ISO Compliance Insights & Best Practices
CART
ISO Compliance Insights & Best Practices
A laboratory can calibrate perfectly, run flawless quality control, and report a result that is completely wrong — because the sample arrived from the wrong patient, or sat on a ward for four hours, or was drawn above an infusion line. Pre-analytical errors are the ones the analyser cannot see, and the published literature has put the majority of laboratory errors in this phase for decades.
This guide covers what actually goes wrong before the examination starts, why the laboratory’s controls are weakest exactly where the risk is highest, and how to measure the problem in a way that leads to action rather than a number in a report.
ISO 15189:2022 calls this the pre-examination phase; most clinical staff call it pre-analytical. Either way it covers everything from the moment a clinician decides to request an examination to the moment the examination begins:
| Step | Characteristic failure |
|---|---|
| Selecting the examination | The wrong test for the clinical question; a duplicate within the retest interval |
| Making the request | Missing clinical information; wrong patient selected from a list |
| Identifying the patient | Wrong patient — the highest-consequence failure in the whole pathway |
| Preparing the patient | Not fasting, wrong time relative to a dose, wrong posture |
| Collecting the sample | Wrong container, wrong order of draw, underfill, haemolysis, contamination from an infusion |
| Labelling | Labelled away from the patient; mismatch with the request |
| Transport | Delay, temperature excursion, pneumatic tube damage, loss |
| Receipt | Accepting a sample that should have been rejected; rejecting one that was irreplaceable |
Pre-analytical errors persist because the examination itself is the best-controlled part of the pathway. It has calibration, internal quality control, external quality assessment, an evaluated measurement uncertainty, and competent staff working to a written method in a controlled environment.
The pre-analytical phase has almost none of that, for a simple reason: most of it is performed by people the laboratory does not employ, in places it does not control, leaving records it does not hold. A ward nurse collecting a sample at 3 a.m. is inside your management system as far as an assessor is concerned, and outside your line management entirely.
That leaves the laboratory four levers, and only four:
The fourth is the cheapest and the most often skipped.
Every other control assumes the sample came from the patient named on it. If that fails, calibration, quality control and uncertainty all operate perfectly on the wrong person’s blood. In transfusion it can kill; elsewhere it produces a treatment decision made on someone else’s physiology.
Two rules do most of the work.
Ask open questions. “Are you Mrs Smith?” is answered yes every day by patients who are unwell, anxious, hard of hearing, or simply being polite. Ask the patient to state their name and date of birth, and check what they say against the request.
Label at the bedside, before leaving the patient. Labelling anywhere else — at the trolley, at the nurses’ station, back in the treatment room — is the mechanism by which almost all mislabelling happens. For transfusion samples most services prohibit pre-printed labels entirely, because a pre-printed label can be applied to the wrong tube.
Where the patient cannot be identified, the emergency identification scheme is used and every subsequent sample carries the same identifier. Reconciling that identifier with the permanent record later is a defined step performed by a named person — merging records incorrectly moves results onto the wrong patient permanently, which is worse than the original problem.
These pre-analytical errors are artefacts of how the sample was taken, not facts about the patient. Each is preventable, each is common, and none of them are visible to the analyser:
| What happened | Effect on the result |
|---|---|
| Tourniquet left on too long | Haemoconcentration — raises protein-bound analytes, and potassium |
| Patient asked to pump their fist | Raises potassium |
| Drawn above an infusion site | Dilution, or contamination with whatever is running |
| Wrong order of draw | Additive carryover between tubes |
| Citrate tube underfilled | Wrong anticoagulant ratio — falsely prolonged clotting times |
| EDTA tube over- or underfilled | Altered cell indices |
| Tube shaken rather than inverted | Haemolysis |
| Difficult draw through a fine needle | Haemolysis |
Haemolysis deserves particular attention because of what it does downstream. A haemolysed potassium reported without a comment can trigger a wholly unnecessary clinical intervention — and the intervention for hyperkalaemia is not benign. Where a collection was difficult, record that it was difficult, so the result is interpreted alongside it. It also matters downstream: a haemolysed potassium is one of the commonest false criticals in critical value reporting.
Two transport failures dominate, and both are measurable with data the laboratory already has.
