Pan Program Meeting 2020: Making Record Linkage Work
Updated: May 11, 2020
On Friday 7 February we welcomed program staff, investigators and collaborators to London for our Annual Pan Program meeting. Here we introduce a key theme from the meeting and share what each of our speakers presented on.
Making Record Linkage Work
Session two of the day focused around the work to link multiple datasets from Public Health England (PHE) and securely release the datasets for analysis by CGCV researchers.
Steven Hardy (PHE), one of our co-investigators, explained (i) the role of the National Cancer Registry Analytics Service (NCRAS), (ii) the methodology through which they securely store pseudonymised genetic data from laboratories across England, and (iii) the two-stage approach through which this data is processed. This includes, linking the genetic data to the cancer registry (where a patient has received both genetic testing and a diagnosis of cancer) and restructuring the genetic data to ensure it is in a standard format for information to be extracted.
This data can then also be linked to further data sets and made available within a secure data environment for researchers with the correct governance approvals to access for analysis.
Eva Morris (University of Oxford) is leading this process for CGCV and she presented the CORECT-R data repository, which is being utilised as a model for this work. CORECT-R collates colorectal cancer data from 20 different sources and has been used for various research purposes. For example, evaluating the NHS bowel screening program and indicating the variations in colonoscopy (the main diagnostic test for colorectal cancer) quality across the UK. Demonstrating how the findings from analysis of national datasets can direct improvements in patient care.
Fiona Gilbert (University of Cambridge) reiterated the impact ‘big data’ analysis can have on patient care through examples linking tumour imaging to patient data. Her talk had a particular focus on artificial intelligence being developed to recognise and diagnose tumours from medical images. Improvements in the technology have seen AI identify cancerous tumours with similar accuracy to clinicians. Developing this technology could have considerable benefits for patient care, by increasing the throughput of imaging review allowing results and diagnoses to be delivered faster and adding an additional layer of imaging review to ensure tumours are not missed.