Identifying children with Cystic Fibrosis in population-scale routinely collected data in Wales: A Retrospective...

Area of Science:

• Medical Informatics
• Rare Disease Epidemiology
• Clinical Data Management

Background:

• Routinely collected electronic health record (EHR) data offers population-scale insights for rare disease cohort identification.
• Challenges exist in accurately identifying specific patient cohorts, such as children with Cystic Fibrosis (CF), within large EHR datasets.
• Validation against specialized registries is crucial for ensuring data accuracy and research integrity.

Purpose of the Study:

• To demonstrate the feasibility of using linked EHR and registry data for Cystic Fibrosis (CF) research.
• To evaluate the benefits of integrating multiple data sources, specifically the UK CF Registry, for enhancing CF research capabilities.
• To establish a proof of principle for utilizing linked data as a valuable resource in CF research.

Main Methods:

• Utilized three distinct EHR data sources within the Secure Anonymised Information Linkage (SAIL) Databank to identify children with CF born in Wales (1998-2015).
• Acquired and linked the UK CF Registry data to the identified EHR cohort for case validation and analysis of misclassification reasons.
• Employed a retrospective review methodology to assess the accuracy of CF case identification across different data sources.

Main Results:

• Identified 352 potential pediatric Cystic Fibrosis (CF) cases across three electronic health record (EHR) data sources, exceeding expected incidence.
• Validation against the UK CF Registry confirmed 257 (73%) of the identified individuals as true CF cases.
• High concordance (98.7%) for true CF cases was observed when identified across all three EHR sources, contrasting with lower accuracy (19.8%) from single-source identification.

Conclusions:

• Accurate identification of health conditions within EHR data necessitates rigorous data quality assurance and validation processes.
• Linking patient data across multiple sources, including specialized registries like the UK CF Registry, substantially improves the quality and reliability of identified cohorts.
• This study highlights the challenges and benefits of using linked EHR and registry data for robust rare disease research, underscoring the importance of data linkage for improving cohort accuracy.