OHDSI Home | Forums | Wiki | Github

Index date from the cohort definition function

(Shinyoung Ju) #1

Hi there,

I hope you are all doing well.
I wanted to ask you a quick question regarding the index dates defined by the cohort definition function in Atlas.
I tried to define the cohort entry as below:

In this case, would the index date be the date of HBP, or would it be the date of Flu as patients would not enter the cohort until they pass the inclusion criteria?

Please let me know what you think! I really appreciate your insights on this.

Best regards,

(Chungsoo Kim) #2

Hi Mariana,
In this case, the index date is the date of the high blood pressure event (cases that meet the condition of occurrence start date and earliest event). Inclusion criteria (in your case, flu) is just qualifying criteria for your initial entry events. they don’t affect the index date. you can find the detailed description in here

Chungsoo Kim

(Chris Knoll) #3

That’s correct, @Chungsoo_Kim that inclusion rules don’t alter the index dates (beyond disqualifying them for inclusion into the cohort)…but i also wanted to point out that the definition above has the following behavior:

  1. Find all events of High Blood Pressure that occurred before 2018-06-30
  2. Filter those events to the earliest event per person note: this is from the 'limit initla events to earliest event per person
  3. For each cohort entry event (remember: this is just 1 per person now), include that event in the cohort if it had prior flu
  4. Select the earliest event per person from the remaining events note, this is from the bottm 'limit qualifying events to earliest event per person’

Sometimes, this is the desired behavior: you only want to let the earliest event be considered for inclusion criteria before any other inclusion rules are applied. But, if that isn’t what is desired here, then you need to change one of the settings: if you want to consider all high Blood Pressure events that occurred before 2018-06-30 for the Flu inclusion rules, but in the end you just want to have one event per person, then you should change the top ‘limit cohort entry events’ to be ‘all events per person’, and leave the lower ‘limit qualifying events’ to ‘earliest per person’.

(Shinyoung Ju) #4

@Chungsoo_Kim and @Chris_Knoll Thank you so much for your replies! This is really helpful.
One question I have from your explanation, @Chris_Knoll is: in #3, you said the above cohort definition will only include the ones that had prior flu. Wouldn’t it check whether the patients had a flu event after the cohort entry date?

@pareen.vora We got an answer --> the index date is the date of High Blood Pressure!

(Shinyoung Ju) #5

To give you more information on how we set the timeframe for checking the flu event, please see the following:

(Chris Knoll) #6

Hi, @byr0sxj,
Yes, #3 should have said ‘Had flu on or after index’. I made the assumption you were looking for prior when looking at the screenshot of your example. My main point was how ‘limit initial events to earliest per person’ vs. ‘limit qualifying events to earliest per person’ are applied with respect to the inclusion rules.