That’s right. We should all be ashamed of the way our supply chain handles drug recalls and we should do something about it. I will explain, but first, what is a drug recall?
Today, when the manufacturer decides that a recall is necessary—either on their own or through a request by the FDA—they issue a recall for it. The FDA website is a great resource for learning about what a recall is (see “What is a recall?”, and “FDA 101: Product Recalls – From First Alert to Effectiveness Checks” and their recalls homepage at “Drug Recalls”).
Recalls can be issued for a number of reasons including (this is my own list and so it may be incomplete),
- unexpected adverse events that are worse or more frequent than discovered in clinical trials;
- a quality problem with one of the ingredients of the finished drug discovered after manufacturing and shipment (e.g., the Heparin recall of 2008);
- manufacturing quality problems discovered after shipment (potency out of spec, contamination, equipment calibration, etc.);
- labeling quality problems discovered after shipment (wrong label, mis-print of critical information, etc.);
- Environmental problems discovered after shipment (storage temperature problems);
- Theft of a quantity of drugs while in the supply chain.
Here are the various levels and types of recalls listed by the FDA:
- Class I recall: a situation in which there is a reasonable probability that the use of or exposure to a violative product will cause serious adverse health consequences or death.
- Class II recall: a situation in which use of or exposure to a violative product may cause temporary or medically reversible adverse health consequences or where the probability of serious adverse health consequences is remote.
- Class III recall: a situation in which use of or exposure to a violative product is not likely to cause adverse health consequences.
- Market withdrawal: occurs when a product has a minor violation that would not be subject to FDA legal action. The firm removes the product from the market or corrects the violation. For example, a product removed from the market due to tampering, without evidence of manufacturing or distribution problems, would be a market withdrawal.
- Medical device safety alert: issued in situations where a medical device may present an unreasonable risk of substantial harm. In some cases, these situations also are considered recalls.
By “issuing a recall”, the manufacturer notifies the FDA of the problem and they work together to determine the recall class. At that point, the FDA and the manufacturer make the public announcement of the recall and begin notifying companies in the supply chain and the public.
The recall can include:
- the entire product line (all NDC’s for that product regardless of lot numbers…could be a “market withdrawal”);
- a subset of one or more NDC’s for that drug (only a certain strength and/or package size for example)
- for all lot numbers, or
- for a given combination of NDC, and specific lot numbers.
Most drug recalls fall into this last sub-category: a specific subset of one or more NDC’s, each with one or more associated lot numbers.
THIS IS WHERE THE EMBARRASSMENT STARTS
There are two serious problems with the way recalls are handled in the supply chain today.
First, recall notices are communicated only in human readable form, whether emails, faxes or websites. This is fine for consumers, but it makes it hard for supply chain companies to automate the process of stopping the forward movement of recalled drugs, keeping them out of the hands of patients and returning them to the manufacturer. Humans must literally read the recall notice and take some manual action. Any mis-reading or delay in action may result in recalled drugs being shipped or dispensed to patients.
Second, the lot number is printed on the product packages but is not in a machine readable form. That is, it is not in the barcode. While the lot number is in barcode form on the manufacturer’s case, most of those are broken down and discarded by the first wholesaler in the supply chain. Subsequent supply chain members can’t get the lot number by reading the package barcode.
Believe it or not, the action those subsequent owners of the drug (drug stores, hospitals, clinics, dental offices, doctor’s offices, secondary wholesalers, etc.) must take is usually to go to the current storage location(s) of the recalled drug and visually look at the lot numbers printed on each package found there to manually compare them with the lot numbers listed on the recall notice. When the number of lot numbers and/or the number of packages is more than a few, this can be very tedious, error prone, and hard on your eyes.
Finally, even when packages are found on this first inspection and are removed from those storage locations, if they aren’t marked clearly and taken completely out of the warehouse, clinic or pharmacy, they can too easily be returned to the forward supply chain and eventually reach patients. This was a major problem with the Heparin recall back in 2008.
According to Virginia Herald, Executive Officer of the California Board of Pharmacy, California investigators documented three instances back then where hospitals discovered recalled units of Heparin, removed them from their supplies as directed by the recall, returned the units to their wholesale supplier for credit, and those units were then mistakenly shipped by that wholesaler to another hospital because the wholesaler failed to recognize the units as being included in the recall.
