UK BioBank to OMOP CDM v5.3.1

The UK BioBank data consists of five sections

Baseline, survey and lab results collected during visits to the assessment centres
Hospital Episode Statistic, inpatient care
General Practitioner, outpatient care
Covid registry
Death registry

Here we specify the mapping of each of these sections to the respective OMOP tables.

Health System data

Care_site

Clinical data

Visit_occurrence

We use a heuristic to ‘calculate’ a unique 16 digit visit_occurrence_id for each of the sources. The id is a concatenation of a source digit (1 digit), the eid (7 digits) and an index unique within the patient (filled to 8 digits). What field is used as the index is specific to the source table, see overview below. Note that the visit_occurence_id has to be a Big Integer to be able to hold 16 digits.

From baseline: 1<eid><instance>, e.g. 1_9876543_00000001

From covid: 2<eid><date>, e.g. 2_9876543_20201231

From hesin: 3<eid><spell_index>, e.g. 3_9876543_00000012

From gp_clinical and gp_prescriptions: 4<eid><date>, e.g. 4_9876543_20201231

Visit_detail

We use a heuristic to ‘calculate’ a unique 11 digit visit_detail_id. As the hesin table is the only source for visit details, we only need to concatenate the eid and an index (the ins_index).

From hesin: <eid><ins_index>, e.g. 9876543_0084

From hesin

Condition_occurrence

Drug_exposure

From gp_prescriptions

Procedure_occurrence

From hesin_oper

Observation

Stem_table

Stem_to_clinical_event

The stem table is mapped to the respective OMOP domains based on the domain_id. The following rules are applied, in this order:

If stem_table.domain_id given, then read target domain from stem_table.domain_id
If stem_table.concept_id not 0, then read target domain from concept.domain_id
Else, the target domain is Observation.

Source table appendix

Metadta

CDM Source