OHDSI Home | Forums | Wiki | Github

What tables are must-haves for a dataset to be CDM v5.3.1 compatible so that it can be loaded into OHDSI on AWS


I’m planning to make a EHR dataset compatible with CDM v5.3.1, so that I can load it into OHDSI on AWS as a custom dataset. Since this is my first time, I’ve few questions that I would appreciate any inputs from the group.


  1. Are all the tables/fields specified in this Github needed ?

  2. What do we do if our custom dataset doesn’t have all the tables listed in the above Github ?

  3. Is creating the table but not loading it with any data helpful OR should we ignore table creation altogether ?

  4. Would opting either of these options create any issues down the line when trying to load the dataset into OHDSI ?

You should create all the tables, and just populate them with the data that you have.

If you’re going to use our analytical tools, I’d say the most important tables are Person and Observation Period. Everything else is basically if you have the data or not. An empty table just means you don’t have data.