HOW CAN THAT HAPPEN?
Look at what supply chain members were up against in early 2008 when the Heparin situation unfolded over a couple of months.
This is a list of NDC’s and lot numbers included in the initial recall by Baxter Healthcare, issued on January 17, 2008 (copied as it appeared in the original notice).
NDC NUMBERS 0641-2440-45, 0641-2440-41, 0641-2450-45 and 0641-2450-41; LOTS: 107054, 117085, 047056, 097081, 107024, 107064, 107066, 107074, 107111
About a month later, Baxter issued a recall for all remaining lots of their heparin products.
About three weeks after that, American Health Packaging issued a recall for the following units (copied as it appeared in the original notice):
APS HEPARIN 10MU/ML (10000 USP units/ml) 1ml SDV 25UD (bag) NDC # 00641-0410-25, lot numbers 074155, 073089, 073391, 073613, 070095A, 073712, 072907, 073454, 070095D and APS HEPARIN SDV 10MU (10000 USP units/ml) 1ml 25UD (box and rod) NDC # 00641-0410-25, lot numbers 070095B, 070095C, 068286, 067755
The next day, B. Braun issued a recall for the following units (copied as it appeared in the original notice):
B. Braun FP Lot # B. Braun FP Material Description NDC Numbers CAN DIN J7D490 P5872 Heparin Sodium 25,000 Units in 5% Dextrose Injection (250mL) 0264-9587-20 N/A J7C684 P5771 Heparin Sodium 25,000 Units in 5% Dextrose Injection (500mL) 0264-9577-10 N/A J7D496 P5771 Heparin Sodium 25,000 Units in 5% Dextrose Injection (500mL) 0264-9577-10 N/A J7C470 P5872 Heparin Sodium 25,000 Units in 5% Dextrose Injection (250mL) 0264-9587-20 N/A J7D580 P5671-00 Heparin Sodium 20,000 Units in 5% Dextrose Injection (500mL) N/A 02209713 J7E420 P5872-00 Heparin Sodium 25,000 Units in 5% Dextrose Injection (250mL) N/A 02209721 J7C611 P8721 Heparin Sodium 1000 Units in 0.9% Sodium Chloride Injection (500mL) 0264-9872-10 01935933 J7C557 P8721 Heparin Sodium 1000 Units in 0.9% Sodium Chloride Injection (500mL) 0264-9872-10 01935933 J7C477 P8721 Heparin Sodium 1000 Units in 0.9% Sodium Chloride Injection (500mL) 0264-9872-10 01935933 J7C705 P8721 Heparin Sodium 1000 Units in 0.9% Sodium Chloride Injection (500mL) 0264-9872-10 01935933 J7D485 P8721 Heparin Sodium 1000 Units in 0.9% Sodium Chloride Injection (500mL) 0264-9872-10 01935933 J7E415 P8721 Heparin Sodium 1000 Units in 0.9% Sodium Chloride Injection (500mL) 0264-9872-10 01935933 J7E416 P8721 Heparin Sodium 1000 Units in 0.9% Sodium Chloride Injection (500mL) 0264-9872-10 01935933 J7E494 P5872 Heparin Sodium 25,000 Units in 5% Dextrose Injection (250mL) 0264-9587-20 N/A J7E500 P5771 Heparin Sodium 25,000 Units in 5% Dextrose Injection (500mL) 0264-9577-10 N/A J7E577 P5771-00 Heparin Sodium 25,000 Units in 5% Dextrose Injection (500mL) N/A 01935941 J7E489 P8721 Heparin Sodium 1000 Units in 0.9% Sodium Chloride Injection (500mL) 0264-9872-10 01935933 J7N556 P5872 Heparin Sodium 25,000 Units in 5% Dextrose Injection (250mL) 0264-9587-20 N/A J7P404 P5771 Heparin Sodium 25,000 Units in 5% Dextrose Injection (500mL) 0264-9577-10 N/A J7N604 P5771 Heparin Sodium 25,000 Units in 5% Dextrose Injection (500mL) 0264-9577-10 N/A J7P476 P5872 Heparin Sodium 25,000 Units in 5% Dextrose Injection (250mL) 0264-9587-20 N/A J7N519 P8721 Heparin Sodium 1000 Units in 0.9% Sodium Chloride Injection (500mL) 0264-9872-10 01935933 J7N676 P8721 Heparin Sodium 1000 Units in 0.9% Sodium Chloride Injection (500mL) 0264-9872-10 01935933
A week later, Covidien issued a recall for the following units (copied as it appeared in the original notice):
Product Lot Numbers REF # 8881580121
Monoject PreFillTM 10U/mL Heparin
Lock Flush Syringe, 10mL7082274
7113214REF # 8881580123
Monoject PreFillTM 10U/mL Heparin
Lock Flush Syringe, 3mL7051524
7113214REF # 8881580125
Monoject PreFillTM 10U/mL Heparin
Lock Flush Syringe 5mL7051524
7082274
7113164
7113174REF # 8881580300
Monoject PreFillTM 10U/mL Heparin
Lock Flush Syringe
