OHDSI Home | Forums | Wiki | Github

DQD errors - Questions about some thresholds

Hi, while running DQD checks, a few errors surged to which I cannot find an explanation:

Tables PERSON and PROVIDER:

  • “The number and percent of records with a NULL value in the CARE_SITE_ID”: if this is not a required field and since the source data does not state a primary care facility for each patient or doctor, this should be NULL.

Tables PROCEDURE_OCCURRENCE and OBSERVATION

  • “The number and percent of records with a NULL value in the VISIT_DETAIL_ID”: a concept_id as such, if populated with 0 will prompt a database error since no row in the corresponding table (visit_occurrence) has a concept_id of 0. Therefore, this should be populated with NULL when there is no visit_detail.

Table OBSERVATION:

  • “The number and percent of persons in the CDM that do not have at least one record in the OBSERVATION table (Threshold=0%)”: why is the threshold 0 in this situation, since, according to the documentation, observation table “captures clinical facts about a Person obtained in the context of examination, questioning or a procedure. Any data that cannot be represented by any other domains, such as social and lifestyle facts, medical history, family history, etc. are recorded here.” Since plenty of the source data corresponds to measurements and procedures, not observations, it is natural that some patients don’t have observations.

Similar thing happens for PROCEDURE_OCCURRENCE. Not all patients will have a procedure so why is the threshold 0%?

Have any of you run into this type of questions?

Hi there, thanks for sharing. Could you please provide the version of DQD you are running, as well as the version of your CDM (and the corresponding threshold files you’re using)?

For care_site_id in person & provider, and visit_detail_id in procedure_occurrence & observation: In the default threshold files these fields are not marked as required fields and thus this check should not be running by default in DQD. I’d recommend checking your threshold file to see whether or not these fields are marked as required.

For observation and procedure_occurrence table-level checks: It looks like the default failure threshold for these checks was 95 in v5.2 and v5.3 threshold files but changed to 0 in v5.4. I agree with you that we don’t expect all patients to have data in these tables. It seems like this may have been a typo/error - I will file an issue to get this fixed!

Note also the threshold files are designed to be editable by the user so that you can adjust thresholds and turn checks on & off based on the requirements for your CDM. See this article for reference: Failure Thresholds and How to Change Them • DataQualityDashboard

2 Likes

Thanks for the info! I did in fact manually change the threshold files and explained these changes in the report as per requested by EHDEN but I appreciate the answer a lot!
I will try to update DQD and see if this was the problem

t