areas, oncology, metabolic diseases, and innovative therapies, Magellan demonstrated its ability to reliably identify clinically meaningful patient populations. External benchmarking in the breast cancer case supported the validity of the algorithm-based approach. The consistent application of a standardized coding framework (ICD-10, CIP/ATC, CCAM) across all three cases further illustrates the reproducibility of this methodology. These findings position Magellan as a credible and multipurpose data source for population identification and for the full spectrum of RWE research in France.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
MSR140
Topic
Methodological & Statistical Research
Disease
Diabetes/Endocrine/Metabolic Disorders (including obesity), Oncology