
TL;DR
Critical value reporting is the process by which a laboratory identifies a result signalling possible immediate risk to a patient, communicates it to a responsible clinician within a time the lab has defined, and documents that communication. ISO 15189:2022 clause 7.4.1.3 requires notification as soon as relevant, a documented record of the action taken, and an escalation procedure for when the responsible person cannot be reached. Those three requirements, not the numeric limits, are where most labs lose points.
There is a kind of failure that keeps laboratory directors awake, and it is not an analytical one. The instrument was calibrated. Quality control passed. The result left the lab inside the promised turnaround time, and it was correct. Then it sat in a queue, or reached a phone nobody answered, or reached someone who wrote down the wrong patient. By the time anyone acted, the window had shut.
Nothing in that sequence shows up on a proficiency testing report. It shows up in an incident review, and increasingly in an accreditation finding, because assessors know the numeric limits in a lab’s policy are the easy part. What they want is the trail: who was told, when, by whom, whether they repeated it back, and what the lab did when the first call went nowhere. This guide covers that end to end, written so you can lift it into your own SOP.
Any lab running finances out of spreadsheets hits the same tipping point eventually. It arrives the moment reconciliation starts takina
A critical value is a result so far outside the expected range that it may signal a condition needing urgent clinical action. The older term, still used in many labs, is a panic value. What matters operationally is not how unusual the number is, but what the lab must do about it.
g longer than the billing cycle itself. Once that happens, the spreadsheet stops being a safety net. It becomes the bottleneck.
An abnormal result falls outside the reference interval and goes out on the report with a flag. Nobody is called, and the only record is the report. A critical value sits at or beyond a critical decision limit the laboratory has established, the term ISO 15189:2022 uses, and crossing it changes four things:
Critical value reporting errors sit in the post-examination phase, also called post-analytical, covering everything after a result is produced: verification, release, communication, follow-up. Our guide to common types of laboratory errors shows how errors distribute across the three phases.
The laboratory director owns the list and sets it with the clinicians the lab actually serves. That second half gets skipped more often than anyone admits, which is why so many lists read like a textbook appendix.
Variation between labs is not a sign somebody is doing it wrong. A 2026 review in Critical Reviews in Clinical Laboratory Sciences found that differences in patient populations and care settings explain much of the variation in alert thresholds, notification delays, communication method, and who receives the call (DOI, 29 August 2026, via PubMed on 28 October 2026). A paediatric referral centre and a standalone outpatient lab should not hold identical lists.
What has to be decided and written down:
ISO 15189:2022 is the framework most laboratories work to, with national rules alongside it. In the United States, CLIA at 42 CFR 493.1291(g) requires the lab to immediately alert the requester when a result indicates an imminently life-threatening condition, or panic or alert values, and CAP’s Laboratory General checklist asks for a read-back policy. Both are local expressions of the same duty, not substitutes for your accreditation body’s wording.
Six steps, in order, every time. Two of them exist only to catch errors introduced by the others.
Direct observation in a cardiac catheterisation laboratory found complete read-back on only 38% of verbal medication orders at baseline, rising to 79% after a five-year improvement programme (DOI). Those were verbal orders, not critical value calls, so the percentages do not transfer. What does transfer is that teams normalise weak confirmation and stop noticing.
ISO 15189:2022 clause 7.4.1.3 applies when results fall within the critical decision limits the laboratory has established. Once they do, subclause c) requires an escalation procedure for when a responsible person cannot be contacted. Not an intention to keep trying. A procedure, written down, with named rungs.
The ladder below is a generic structure. Set your own trigger intervals in policy, because no standard supplies a number, and an interval suited to an inpatient ward will not suit an outreach client.
| Step | Who is contacted | Trigger to move up | What is recorded |
|---|---|---|---|
| 1 | Ordering clinician or care team | No acknowledgement within the interval set in lab policy | Time, name, method of contact |
| 2 | Covering or on-call clinician | Ordering clinician unreachable | Reason for escalation, time, name |
| 3 | Medical director or laboratory director | Covering clinician unreachable | Time, name, decision taken |
| 4 | Facility or patient safety contact, per policy | No acknowledgement at any rung | Full log plus an incident entry |
Four situations break most ladders:
An assessor should be able to follow any critical result from production to acknowledgement using the record alone, with nobody explaining what usually happens.
