20.1 Quality Assessment and Troubleshooting

20.1 Quality Assessment and Troubleshooting

Quick Answer: Classify every blood-bank failure as preanalytical, analytical, or postanalytical. Contain first (retrieve the unit, stop a transfusion in progress, quarantine related components), then document, run root-cause analysis, and close CAPA. A unit issued to the wrong patient is an occurrence and, once it left the laboratory, a potential biological product deviation. Computer downtime is a validated SOP, not improvisation. Discrepancy algorithms (Chapters 6.2 and 13.1) are controlled documents. A deviation is unplanned; a planned deviation is approved before anyone leaves the SOP. Recalls pull distributed product by FDA class.

The June 9, 2026 BB outline parks this work at VI.A. Chapters 16.1–16.2 already taught sample and reagent QC. This section is the system that catches an event after it happens, decides whether a product can still be used, and proves the same event will not recur.

Preanalytical, analytical, and postanalytical errors

A quality program is useless if you cannot name the phase. The exam loves a scenario that looks like a reagent failure and is actually a wristband failure.

Preanalytical events happen before a drop of reagent touches the sample or before a unit is selected correctly. Classic blood-bank examples:

• Wrong blood in tube (WBIT) — the tube is labeled with patient A, the cells belong to patient B. Two identifiers were not checked at the bedside. This is still the leading cause of ABO mistransfusion.
• Unlabeled or mislabeled specimen; a tube “fixed” at the bench from memory.
• Draw from a line running IV fluid, so the reverse type is weak or the hematocrit is nonsense.
• Sample older than the AABB 3-day window after recent transfusion or pregnancy (Chapter 16.1).
• Historical type never compared, so a new WBIT is not caught.
• Unit selected from the wrong bin because the pick ticket was for a different patient.

Analytical events happen during testing or manufacturing:

• Expired or visually failed reagent used anyway.
• Wrong cell suspension, wrong incubation time or temperature, uncalibrated serofuge, residual saline left in a cell washer that then dilutes AHG.
• Mixed-field missed on an A3, a recently transfused patient, or an HPC transplant chimera.
• QC failure released; results filed while the AHG vial was already precipitating.
• Computer result entered on the wrong accession.

Postanalytical events happen after a valid result exists:

• Compatible crossmatch reported on the wrong patient.
• Unit issued to the wrong courier or the wrong operating room.
• Face-to-face clerical check skipped at issue or at the bedside.
• Critical result (incompatible unit already sent, newly identified alloantibody, hemolytic-reaction workup) not called to the responsible clinician.
• Interface transmission posts yesterday’s type on today’s visit.

If the stem asks “which error is most likely to cause ABO mistransfusion,” the answer is still preanalytical WBIT, not a mysterious monoclonal failure.

Occurrence management, root cause, and CAPA

An occurrence (nonconformance, incident, variance — facilities use different labels) is any event that was not supposed to happen: a mislabeled tube that reached the bench, a unit issued to the wrong location, a refrigerator that drifted to 7.2 °C, a tech who skipped check cells. The loop is always the same.

1. Detect — someone notices, a computer hard-stop fires, a nurse calls, or a lookback arrives.
2. Contain — pull remaining components with the same DIN, stop a transfusion, quarantine the refrigerator shelf, lock the LIS from issuing that product code.
3. Document — what happened, who, when, which units and patients, what was already transfused. Facts first; blame later.
4. Investigate / root cause — 5 Whys or a fishbone (people, process, reagents, equipment, environment, computer). “Tech error” is not a root cause. “The issue screen does not display the second identifier unless you scroll” is a root cause.
5. CAPA — Corrective action fixes this event (retrieve unit W204712, redraw Mr. Hale, notify the surgeon). Preventive action stops the class of event (force a barcode scan of both identifiers before the issue label prints; add a historical-type hard-stop).
6. Trend and close — QA reviews open CAPAs, repeat events by shift or product, and whether the preventive action actually reduced the rate.

Effectiveness check is part of CAPA, not optional paperwork. If you retrained one person and the same wrong-bin pick happens next Tuesday, the action failed.

When a unit is issued incorrectly

Treat this as two different stems.

Not yet transfused. Call the floor or courier immediately. Retrieve the unit and the compatibility tag. Quarantine it. Confirm the intended patient still has an un-crossed or correctly crossed unit available. File the occurrence. Do a clerical reconstruction: which pick ticket, which LIS screen, which second check was skipped. Do not put that unit back on the shelf until identity and appearance are re-verified.

