OHDSI Home | Forums | Wiki | Github

How to define this cohort entry: 1 diagnosis

Researchers
#1

I’m trying to replicate a toy exercise with Atlas.

I’m asked to create a cohort with:
“Patients with 1 inpatient diagnosis of XX or 2 outpatient diagnoses”

How would you do it?

I have tried to do it by including in the cohort entry events section just the first occurrence of XX, and then adding inclusion criteria.

Is it better to include the condition occurrence and the visit altogether in the criteria?

Or is it better to include there just the visit and link it somehow to the events defined in the events section? (not easy, maybe saying that the visit occurred from one day before to one day after).

Is the “restrict to the same visit occurrence” doing what I need here?

1807×1881 344 KB

#2

No, the second condition is looking for at least 2 diagnosis of corvid at the same visit.

You are limiting your initial events to earliest per person, so the only chance you are letting people enter your cohort is on the first diagnosis.

When dealing with 2 diagnosis, when should you enter the cohort? At the time of the first dx or the second?

Why not structure the cohort around the logic you describe it: enter the cohort at the covid diagosis, where the dx is during an ER or there was a prior dx sometime prior to the entry dx?

To do this:

  • enter chort at covid dx
  • allow all events per person
  • Inclusion rule looks for an inpatient visit at the same visit as the dx OR a prior dx between 365d prior to 1d prior to the entry DX date. I’m assuming that some temporal proximity matters: should it be within a year? a week? a month? You need to think about this critically to make a meaningful decision.
  • Finally, if you just want 1 event per person, limit your qualifying events to earliest per person.
1 Like
#3

I also have an “age at first diagnosis >= 18” inclusion criteria.
How would add that criterion?

Age >= 18
and
Condition occurrence of Coronavirus
and restrict to same visit
and something about days before and after the index

but I don’t know how to specify that the condition must be applied only to the first event.

If specify “for the first time in person history in the inclusion criteria”… will it take the whole history for that person?, or just the time interval filtered in the cohort entry events section? (that’s all events within my date limits and the 180 days continuous observation)

#4

Just indicate that the age requirement in the criteria…so if you want covid diagnosis at a certain age, then say ‘diagnosis of covid where age >= 18’… Does it need to be a dedicated inclusion rule for that?

If you use that criteria attribute, then it will only use the very first record that matches the code specified in the concept set (ie: covid diagnosis codes) from the condition_occurrence table. It’s not ‘first in history after other criteria are applied’ it is ‘first in history, period’.

Not sure what you mean, and It is not clear why that’s a requirement for your cohort definition. Could you clarify?

t