UDI And The Approaching End Of The NDC

UDI Barcode Cake
UDI Barcode Cake from last week’s UDI Conference

To be honest, I was somewhat surprised when Jay Crowley announced during last week’s UDI Conference that the U.S. FDA had published the final rules for Unique Device Identification (UDI).  Its publication starts the clock on a host of new requirements designed to identify most medical devices in the U.S. market with the same standard approach.  That is expected to result in the reduction of errors made by medical professionals and also increase the accuracy and efficiency of the supply chain.  I’ve become a little jaded about healthcare Automatic Identification and Data Capture (AIDC) deadlines lately and so the fact that one actually happened caught me slightly off-guard.  Of course, UDI was due most recently in June, but a 3 month delay is tiny compared with those we are used to in the drug supply chain.

I have written a number of essays about the proposed rule and they remain pertinent in light of the final rule.  See:

August 20th, 2012 FDA Proposed UDI: A Revolution In Number Assignment
August 27th, 2012 FDA Proposed UDI: AIDC Requirements
September 10th, 2012 FDA Proposed UDI: The GUDID Database
September 16th, 2013 What The UDI Date Format Says About FDA’s Direction

That first one deserves another look.  It talked about the history of the National Drug Code (NDC) and the National Health Related Item Code (NHRIC) and the differences between how those codes were defined and how the new UDI codes would be defined.  Those two older codes required the FDA to assign each manufacturer a “Labeler Code”.  From that they could build their own product-specific, but U.S.-specific, codes.  UDI is different.  You simply register with the FDA the codes that you build yourself from one of a number of FDA-accredited number issuing agencies.  These are currently expected to be GS1, HIBCC and the ICCBBA but there could be more in the future.

In the conclusion to that essay I asked the question, “Could it work for pharmaceuticals too?”.  I concluded:

“I don’t think it would be very hard or very disruptive for the FDA to start an accreditation process for unique identifiers for drugs.  GS1 US, HIBCC and ICCBBA would certainly apply and assuming they achieve accreditation all existing NDCs could be grandfathered in.  From then on any new drug presented to the FDA for approval would first need to be assigned a unique identifier using one of the approaches defined by the issuing agency chosen by the manufacturer or repackager.  For example, in the case of GS1 US that number would likely be a full 14-digit GTIN.”


I have heard rumors that, at some point, the FDA lost track of exactly which 5-digit labeler codes were assigned and that explained why they began assigning codes that start with the digit “9”.  When the NDC was being designed back in the late 1960’s and early 1970’s it was probably hard to conceive that there might be more than 9,999 drug manufacturers selling drugs into the U.S. market.  Within a few years, they realized that was shortsighted and so they increased the Labeler Code from 4 to the 5 digits it is today.  Labeler codes are never reused, even if a company goes out of business, so today’s FDA Labeler Code database contains quite a few unusable codes and the rate of assignment is on the increase due to globalization and the rising complexity of the pharma manufacturing sector.

There is an unknown point out in the future when all available 5-digit Labeler Codes will be assigned.  What will the FDA do then?  The length of the NDC is fixed at 10-digits and the opportunity to leap to a 6-digit Labeler Code has been lost with the assignment of every possible first digit (0-9).  This would have been a bad solution anyway because it would shorten to only 4-digits the length of the combination of the remaining two fields, the product code and the package size (see “Anatomy Of The National Drug Code”).

When will the FDA run out of 5-digit Labeler Codes?  I don’t know, but I hope someone in the FDA is running that calculation at least every quarter so they have enough time to figure out what they will do next before it becomes an industry showstopper.  Transitioning to a new system could take years.


The approach taken with UDI for devices is clearly the direction that the FDA will eventually take with drugs and biologics.  Now that it has been fully adopted in final FDA rules I expect the FDA to turn their sights on the transformation of the NDC system into one that is based on assignments made through accredited number issuing agencies similar to UDI.  In fact, they issued a request for comments back in the middle 2000’s that implied they were thinking about what to do next, but I can’t find a link to it.  Maybe some of you remember it.

Pharmacists and claims processors have a good wish list for improving on the NDC system which include some industry standard way to represent the chemical and the strength within the code along with the manufacturer.  This is a little different from the requirements filled by UDI so the drug identification system might have some differences overall.  However, opening the number base to a global system that is much more flexible and where the limits are so huge that it is unlikely the FDA would face the need for another transformation anytime soon is what is needed.  In short, the FDA needs to get out of the number assigning business as soon as practical.


There is another thing on the horizon that spells trouble for the existing NDC system:  a federal serialization mandate.  RxTrace readers are well aware that we are currently awaiting Congress to act on one of multiple proposed nationwide serialization and track & trace bills (see “Federal Pedigree: Caught In A Web Of Politics”).  What a catastrophe in would be if the industry goes through a hugely expensive period of deployments to meet a federal serialization mandate, only be forced to go through another one immediately afterward to transform the NDC to some next-generation identification scheme.