2.5mL in 3mL syringe7051444 REF # 8881581125
Monoject PreFillTM 10U/mL Heparin
Lock Flush Syringe 5mL, with BLUNTIP plastic cannula7082274 REF # 8881590121
Monoject PreFillTM 100U/mL Heparin
Lock Flush Syringe 10mL7113064 REF # 8881590123
Monoject PreFillTM 100U/mL Heparin
Lock Flush Syringe 3mL7041194
7072154
7113034
8010194REF # 8881590125
Monoject PreFillTM 100U/mL Heparin
Lock Flush Syringe 5mL7041194
7102804
7041204
7113034
7051534
7113044
7051544
7113054
7051554
7113104
7071924
7113114
7072034
7113154
7072044
8010064
7072054
8010114
7072064
8010134
7072154
8010174
7082284REF # 8881591125
Monoject PreFillTM 100U/mL Heparin
Lock Flush Syringe 5mL, with BLUNTIP plastic cannula7082284
The first three recalls were probably manageable. In fact, when Baxter Healthcare issued their “all remaining lots” recall, it would have been a little easier for a while. But the problem is, heparin is so commonly used in hospitals today that the other companies couldn’t follow suit. Recalling every lot would have taken out all heparin from the system and that probably would have caused an even more significant health problem in the U.S. Only Heparin vials that actually had the adulteration problem could be recalled and all good lots—those that were not adulterated—needed to stay in the system to treat patients. The supply of good Heparin started to dry up.
With so much Heparin coming and going, and because the only way to check it for recall once it was out of the manufacturer’s case was to look at each vial, read the lot number and then look at the sheets of all the recalled lots shown above to look for a match—and do that every time the vial changed hands—it would have been a nightmare. The “human readable” characteristic of lot numbers became too much for the system as a whole to handle reliably and clearly. Nineteenth century bookkeeping techniques collided with the massive quantity of drugs that flow through our twenty-first century supply chain, and they crumpled like paper-thin ice under your boot in late winter. Apparently mistakes were made.
Shame on us.
SO WE NEED SERIALIZATION AND TRACK AND TRACE TO FIX IT, RIGHT?
This Heparin debacle is often cited as a shining example of why we need a nationwide serialization and track and trace solution. That was one of the conclusions of “After Heparin: Protecting Americans from the Risks of Substandard and Counterfeit Drugs”, a report published by the Pew Trusts in 2011 (see my full analysis of the report in my RxTrace essay, “Pew Prescription Project: After Heparin”).
However, in my view, this incident is a better example of why we need to standardize and automate the recall detection process that each trading partner in the supply chain uses so that it can be done efficiently, repeatably and reliably with barcodes and computers. If all we were trying to do is fix the recall problem, a full serialization and track and trace solution goes well beyond what is needed.
Now I know that the recall problem is not the only problem we have, so it is not the only thing that we need to fix, but if we were to prioritize the problems we do face, I think this one would be on top. Fortunately, it can be solved with some comparatively simple changes (compared with a full track and trace system). And, those changes would also go a long way toward helping supply chain members detect and stop stolen drugs from being reintroduced into the legitimate supply chain.
What we would need to do is add the lot number to the NDC barcode on each package of drugs and then standardize the way recall data is transmitted to supply chain members. You may recognized that first part as one of the technology changes that were proposed in part of the RxTEC proposal that the Pharmaceutical Distribution Security Alliance (PDSA) published earlier this year (for more on that proposal, see my essay “What If RxTEC Isn’t Adopted?”).
