OHDSI Home | Forums | Wiki | Github

Time of Observation vs time Observation happened


(Mark Wlodarski) #1

Hello, I have a probably simple question. Looking at the time of observation field on the OBSERVATION table I am unsure exactly what this value is recording.

Is this the time the Observation was recorded? Or is this the time the observation event happened? If the first then, how do I record when an observation event occurred. If the second how do I record the time the observation was recorded?

Thank you!

(Vojtech Huser) #2

It is a classic informatics problem.
OMOP CDM seems to be “avoiding it” and not embracing it.

Take measurement of blood group. You blood group happened at conception. But in the table you put time of measurement. Can you provide an example where you need to pick one or the other?

(Melanie Philofsky) #3


Some facts are inherently hard to report the exact start date. Examples include tobacco usage and other social history facts, medical or surgical history, family history, and all ‘history’ type facts. So, in the Observation table, the date for these data are the date it was recorded.

There are some facts recorded in the Observation table with a date the event occurred. Usually these data come from coded data where the standard concept_id has a domain_id = Observation.

So, the date and dttm fields in the Observation table have different meanings depending on the source of the data. Which is not the answer anyone wants to hear :slight_smile:

(Christian Reich) #4

Correct. But we are not avoiding it, we just usually don’t have the data. We tend to only have the recording. What happens in the past is murky, which is why we use the mechanism @MPhilofsky alluded to. Do you remember exactly when your appendix was taken out? I don’t. In the 1970ies.

Generally, we don’t really care that much about the time when things were recorded. We really need the time things happened. That’s where the use cases lie. But, as I said, we rarely have the luxury to have that information.