I’m running into an issue with certain types of cohorts in ATLAS, not sure if this is an implementation issue. The following two cohorts are only one example of this occurrence. I saw the same problem with different conditions and different drugs.
Cohort 1:
Entry event: condition occurrence of hypothyroidism (140673 and all descendants)
Inclusion criteria: drug exposure of levothyroxine (1501700 and all descendants)
SynPUF record count: 0
Cohort 2:
Entry event: condition occurrence of hypothyroidism (140673 and all descendants)
Inclusion criteria: drug exposure of levothyroxine (40169766 only)
SynPUF record count: ~6,000
The strange thing is, the levothyroxine concept 40169766 is a descendant of 1501700, so I would expect the record count in Cohort 1 to be at least the same as Cohort 2, if not higher. When run against a different database, this does occur (Cohort 1 having a higher record count than Cohort 2). Any advice would be greatly appreciated.