Addressing the Inherent Challenges of the French National Claims Database (SNDS) Through Algorithms to Identify Patients With Diffuse Large B-cell Lymphoma (DLBCL) and Their Chemotherapy Treatments
Author(s)
Camille Nevoret, PhD1, Grégoire Mercier, MD, PhD2, Guillaume Cartron, MD, PhD3, Cyril Esnault, MSc4, Sophie MICON, MSc4, Julien Vercruyssen, MSc4, Elodie Torreton, MSc1, Fanny Cherblanc, PhD5, Hervé Ghesquieres, MD, PhD6, Stephane Bouee, MD, MSc, MPH1.
1CEMKA, Bourg-La-Reine, France, 2University Hospital of Montpellier, Public Health, Montpellier, France, 3University Hospital of Montpellier, Clinical Hematology, Montpellier, France, 4Roche SAS, Boulogne-Billancourt, France, 5Lysarc, Hôpital Lyon Sud, Pierre-Bénite, France, 6Hematology, CHU de Lyon, Lyon, France.
1CEMKA, Bourg-La-Reine, France, 2University Hospital of Montpellier, Public Health, Montpellier, France, 3University Hospital of Montpellier, Clinical Hematology, Montpellier, France, 4Roche SAS, Boulogne-Billancourt, France, 5Lysarc, Hôpital Lyon Sud, Pierre-Bénite, France, 6Hematology, CHU de Lyon, Lyon, France.
OBJECTIVES: Diffuse large B-cell lymphoma (DLBCL) is the most common and aggressive form of non-Hodgkin lymphoma. The French National Claims Database (SNDS), which links comprehensive claims data, hospital discharge summaries, and the national death registry, is a valuable resource for pharmacoepidemiology studies. However, its claims-based nature, primarily for reimbursement, makes accurate identification of ICD-10-coded DLBCL patients and specific chemotherapy regimens challenging. This study aimed to develop and validate algorithms for this purpose.
METHODS: DLBCL cases were identified in the SNDS using an expert-driven algorithm based on the inclusion and exclusion criteria for ICD-10 codes. Patients diagnosed with a new case of DLBCL from 2018 to 2021 were included. Chemotherapy regimens were determined through an expert-developed algorithm that analyzed the frequency of rituximab injections, hospitalizations, and chemotherapy sessions. The identified treatment populations were compared to REALYSA, a national lymphoma patient cohort, to evaluate the algorithm's accuracy.
RESULTS: We identified 17,830 newly diagnosed DLBCL patients (mean age 68.4 years, 55.8% male). Within two months post-diagnosis, 26.3% of patients (n=4,693) received no treatment, and 22.3% of these patients died within the same period. Additionally, 69.2% (n=12,342) received rituximab-based therapies, while 4.5% (n=795) received chemotherapy only. In patients treated with rituximab, 23.4% underwent pre-phase therapy, similar to the 25% observed in the REALYSA cohort. Comprehensive details regarding R-Chemo regimens (such as R-ACVBP vs. R-CHOP-14 vs. -21, with or without chemotherapy-based pre-phase treatment +/- Rituximab) will be shared at the conference.
CONCLUSIONS: This study supports the feasibility and utility of expert-driven algorithms in the SNDS for identifying DLBCL patients and their complex chemotherapy regimens, overcoming inherent limitations of claims data.
METHODS: DLBCL cases were identified in the SNDS using an expert-driven algorithm based on the inclusion and exclusion criteria for ICD-10 codes. Patients diagnosed with a new case of DLBCL from 2018 to 2021 were included. Chemotherapy regimens were determined through an expert-developed algorithm that analyzed the frequency of rituximab injections, hospitalizations, and chemotherapy sessions. The identified treatment populations were compared to REALYSA, a national lymphoma patient cohort, to evaluate the algorithm's accuracy.
RESULTS: We identified 17,830 newly diagnosed DLBCL patients (mean age 68.4 years, 55.8% male). Within two months post-diagnosis, 26.3% of patients (n=4,693) received no treatment, and 22.3% of these patients died within the same period. Additionally, 69.2% (n=12,342) received rituximab-based therapies, while 4.5% (n=795) received chemotherapy only. In patients treated with rituximab, 23.4% underwent pre-phase therapy, similar to the 25% observed in the REALYSA cohort. Comprehensive details regarding R-Chemo regimens (such as R-ACVBP vs. R-CHOP-14 vs. -21, with or without chemotherapy-based pre-phase treatment +/- Rituximab) will be shared at the conference.
CONCLUSIONS: This study supports the feasibility and utility of expert-driven algorithms in the SNDS for identifying DLBCL patients and their complex chemotherapy regimens, overcoming inherent limitations of claims data.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
CO7
Topic
Clinical Outcomes, Methodological & Statistical Research
Topic Subcategory
Clinical Outcomes Assessment
Disease
Oncology