Cross TA Teams Head, Global Epidemiology Janssen Allschwil, Switzerland
Background: Rare disease real-world (RW) datasets often have small patient populations and are geographically scattered. Addressing gaps in natural history may require leveraging multiple disease-specific fit for purpose data sources.
Objectives: To develop fit for purpose data source framework for a rare disease, such as portopulmonary hypertension (PoPH), to address a research question which covers all stages of natural history from diagnosis to liver transplant, and survival.
Methods: 1. Articulate research question and associated data source requirements 2. Conduct disease specific landscape scan 3. Apply fit for purpose data source assessment to identify relevant RW datasets 4. Assess each data source on relevance to address research question, clinical data granularity, reliability, and accessibility through feasibility analyses The framework consists of systematic development of study variable capture and its validity, data type & provenance, and comprehensive assessment of data reliability (conformance, completeness, plausibility), relevance to the research question, and accessibility within existing data landscape.
Results: A systematic approach to data landscape scan identified 16 population-based data sources with PoPH patients as potential data sources, 7 were finally included for fit for purpose assessment. Data Sources (region; data granularity; data completeness) included 1 federated electronic health record (EHR) (Europe [EUR] and North America [NA]; low; high), 1 single EHR (NA; low; high), 1 disease registry (EUR; high; high), 1 drug registry (NA; medium; medium), 3 hospital networks (NA; high; high), 1 national transplant database (NA; medium; medium) and 1 retrospective chart review (NA; medium; medium). PoPH patient numbers ranged from 86 - 1182, with 0 - 420 transplanted patients. All data sources were determined to be highly representative of the population, except for the hospital networks (moderately representative); however, none could fully address the research question. Feasibility results show differences (dates of data collection, geography, treatments used, data capture in terms of stages of disease course) and similarities (demographic breakdown, disease severity of patients) across data sources, reflecting complementary nature of data sources in this use case.
Conclusions: In PoPH, similar to other rare disease RWE research, no single data source is fit for purpose to address all aspects of the research question due to: small populations, diverse geography, delayed and complicated diagnostic journey, complex disease specialist care or reference center networks. Documentation of fit-for-purpose assessment are key and add credibility in rare disease RWE research.
