The traceability code fails at the carton, not at the counter
Scanning traceability codes at insurance settlement has moved the discovery of code errors to hospital pharmacies and dispensing windows. Most of those errors are created earlier, when packs are associated with cartons, and distributors should find them there.
For most of the past decade, a drug traceability code in China was something a manufacturer printed and an inspector might one day check. That changed when China's National Healthcare Security Administration began requiring, in phases from 2024, that designated hospitals and pharmacies capture the code of each pack dispensed under medical insurance. A code now has a consequence at the moment a patient collects a medicine. A pack whose code is missing, unknown to the system or already recorded somewhere else becomes a question for the pharmacist, for the hospital's insurance office and, sometimes, for the fund's inspectors.
The policy has a sound purpose. Codes that appear twice at settlement can reveal medicines resold after being dispensed, a long-standing form of insurance fraud, and a complete record makes recalls faster and more precise. I support it. My argument is about where the work of making codes reliable should sit. At present too much of it is discovered at the counter, by the people least able to fix it, when it should be done at the carton, by the people who can.
Where the error is created
A traceability code is applied to each pack on the manufacturer's packing line. Packs are then placed in cartons, and the manufacturer's system records which pack codes sit inside which carton code. That association is what allows a warehouse to scan one carton instead of opening it and scanning sixty packs. It is efficient, and it is the weak point of the whole arrangement.
Associations go wrong in ordinary ways. A packing line stops, packs are removed for inspection and returned to a different carton. A carton is crushed in transit and its contents are repacked at a depot. A contract packer uploads associations late, or to the wrong platform. In each case the carton looks perfect and its code scans. The error is inside, and it stays there until somebody opens the carton and compares what is in it with what the record says.
In the first half of 2024, before we began checking systematically, about 0.84 per cent of the cartons arriving at our nine logistics centres carried associations that did not match their contents. That sounds small. At roughly 1.6 million cartons a month, it meant more than thirteen thousand cartons a month carrying a wrong record towards hospitals and pharmacies. Some manufacturers were nearly perfect. A minority, often smaller domestic companies with older packing lines, accounted for most of the errors.
Why the hospital should not find it first
When a distributor passes an unchecked association along, the error surfaces in one of two places. The first is the hospital's receiving store, if the hospital scans packs on arrival. The second, and worse, is the dispensing window, where a pharmacist scans a pack for an insured patient and the system rejects it. The pharmacist must then choose between delaying the patient, dispensing a different pack or recording an exception that someone in the insurance office will later have to explain.
Neither place is suited to fixing the problem. A hospital pharmacist cannot see the manufacturer's packing record. A hospital's technical team cannot correct an association it did not create. What they can do is refuse the delivery, and in March 2024 three tertiary hospitals told us they would stop accepting our electronic deliveries if code rejections, then 1.9 per cent of our delivery lines, did not fall. They were right to say so. They were being asked to absorb errors made two steps upstream.
There is also a question of fairness. The burden of discovery falls hardest on those with the fewest resources: county hospitals with two pharmacists on duty, village clinics that dispense a handful of insured medicines a day, independent pharmacies without anyone technical on the payroll. A tertiary hospital can build an exception workflow. A township health centre cannot. If errors are allowed to travel, they cost most where they can least be afforded.
What a distributor owes the codes it passes on
A distributor is not the author of a code association, and some colleagues argue that we should therefore not be answerable for it. I understand the argument. But a distributor is the last party to handle the carton in bulk before it is broken up, and the only party in the chain that sees every manufacturer's data. That position creates an obligation to check, even if it creates no liability for the original error.
At Shuheng we now verify at receipt. For each batch, sample cartons are opened and every pack is scanned against the association record. Manufacturers whose error rate exceeds 0.3 per cent move to full checking until they improve. Deliveries without code data are not received electronically. Each manufacturer receives a monthly scorecard showing its error rate and the cartons concerned.
None of this was cheap. We spent RMB 46 million on scanning equipment, added about 260 people and accepted a 14 per cent fall in receipt throughput for six months. Two small manufacturers moved products to distributors with lighter checks. Our logistics leadership opposed the programme, for good reason, because the dock is where a distributor's service promises are kept or broken.
The results have justified it. By mid-2025 errors in received cartons were down to 0.05 per cent, and code-related rejections at hospital receipt had fallen to 0.2 per cent of delivery lines. More than a hundred manufacturers corrected their packing processes once they could see their own error rates. The scorecard achieved more than any contract clause had.
I would ask three things of whoever designs the next stage. The authorities could publish association error rates by manufacturer, as they already publish other quality information, so that the pressure to fix a packing line does not depend on how much business a distributor happens to control. Manufacturers could treat aggregation as part of packaging validation rather than as an IT task handed to a code vendor. And distributors could agree a common verification standard, so that a manufacturer is not asked to pass a different test at every warehouse it ships to.
The traceability code is a good idea. Its value depends on whether the record matches the pack, and that is decided long before a patient reaches the counter.