OMOP Community Queries -- how strict on "no local changes"?

[mgkahn (Michael Kahn)]

I have not yet participated in a community-wide study (I can feel Patrick’s burning eyes…). When a project determines the OMOP CDM query to use across participating sites, are sites allowed to make adjustments to the query to fit local CDM features, such as altering normal ranges or the location of various facts, such as different decisions made on which facts go into Observations versus Measurement? If site-specific alterations are allowed, how are these changes captured and made “visible” to the study results.

Trying to figure out how religiously intolerant, as is becoming more common in the USA these days, I should be in allowing/disallowing sites to alter an OMOP-based query in a multi-site project and how best to manage site-level changes if they are allowed.

[CRoeder (Chris Roeder)]

Scary thoughts.

Would the query modifications be transparent, and would the basis for those changes be available? If I cared to, could I discover the alterations? For them to mean anything or to be evaluated, I would need to know how the site’s data is organized differently.

I’d question the deviations from a standard and look to see how it could be improved while trying to forget other file-conversion contexts where you ultimately had to know where the data came from.

[Christian_Reich]

@mgkahn:

NOOO!! There is only one God! No fiddling with the queries.

This is where we want to be. I realize we are not there yet. Therefore, we have the THEMIS initiative to tighten the model and conventions in such a way that you can rely on queries blindly.

So, please please submit your use cases (like the ranges) to THEMIS. They will figure it out.