OHDSI Home | Forums | Wiki | Github

Can Atlas or other ODSI tools still being used if

Hi community,

Many standard clinical tables have one or multiple ‘DATE’ field(s), say, in condition_occurrence table, ‘condition_start_date’, it is a required field.

My data is kind of REDCap data, some the text question/answer data, we let it go to Observation table, but there are quite amount of data can go to other tables like ‘drug_exposure’, ‘Measurement’ etc but without dates information.

My questions are:

  1. If the date(s) is missing from source data, can you fake the dates?

  2. If you fake the missing date info, what analysis are not affected and what are affected from Atlas?

  3. If the date info is missing, can you still use OHDSI/OMOP tools like Atlas to do some type of analysis and get meaningful results?

Thanks in advance.
Ray

@zohdsi:

I think the timing is the very thing that relates various entities to each other. ATLAS analytics use it for characterization (things happen before or after cohort membership), estimation (things happen causally and therefore after some intervention) and prediction (things happen at some point in the future). Without the time, the records become useless. I would not write them unless you can time stamp them correctly. Fake times will get you around the database constraints, but you are not doing yourself a favor.

Generally, observational data have a problem with question-answer pairs. They tend not to be standardizable, the answers lack semantic content unless in conjunction with the question (e.g. what does “Yes” mean?), they usually miss timing (“Do you have high blood pressure?”) and they are hard to find (what is in the question and what is in the answer).

Use Redcap for data that make sense in a local context. That’s what it is for. OMOP is for standardizing data and putting them into a research network. The context is given by the standard.

Thank you very much for your explanation of the time fields and the comments on Redcap vs OMOP, Christian. I appreciate it.

t