Already hanging or already in the patient. Stop the transfusion. Keep the bag and attached tubing. Clerical check at the bedside and in the laboratory (patient identifiers, unit DIN, ABO/Rh, compatibility tag, computer record). Start the transfusion-reaction workup if any blood entered the patient (Chapter 18.1). Notify the responsible physician. Quarantine other components from the same donation if the error was a manufacturing or labeling problem rather than a simple mis-pick. Because the unit left controlled inventory, evaluate 21 CFR 606.171 biological-product-deviation reporting (Chapter 20.2). If the wrong ABO went into the patient, this is also a potential FDA recall / consignee-notification event, not just an internal incident.

Do not wait for hemolysis to “declare itself” before you retrieve a unit sitting in the wrong OR warmer.

Computer downtime procedures

Downtime is not “write the type on a sticky note.” A validated downtime SOP is a quality-system document. It must already exist, staff must be competent on it, and it must be used as written.

While the LIS is down:

• Accept only complete paper requisitions with two identifiers.
• Perform testing from written worksheets. A second qualified person reviews ABO/Rh and the unit DIN before issue.
• Issue with a paper compatibility tag. Record the unit number, product code, expiration, ABO/Rh, recipient identifiers, date/time, and both signatures.
• Limit issue to what the medical director’s downtime policy allows if the outage is prolonged (emergency and OR first).
• Keep every paper record in a single controlled location so nothing is typed from memory later.

When the system returns:

• Enter every downtime transaction.
• Reconcile physical inventory to the computer unit by unit before anyone uses electronic release again.
• Investigate any DIN that exists on a shelf but not in the computer, or the reverse.
• Do not turn electronic issue back on because “the printer works now.” Turn it on when the inventory and the pending-work queues match reality.

A downtime that produced a wrong-unit issue is an occurrence. The root cause may be the downtime procedure itself, not the outage.

Discrepancy algorithms as a QA system

Forward/reverse mismatch, mixed-field ABO, and historical-type disagreement are taught as serology in Chapters 6.2 and 13.1. Here they are a quality algorithm: a version-controlled SOP that tells every tech the same stop points.

• Do not report an ABO or issue type-specific red cells until the discrepancy is resolved or a documented emergency path is used.
• Emergency path is group O red cells (D-negative for women of childbearing potential when inventory allows) and group AB plasma, not a guessed type.
• Historical type that does not match today’s type is WBIT or a previous error until proven otherwise. Redraw. Do not “correct” the history from the new tube.
• QA reviews how long discrepancies stay open, how often emergency O is issued while a type is unresolved, and whether the written algorithm was followed. A tech who invents a one-off workup has created an undocumented deviation.

The algorithm is a controlled document. If the medical director changes the emergency-O rule, that is a document revision, not a hallway decision.

Document control, deviation versus planned deviation, and recalls

Document control means only the current approved SOP, form, and job aid are at the bench. Each document has an identifier, version, effective date, and approver. Obsolete copies are marked and removed. Annual (or otherwise defined) review is recorded. A laminated 2019 “quick card” taped to the serofuge is an uncontrolled document and a finding.

A deviation is an unplanned departure from an approved procedure or specification: the 37 °C block was 34 °C for two hours and three antibody screens were already reported; a unit was issued on a verbal request without two identifiers. Investigate, contain, CAPA, and ask whether a distributed product’s safety, purity, or potency may have been affected.

A planned deviation (planned process deviation, temporary change — the label varies) is approved before anyone leaves the SOP. It is time-limited, names the units or dates it covers, records the risk assessment and who approved it, and expires. It is not a silent rewrite of the SOP. If the change should become permanent, revise the document through change control. Using “planned deviation” after the fact to paper over a mistake is itself a documentation failure.

Recall is retrieval of already distributed product that violates the law or may be hazardous. FDA classifies recalls by health risk. Class I: reasonable probability of serious adverse health consequences or death (an HIV-reactive unit that left the building). Class II: temporary or medically reversible harm, or remote probability of serious harm (under-irradiated unit issued to an at-risk patient, some labeling defects that could cause misuse). Class III: not likely to cause adverse health consequences (minor labeling that does not affect safety). Market withdrawal is a firm’s removal of product for a minor violation not subject to legal action. Blood establishments notify consignees, quarantine remaining stock, document disposition, and keep the file. Lookback is a different pathway: a donor later tests reactive for a relevant infection, and prior collections must be interdicted and recipients offered testing. Do not call lookback a Class II recall unless the stem is actually about retrieving distributed product under the recall rule.

If the product never left the establishment, you quarantine and discard — that is not a recall. If it left and SPP may be affected, you are in recall and/or BPD territory at the same time.