| OHDSI Home | Forums | Wiki | Github |
Greetings, OHDSI Community!
Do you include patients with no clinical data into your Person table? We are trying to decide if there is any value of including orphan persons or if we should limit the Person table to only the patients with clinical data.
Can you please share your thoughts and practices?
Thank you!!!
If they have no clinical data, are they actually a patient? I would argue that they are not, but this may depend on your population and the services your institution provides.
That was a long way of saying we do not.
What is the source for you data?
If you have claims data and a person was enrolled in the healthcare plan, but they do not have any records, then you should leave the person in the Person table. Healthy people will have enrollment data, but not clinical event records.
Now if you have EHR data and the person is in your patient table, but lacks clinical data, then I would remove these persons. At the University of Colorado, we are currently in the process of removing persons with have sparse clinical data. Sparse clinical data are defined by less than 3 clinical event records for a person in the OMOP CDM. Clinical events consist of records in the Condition Occurrence, Device Exposure, Drug Exposure, Measurement, Observation, or Procedure Occurrence domains. Since the OMOP CDM is designed for longitudinal research studies, persons with sparse or no clinical data are not suitable for research.
You should keep the patient if you believe they are legit, but just never sick. You need them as a denominator for incidence rates. If however you believe those sparse patients are bogus (e.g. captured at other providers not in the database) as @MPhilofsky seems to have it, and they may indeed have plenty of records you just cannot see, you should remove them.