Quality Assurance

External quality assessment sends a laboratory unknown specimens that must be worked up exactly as a patient specimen would be. The point is to measure the whole system — media, staff, instruments, judgement — so the challenges go through the normal bench rotation, on the normal schedule, with no extra replicates and no telephone calls to neighbouring laboratories. Results are submitted by the deadline and the graded report is reviewed when it returns. Any challenge graded unacceptable triggers a documented investigation, a corrective action, and a check on whether patient results from that period were affected.

Errors come in two shapes and they are managed differently. Random error scatters results unpredictably in both directions and usually traces to technique variation — inconsistent reading times, pipetting, uneven incubation. Systematic error pushes results consistently one way and usually traces to something that changed: a new reagent lot, a drifting incubator, a recalibrated reader. Plotting control data on a Levey-Jennings chart makes the difference visible, because a shift or a trend looks nothing like scatter. The technologist's job is to notice the pattern early, hold reporting if the run is compromised, find the cause, and record what was done.

Nothing new goes into patient testing on the manufacturer's word alone. Each new lot of media, stain, reagent or identification card is checked against the lot already in use with organisms of known reaction, and the parallel results are recorded before the new lot is released. Media are also examined for sterility, moisture, thickness and pH. New or replacement instruments need a documented verification that they perform as claimed in this laboratory, with this workload and these staff — installation by the vendor is not verification. All of it is retained as evidence, not merely performed.

A stain that is read without first being judged is a guess. Every batch of Gram, acid-fast, fluorochrome and fungal stains is run with control smears containing organisms of known staining behaviour, and the control is read before any patient slide. The technologist checks that positives hold their colour, that negatives take up the counterstain, that the background is clean and that the smear is neither too thick nor washed away. When a control fails — most often over-decolourisation turning Gram-positive cocci pink — the patient smears from that batch are not reported. They are stained again and the failure is recorded.

Every technologist must be working from the same current version of a procedure. Controlled documents carry an identifier, a version number, an effective date and evidence of who approved them, and only the copy issued through that system may be followed at the bench. Superseded versions are withdrawn and archived so they cannot be picked up by mistake, and informal photocopies taped to shelves have no standing at all. When a procedure changes, staff are trained and their review is recorded before the new version takes effect. Retention periods for withdrawn versions are set by the laboratory's quality manual.