What Is Inference?
This is a great question, and an important concept in the pharmaceutical industry, but…. In order to grasp inference, we should probably first talk about aggregation.
If you’re a regular reader of my blog, you’re probably familiar with aggregation – a serial number-based packaging hierarchy that captures the parent-child relationship between a large container (the parent) and the smaller container within (the child). Here’s an example: Six bottles packaged into a case; 12 cases packaged into a pallet. Aggregation, then, is the virtual structure of the serial data that mimics the physical structure of the packaging.
When someone scans the serial number on the pallet, they retrieve all the subsequent child-level serial numbers based on the aggregation data. The concept of knowing the child serial numbers through the parents’ serial number is called “inference.”
Inference is vital if you need to know all the child-level serial numbers without opening the parent packages. For example, when you scan a case that contain 12 bottles, the scan captures the case serial number and, by inference, also captures all 12 child serial numbers. Without inference, such data can only be obtained by opening the case and scanning all the bottles in the case – a highly inefficient method that no business wants to undertake.
Why Wholesale Distributors Depend on Inference
Inference is crucial to wholesale distributors. They need to track the movement of every child serial number they possess, and they don’t want to open cases – or worse, pallets – in order to do it. Wholesale distributors receive, move, and ship millions of units a day. If they have to open cases and pallets to scan individual packages, the healthcare supply chain as we know it would come to a halt, resulting in pharmacies with empty shelves and patients without their life-saving drugs.
The big three announced in May that they are asking manufacturers to send serialized shipment data. With that data, wholesale distributors can receive, move, and ship inventory the way they are now, and use inference to tie transactions to child-level packages. This is, of course, very useful in helping them meet DSCSA compliance, especially the FDA’s requirement regarding returned resalable products. What this means is that, with inference, the wholesale distributor can ship in cases and process returns in units.
Of course all of these inference benefits are reliant on the assumption that inference data is 100% accurate. But is that a safe assumption?
Good Product, Bad Data
Inference is derived from aggregation data. Aggregation data is generated by automated or manual packaging lines. And packaging lines have changed and improved a lot in the decade since the California ePedigree era. One of the major motivations for making those improvements was to gain accuracy in aggregation data. While our current packaging lines are a big step up from what they were in the early 2000s, they are far from perfect, and neither is manual packaging. Anyone who runs a packaging line knows all too well that aggregation mistakes happen, and when they do, they’re hard to detect. There are three aggregation error categories:
- Missing – the aggregation data is incomplete.
- Duplicate – the aggregation data has duplicate serial numbers.
- Mismatch – the aggregation data does not match the physical packaging.
In short: garbage in, garbage out. When aggregation data is passed from manufacturer to wholesaler, the errors are passed on as well, with serious implications at every link in the chain.
Inference Errors and their Ramifications
When you scan a product and the system can’t find that serial number, the entire transaction has to stop. That means pausing the receiving, inventory movement, shipping, or returning, and initiating a suspect product investigation. When a product can’t be verified by the distributor, they’ll need to turn to the product origin – the manufacturer. But what if the manufacturer can’t verify the number either? Given that aggregation initiates at packaging, the responsible party might very well be the CMO. If CMOs send bad data, manufacturers get bad data.
So, there’s a lot riding on trustable aggregation and inference data – including time, cost, supply chain movement, and reliable access to the drugs patients need. Plus, given the FDA’s 24-hour requirement on suspect product reporting, the roles and responsibilities around this issue need clarification among partners.
How LSPediA Can Help
Inference, aggregation, packaging processes, barcode labeling, IT systems – there are a lot of details to manage when you undertake serialization implementation. LSPediA is experienced and trusted by manufacturers and distributors large and small to ensure a smooth, cost efficient path to DSCSA compliance. Call me today and I can help you overcome your regulatory complexity, technical hurdles, and operational challenges. Starting for as little as $1000 a month, we can give your company compliance piece of mind – and a competitive edge.
LSPediA helps pharmaceutical manufacturers and distributors plan, design, and implement processes and solution to meet serialization global regulations, DSCSA requirements, and future track-and-trace mandates. Our services include serialization gap analysis, strategic planning, solution architecture, vendor selection, line execution, CMO management, supplier management, implementation, and more.
We value long-term relationships and work with our clients’ internal teams to properly define roadmaps, create architectures, and implement systems that align to vital business goals, ultimately helping them derive maximum value from their investments, both now and well into the future.