Time to receipt. Collection time minus receipt time is available for every sample at no cost, and it is the single most useful pre-analytical indicator there is. Potassium, glucose and lactate all move with delay; a collection point visited twice a day cannot support an examination with a four-hour stability limit, and the honest answer is either to change the route or to stop offering the examination from that location.
Pneumatic tube systems cause haemolysis. Acceleration and impact vary between systems and between routes within the same building. A laboratory using a tube system should validate it — compare tube-transported samples against hand-carried ones — and record which examinations are excluded from it. Blood gases, ammonia, lactate and cold agglutinins are the usual exclusions, and every one of them gets sent by the standard route eventually by someone who did not know.
Criteria should be quantified rather than descriptive. “Insufficient volume” means below a stated figure. “Delayed” means beyond a stated interval from collection. “Haemolysed” means above a stated index — and that index should be set per examination, because a haemolysed sample can be unusable for potassium and perfectly usable for calcium.
Some failures admit no exception. An unlabelled sample, or one whose label does not match the request, is rejected: there is no clinical urgency that makes a result on an uncertain patient safer than no result.
Others are genuinely irreplaceable — a cerebrospinal fluid, an intra-operative specimen, a sample from a patient since discharged or died. Where the laboratory examines one of those anyway, three things have to happen: a named authorised person makes the decision rather than the bench, the reason is recorded, and the limitation appears on the report. A result the clinician does not know is compromised is treated as though it is not.
Most laboratories measure a sample rejection rate, and most of those rates change nothing, because they are reported at laboratory level. Pre-analytical errors are generated in specific places by specific circumstances, and a single site-wide figure hides both.
Report by requesting location instead. A laboratory-wide rejection rate of 2% is a statistic. A ward with a rate of 6% against a median of 1% is a conversation with the person who can fix it, and that conversation is usually short — it is a trolley without the right tubes, a rota with no trained collector, or a shift where samples wait for a porter.
A useful monthly set:
Two cautions on reading these. A rate that falls to almost nothing may mean collection improved — or that criteria stopped being applied. And near misses are the more valuable record: a wrong-blood-in-tube caught at the receipt desk tells you exactly what a wrong-blood-in-tube that reached a patient would have, at no cost to anyone.
Pre-analytical errors are fixed at the collection point or not at all. Feedback works when it is specific, routine and directed at someone who can act. A monthly one-page summary to each requesting location, showing its own numbers against the median, does more than any amount of general guidance.
The same applies to samples collected at point-of-care testing sites, where the person collecting is usually the person treating the patient. Where a location’s collection quality stays poor, escalating it as a risk to patients is legitimate and appropriate. Rejecting the samples and moving on is not a control — it is a laboratory absorbing another department’s failure indefinitely, and the patients are the ones being re-bled.
ISO 15189 calls these pre-examination requirements. Clause 7.2 covers the phase and requires documented procedures for the request, patient identification and preparation, collection, transport, and receipt with defined acceptance criteria. Our guide to ISO 15189 covers how that fits with the rest of the standard, and how accreditation differs from certification.
Our ISO 15189 Toolkit gives the pre-examination phase seven documents rather than one — request content, patient identification and preparation, collection, transport, and receipt each with their own procedure, plus a sample acceptance and rejection log that reports by requesting location and records conditional acceptances with the wording used on the report.
An error occurring before the examination begins — in the request, patient identification, preparation, collection, labelling, transport or receipt. ISO 15189 calls this phase pre-examination. These errors are invisible to the analyser and to quality control, which is why they reach patients.
Because most of the steps are performed outside the laboratory, by staff it does not employ, in conditions it does not control, and they leave few records. The examination itself has calibration, quality control, competence assessment and a written method behind it; a bedside collection at 3 a.m. has none of those.
No — it depends on the examination. Haemolysis makes potassium and LDH unusable while leaving other analytes unaffected, so acceptance criteria should be set per examination against a measured index rather than per sample by eye.
By requesting location and by cause, monthly, and sent back to those locations. A single laboratory-wide figure identifies that a problem exists but not where it is, which is why it rarely leads to any change.
It can. Acceleration and impact vary between systems and between routes, and haemolysis is the usual consequence. Validate your own system against hand-carried samples and record which examinations must not be sent through it.
A named authorised person decides whether to examine it, the reason and the extent of the compromise are recorded, and the limitation is stated on the report. What must not happen is a compromised result being issued as though it were routine.
August 13, 2026
August 13, 2026