Wouldn’t you expect the costs and risks to be lower if the industry goes through these transformations in the opposite order?  I would.  In fact, transforming the NDC system first just might make the transformation to serialization easier for some companies.  Considering that the latest Senate draft bill would mandate package-level serialization of all drugs within 4 years of adoption and the House bill that passed would mandate it within 5 years, the FDA would need to get moving very soon to achieve this sequence.

And let’s not forget about the transition from linear to 2D barcodes (and/or RFID), which also needs to happen before serialization.  So that makes the sequence:

  1. Replace NDC with a nextGen scheme like UDI,
  2. Replace today’s linear barcode requirement with 2D barcodes (and/or RFID), and finally,
  3. supply chain-wide serialization and track & trace.

You could (probably should) combine the first two steps so that production identifiers would fit, but any other sequence will likely cost more.

Agree?  Disagree?  Leave a comment below.



7 thoughts on “UDI And The Approaching End Of The NDC”

  1. The NDC should have been replaced or fixed years ago. Ironically the FDA has seemed to ignore suggestions on how to fix it. A simple fix would add delimiters between the numbering segments which would allow expansion without replacing existing assignments. The root cause of failure with many numbering schemes is that they try to encode attributes in a number without delimiters and the fixed length segments run out if numbers. Another point: Hibcc standards can encode the chemical strengths for a drug in a 2d bar code. CDER’s SPL group issues unique identifiers for ingredients.
    Harry Manolopoulos

  2. Dirk, as always thanks for the essay! Here is my question/concern: Will the issuing entities “control” the number? HIBCC today doesn’t allow non-subscribers to use the HIN which it controls. If this is going to be the case with the new numbering scheme then it is going to be a nightmare. What is in it for the companies that administer the numbers? If they charge isn’t that a death trap of rising cost? While I agree that the UDI seems a good solution on the surface I’m a little weary of turning it over to those that have their best interest in mind and not the consumer.

    1. Al,
      In fact, yes, the issuing agency will control the numbers they issue. As I understand it, if you don’t continue to meet their requirements–including whatever fees they impose–you will lose the right to continue using their number in their barcode symbologies. In theory, we are all paying the FDA some small fees in the form of taxes to maintain the current NDCs. Now each company will pay an issuing agency. The current competition–and the opportunity for future competition–is intended to keep the costs within reason. It appeared to me that HIBCC is charging a very reasonable fee for their numbers and I’ve also heard that GS1 US is lowering the price of their GCPs starting on October 1, 2013.

      The thing that caught my ear was that between last year’s draft UDI language and the final language they removed the requirement that an issuing agency be a non-profit organization. Luckily the initial three incumbents are all non-profit.


  3. Hi Dirk,

    I agree, the FDA Should get out of the Number assigning Business. Per your comment “When will the FDA run out of 5-digit Labeler Codes” I think it will be some time. On the FDA website


    you will notice there are 7135 labeler codes are assigned. With the 4 digit code they were getting close to the limit. If we exclude the code 0000 then there are 10x10x10x10 = 10,000 -1 =9,999 Codes with the 4 digit system. With the 5 digit code they will have 10x10x10x10x10=100,000-1=99,999 Codes. It could be some time

    Thanks for another great article!

    1. Sam,
      Thanks for that correction, although you forgot to eliminate the labeler codes 1000 through 9999 that cannot be assigned. That’s not that great a loss. I also think that the number of rows in the FDA’s spreadsheet is an incomplete picture of the number of labeler codes that have been consumed. There are a lot of “skipped” numbers that don’t take up a row. I assume these are companies that closed without selling their assets–including their labeler code–to anyone else. For example, 0011 and 0012. The codes appear to have been assigned sequentially through about 89160 but then there are five other numbers that are higher but not contiguous. What are these, “prestige labeler codes” like license plates?

      The highest labeler code is 99528 in the spreadsheet I pulled down from the URL you provided. If the FDA cannot go back and “reclaim” the labeler codes that are not in the list today then they are in worse trouble than I thought. So, apparently, they can go back and start (re)assigning codes like 0011 and 0012 and the many many others that are not present in that list???


  4. Dirk,
    Insightful as always, and it highlights a more general point about product data: in our experience the need to improve master data management is often neglected in the “scramble to serialise”. Small and mid-size companies, especially, can have some esoteric SKU numbering schemes that require a lot of work to map to GTINs. Your conclusion is right. Until the USA, EU and others fix on unambiguous numbering structures and data sharing protocols (ideally the same but at least globally compatible) then serialisation and e-pedigree will remain costlier and riskier propositions than they need to be. We all, as patients, will pay the bill.

    1. Mark,
      Right. And you bring up the best reason for the FDA and other countries to get out of the number assignment business. The numbers they assign have local scope so they can only be used in that country. The number issuing agency concept like the U.S. FDA has just adopted for UDI turns that role over to organizations who provide and administer numbers that have global scope so they can be used globally. That’s probably slightly more important for medical devices than it is for drugs.

      The problem is, there isn’t any discussion that I’m aware of in the U.S. to address this problem before serialization is mandated. We need to raise some voices.


Comments are closed.