Electronic witnessing is a traceability problem, not a barcode problem

Every embryology lab witnesses. A second embryologist confirms the identity of dishes and samples at each transfer point, initials a sheet, and moves on. It is a genuine safety control and it catches real errors. It is also, on its own, a remarkably thin record of what actually happened.

What a witness signature proves, and what it does not

A signature on a witnessing sheet proves that at some point a second person agreed the labels matched. It does not record when, which dishes were open at the same time, which incubator the dish came from, or whether the witness was physically present or signing a batch of sheets at the end of a session.

None of that is an accusation of bad practice. It is a statement about what a paper sheet is capable of storing. When a clinic is asked — by a regulator, a couple’s lawyer, or its own quality lead — to reconstruct a specific cycle eighteen months later, the sheet supports one sentence: two people said it was right.

Traceability is the actual requirement

The thing worth building is a continuous chain of custody for every sample from collection to transfer or storage. In practice that means each handling event carries five pieces of state:

• What — the sample or dish, by its own identifier, not by the couple’s name.
• Who — the operator and the witness, as authenticated users rather than initials.
• When — a system timestamp, not a hand-written one.
• Where — bench, incubator, tank, position.
• Why — the procedure step that occasioned the handling.

Once those five exist for every event, witnessing becomes a property you can query rather than a stack of paper you have to find. “Show me every dish this embryologist handled on the morning of the fourteenth” stops being a two-day archive exercise.

Where barcode systems stop short

Barcode and RFID witnessing is a real improvement on paper, and clinics that adopt it generally do not go back. But a scanner enforces a check at a moment; it does not necessarily write that moment anywhere durable, and it rarely knows anything about the clinical context around it.

Two questions separate a witnessing gadget from a traceability system. First: if the scanner is bypassed, does the record show a gap, or does it show nothing at all? A system that silently tolerates missing events cannot be used as evidence. Second: does the handling event attach to the patient record, or does it live in a parallel lab database that has to be reconciled by hand?

The mismatch nobody plans for

Most mix-up near-misses reported in the literature do not happen at the moment of witnessing. They happen around it — a dish relabelled during a machine fault, a sample moved between tanks during maintenance, a cycle cancelled and restarted with paperwork carrying the old identifiers.

These are exactly the events a witnessing sheet is not designed to capture, because they are not scheduled procedure steps. A traceability model that only records the happy path will be silent precisely when you need it to speak.

A reasonable standard to hold yourself to

Pick a completed cycle at random from a year ago. Without phoning anyone, reconstruct: every handling event on that couple’s samples, in order, with operator, witness, time and location; every deviation from the standard protocol and who authorised it; and every person who viewed or amended the lab record afterwards.

If that takes more than a few minutes, the gap is not in your witnessing discipline. It is in where the witnessing goes.