Description
Proposal
Create a data mart that maps a product-level RXCUI (TTY in [SCD, SBD, GPCK, BPCK]) to a "representative NDC" which means a somewhat random (but intentional) NDC that can be used as a surrogate identifier in downstream systems.
Likely use the NDC data mart (do we need to make this into an intermediate table and make the current mart into just a select * from type mart?) which already maps NDCs to RXCUIs (where possible). Maybe could use the labeler column to join to a count of how many products that labeler currently has available (or some other way to rank NDCs by "stability" or "representativeness").
- NDC 1 is made by labeler A and labeler A has 10 NDCs available (not just of this specific product, but of all products) in FDA NDC Directory
- NDC 2 is made by labeler B and labeler B has 5 NDCs available in FDA NDC Directory
- Therefore, we would use NDC 1 as our representative NDC
Could also look to FDASTARTMARKETINGDATE.
Likely create some sort of algorithm to weight both of these.
Rationale
NDC codes become inactive more rapidly than RxNorm has releases. People depending on RxNorm for NDC information are going to be days if not weeks behind NDC churn.
Representative NDCs are required by Surescripts / NCPDP and are likely useful to have for other reasons as well.