It has been noted that the output for Drug groupings (based on ATC classes in the current implementation) is quite limited as:
- not all ATC levels are useful - ATC 1 and 2 are too broad and 3 is generally not that useful.
- The groupings create a large amount of junk data that can be a challenge to sift through
- It is not clear which group the categories correspond to, making it difficult to filter categories
proposed change
Add in options within the default covariate builders for all different grouping levels.
The default useDrugEraGroup can be maintained but, in addition, it would be nice to have
useDrugEraGroupAtc1
useDrugEraGroupAtc2
useDrugEraGroupAtc3
useDrugEraGroupAtc4
useDrugEraGroupAtc5
I expect CohortDiagnostics will only use 3 and 4, with 5 being less useful than the RxNorm ingredient level mapping.
The names of covariates could also be modified to include the ATC level within them, which would make filtering the output more straightforward.
It has been noted that the output for Drug groupings (based on ATC classes in the current implementation) is quite limited as:
proposed change
Add in options within the default covariate builders for all different grouping levels.
The default
useDrugEraGroupcan be maintained but, in addition, it would be nice to haveI expect CohortDiagnostics will only use 3 and 4, with 5 being less useful than the RxNorm ingredient level mapping.
The names of covariates could also be modified to include the ATC level within them, which would make filtering the output more straightforward.