For the second part, fortunately GS1 developed a standard for the electronic formatting of product recall notifications and data so that they can be shared between the computers of supply chain members. It is called “Business Message Standard (BMS), Product Recall” and it includes a separate “Product Recall Implementation Guideline”. It was published in January of 2012 and it is so new that I’m not aware of its use anywhere yet. Pharma should leap on it.
BUT THAT’S NOT ALL
But that’s not all you would need to do. Every company in the supply chain would need to deploy recall monitoring functions that would be invoked whenever a drug barcode—case, pallet or unit—is scanned. The function would have to check the current local recall database on every scan to make sure that the NDC and lot number just scanned is not in a recall. If it is, the user would have to be informed and only allowed to proceed if the business step being performed was related to the unit’s removal from the forward supply chain, like a return shipment. All other business processes would be blocked.
Perhaps this is what the PDSA had in mind when they proposed that each segment in the supply chain…
“…Have systems and processes in place to support lot level product tracing upon change of ownership, lot level recall of a product, and lot level verification of a suspect product…”
…on a staggered deployment schedule.
As long as all supply chain members (manufacturers, wholesalers, repackagers and pharmacies of all kinds) have those functions deployed, we will have brought the handling of drug recalls into the twenty-first century and there wouldn’t be another debacle like the one in 2008.
Like I said, recall execution isn’t the only problem in the pharma supply chain, so this should be considered the first step (or “plateau” as I called it in my RxTrace essays “Plateaus of Pharma Supply Chain Security” and “SNI’s Are Not Enough In a Plateau-Based Supply Chain Security Approach”). One of the problems with the original RxTEC proposal was that it wasn’t clear that what was being proposed was only the first in a series of steps over time that would bring the industry to the point where all of the problems experienced in the supply chain today are addressed.
Until we get an agreement on the other plateaus and begin developing the changes needed for this first one, we are vulnerable to another occurrence like the one in 2008, and for that, we should truly be ashamed.
Dirk.
Good article–helps make the complexity of today’s recalls concrete and points up the need for automated checking.
Thanks–Lew
Dirk, as always I enjoy your article and I really like the idea of have a better recall process. LONG overdue.
I do think you went soft, very soft, on your former employer, the wholesaler. Come on, you know they have fought forever to not have to track lot numbers. Manufacturers track lot numbers because they are required to. Wholesalers lobbied to not have to do this and won. If they tracked them the recalls would be faster and easier. One other thing, Manufacturers have had the ability and have shared lots on all shipments electronically with wholesalers for years via the 856 and could have included it with other EDI documents.
This is NOT to place blame on any one segment, but clearly this refusal to handle a critical part of the medical identification has and does play a part in the problems.
Al,
Regardless of whatever happened in the past to create the recall situation we are in right now I think you will find that, today, wholesalers are generally more interested in dealing with lot-level track and trace than they might have been in the past. This turnabout has occurred because they are now opposed to unit-level serial number-based track and trace, and see lot-level as being more practical. The only thing holding back change in this area seems to be lack of an organized effort. We just need some way to establish the over-arching architecture to ensure interoperability. GS1 already has the product recall standard. Is an FDA mandate the only way to kick off its use in U.S. pharma? Maybe what we need is a consortium of industry associations to take this on: PDSA perhaps?
Dirk.
Dear Dirk,
Thanks for a truly excellent article. It should be required reading of all. Way back in 2001 we gave distributors the means to track every single tablet of mifepristone and after reading this I have to wonder if anybody ever did anything with that capabilitity in the real world or was it just done to keep the FDA and the Bush administration happy. We all as solution providers have the means. It is time for the FDA to show it is serious about dealing with these issues.
Thanks again for a great article.
Robert
But that’s not all you would need to do.
“Every company in the supply chain would need to deploy recall monitoring functions that would be invoked whenever a drug barcode—case, pallet or unit—is scanned. The function would have to check the current local recall database on every scan to make sure that the NDC and lot number just scanned is not in a recall”
Great work Dirk. However the above suggestion assumes the recall notice is issued *before* the downstream supply chain has shipped its product. In some cases, aren’t recalls issued after “the horse has bolted the stable”? If so, the verification would have to accommodate a process for downstream products that are already sent. Which shouldn’t be too onerous but I just thought I’d mention it. Agree?