I have been conducting an OMOP-based ETL on a set of data from one EHR for the last few months and I find myself stuck on a number of issues, namely the the setup of the Observations table. I have gone through all the tables in my source data and identified about 20 tables which have at least one column that I believe adhere to Observations. One of these tables contains the recorded blood group types of patients (e.g. A+, B-, AB+, 0-, etc). The problem is this table does not contain any dates, and from what I can see in the OMOP Observation table (v5.4), observaton_date is a required, non-nullable column.
So, in order to get these date values, I tried searching in the lab results table and notes table in order to find when patients had their blood group tested and/or noted. While I did manage to find some overlap between the patients in the original blood group types table and the lab results/notes tables, I also realized that there many other patients in the two latter tables whose blood group was not recorded in the original blood group types table.
What this means is that there are multiple observations of the same concept ID (blood group types) dispersed between multiple tables that do not always include a date that I can use in the observation_date column.
So how best should I proceed? I assumed that maybe I could just use the birth date or registration date of the patient instead (since a patient’s blood group type does not change throughout their life), but I feel like this is might be a very tenuous approach. Alternatively, I could leave out the data, but I would prefer not to. Does anybody have any recommendation for how best to proceed in this scenario?