Stream: Da Vinci PDex Drug Formulary
Topic: NDC, RXCUI and RXNorm
Courtney Bland (Mar 19 2021 at 01:22):
There are times that we add drugs to our formulary before an RXCUI is created. When the IG indicates that we must return back an RxNORM ID, we are interpreting that to mean an RXCUI. For those cases in which we have added a drug prior to the creation of an RxCUI, we were planning to report back the NDC. There are two problems with this approach.
- Version 1.0.1 makes RxNorm required instead of extensible. What is the recommended approach for these drugs that don't have an RXCui?
- For those drugs that don't have an RXCUI, we have a proxy NDC to represent that drug. That proxy NDC could end up changing. Does the FHIR logical ID assigned to the first NDC have to be used for the new NDC? For example, NDC of 59148003085 gets a FHIR logical ID 12345. It gets a new proxy NDC before it is assigned an RXCUI. Should the new proxy NDC of 4444900308 get a new FHIR logical ID or should it get 12345? Does it matter? When the RXCUI gets assigned, should it reflect a FHIR logical ID of 12345?
Thank you for your time.
Saul Kravitz (Mar 19 2021 at 13:27):
Hi @Courtney Bland
The issue of drugs that do not yet have an RXCUI has never been previously discussed in the context of the Formulary IG.
The Value Set used by the IG for the medication code defines the subset of RxNorm codes as "Include codes from http://www.nlm.nih.gov/research/umls/rxnorm where TTY in SCD,SBD,GPCK,BPCK,SCDG,SBDG,SCDF,SBDF".
I think you raise several issues questions:
1) In US Core Medication, the binding to the same Value Set is 'extensible'. Shouldn't it be extensible here to deal with drugs that do not yet have an RxNorm code? -- https://jira.hl7.org/browse/FHIR-31579
2) Assuming an extensible binding, for a given drug how should the not-so-stable NDC code be managed relative to the FHIR id?
I think both of these would benefit from discussion at the Friday PDEX call. (My personal take on #2 is that the only stable element of the FormularyDrug is its code, so if the code is unstable, I would suggest allocating a new FHIR id for each incarnation. But I would like to have this confirmed by wiser folks.)
Courtney Bland (Mar 19 2021 at 13:51):
Saul Kravitz said:
Hi Courtney Bland
The issue of drugs that do not yet have an RXCUI has never been previously discussed in the context of the Formulary IG.
The Value Set used by the IG for the medication code defines the subset of RxNorm codes as "Include codes from http://www.nlm.nih.gov/research/umls/rxnorm where TTY in SCD,SBD,GPCK,BPCK,SCDG,SBDG,SCDF,SBDF".I think you raise several issues questions:
1) In US Core Medication, the binding to the same Value Set is 'extensible'. Shouldn't it be extensible here to deal with drugs that do not yet have an RxNorm code? -- https://jira.hl7.org/browse/FHIR-31579
2) Assuming an extensible binding, for a given drug how should the not-so-stable NDC code be managed relative to the FHIR id?I think both of these would benefit from discussion at the Friday PDEX call. (My personal take on #2 is that the only stable element of the FormularyDrug is its code, so if the code is unstable, I would suggest allocating a new FHIR id for each incarnation. But I would like to have this confirmed by wiser folks.)
Thank you for your reply. I will try to attend the call today. Appreciate your help. @Mark Neumuth
Saul Kravitz (Mar 19 2021 at 14:15):
@Dave Hill @Corey Spears
Saul Kravitz (Apr 02 2021 at 16:18):
@Courtney Bland do you have an example of a drug that was on formulary that lacked an RxNorm code.
Dave Hill (Apr 14 2021 at 15:41):
Just wanted to follow up with the @Saul Kravitz question earlier in this thread. Does anyone @Courtney Bland have an example of a drug that was on formulary that lacked an RxNorm code? If not, we may be chasing a solution for something that really isn't a problem. @Corey Spears
Courtney Bland (Apr 14 2021 at 19:44):
Dave Hill said:
Just wanted to follow up with the Saul Kravitz question earlier in this thread. Does anyone Courtney Bland have an example of a drug that was on formulary that lacked an RxNorm code? If not, we may be chasing a solution for something that really isn't a problem. Corey Spears
Hello @Saul Kravitz and @Dave Hill . Thank you for the message.
Unfortunately, I do not have an example to share. We don't currently store, in any of our databases, when an RxCUI becomes available. RxCUI is not considered a critical data element in our systems as they stand today. The feedback was provided by our business partners based on their experiences, but they did not have a precise example to supply. Since we will be going live relatively soon, we have decided to record an error for any drug in our formulary that does not have an RXCUI at the time the record is supposed to be created for the Formulary Profile or the Coverage Profile in our database. That drug will not be returned back in any API response until an RXCUI is generated. If the consuming app asks for that specific drug by name or NDC, they will given the standard message indicating that a record was not found. If this situation occurs after we go live, then we will provide the examples that you seek from our error log. It would be great if we could see what other payers might think on the matter. cc @Mark Neumuth
Tun Naing (Apr 15 2021 at 20:47):
I'm running into the same situation as well. We covered this proxy NDC 50924088401 but there is no associated RxCUI yet. We are also considering not creating a FormularyDrug resource for drug without RxCUI.
