Specimen Labeling Error Prevention: Taxonomy, Controls, and the Rejection Policy

A specimen that reaches the bench with the wrong name on it does not announce itself. It looks like every other tube in the rack, it runs on the analyzer without complaint, and the result posts to a chart that may belong to someone else entirely. Specimen labeling error is one of the few patient-safety failure modes where the error, the harm, and the point where it could have been caught are separated by hours and by several different staff members — which is exactly why it needs a named taxonomy, a point-of-collection control, and a rejection policy that does not depend on any one person noticing.

How big the problem actually is

A 24-month review of 4.29 million specimens found mislabeled specimens at a rate of 1.0%, unlabeled specimens at 4.6%, and specimen-requisition mismatches at 6.3% — a combined identification-error exposure well above what any single downstream check reliably catches (ADLM Clinical Laboratory News). The College of American Pathologists’ own Q-Probes work puts the aggregate identification-error rate at up to 379 per million billable tests, with roughly half of that total attributable to primary specimen labeling errors specifically, and about 6% of identification errors going on to generate an actual adverse event rather than being caught before reaching a patient. Blood-bank literature is blunter still: mislabeled or misidentified blood-bank samples — wrong blood in tube, or WBIT — are implicated in roughly 11% of transfusion-related deaths, which is why transfusion services generally run the strictest labeling controls of any specimen type in the building.

None of that is abstract. A hospital running, say, 3,000 billable tests a day is looking at more than one identification error a week at the CAP-reported rate, most of them caught by downstream checks that exist for exactly this reason — and a program built to prevent the error in the first place, rather than rely on catching it, closes the gap those checks are only ever a backstop for.

The error taxonomy: what “mislabeled” actually covers

Treating every labeling problem as one undifferentiated category makes it impossible to target a fix. The literature and common laboratory error-reporting taxonomies split it into distinct failure modes, each with a different root cause and a different control:

• Unlabeled specimen — no label present at all. Almost always a collection-workflow gap (label printed but not applied, or applied to the wrong tube in a multi-draw).
• Mislabeled specimen — a label is present but the information on it is wrong: wrong patient name, transposed date of birth, or a label carried over from the previous patient’s draw.
• Specimen–requisition mismatch — the label and the physical specimen agree with each other but disagree with the order (wrong test ordered against the tube, or the requisition and label were generated from two different registrations).
• Illegible or damaged label — smudged, wet, or partially printed; functionally unlabeled even though a label was applied.
• Wrong blood in tube (WBIT) — the label is correct, but the blood inside the tube is not the labeled patient’s. This is the wrong-patient-draw scenario specifically, not a printing failure, and it is the one a label-content standard alone cannot fix — only a correct-patient-verification step at the moment of draw can.

The taxonomy matters operationally because unlabeled and mislabeled specimens are catchable at accessioning (no result should ever post against them), while a specimen-requisition mismatch or a WBIT specimen can look completely correct through accessioning and testing and only surface as a clinically implausible result — or not surface at all until the wrong therapy has already been given. A prevention programme has to cover both ends: hard-stop label content requirements at the front, and a verification step that catches a correctly-labeled-but-wrong specimen.

Point-of-collection controls: what has to happen before the tube leaves the room

Two separate requirements operate at the point of collection, and conflating them is a common design mistake:

Patient identification uses two patient-specific identifiers — typically full name and date of birth, never room number or bed location — confirmed against the order before the draw. CASRAI covers the identifier list and where the requirement applies in detail on Two Patient Identifiers: The Approved List, and Where the Requirement Applies; a labeling-error-prevention programme should point to that page rather than re-litigate the identifier list.

Label content and placement is a separate, format-level requirement: CLSI’s AUTO12 — Specimen Labels: Content and Location, Fonts, and Label Orientation standard specifies which human-readable elements must appear on a laboratory specimen label, where barcodes are positioned relative to those elements, truncation rules for long patient names, and label orientation on the tube so the barcode scans reliably on automated line equipment. A label that satisfies AUTO12 is legible and machine-readable in a consistent, predictable place — but AUTO12 governs what the label says and where, not whether the label was ever checked against the right patient. Both requirements have to hold; a perfectly formatted label on the wrong patient’s tube is still a WBIT event.

The sequencing rule that actually prevents WBIT is: label at the bedside, in the presence of the patient, after collection, never before. Pre-printing and pre-affixing labels ahead of the draw — a workflow shortcut some phlebotomy programmes still use for efficiency — removes the last verification opportunity, because the tube is already labeled before the final patient-identity check happens; if the wrong tube was picked up for the wrong patient’s pre-printed label, nothing downstream catches it (Ernst, Annals of Blood). A programme that permits pre-labeling for throughput reasons is trading a known, quantifiable WBIT risk for a small time saving.

The rejection and recollection policy that closes the loop

Prevention controls will not catch everything, so the policy question that actually determines outcomes is what happens to a specimen that arrives at the laboratory already mislabeled, unlabeled, or mismatched. A working policy specifies, in writing:

• Automatic rejection criteria — unlabeled specimens, specimens with two or more discrepant identifiers against the requisition, and illegible labels are rejected outright and recollected; there is no acceptable corrective-documentation path around a fully unlabeled tube.
• The single-discrepancy exception, narrowly scoped — some laboratories permit a specimen with exactly one minor, correctable discrepancy (a single-character name misspelling that is otherwise unambiguous, for example) to proceed with a documented correction by the collector, rather than mandatory recollection. This path should be defined narrowly and in writing, not left to bench-level judgment call by call — the risk of the exception swallowing the rule is real.
• Irreplaceable-specimen handling — some specimens genuinely cannot be safely or practically recollected (a surgical pathology specimen, a one-time cerebrospinal fluid draw, an already-administered timed collection). For these, the policy needs an explicit escalation path — ordering-provider notification, a documented risk acceptance, and often a pathologist or laboratory director sign-off — rather than a default to either automatic rejection or automatic acceptance.
• Blood-bank specimens held to a stricter bar — given the WBIT mortality data above, most transfusion services reject on any labeling discrepancy without exception, including the single-character case that a general chemistry specimen might tolerate with correction.
• Reporting, not just rejecting — every rejection is a near-miss and belongs in the same incident-reporting and trending pathway as any other patient-safety event, not a silent recollection request. Trending the labeling-error rate by collecting unit, shift, and collector is what turns individual rejections into a program-level fix rather than a repeating one-off.

Where a labeling error does reach a patient — a WBIT transfusion, a result acted on for the wrong patient — it is a reportable patient-safety event in its own right, not just a laboratory quality metric. CASRAI covers how that severity call gets made on Patient Safety Event Severity Classification, and the investigation mechanics on Root Cause Analysis in Healthcare and Root Cause Analysis for CAPA.

Where labeling-error prevention fits alongside the rest of the identification chain

Specimen labeling is one link in a longer patient-identification chain, and a page-by-page CASRAI reader building out that chain should also see: Point-of-Care Testing (POCT) for how identification requirements change when testing happens at the bedside rather than in the central lab; CLIA-Waived Point-of-Care Testing and IQCP for the quality-control framework specimen-handling controls sit inside; and CLIA Certification for how a laboratory’s complexity category shapes which of these controls are regulatory requirements versus voluntary best practice. Universal Protocol and the Surgical Safety Checklist covers the parallel identification chain for surgical specimens and site verification specifically.

This page sits in CASRAI’s Patient Safety & Infection Prevention cluster, under the clinical-risk-and-regulatory-operations area covering patient identification and the daily operational controls that carry it out.