search

National-COVID-Cohort-Collaborative / Phenotype_Data_Acquisition

The repository for code and documentation produced by the N3C Phenotype and Data Acquisition workstream
60 stars 35 forks source link

OMOP vs TrinetX cohorts are not comparable #112

Closed vojtechhuser closed 4 years ago

vojtechhuser commented 4 years ago

Looking at the SQL, the logic in OMOP is not using any encounter criteria.

Are we OK with the cohorts not being fully the same accross CDMs?

This logic in TrinetX is not in OMOP image

kmkostka commented 4 years ago

@vojtechhuser An easy-to-read version of the OMOP phenotype can be found here: http://atlas-covid19.ohdsi.org/#/cohortdefinition/1148

The weak diagnosis occurring on the same visit is at this point: image image

It's harder to trace through the SQL because the references occur in multiple places as the concepts are pulled, primary events are created and observation periods are constructed. The fastest way to find the 2 distinct occurrences if you look for the annotations that say "--Criteria Group" relating to the construction of the nested criteria for this logic.

In the older version of the Phenotype, we had this logic slightly different: http://atlas-covid19.ohdsi.org/#/cohortdefinition/1119 In testing, the numbers were identical though programmatically these are slightly different.