You’d use nested criteria for this:
Having at least 1:
having at least 1: <-- this is the nested criteria
betwwen 1 days after and all days after <- this index refers to the INR test
between 7d before index and 0d before index <- this index refers to the warfarin exposure
You could be looking to do something later such as ‘and some change between the two measurements’ You can’t do this using the Atlas cohort definition tool.
If the change in measurement value would lead to a condition diagnosis, then use that fact instead of looking for a change in measurement value.
I may not have understood your question properly: are you saying you want to find warfarin users that have had 2 distinct INR measurements, or are you just trying to define an outcome cohort of ‘2 distinct INR measurements’? The way I interpreted your question was that you were looking for warfarin exposed with INR measurements.
If you are looking for an outcome of INR measurements, then you can just define a cohort of the INR measurements, but you need to be clear about which date (the first INR or the second INR) is the ‘cohort start date’ of your outcome.