Every unit of blood is tracked from the donor to the arm, by law, for thirty years. Ours goes out to a named patient, comes back unused, and the label on it names a patient it was never cross-matched for.
I am a biomedical scientist and I run transfusion at a district general.
The traceability is the point of everything here so it is worth being exact.
A unit of blood arrives from the blood service with a unique donation number and a group. It goes into the issue fridge and its location is recorded.
When a patient needs blood we cross-match: we take their sample, we group them, and we test their plasma against the donor cells to be sure the two will not fight. If it is compatible, that unit is assigned to that patient, and a label goes on the bag with the patient's name, date of birth, hospital number, and the unit's number.
That label is the safety system. A nurse at the bedside reads it against the patient's wristband, out loud, with a second person, before anything is connected.
If the unit is not used it comes back to us, and if it has been out of the fridge too long it is discarded, and if it has not it goes back into stock. Every one of those steps is a scanned event.
About once a year, a unit comes back with the wrong label.
Not a smudge. A properly printed label from our own printer, on our own stationery, with our own barcode, naming a patient.
The patient it names has never been cross-matched against that unit. There is no record of a request, a sample, a cross-match or an issue for them.
What that should be impossible for is worth spelling out.
Our labels are printed by the laboratory information system at the moment of issue. The system will not print one without a completed cross-match, because the print is triggered by the cross-match result being authorised.
The label carries a barcode encoding the patient's hospital number and the donation number, and the two are bound together in the database at that moment.
So a label naming patient B on a unit cross-matched for patient A cannot be printed by our system, and I have had the supplier confirm that in writing, twice.
The patients.
I have eleven of these across eleven years and I have looked all eleven up, which I am allowed to do, because a transfusion incident is exactly what my access is for.
All eleven are real patients who were in this hospital. Not made up names, not staff, not tests.
All eleven were inpatients at the time the unit went out. Different wards, different specialties, no pattern I can find in age or diagnosis or consultant.
And not one of them was ever prescribed blood. No request, no group and screen, no sample in the laboratory. They were in the building and they had nothing to do with transfusion at all.
What happens to the unit.
It comes back within the four hour window, which means it can go back into stock, and that is what happens, after I have removed the label and destroyed it and made a record.
The unit's own integrity is not in question. It is a sealed bag that went to a ward and came back.
I file each one as a labelling incident, which is the correct category, and each has been investigated, and each investigation has closed as no cause identified with a recommendation about label handling.
I have written four of those investigation reports myself, which is a thing that has started to trouble me.
The ward end.
I have spoken to the nurse in charge on eight of the eleven, always the same day, always in the way you speak to a colleague about a labelling error, which is carefully, because a transfusion error is a career.
Six of the eight had no memory of the unit at all. That is normal: a unit that comes back unused is a non-event on a busy ward.
Two remembered it, and both said the same thing without prompting, which is that the blood had been brought to them for a patient who had already gone.
Gone in the sense of discharged, is what I assumed the first time, and I said so, and she said no, and did not elaborate, and I did not ask, because we were standing in a corridor and I had understood her perfectly well.
The eleven names.
I did the thing I should not have done, in 2023, and it took me four hours in an office with the door shut.
All eleven died in this hospital. Not on the day the unit went out: between nine and forty one days afterwards, of the things people die of in a district general at the ages they were.
That is not the finding it sounds like. Eleven inpatients over eleven years, in the age range these were, dying within six weeks of an admission is not statistically remarkable and I have had that checked by somebody who does audit properly, without telling her why I was asking.
Her answer was that it is a completely unremarkable number.
What I do about it.
I do exactly what the procedure says. The label comes off, it is destroyed, the incident is raised, the unit is returned to stock if it is within window, and the investigation is written.
I have never once put in a report that the patient named on the label had no cross-match. I write that the label was incorrect, which is true, and the investigation closes.
The reason I have never written the other sentence is that the first question would be how a label naming an un-cross-matched patient came out of a system that cannot print one, and the second question would be who printed it, and the person with access to that printer, in that laboratory, at that hour, on all eleven occasions, is me.