OHDSI Home | Forums | Wiki | Github

Atlas 2.7.2 - Records outside observation period

Hello everyone,

I am using the new version 2.7.2 of Atlas. I see that we have new field added under inclusion criteria

“records outside observation period” is similar to have “all days before and all days after index date”. Is my understanding correct?

Whereas 0 days before and 0 days after index date, checks for presence of event only on the date of diagnosis(index date)

Request your inputs to make sure I got it right

No. The option allows you to ignore observation period boundaries. So, if you have the checkbox on, then ‘all days before and all days after’ means any time, ever. If the checkbox is off, then it means all days before and all days after within the observation period of the index event.

Okay, an example scenario

We don’t wish to limit our records for analysis. So our primary criteria says that continuous observation is 0 days before and 0 days after index date. This gives us maximum records.

In this case, let’s say I have a inclusion criteria to filter out people who aren’t pregnant.

  1. Check box on (records outside OP)

    Based on our primary criteria of 0 days of continuous observation before and after index date and old inclusion criteria of "All days before and All days after index date " is same as selecting the checkbox on. I can just go with any one of these inclusion criteria

Am I right?

No, I’m sorry. I think you’re just over-thinking it.

The ‘continuous observation’ applies to the cohort entry events. Period. End of Story. You’re trying to associate the 0 days of continuous observation to inclusion criteria, but they are separate things.

As I said above: the ‘all days before and all days after’ establishes the window of time around the index event, which could mean the entire observation period if you do not check the checkbox or it could mean include time outside the observation period if you do check the checkbox.

Maybe I am missing a point you are making, but can you clarify for me why you think that 0 days of continuous observation has some association to specifying ‘30d befoore and 40d after the index event’ when looking for the inclusion criteria?

Is it possible you are mixing up “continuous observation before and after” of a index criterion with the Observation Period defined in the CDM, @SELVA_MUTHU_KUMARAN? The former is the period you are setting for your cohort, the second is a table in the CDM telling you when you have reliable data for that patient.

Hi,

Is it possible to ignore the observation boundaries for cohort entry events as well?

I understand’s it’s used to set the period for the cohort…

Where I am coming from is, we have for example public datasets, where people just do retrospective study by analyzing the data collected based on research question.Just run some ML algorithms and try to do prediction/exploration etc. where people focus mainly on getting the cohort of people who satisfy their conditions rather than trying to know when did they satisfy the condition.

Is it possible to prepare a cohort for this through Atlas? I understand we can set atleast 0 days of continuous observation in Atlas but I guess that still has observation period. So for studies like above, can’t Atlas be helpful?

t