Dear FDA:
Thank you for inviting interested parties like me to provide our thoughts on the new NDC format that you think will be necessary in 10 to 15 years (see “FDA Seeks Input On The Future Format of the National Drug Code”).  I hope you can make it that long, but regardless, now is certainly the time to begin working on a replacement.

I am not currently a true stakeholder of the National Drug Code (NDC).  That is, I do not currently work within the US pharma supply chain, but I have a long history of working with the NDC, both within and outside the supply chain.  Back around 1990 I wrote software to read UPC barcodes on drug packages and extract the NDC as part of a large solution for a consolidated Rx Returns system for Walgreens.  Four years later I oversaw the development of a full warehouse automation system for Cardinal Health.  Again, our system needed to scan barcodes on drug packages and extract the NDC.

From these experiences, I knew that the encoding of NDCs in linear barcodes on drugs was a mess.  In 2002 I started working in the IT department of Cardinal Health in the group that developed and maintained their warehouse automation systems.  I was there in 2006 when the FDA finalized, what we called “The Linear Barcode Rule”, but not much really changed after that.  Manufacturers continued to make mistakes in encoding the NDC so the best we could do was use a cross-reference table to search for the scanner output, and if we got a hit, extract the corresponding NDC from the table.  You could not reliably decode the scanner output to find the NDC directly.

In 2009 I founded RxTrace.com and began writing essays about serialization, track and trace technologies, standards and regulatory compliance.  It’s mostly an educational website, with an occasional opinion thrown in, here and there, to make it entertaining.  The NDC format and its encoding in barcodes are a reoccurring topic of mine.  In fact, “Anatomy Of The National Drug Code”, written in 2012, is still the most popular essay on the entire site.  The second most-read essay is, “Depicting An NDC Within A GTIN”, and the ninth most-read essay is “Anatomy of a GTIN”.  I also wrote “Anatomy Of An FDA SNI”, and more recently, “How To Properly Define GTINs For Your NDCs”.

I’ve been an advocate of making changes to the NDC system to address its deficiencies for a long time.  I wrote “An Open Letter To The FDA, RE: Please Fix The National Drug Code Soon”, and, “Product Identification And National Registration Codes”, and, “Why NOW Is The Time To Move Away From Linear Barcodes”, and the classic, “NDC Nearing Its End, Afflicted by ‘Identifier Failure’”.  So thank you for focusing on this important issue and asking for ideas.

Today, I am the Global Regulatory Strategist for Systech International, a solution provider that offers solutions for placing and verifying barcodes and human readable content on drug packages, among other things, in all markets around the world.  We will adjust our US solutions to accommodate whatever changes result from this effort, regardless how complex they are, but, we can offer help with assessing the impact of various technical approaches.

The core problem with today’s NDC is not just that you are running out of 5-digit Labeler Codes.  The more important problem is, that the NDC—as defined today, or in any of the Options A, B, C or D shown on your website—is not the identifier that supply chain stakeholders use in their computer systems to refer to drug products (see “DSCSA Uniqueness: SNI vs SGTIN”).  Instead, these systems only work with the GS1 GTIN encoded in the barcodes—whether linear or 2D.

Up through today, the NDC could be wedged into the middle of the GTIN structure, and that is possible only because the Universal Grocery Code Council (UGCC), back in 1973, reserved Universal Product Code (UPC) Company Prefixes that matched all possible 4- and 5-digit FDA Labeler Codes (see “Anatomy Of The National Drug Code” and “Anatomy of a GTIN”).  Today, all drugs that meet the 2006 FDA Linear Barcode Rule and very soon will meet the Drug Supply Chain Security Act (DSCSA) do so by taking advantage of that 1973 decision by the grocery folks.

You can see this discrepancy between the identifier the supply chain uses and the identifier the FDA uses, in the recent draft DSCSA Q&A guidance document (see “FDA’s Late Recommendation On Human Readable”).  Question B.5 in that document asks if the GS1 GTIN can be used in place of the NDC to comply with the requirements for a human-readable NDC as part of the DSCSA product identifier?  FDA’s draft answer is, no, you cannot use the identifier the supply chain uses to identify the product, when the FDA expects the identifier they use to identify the product.  Key to this discrepancy is that the identifier the FDA requires cannot be encoded into a barcode by itself in a standardized way.  Of course not, that’s what the GTIN is for.

Once the FDA expands the Labeler Code to six digits, the approach defined in 1973 by the grocery folks will no longer work, and a new approach will have to be defined.  The approach I recommend is not described in your Options A, B, C, or D, because none of those approaches address the core problem.  What we need is a solution that allows the same identifier to be used as the regulated product identifier, and the supply chain product identifier.

This is exactly the issue the industry is facing in Europe.  Many countries used to have their own identifier for prescription drugs, but once the Falsified Medicines Directive, Delegated Regulation was adopted—which required standardization on datamatrix barcodes across the EU—it made more sense for those countries to abandon their national drug codes and just use the GS1 GTIN instead (see “The FMD Product Code”).

Yes, a handful of countries did not have time or the will to switch their legacy regulatory IT systems to accept the GTIN in their drug registration systems, so the industry has gotten GS1 to define additional Application Identifiers to accommodate those codes in barcodes, but others did make the change to the GTIN.  I am confident that those handful of countries will eventually switch to the GTIN as their official drug registration number because, having two different numbers to identify drugs at different points will introduce confusion, complexity, and therefore it will introduce additional risks to patients.

And isn’t that what it’s all about?  Lowering the risks to patients?  I’m a patient.  We’re all patients.  When I read the four options, A, B, C and D, as a patient, I shudder at the risk that is inherent in all of them, because I know that they all result in two different identifiers for each drug.  Yes, you can make that work, most of the time, by adding extra layers of technology, but from a patient perspective—particularly a patient who understands the underlying technologies—that’s crazy.

I urge everyone to go back and re-read their copy of the Institute of Medicine’s (IOM) classic paper, “To Err is Human:  Building a Safer Health System” from 1999.  Despite the implication of the title, the main finding is that most medical errors are the result of system design problems.  That is, healthcare systems that result in confusion and delay just because of their design.  You now have a once-in-a-career opportunity to re-design one of those systems that will have a direct impact on patient safety.  What you come up with to replace today’s 10-digit NDC can either raise the risks or lower them—for the next 30 to 50 years.

Now, you tell me, from a patient safety perspective, does it raise or lower the risks if you have two identifiers for every drug rather than one?  I think the answer is clear.  Two identifiers will result in more errors in recalls, more ordering errors that result in spot shortages, more errors in prescribing and administration.  Over the expected life of your decision, if we end up with two identifiers for every drug, the greater number of errors made by healthcare and supply chain workers will result in more suffering and death than if we end up with just one.  That one idea should be paramount in your decision.

So let’s resolve today to find a solution that results in a single identifier for every drug.  That’s my recommendation, and it isn’t Options A, B, C or D.

Sincerely, Your Friend,

Dirk.