Presentation (CTI)

OBJECTIVES: Administrative health databases are widely used to inform HTA, access to medications, and healthcare-governance decisions. However, in rare diseases, drug utilization patterns may not be clinically interpretable without disease-specific endpoints. Then, this study examined whether prescription-based administrative data can generate decision-grade real-world evidence (RWE) on treatment effectiveness by using chronic myeloid leukemia (CML) as a proof-of-concept.
METHODS: We analysed a population-based Italian administrative database covering approximately 5.9 million inhabitants. CML patients who started treatment with BCR::ABL1 tyrosine kinase inhibitors (TKIs) between 2017 and 2023 were identified through regional treatment plans and prescription registers. Longitudinal treatment pathways were reconstructed, including first-line TKI, switching, treatment gaps, interruption and restart. recorded administrative events were linked to clinically significant milestones in CML, including BCR::ABL1 response kinetics, major and deep molecular response, treatment-free remission (TFR) eligibility, molecular relapse, and long-term outcomes.
RESULTS: Overall, 609 incident CML patients were identified and followed for up to six years. Administrative data described population-level treatment trajectories. Annual incidence ranged from 1.0 to 2.0 cases per 100,000 inhabitants, 29.2% of patients switched TKI at least once, 25.3% experienced treatment interruption, and 90.3% of those interrupting therapy subsequently restarted after a mean interval of approximately 112 days. However, same administrative trajectories could represent clinically distinct scenarios, including toxicity-driven interruption, planned TFR attempt with molecular relapse, or unintentional non-adherence. Molecular response, TFR eligibility, relapse after treatment stop and reasons for treatment change were not retrievable.
CONCLUSIONS: Administrative databases are essential to create drug-utilization patterns, but insufficient alone to support decision-grade RWE on effectiveness in CML. The main challenge is not data volume, but clinical interpretability. Rare oncology requires the linking of administrative and clinical molecular data, whilst respecting privacy, in order to generate adequate evidence for HTA, access to care and policy decisions.