OHDSI Home | Forums | Wiki | Github

Proposal - Add country field in Location table

(Ajit Londhe) #1

Hello all,
Here is a proposal to add a field for Country to the Location table in CDM.
Thanks, Ajit


The CDM’s Location table includes physical location fields such as street address, state, zip code, and county. However, there is no field for capturing the country.

Use cases, analytical questions

For data sets with data originating from multiple countries, there is no way to represent this difference in country; you could not stratify an analysis on country or include country as a covariate.

Exploration of the state of the field, other implementations-

The only way to represent data from multiple countries and make that distinction clear now is to have multiple CDM databases or force-feed the country information into a field, causing the CDM to be non-standard.

Importance, also with respect to other projects

Not critical, and actually no use case exists within our team currently, but it could be useful in the future.

Consequences of doing it and not doing it including technical (e.g. implications on vocab, existing software), resources

Limited impact, as it would not be reliant upon the Vocabulary. Standardized representations of country names, such as the ISO Alpha-3 code, could be used in order to constrain the field to 3 characters and prevent the field from becoming unwieldy. Country would be an optional field, so no existing CDMs would be affected.

(Christian Reich) #2


Please fill out the request here. I also added your proposal to the list of change requests. Use this page for the discussion.

Question: Do you have a standardized vocabulary you want to propose for the countries?

(Ajit Londhe) #3

Thanks @Christian_Reich. I’ve added content to the form. I proposed we use the ISO 3166-1 alpha-3 vocabulary, which officially designates country codes, and is leveraged by many organizations including the United Nations. It’s also lightweight in terms of size.

(Melanie Philofsky) #4

We are ETL’ing our location data now. Our source data contains a country field. Our researchers contact patients for potential clinical trials or research surveys. And also to contact patients for follow up surveys.

Phone number and email also go along with the above. We have multiple data sources and we use phone number, email and address along with the usual suspects for fuzzy matching.

Would we be able to move forward with adding the country field to address? And use the ISO 3166-1 alpha-3 vocabulary that @Ajit_Londhe proposed above? And as long as I am asking, is there a spot for email and phone number?