I have a column in one of my tables called AlcoholStartDate, which indicates the self reported date for when the patient started consuming alcohol.
According to Athena, the closest concept I could find to this was “Age at starting drinking” (Athena), and so, using the patients’ birthdates, I extrapolated the age at which they started drinking into a new column, which I then mapped to the aforementioned concept.
Now I will be placing this into the Observations table, but I’m uncertain how best I can set up my table to fit the OMOP CDM conventions.
I have ‘AgeWhenStartedDrinking’ placed into observation_source_value, 4038568 in observation_concept_id, the age in value_as_number, and 9448 (concept ID for years as units) into unit_concept_id. But how should I mention that ‘AlcoholStartDate’ in this context? Or would it be enough to just mention it in my ETL documentation?
Do you really have a use case for this data? What exactly do you want to study?
I would follow the “Age at starting drinking” approach and not overengineer it (mentioning AlcoholStartDate in your ETL documentation should be enough).
Hello @zhuk . I do not have a specific use case for this data quite yet, but I am working with other researchers who will be studying a few different chronic diseases, some of which take alcoholism into account.
If you think that merely mentioning AlcoholStartDate in my ETL documentation should be enough, then I will take that into consideration. Thanks so much.
OHDSI likes the use-case driven approach and large-scale analytics, so talk to your researches and consider how you would later define a patient cohort. Preserving AlcoholStartDate may certainly help you identify patients, who did not complete dry January in 2025, but is it really the cohort you are looking for?
Also, we have a psychiatry Working group, so you maybe interested to check that out. Tagging @Dymshyts@TatianaSkugarevskaya