Registry‐related Data - OHDSI/ETL--PulmonaryHypertensionRegistries GitHub Wiki

In general, our recommendations for storing registry-related data in the CDM coincide with those summarized in the Clinical Trials Work Group conventions.

Enrollment, eligibility, informed consent and study termination

If a source dataset contains records about dates when a patient was enrolled in a study, gave their consent, and terminated the study, then the good news is that such information can be stored in the CDM. The following concepts are suggested for use:

concept_id	vocabulary_id	concept_name	domain
4163733	SNOMED	Patient consented to clinical trial	Observation
4160469	SNOMED	Patient eligible for clinical trial	Observation
44811245	SNOMED	Eligible for participation in research study	Observation
44811374	SNOMED	Not eligible for participation in research study	Observation
4090379	SNOMED	Patient entered into trial	Observation
40482840	SNOMED	Completion of clinical trial	Observation
40480198	SNOMED	Subject in clinical trial unable to be contacted	Observation
44811247	SNOMED	Lost to clinical trial follow-up	Observation
4087907	SNOMED	Patient withdrawn from trial	Observation

As seen from the domain column of the table above, all such facts are stored in the Observation table.

Study information

Study information, i.e., description of inclusion/exclusion criteria, trial phase, protocol title, and study objectives, relates to the study itself but not to a patient, so it cannot be stored anywhere in the event tables since they require person_id. A legitimate way to keep this data is to store it in the two metadata tables CDM offers: CDM source and Metadata.

How to populate the former should be clear since it is a one-record table. The latter is not tricky either. Each fact about the study goes into one row of the Metadata table. Usually, it is not worth mapping such attributes to appropriate concept ids (unless some of them have already been mapped while working on patients' events), so 0 is used.

Cohort assignment

Trial arms that are presented in the source data can be translated into CDM cohorts. It is a semi-manual process: the cohort definition table is populated manually with a description of trial arms, then the cohort table is populated based on those definitions and flags indicating a patient's assignment to a trial arm. Cohorts can be created both for planned and actual trial arms.