DERIVING CLINICALLY PLAUSIBLE IDH-MUTANT LOW-GRADE GLIOMA COHORTS FROM GERMAN CLAIMS DATA USING A MULTI-STAGE ALGORITHM
Author(s)
Constantin Heidecke, M.A.1, Lisa-Marie Müller, M.Sc.1, Katrin Strassen, M.Sc.1, Isabel Karkossa, Dr. rer. nat.1, Philipp Karge, MBE2.
1Gesundheitsforen Leipzig GmbH, Leipzig, Germany, 2Servier Deutschland GmbH, Munich, Germany.
1Gesundheitsforen Leipzig GmbH, Leipzig, Germany, 2Servier Deutschland GmbH, Munich, Germany.
OBJECTIVES: IDH-mutant low-grade gliomas (LGG) cannot be directly identified in claims data because molecular and histopathological information are unavailable. This study evaluated whether a multi-stage, treatment-informed claims algorithm could refine a broad incident brain tumour cohort into plausible subpopulations.
METHODS: This retrospective cohort study used the German Analysis Database for Evaluation and Health Services Research (DADB, ~4.4 million individuals, 2013-2023), a subset of the German statutory health insurance (SHI) population. Adults with a brain tumour diagnosis in 2015-2020 were considered incident after a 2-year diagnosis-free period and required follow-up up to 3 years after index or until death. Based on German glioma guidelines, eligibility required reconfirmed malignant brain tumour coding and exclusion of differential diagnoses within 12 months after index. Sequential refinement considered diagnostic, treatment-related, and longitudinal proxies, including neurosurgical procedures, radiotherapy/chemotherapy sequencing, and survival patterns to approximate LGG subgroups and care pathways.
RESULTS: The initial incident brain tumour cohort comprised 284-364 patients annually in DADB, corresponding to 10,324-12,329 patients extrapolated to the German SHI population. These counts exceeded epidemiological expectations, prompting further cohort refinement. Stepwise refinement enabled identification of more specific treatment patterns. Ultimately, a post-surgical population without radiochemotherapy within 24 months after surgery was derived as an approximation of watch-and-wait, consistent with care pathways observed in IDH-mutant LGG (272 distinct patients across the study period in DADB; 8,743 extrapolated patients). A population receiving early post-surgical radiochemotherapy (26 distinct patients across the study period in DADB; 768 extrapolated patients) was used to compare treatment patterns and explore implications of early therapy escalation.
CONCLUSIONS: This algorithm enabled refinement of a broad cohort with incident coded brain tumour diagnoses into clinically interpretable subpopulations relevant to IDH-mutant LGG despite absent molecular data, supporting future analyses of treatment patterns, outcomes, and healthcare delivery in German SHI data.
METHODS: This retrospective cohort study used the German Analysis Database for Evaluation and Health Services Research (DADB, ~4.4 million individuals, 2013-2023), a subset of the German statutory health insurance (SHI) population. Adults with a brain tumour diagnosis in 2015-2020 were considered incident after a 2-year diagnosis-free period and required follow-up up to 3 years after index or until death. Based on German glioma guidelines, eligibility required reconfirmed malignant brain tumour coding and exclusion of differential diagnoses within 12 months after index. Sequential refinement considered diagnostic, treatment-related, and longitudinal proxies, including neurosurgical procedures, radiotherapy/chemotherapy sequencing, and survival patterns to approximate LGG subgroups and care pathways.
RESULTS: The initial incident brain tumour cohort comprised 284-364 patients annually in DADB, corresponding to 10,324-12,329 patients extrapolated to the German SHI population. These counts exceeded epidemiological expectations, prompting further cohort refinement. Stepwise refinement enabled identification of more specific treatment patterns. Ultimately, a post-surgical population without radiochemotherapy within 24 months after surgery was derived as an approximation of watch-and-wait, consistent with care pathways observed in IDH-mutant LGG (272 distinct patients across the study period in DADB; 8,743 extrapolated patients). A population receiving early post-surgical radiochemotherapy (26 distinct patients across the study period in DADB; 768 extrapolated patients) was used to compare treatment patterns and explore implications of early therapy escalation.
CONCLUSIONS: This algorithm enabled refinement of a broad cohort with incident coded brain tumour diagnoses into clinically interpretable subpopulations relevant to IDH-mutant LGG despite absent molecular data, supporting future analyses of treatment patterns, outcomes, and healthcare delivery in German SHI data.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
EPH185
Topic
Epidemiology & Public Health
Topic Subcategory
Disease Classification & Coding
Disease
Oncology