Phenotypes

We used phenotypic data from the UK Biobank including data from Hospital Episode Statistics (HES) and General Practice (GP). HES is a database containing details of all admissions at NHS hospitals in England. All clinical data are coded using International Classification of Disease (ICD) codes and all operations and procedures are coded using the Office of Population Censuses and Surveys (OPCS); Classification of Surgical Operations and Procedures (4th revision). The most recent ones are ICD-10, which are the ones we used, including 11,726 unique ICD-10 codes for 410,321 unique participants.

GP data contain the vast majority of a person’s health care information; 40% of GP records are linked to UK Biobank cohort up to now. Multiple clinical coding data are provided using Read codes; a standard vocabulary for clinicians to record patient findings and procedures. The most recent is version 3 CTV3 or Read v3), which is the one we used. Regarding the participants, 999 were excluded because the clinical event or prescription date precedes participant date of birth and 66 were removed because the event date is a date in the future and is presumed to be a place-holder or other system default) [29]. However, the 80,982 unique Read v3 codes, were converted to ICD-10 codes [30], in order to be subsequently used from the PheWAS library [20] in R [25]. As a result, 9,492 unique ICD-10 codes for 161,333 unique participants.