ISO 15189:2022 clause 7.4.1.3 is specific. Actions must be documented, including the date, the time, the responsible person, the person notified, the results conveyed, verification that the communication was accurate, and any difficulties encountered. That last item is the one labs omit, and the one that makes a record credible, because a log with no failed attempts describes a lab that never has a bad night.
So the fields per event are: patient identifier, analyte, result and unit, time the result became available, time of notification, notifier, recipient, read-back confirmation, escalation steps, and comments or follow-up. Two principles make that record defensible. Timestamps are system-generated, never typed, because a typed time is an estimate and everyone knows it. And there is one record per event with an audit trail, because rebuilding a timeline from a phone log, an email, and somebody’s memory is how a minor finding becomes a major one.
Then sample events monthly: required fields completed, notification inside the lab’s own target, read-back documented where policy requires it, escalation used correctly. An assessor will sample. Better you found it first.
Almost none of these failures are about the result. They are about the process around it, which makes them preventable with controls rather than vigilance.
A missed overnight call usually means a stale on-call list or an unclear boundary between lab and client site, not carelessness, so review the system and not just the person. Our guide to common laboratory problems and their solutions covers the handoff gaps behind them.
The software should flag, time-stamp, remind, and record. People should make the call, hear the read-back, and exercise judgment when the ladder runs out. Getting that division wrong either way causes harm: automate the conversation and you lose the error catch; automate nothing, and you depend on memory.
A laboratory information system should:
A person should still set and review the list with clinical input, speak to the clinician and hear the read-back, and decide what to do when every rung has failed.
Three numbers, reviewed monthly, against targets the lab set itself:
Then break each down by department before you believe the total. According to PubMed, a retrospective study of 53,639 critical value records at a large Chinese tertiary hospital across 2023 found a median acknowledgement time of 35 seconds and a timely acknowledgement rate of 94.57%, which reads like a solved problem. Split by department, untimely acknowledgement reached 9.53% in the emergency department, 6.98% in intensive care, and 6.39% in the haematology outpatient clinic (DOI). Clinical interventions and outcomes were not evaluated, so this measures acknowledgement inside the hospital system, not patient benefit. Still, a healthy aggregate hid a wide gap between best and worst areas, concentrated where patients are sickest.
One more indicator earns a place, the one that the 2026 review argues for: measure the timeliness of the clinical response, not only the lab’s notification. A call made in ninety seconds and acted on six hours later is not a success. For the wider set, our laboratory KPI framework and turnaround time guide sit beside this one.
On the schedule the laboratory approves, and earlier whenever a method, patient population, service model, or clinical guideline changes. Every revision needs documented approval, a version number, retirement of the superseded limits in the system, and notice to affected clinicians.
The post-examination phase, also called post-analytical, covering everything after a result is produced: verification, release, communication, and follow-up. Failures here include missing critical flags, unreachable clinicians, skipped read-back, and notification times that were never recorded.
The responsible clinician or care team who can act on it. When they cannot be reached, the laboratory follows its escalation ladder: covering or on-call clinician, then medical or laboratory director, then the facility or patient safety contact named in policy.
The recipient repeats the patient identifier, the analyte, and the value back to the caller, so transcription and identification errors surface during the call rather than afterwards. ISO 15189:2022 requires verification that communication was accurate, and most labs use read-back to provide it.
No international standard sets a universal number. ISO 15189:2022 requires notification as soon as clinically relevant, based on available information, and the laboratory sets its own target within its accreditation body’s rules, then measures performance against that target monthly.
Partly. A laboratory information system can flag results against your limits, hold open events in a worklist, send escalation reminders, apply system timestamps,s and keep an audit trail. A person still makes the call, hears the read-back and decides when escalation fails.
Your critical values list is the part an assessor reads in two minutes. The escalation ladder, the read-back field and the event record take a year to build, and they prove whether the list means anything. So the decision is not whether your limits are right. It is whether you could pull one critical result from last month and walk somebody through it, from the analyser to the clinician who acknowledged it, without filling a gap from memory. If not, start with the timestamps, because every other control depends on them.
CrelioHealth’s Smart Reports and lab reporting capabilities cover the flagging, timestamping, and audit trail side- the part that has to be systematic rather than remembered. The call and the read-back stay with your staff. To see it working rather than read about it, book a walkthrough of critical result escalation.