Linkage of routinely collected genetic data to investigate cystic fibrosis in Wales

Maddison, Robert 2026. Linkage of routinely collected genetic data to investigate cystic fibrosis in Wales. PhD Thesis, Cardiff University. Item availability restricted.

	PDF - Accepted Post-Print Version Restricted to Repository staff only until 21 May 2027 due to copyright restrictions. Download (6MB)
	PDF (Cardiff University Electronic Publication Form) - Supplemental Material Restricted to Repository staff only Download (1MB)

Abstract

Background Trusted research environments enable the integration of healthcare and administrative datasets for research through pseudonymised linkage. Routinely collected genetic data (RCGD) offer substantial research value, but linkage pathways for these data are underdeveloped. Aims I sought to evaluate the feasibility of integrating RCGD with pseudonymised linkage systems, and to investigate their research value using cystic fibrosis (CF) population genetics and epidemiology as a case study. Methods CF genetic test reports held by the All-Wales Medical Genomics Service (AWMGS) were adapted for import to the Secure Anonymised Information Linkage (SAIL) Databank. The impact of related individuals on allele frequency estimates was modelled using simulations. Linked data were used to infer relatedness in the pseudonymised data. Linear models were used to examine the geographic and socioeconomic patterns associated with CFTR allele distribution. The CF carrier phenotype was investigated in a phenome-wide association study (PheWAS). Results The Welsh Cystic Fibrosis Test Record (W-CFTR) was prepared, containing pseudonymised testing data for 331 people with CF (pwCF), 1824 CF carriers and 10,494 wildtype individuals tested from 1990-2023. Undetected related individuals in W-CFTR decreased the accuracy of allele frequency estimates. Linked data enabled relatedness detection and filtering to improve accuracy. W-CFTR captured a broader range of CFTR variants than prior data, but disclosure. Summary risks from small numbers meant that frequencies could not be reported directly. CFTR alleles were distributed in an urban-rural manner, with a higher frequency and diversity of alleles observed in South Wales than elsewhere. PheWAS of CF carriers and wildtype individuals found that phenotypes associated with CF heterozygosity were due to ascertainment bias. Conclusion My study has established that RCGD can be integrated with pseudonymised linkage systems and provide substantial research value, particularly for rare disease. Key challenges include scaling, limited processing tools, undetected relatedness, ascertainment bias and restrictions on reporting information about rare variants.

Item Type:	Thesis (PhD)
Date Type:	Completion
Status:	Unpublished
Schools:	Schools > Medicine
Date of First Compliant Deposit:	21 May 2026
Last Modified:	21 May 2026 15:29
URI:	https://orca.cardiff.ac.uk/id/eprint/187158

Actions (repository staff only)

Edit Item