Kate Dech (Apr 20 2021 at 15:19):
From a PBM perspective, we are having to leave off products that do not have an RxCUI which diminishes the reliability/value of this resource for communicating covered products.
Corey Spears (Apr 20 2021 at 16:24):
The project group has been looking for examples where covered products that would show up on a formulary do not have an RxCUI assigned for an extended period of time (for months), but we have not been able to find solid examples. Are there some examples you could share @Kate Dech ? Might you be able to join our weekly PDex call to discuss?
Robert McClure (Apr 20 2021 at 22:58):
@Tun Naing There actually is information about NDC code 50924088401 in RxNav but it's something I have not seen before so I need to see how we access this information. That NDC is linked to RxCUI 2004255 and all it says is that the link is not curated by NLM and the concept is "PROPRIETARY" and is not current. It is not even clear if this is used as a placeholder for a specific thing or a general thing, so stay tuned. Please give us more information on what that drug actually is. Is it a pre-release drug, investigational, a special compounding?
Robert McClure (Apr 20 2021 at 23:01):
@Kate Dech The point of this exercise is to find specific situations where actual products on a formulary cause problems with RxNorm. So far, we've not found any products you would have to "leave off" so please provide an example.
Robert McClure (Apr 21 2021 at 14:30):
Here is the update on @Tun Naing 's NDC - it represents a glucose test strip which is not in scope for RxNorm. This is a great example. So we are going to have an issue given the current Required binding to USCoreMedicationCodes if formularies are going to include things defined as out of scope, like test strips. Yes, we could make the binding extensible but that is a big opening (as I noted before) when I'd rather determine how to represent the few known types of things that are out of scope. I'm setting up a discussion with NLM, but we need more examples of these non-drug items that are on formularies. @Dave Hill
Kate Dech (Apr 21 2021 at 19:35):
@Robert McClure - I do want to say that I think this is the exception and not the (Rx)Norm (sorry, couldn't resist!). But it does seem to happen more with OTC products, like the Test Strips mentioned above. Usually, only diabetic supply OTCs are pulled into the formulary. But we have seen with some Medicaid formularies a lot more OTC products that are added. Not all of these even have NDC product types. They only have UPC codes. But we do put the UPC product ID on the formulary.
I found an example from one Medicaid Formulary of an OTC that only has UPC products that I can't find in RxNav.
CALCIUM CARBONATE-CHOLECALCIFEROL TAB 333 MG-133 UNIT
I believe this would use ingredient calcium carbonate / cholecalciferol [RxCUI = 608343]
But I don't see this strength in the list of Clinical Drug Components [SCDC]
The UPC codes are 30768000498 & 30768002498.
Robert McClure (Apr 22 2021 at 14:46):
Based on discussion with the RxNorm folks, @Kate Dech they are also finding UPC codes in submitted information masquerading as NDC codes and they ignore those. They reconfirmed that OTC content will remain out of scope.
They shared a detailed analysis:
We recently conducted a review of 1.7B prescriptions from claims data collected from private insurance companies. Here is the breakdown of the NDCs after analysis with the RxNorm API.
The status of the API returned by the API is one of:
ACTIVE. The NDC is recognized by RxNorm as active and is associated with an active concept. The RxCUI associated with the NDC is contained in the returned history data.
OBSOLETE. The NDC previously existed in RxNorm, but is no longer recognized as active. The history data will indicate the time periods and RxCUI(s) associated with the NDC.
ALIEN. The NDC is or was contained only in a source vocabulary other than RxNorm. This status means that the NDC has never been active in RxNorm.
UNKNOWN. The NDC is unknown or invalid. This means that the NDC has never been active in RxNorm, is not in a source vocabulary other than RxNorm, and no history data is available.
Active and obsolete NDCs are curated by RxNorm, as opposed to alien NDCs, only known from some sources, but not curated by RxNorm. Unknown NDCs are absent from RxNorm.
[omitted table]
The vast majority of the NDCs (80%) and even more of the prescriptions (89.3%) correspond to NDCs curated by RxNorm, active or obsolete. Alien NDCs tend to correspond to OTCs, supplements and supplies. They represent a significant proportion of NDCs (14.9%), but relatively few prescriptions (1.6%). Unknown NDCs are uncommon and negligible (0.1%) in terms of prescriptions.These results are consistent with similar analyses performed on the Medicare dataset over several billion prescriptions.
This means that if we must represent ALL CONTENT OF FORMULARIES we can not bind required to only RxNorm. The best approach would be to find a code system we can use to represent types that the UPC codes or NDC codes could be mapped to for exchange, similar to what we do for drugs. I have not done an analysis to determine what we might use for an OTC code system.
I must note that the analysis does show there are rare, but existing drug formulary items that are not in RxNorm. So, a decision that we MUST have all formulary items encoded when exchanged may force us to open the binding up to Extensible. Perhaps an alternative would be to come up with a one or more concepts to be used when not in RxNorm and then the actual item can be sent in text along with the generic code we create.
@Dave Hill Perhaps for discussion Friday?
Last updated: Apr 12 2022 at 19:14 UTC
