Consider using var name instead of var label to identify unique record

[aclark02-arcus]

Right now, item_names are used to define a unique record in the review db. Currently, item_name is populated with a variable label. This can be problematic if variable labels change slightly mid-study since changing the item_name will trigger a re-review. Just thought we should consider if that's a desirable trait & whether we should consider the alternative, variable name?

[LDSamson]

Hi can you elaborate a bit? Your definitions are a bit confusing for me. You are using the term item_name, variable_label variable_name while only item_name is defined in clinsight. There is also the var column defined in the metadata which in our EDC that maps to ItemId. Is the ItemId the same as what you name variable name?

[aclark02-arcus]

Hi @LDSamson, yes sorry. Let me provide an example. Running clinsight::create_clinsight_memtadata() produces this metadata.xlsx file with the following common_forms tab:

In ClinSight, item_name is used to identify unique rows in the review db via key_columns and item_name is populated by variable labels pulled from ItemExportLabel in the raw CSV data. We have an upcoming migration that is going to change a bunch of labels in our EDC system and we don't think it should trigger a re-review... but I guess whether it should or should not is debatable.

Would it make sense for item_name to be replaced with ItemId? That way, review would be tied to a specific variable, which would likely remain unchanged (at least more so than labels).

[LDSamson]

That is clearer, thanks for the explanation. But I am still a bit confused. item_name is not populated by variable labels pulled from ItemExportLabel in the raw CSV data. It is only populated by the metadata defined in the metadata Excel file. Check for example this raw CSV data, there is no variable called ItemExportLabel: https://github.com/openpharma/clinsight/blob/main/inst/raw_data/clinsight_raw_AE.csv

[LDSamson]

Regardless of that, I see what you mean. Few thoughts:

1. How do you envision this to work with derived variables with multiple ItemIds? How would you map this to the database?
2. It sounds like such a change is not backwards compatible, sort of a breaking change with ongoing studies. Can still be considered though.
3. Would an additional database update function also work? with arguments the database, the old metadata and the new metadata, with output the database with adjusted item_name? Maybe that is easier to realize and then point 1) and 2) would not be an issue?

[aclark02-arcus]

• How do you envision this to work with derived variables with multiple ItemIds? How would you map this to the database?

hey @LDSamson, what do you mean by multiple ItemIds for one variable? I would assume whatever the name of the new derived variable is would become the new ItemId, no? For example, if you create a new variable called "derived_age" at screening based on birthdt, the ItemId would be "derived_age". Right?

• It sounds like such a change is not backwards compatible, sort of a breaking change with ongoing studies. Can still be considered though.

Agree, not backwards compatible.

[aclark02-arcus]

That is clearer, thanks for the explanation. But I am still a bit confused. item_name is not populated by variable labels pulled from ItemExportLabel in the raw CSV data. It is only populated by the metadata defined in the metadata Excel file. Check for example this raw CSV data, there is no variable called ItemExportLabel: https://github.com/openpharma/clinsight/blob/main/inst/raw_data/clinsight_raw_AE.csv

Sorry, I must be referring back to my memory of some older csv artifacts. But yes, you're right... however item_name is populated now-a-days.